1 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
3 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
4 vector mode and that that mode matches the mode of the data
6 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
7 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
8 directly using expand_vec_perm_1 when forcing selectors into
10 (expand_vec_perm_var): New function, split out from expand_vec_perm.
12 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
14 * optabs-query.h (can_vec_perm_p): Delete.
15 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
16 * optabs-query.c (can_vec_perm_p): Split into...
17 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
18 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
19 particular selector is valid.
20 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
21 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
22 (vect_grouped_load_supported): Likewise.
23 (vect_shift_permute_load_chain): Likewise.
24 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25 (vect_transform_slp_perm_load): Likewise.
26 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
27 (vectorizable_bswap): Likewise.
28 (vect_gen_perm_mask_checked): Likewise.
29 * fold-const.c (fold_ternary_loc): Likewise. Don't take
30 implementations of variable permutation vectors into account
31 when deciding which selector to use.
32 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
33 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
34 with a false third argument.
35 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
36 to test whether the constant selector is valid and can_vec_perm_var_p
37 to test whether a variable selector is valid.
39 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
41 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
42 * optabs-query.c (can_vec_perm_p): Likewise.
43 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
44 instead of vec_perm_indices.
45 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
46 (vect_gen_perm_mask_checked): Likewise,
47 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
48 (vect_gen_perm_mask_checked): Likewise,
50 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
52 * optabs-query.h (qimode_for_vec_perm): Declare.
53 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
54 (qimode_for_vec_perm): ...this new function.
55 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
57 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
59 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
60 does not have a conditional at the top.
62 2018-01-02 Richard Biener <rguenther@suse.de>
64 * ipa-inline.c (big_speedup_p): Fix expression.
66 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
69 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
72 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
76 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
77 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
78 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
79 cond_taken_branch_cost 3->4.
81 2018-01-01 Jakub Jelinek <jakub@redhat.com>
83 PR tree-optimization/83581
84 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
85 TODO_cleanup_cfg if any changes have been made.
88 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
89 convert_modes if target mode has the right side, but different mode
93 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
94 last argument when extracting from CONCAT. If either from_real or
95 from_imag is NULL, use expansion through memory. If result is not
96 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
97 the parts directly to inner mode, if even that fails, use expansion
101 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
102 check for bswap in mode rather than HImode and use that in expand_unop
105 2017-12-31 Jakub Jelinek <jakub@redhat.com>
108 * config/i386/i386.c (ix86_attribute_table): Remove excess
109 initializer for "shared" attribute.
111 2017-12-30 Tom de Vries <tom@codesourcery.com>
114 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
116 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
117 offload_funcs entries that no longer have a corresponding cgraph_node.
118 Mark the remaining ones as DECL_PRESERVE_P.
119 (output_lto): Call prune_offload_funcs.
121 2017-12-30 Jakub Jelinek <jakub@redhat.com>
123 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
124 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
125 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
126 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
127 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
128 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
129 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
130 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
131 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
132 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
133 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
134 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
135 vaesenc_<mode>, vpclmulqdq_<mode>,
136 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
138 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
140 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
141 _Float<N> and _Float<N>X variants for rounding built-in
143 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
144 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
145 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
146 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
147 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
148 * builtins.c (mathfn_built_in_2): Likewise.
149 * internal-fn.def (CEIL): Likewise.
151 (NEARBYINT): Likewise.
155 * convert.c (convert_to_integer_1): Likewise.
156 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
157 (integer_valued_real_call_p): Likewise.
158 * fold-const-call.c (fold_const_call_ss): Likewise.
159 * gencfn-macros.c (print_case_cfn): Change CFN and operator
160 printers to take a const char * suffix instead of a bool.
161 (print_define_operator_list): Likewise.
162 (fltall_suffixes): New list of suffixes, that include the
163 traditional suffixes as well as all of the _Float<N> and
165 (main): For _Float<N> and _Float<N>X functions, emit both
166 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
167 has the _Float<N> and _Float<N>X case names or operators. The
168 <name>_ALL variant has both the traditional and the
169 _Float<N>/_Float<N>X case names or operators.
170 * match.pd (COPYSIGN optimizations): Provide optimizations for
171 _Float<N> and _Float<N>X types where possible.
172 (MIN/MAX optimizations): Likewise.
173 (sqrt optimizations): Likewise.
174 (rounding optimizations): Likewise.
176 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
178 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
179 gen_int_mode rather than GEN_INT.
180 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
182 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
184 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
185 instead of CONSTANT_P.
186 (gen_vec_series): Likewise.
187 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
189 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
191 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
194 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
196 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
197 (aarch64_ld1x2<VDC:mode>): Likewise.
198 (aarch64_simd_ld1<mode>_x2): Likewise.
199 (aarch64_simd_ld1<mode>_x2): Likewise.
200 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
201 (vld1_s8_x2): Likewise.
202 (vld1_u16_x2): Likewise.
203 (vld1_s16_x2): Likewise.
204 (vld1_u32_x2): Likewise.
205 (vld1_s32_x2): Likewise.
206 (vld1_u64_x2): Likewise.
207 (vld1_s64_x2): Likewise.
208 (vld1_f16_x2): Likewise.
209 (vld1_f32_x2): Likewise.
210 (vld1_f64_x2): Likewise.
211 (vld1_p8_x2): Likewise.
212 (vld1_p16_x2): Likewise.
213 (vld1_p64_x2): Likewise.
214 (vld1q_u8_x2): Likewise.
215 (vld1q_s8_x2): Likewise.
216 (vld1q_u16_x2): Likewise.
217 (vld1q_s16_x2): Likewise.
218 (vld1q_u32_x2): Likewise.
219 (vld1q_s32_x2): Likewise.
220 (vld1q_u64_x2): Likewise.
221 (vld1q_s64_x2): Likewise.
222 (vld1q_f16_x2): Likewise.
223 (vld1q_f32_x2): Likewise.
224 (vld1q_f64_x2): Likewise.
225 (vld1q_p8_x2): Likewise.
226 (vld1q_p16_x2): Likewise.
227 (vld1q_p64_x2): Likewise.
229 2017-12-27 Martin Liska <mliska@suse.cz>
231 PR tree-optimization/83552
232 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
233 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
235 2017-12-27 Tom de Vries <tom@codesourcery.com>
237 * config/nvptx/nvptx.c (nvptx_option_override): Disable
238 -gstatement-frontiers.
240 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
242 PR rtl-optimization/83513
243 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
244 before priority comparison.
246 2017-12-25 Jakub Jelinek <jakub@redhat.com>
249 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
250 ix86_isa_flags2 to ix86_isa_flags.
251 * config/i386/i386-c.c (ix86_target_macros_internal): Test
252 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
253 isa_flags rather than isa_flags2.
254 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
255 and -mavx512bitalg from isa2_opts to isa_opts.
256 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
257 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
258 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
260 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
261 OPTION_MASK_ISA_AVX512F ored with another option similarly to
262 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
263 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
264 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
265 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
266 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
267 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
268 builtins from bdesc_args2 to bdesc_args section.
269 (__builtin_ia32_compressstoreuqi512_mask,
270 __builtin_ia32_compressstoreuhi512_mask,
271 __builtin_ia32_compressstoreuqi256_mask,
272 __builtin_ia32_expandloadqi512_mask,
273 __builtin_ia32_expandloadqi512_maskz,
274 __builtin_ia32_expandloadhi512_mask,
275 __builtin_ia32_expandloadhi512_maskz,
276 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
277 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
278 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
279 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
280 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
281 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
282 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
283 __builtin_ia32_vpshldv_v32hi_maskz,
284 __builtin_ia32_vpopcountb_v64qi_mask,
285 __builtin_ia32_vpopcountw_v32hi_mask,
286 __builtin_ia32_vpshufbitqmb512_mask,
287 __builtin_ia32_vpshufbitqmb256_mask): Add
288 " | OPTION_MASK_ISA_AVX512BW".
289 (__builtin_ia32_expandloadqi256_mask,
290 __builtin_ia32_expandloadqi256_maskz,
291 __builtin_ia32_vpopcountb_v32qi_mask): Add
292 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
293 (__builtin_ia32_expandloadhi256_mask,
294 __builtin_ia32_expandloadhi256_maskz,
295 __builtin_ia32_expandloadqi128_mask,
296 __builtin_ia32_expandloadqi128_maskz,
297 __builtin_ia32_expandloadhi128_mask,
298 __builtin_ia32_expandloadhi128_maskz,
299 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
300 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
301 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
302 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
303 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
304 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
305 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
306 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
307 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
308 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
309 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
310 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
311 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
312 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
313 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
314 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
315 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
316 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
317 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
318 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
319 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
320 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
321 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
322 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
323 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
324 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
325 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
326 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
327 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
328 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
329 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
330 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
331 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
332 __builtin_ia32_vpopcountw_v8hi_mask): Add
333 " | OPTION_MASK_ISA_AVX512VL".
334 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
335 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
336 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
337 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
338 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
339 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
340 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
341 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
342 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
343 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
344 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
345 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
346 _mm_mask_bitshuffle_epi64_mask): Likewise.
347 * common/config/i386/i386-common.c
348 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
349 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
350 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
351 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
352 OPTION_MASK_ISA_AVX512BITALG_UNSET.
353 (OPTION_MASK_ISA2_AVX512F_UNSET,
354 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
355 (ix86_handle_option): For -mno-general-regs-only, clear from
356 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
357 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
358 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
359 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
360 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
361 for moving from ix86_isa_flags2 to ix86_isa_flags.
363 2017-12-23 Jakub Jelinek <jakub@redhat.com>
366 * fold-const.c (struct contains_label_data): New type.
367 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
368 inside of a SWITCH_BODY seen during the walk.
369 (contains_label_p): Use walk_tree instead of
370 walk_tree_without_duplicates, prepare data for contains_label_1 and
373 2017-12-22 Jakub Jelinek <jakub@redhat.com>
376 * tree-iterator.c (alloc_stmt_list): Start with cleared
377 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
381 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
382 enable avx512f explicitly in #pragma GCC target.
383 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
384 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
385 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
386 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
387 __builtin_ia32_vpdpbusds_v8si_mask,
388 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
389 __builtin_ia32_vpdpbusds_v4si_mask,
390 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
391 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
392 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
393 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
394 __builtin_ia32_vpdpwssds_v8si_mask,
395 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
396 __builtin_ia32_vpdpwssds_v4si_mask,
397 __builtin_ia32_vpdpwssds_v4si_maskz): Use
398 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
399 just OPTION_MASK_ISA_AVX512VNNI.
401 2017-12-22 Martin Jambor <mjambor@suse.cz>
404 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
407 2017-12-22 Julia Koval <julia.koval@intel.com>
408 Sebastian Peryt <sebastian.peryt@intel.com>
410 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
411 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
412 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
413 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
414 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
415 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
416 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
417 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
418 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
419 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
420 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
421 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
422 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
423 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
424 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
425 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
426 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
427 _mm256_maskz_popcnt_epi64): New intrinsics.
428 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
429 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
430 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
431 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
432 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
433 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
434 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
435 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
436 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
437 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
438 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
439 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
440 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
441 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
442 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
443 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
444 __builtin_ia32_vpshufbitqmb256_mask,
445 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
446 * config/i386/i386-c.c (__AVX512BITALG__): New.
447 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
448 (ix86_valid_target_attribute_inner_p): Ditto.
449 (ix86_expand_args_builtin): Handle new types.
450 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
451 * config/i386/i386.opt: Add -mavx512bitalg.
452 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
453 avx512bitalgintrin.h.
454 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
455 (vpopcount<mode><mask_name>): Add more types.
456 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
457 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
459 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
461 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
462 Or in OPTION_MASK_ISA_AVX512F_SET.
463 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
464 OPTION_MASK_ISA_AVX512VNNI_UNSET.
465 (ix86_handle_option): Adjust for
466 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
467 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
469 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
470 isa_flag instead of isa_flag2.
471 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
472 isa_opts2 to isa_opts.
473 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
476 2017-12-22 Mike Stump <mikestump@comcast.net>
477 Eric Botcazou <ebotcazou@adacore.com>
479 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
481 2017-12-21 Jakub Jelinek <jakub@redhat.com>
484 * config/i386/i386.c (ix86_function_arg_boundary): Return
485 PARM_BOUNDARY for TYPE_EMPTY_P types.
488 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
489 if navail is >= dir.len.
491 2017-12-21 Steve Ellcey <sellcey@cavium.com>
493 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
496 2017-12-21 Jakub Jelinek <jakub@redhat.com>
498 PR rtl-optimization/80747
499 PR rtl-optimization/83512
500 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
501 succ edge from ENTRY, copy partition from e->dest to the newly
503 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
504 ENTRY, use BB_PARTITION of its successor block as current_partition.
505 Don't copy partition when splitting succ edge from ENTRY.
507 PR tree-optimization/83523
508 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
509 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
510 (convert_mult_to_fma): Likewise.
512 PR tree-optimization/83521
513 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
514 gimple_build_assign without code on result of
515 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
518 2017-12-21 Andrew Pinski <apinski@cavium.com>
519 Steve Ellcey <sellcey@cavium.com>
521 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
522 multi-arch for ilp32.
524 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
527 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
528 constraints to operand 2.
529 (*ashl<mode>3_mask_1): Ditto.
530 (*<shift_insn><mode>3_mask): Ditto.
531 (*<shift_insn><mode>3_mask_1): Ditto.
532 (*<rotate_insn><mode>3_mask): Ditto.
533 (*<rotate_insn><mode>3_mask_1): Ditto.
535 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
537 * reorg.c (make_return_insns): Reemit each insn with its own location.
539 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
541 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
542 generation for cases where splatting a value is not useful.
543 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
544 across a vec_duplicate and a paradoxical subreg forming a vector
545 mode to a vec_concat.
547 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
549 * combine.c (simplify_set): Do not transform subregs to zero_extends
550 if the destination is not a scalar int mode.
552 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
555 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
556 shared zero if the input has overflowed.
558 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
560 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
561 support for Cortex-A55 and Cortex-A75.
563 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
565 * common/config/arm/arm-common.c (compare_opt_names): Add function
566 comment. Use strcmp instead of manual loop.
568 2017-12-21 Martin Liska <mliska@suse.cz>
570 PR gcov-profile/83509
571 * gcov-dump.c (dump_gcov_file): Do not read info about
572 support_unexecuted_blocks for gcda files.
574 2017-12-21 Jakub Jelinek <jakub@redhat.com>
576 PR rtl-optimization/82973
577 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
578 (valid_for_const_vector_p): ... this.
579 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
580 (valid_for_const_vector_p): ... this. Adjust function comment.
581 (gen_vec_duplicate): Adjust caller.
582 * optabs.c (expand_vector_broadcast): Likewise.
583 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
584 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
586 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
587 instead of GET_CODE == CONST_FIXED.
588 (simplify_subreg): Use CONST_FIXED_P macro instead of
589 GET_CODE == CONST_FIXED.
592 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
593 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
594 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
595 (ix86_option_override_internal): Adjust for
596 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
597 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
598 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
599 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
600 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
601 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
602 def_builtin2 for CET builtins.
603 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
604 up formatting in IX86_BUILTIN_RDPID code.
605 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
606 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
607 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
608 ix86_isa_flags2 to ix86_isa_flags.
609 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
611 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
612 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
613 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
615 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
616 Or in OPTION_MASK_ISA_AVX512F_SET.
617 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
618 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
619 (ix86_handle_option): Adjust for
620 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
621 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
623 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
624 Alan Hayward <alan.hayward@arm.com>
625 David Sherwood <david.sherwood@arm.com>
627 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
628 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
629 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
630 Track polynomial offsets.
632 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
633 Alan Hayward <alan.hayward@arm.com>
634 David Sherwood <david.sherwood@arm.com>
636 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
637 Treat drb->init as a poly_int. Fail if its misalignment wrt
638 vector_alignment isn't known.
640 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
641 Alan Hayward <alan.hayward@arm.com>
642 David Sherwood <david.sherwood@arm.com>
644 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
646 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
647 (vect_loop_versioning): Take the loop versioning threshold as a
649 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
650 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
651 versioning_threshold.
652 (vect_analyze_loop_2): Compute the loop versioning threshold
653 whenever loop versioning is needed, and store it in the new
654 field rather than combining it with the cost model threshold.
655 (vect_transform_loop): Update call to vect_loop_versioning.
656 Try to combine the loop versioning and cost thresholds here.
658 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
659 Alan Hayward <alan.hayward@arm.com>
660 David Sherwood <david.sherwood@arm.com>
662 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
663 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
664 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
665 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
666 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
667 (group_compare_offset): Update accordingly.
668 (split_small_address_groups_p): Likewise.
669 (record_use): Take addr_offset as a poly_uint64 rather than
670 an unsigned HOST_WIDE_INT.
671 (strip_offset): Return the offset as a poly_uint64 rather than
672 an unsigned HOST_WIDE_INT.
673 (record_group_use, split_address_groups): Track polynomial offsets.
674 (add_iv_candidate_for_use): Likewise.
675 (addr_offset_valid_p): Take the offset as a poly_int64 rather
676 than a HOST_WIDE_INT.
677 (strip_offset_1): Return the offset as a poly_int64 rather than
680 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
681 Alan Hayward <alan.hayward@arm.com>
682 David Sherwood <david.sherwood@arm.com>
684 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
685 rather than a HOST_WIDE_INT.
686 * tree.c (get_binfo_at_offset): Likewise.
688 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
689 Alan Hayward <alan.hayward@arm.com>
690 David Sherwood <david.sherwood@arm.com>
692 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
693 rather than a HOST_WIDE_INT.
694 * tree-sra.c (build_ref_for_offset): Likewise.
696 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
697 Alan Hayward <alan.hayward@arm.com>
698 David Sherwood <david.sherwood@arm.com>
700 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
702 * tree.c (mem_ref_offset): Likewise.
703 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
704 * builtins.c (get_object_alignment_2): Likewise.
705 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
706 * gimple-fold.c (get_base_constructor): Likewise.
707 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
708 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
710 * ipa-polymorphic-call.c
711 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
712 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
713 (get_ancestor_addr_info): Likewise.
714 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
715 * match.pd: Likewise.
716 * tree-data-ref.c (dr_analyze_innermost): Likewise.
717 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
718 * tree-eh.c (tree_could_trap_p): Likewise.
719 * tree-object-size.c (addr_object_size): Likewise.
720 * tree-ssa-address.c (copy_ref_info): Likewise.
721 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
722 (indirect_refs_may_alias_p): Likewise.
723 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
724 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
725 (non_rewritable_mem_ref_base): Likewise.
726 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
727 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
728 * varasm.c (decode_addr_const): Likewise.
730 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
731 Alan Hayward <alan.hayward@arm.com>
732 David Sherwood <david.sherwood@arm.com>
734 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
735 offsets for MEM_REFs.
737 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
738 Alan Hayward <alan.hayward@arm.com>
739 David Sherwood <david.sherwood@arm.com>
741 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
742 as a poly_uint64 rather than an unsigned int.
743 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
745 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
746 Alan Hayward <alan.hayward@arm.com>
747 David Sherwood <david.sherwood@arm.com>
749 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
751 (decode_addr_const): Update accordingly.
753 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
754 Alan Hayward <alan.hayward@arm.com>
755 David Sherwood <david.sherwood@arm.com>
757 * tree.h (bit_field_size, bit_field_offset): New functions.
758 * hsa-gen.c (gen_hsa_addr): Use them.
759 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
760 (simplify_vector_constructor): Likewise.
761 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
762 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
763 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
764 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
767 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
768 Alan Hayward <alan.hayward@arm.com>
769 David Sherwood <david.sherwood@arm.com>
771 * expr.h (emit_group_load, emit_group_load_into_temps)
772 (emit_group_store): Take the size as a poly_int64 rather than an int.
773 * expr.c (emit_group_load_1, emit_group_load): Likewise.
774 (emit_group_load_into_temp, emit_group_store): Likewise.
776 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
777 Alan Hayward <alan.hayward@arm.com>
778 David Sherwood <david.sherwood@arm.com>
780 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
781 an unsigned int to a poly_uint64.
782 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
783 sizes as poly_uint64s rather than unsigned ints.
784 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
787 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
788 Alan Hayward <alan.hayward@arm.com>
789 David Sherwood <david.sherwood@arm.com>
791 * lra-constraints.c (emit_inc): Change inc_amount from an int
794 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
795 Alan Hayward <alan.hayward@arm.com>
796 David Sherwood <david.sherwood@arm.com>
798 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
800 (add_stack_var, stack_var_cmp, partition_stack_vars)
801 (dump_stack_var_partition): Update accordingly.
802 (alloc_stack_frame_space): Take the size as a poly_int64 rather
803 than a HOST_WIDE_INT.
804 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
805 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
806 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
807 rather than a HOST_WIDE_INT.
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 * target.def (return_pops_args): Treat both the input and output
814 sizes as poly_int64s rather than HOST_WIDE_INTS.
815 * targhooks.h (default_return_pops_args): Update accordingly.
816 * targhooks.c (default_return_pops_args): Likewise.
817 * doc/tm.texi: Regenerate.
818 * emit-rtl.h (incoming_args): Change pops_args, size and
819 outgoing_args_size from int to poly_int64_pod.
820 * function.h (expr_status): Change x_pending_stack_adjust and
821 x_stack_pointer_delta from int to poly_int64.
822 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
823 (ARGS_SIZE_RTX): Update accordingly.
824 * calls.c (highest_outgoing_arg_in_use): Change from int to
826 (stack_usage_watermark, stored_args_watermark): New variables.
827 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
828 (emit_call_1): Change the stack_size and rounded_stack_size
829 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
831 (save_fixed_argument_area): Check stack_usage_watermark.
832 (initialize_argument_information): Change old_pending_adj from
833 a HOST_WIDE_INT * to a poly_int64_pod *.
834 (compute_argument_block_size): Return the size as a poly_int64
836 (finalize_must_preallocate): Track polynomial argument sizes.
837 (compute_argument_addresses): Likewise.
838 (internal_arg_pointer_based_exp): Track polynomial offsets.
839 (mem_overlaps_already_clobbered_arg_p): Rename to...
840 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
841 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
842 Check stored_args_used_watermark.
843 (load_register_parameters): Update accordingly.
844 (check_sibcall_argument_overlap_1): Likewise.
845 (combine_pending_stack_adjustment_and_call): Take the unadjusted
846 args size as a poly_int64 rather than an int. Return a bool
847 indicating whether the optimization was possible and return
848 the new adjustment by reference.
849 (check_sibcall_argument_overlap): Track polynomail argument sizes.
850 Update stored_args_watermark.
851 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
852 (expand_call): Likewise. Maintain stack_usage_watermark and
853 stored_args_watermark. Update calls to
854 combine_pending_stack_adjustment_and_call.
855 (emit_library_call_value_1): Handle polynomial argument sizes.
856 Call stack_region_maybe_used_p and mark_stack_region_used.
857 Maintain stack_usage_watermark.
858 (store_one_arg): Likewise. Update call to
859 mem_overlaps_already_clobbered_arg_p.
860 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
862 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
863 * config/microblaze/microblaze.c (microblaze_function_prologue):
865 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
866 TARGET_RETURN_POPS_ARGS interface.
867 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
869 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
870 * config/i386/i386.c (ix86_return_pops_args): Update for new
871 TARGET_RETURN_POPS_ARGS interface.
872 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
873 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
874 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
875 TARGET_RETURN_POPS_ARGS interface.
876 * config/vax/vax.c (vax_return_pops_args): Likewise.
877 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
878 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
879 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
880 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
881 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
882 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
883 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
884 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
885 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
886 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
887 and x_stack_pointer_delta from int to poly_int64.
888 * dojump.c (do_pending_stack_adjust): Update accordingly.
889 * explow.c (allocate_dynamic_stack_space): Handle polynomial
890 stack_pointer_deltas.
891 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
892 (pad_to_arg_alignment): Track polynomial offsets.
893 (assign_parm_find_stack_rtl): Likewise.
894 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
895 * toplev.c (output_stack_usage): Update reference to
896 current_function_pushed_stack_size.
898 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
899 Alan Hayward <alan.hayward@arm.com>
900 David Sherwood <david.sherwood@arm.com>
902 * function.c (in_arg_offset, var_offset, dynamic_offset)
903 (out_arg_offset, cfa_offset): Change from int to poly_int64.
904 (instantiate_new_reg): Return the new offset as a poly_int64_pod
905 rather than a HOST_WIDE_INT.
906 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
907 (instantiate_virtual_regs_in_insn): Likewise.
909 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
910 Alan Hayward <alan.hayward@arm.com>
911 David Sherwood <david.sherwood@arm.com>
913 * rtl.h (get_args_size, add_args_size_note): New functions.
914 (find_args_size_adjust): Return a poly_int64 rather than a
916 (fixup_args_size_notes): Likewise. Make the same change to the
917 end_args_size parameter.
918 * rtlanal.c (get_args_size, add_args_size_note): New functions.
919 * builtins.c (expand_builtin_trap): Use add_args_size_note.
920 * calls.c (emit_call_1): Likewise.
921 * explow.c (adjust_stack_1): Likewise.
922 * cfgcleanup.c (old_insns_match_p): Update use of
923 find_args_size_adjust.
924 * combine.c (distribute_notes): Track polynomial arg sizes.
925 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
926 end_true_args_size, beg_delay_args_size and end_delay_args_size
927 from HOST_WIDE_INT to poly_int64.
928 (add_cfi_args_size): Take the args_size as a poly_int64 rather
929 than a HOST_WIDE_INT.
930 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
931 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
932 polynomial arg sizes.
933 * emit-rtl.c (try_split): Use get_args_size.
934 * recog.c (peep2_attempt): Likewise.
935 * reload1.c (reload_as_needed): Likewise.
936 * expr.c (find_args_size_adjust): Return the adjustment as a
937 poly_int64 rather than a HOST_WIDE_INT.
938 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
939 to a poly_int64 and change the return type in the same way.
940 (emit_single_push_insn): Track polynomial arg sizes.
942 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
943 Alan Hayward <alan.hayward@arm.com>
944 David Sherwood <david.sherwood@arm.com>
946 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
947 from HOST_WIDE_INT to poly_int64.
948 * expr.c (push_block, emit_push_insn): Likewise.
950 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
951 Alan Hayward <alan.hayward@arm.com>
952 David Sherwood <david.sherwood@arm.com>
954 * function.h (frame_space): Change start and length from HOST_WIDE_INT
956 (get_frame_size): Return the size as a poly_int64 rather than a
958 (frame_offset_overflow): Take the offset as a poly_int64 rather
959 than a HOST_WIDE_INT.
960 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
961 (assign_stack_temp): Likewise for the size.
962 * function.c (get_frame_size): Return a poly_int64 rather than
964 (frame_offset_overflow): Take the offset as a poly_int64 rather
965 than a HOST_WIDE_INT.
966 (try_fit_stack_local): Take the start, length and size as poly_int64s
967 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
968 rather than a HOST_WIDE_INT.
969 (add_frame_space): Take the start and end as poly_int64s rather than
971 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
972 (assign_stack_temp): Likewise for the size.
973 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
975 (find_temp_slot_from_address): Handle polynomial offsets.
976 (combine_temp_slots): Likewise.
977 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
979 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
980 poly_int64 rather than a HOST_WIDE_INT.
981 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
982 than a HOST_WIDE_INT.
983 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
984 polynomial frame offsets.
985 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
986 as a poly_int64 rather than an int.
987 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
988 * config/v850/v850-protos.h (compute_frame_size): Likewise.
989 * config/v850/v850.c (compute_frame_size): Likewise.
990 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
991 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
992 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
993 * config/pa/pa.c (pa_compute_frame_size): Likewise.
994 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
995 rather than a HOST_WIDE_INT.
996 * explow.c (get_dynamic_stack_base): Likewise.
997 * final.c (final_start_function): Use the constant lower bound
998 of the frame size for -Wframe-larger-than.
999 * ira.c (do_reload): Adjust for new get_frame_size return type.
1000 * lra.c (lra): Likewise.
1001 * reload1.c (reload): Likewise.
1002 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1003 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1004 * rtlanal.c (get_initial_register_offset): Return the offset as
1005 a poly_int64 rather than a HOST_WIDE_INT.
1007 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1008 Alan Hayward <alan.hayward@arm.com>
1009 David Sherwood <david.sherwood@arm.com>
1011 * reload1.c (elim_table): Change initial_offset, offset and
1012 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1013 (offsets_at): Change the target array's element type from
1014 HOST_WIDE_INT to poly_int64_pod.
1015 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1016 (elimination_costs_in_insn, update_eliminable_offsets)
1017 (verify_initial_elim_offsets, set_offsets_for_label)
1018 (init_eliminable_invariants): Update after above changes.
1020 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1021 Alan Hayward <alan.hayward@arm.com>
1022 David Sherwood <david.sherwood@arm.com>
1024 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1025 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1026 (decomposition): Change start and end from HOST_WIDE_INT
1028 (decompose, immune_p): Update accordingly.
1029 (find_inc_amount): Return a poly_int64 rather than an int.
1030 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1033 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1034 Alan Hayward <alan.hayward@arm.com>
1035 David Sherwood <david.sherwood@arm.com>
1037 * tree.h (get_inner_reference): Return the bitsize and bitpos
1038 as poly_int64_pods rather than HOST_WIDE_INT.
1039 * fold-const.h (ptr_difference_const): Return the pointer difference
1040 as a poly_int64_pod rather than a HOST_WIDE_INT.
1041 * expr.c (get_inner_reference): Return the bitsize and bitpos
1042 as poly_int64_pods rather than HOST_WIDE_INT.
1043 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1045 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1046 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1047 (optimize_bit_field_compare): Update call to get_inner_reference.
1048 (decode_field_reference): Likewise.
1049 (fold_unary_loc): Track polynomial offsets and sizes.
1050 (split_address_to_core_and_offset): Return the bitpos as a
1051 poly_int64_pod rather than a HOST_WIDE_INT.
1052 (ptr_difference_const): Likewise for the pointer difference.
1053 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1054 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1055 * dbxout.c (dbxout_expand_expr): Likewise.
1056 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1057 (loc_list_from_tree_1, fortran_common): Likewise.
1058 * gimple-laddress.c (pass_laddress::execute): Likewise.
1059 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1060 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1061 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1062 * tree-affine.c (tree_to_aff_combination): Likewise.
1063 (get_inner_reference_aff): Likewise.
1064 * tree-data-ref.c (split_constant_offset_1): Likewise.
1065 (dr_analyze_innermost): Likewise.
1066 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1067 * tree-sra.c (ipa_sra_check_caller): Likewise.
1068 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1069 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1070 (instrument_bool_enum_load, instrument_object_size): Likewise.
1071 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1072 to get_inner_reference.
1073 * hsa-gen.c (gen_hsa_addr): Likewise.
1074 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1075 * tsan.c (instrument_expr): Likewise.
1076 * match.pd: Update call to ptr_difference_const.
1078 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1079 Alan Hayward <alan.hayward@arm.com>
1080 David Sherwood <david.sherwood@arm.com>
1082 * fold-const.c (fold_comparison): Track sizes and offsets as
1083 poly_int64s rather than HOST_WIDE_INTs when folding address
1086 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1087 Alan Hayward <alan.hayward@arm.com>
1088 David Sherwood <david.sherwood@arm.com>
1090 * expr.h (get_bit_range): Return the bitstart and bitend as
1091 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1092 as a poly_int64 rather than a HOST_WIDE_INT.
1093 * expr.c (get_bit_range): Likewise.
1094 (expand_assignment): Update call accordingly.
1095 * fold-const.c (optimize_bit_field_compare): Likewise.
1097 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1098 Alan Hayward <alan.hayward@arm.com>
1099 David Sherwood <david.sherwood@arm.com>
1101 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1102 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1103 as an optional final parameter.
1104 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1105 (aarch64_legitimate_address_p): Likewise.
1106 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1107 instead of an rtx code.
1108 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1109 (aarch64_legitimate_address_hook_p): Likewise.
1110 (aarch64_print_ldpstp_address): Likewise.
1111 (aarch64_print_operand_address): Likewise.
1112 (aarch64_address_cost): Likewise.
1113 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1114 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1115 (aarch64_mem_pair_lanes_operand): Likewise.
1117 2017-12-20 Richard Biener <rguenther@suse.de>
1119 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1120 update_stmt_if_modified.
1122 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1124 PR tree-optimization/83491
1125 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1126 before walking uses. Improve coding style and comments.
1128 2017-12-20 Tom de Vries <tom@codesourcery.com>
1130 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1132 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1135 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1137 (pass_ipa_free_fn_summary): Add small_p private data member,
1138 initialize to false in the ctor.
1139 (pass_ipa_free_fn_summary::clone,
1140 pass_ipa_free_fn_summary::set_pass_param,
1141 pass_ipa_free_fn_summary::gate): New methods.
1142 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1143 | TODO_dump_symtab if small_p.
1144 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1145 entry and add another instance of the pass with false parm after
1147 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1148 ipa_free_fn_summary here.
1150 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1152 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1154 2017-12-20 Martin Sebor <msebor@redhat.com>
1157 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1158 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1159 (expand_builtin_strncmp): Same.
1161 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1164 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1165 allowed debug stmts before labels.
1166 (expand_gimple_basic_block): Likewise.
1167 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1168 * gimple-iterator.h (gsi_after_labels): Likewise.
1169 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1170 rename reused variable, and simplify using gsi_move_before.
1171 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1172 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1173 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1174 (gimple_verify_flow_info, gimple_block_label): Likewise.
1175 (make_blocks): Move debug markers after adjacent labels.
1176 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1177 allowed debug insns outside blocks.
1178 * df-scan.c (df_insn_delete): Likewise.
1179 * lra-constraints.c (update_ebb_live_info): Likewise.
1180 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1181 (vt_initialize, delete_vta_debug_insns): Likewise.
1182 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1184 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1185 Alan Hayward <alan.hayward@arm.com>
1186 David Sherwood <david.sherwood@arm.com>
1188 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1189 (round_up_to_byte_boundary): Likewise.
1190 * expr.h (get_bit_range): Add temporary shim.
1191 * gimple-ssa-store-merging.c (store_operand_info): Change the
1192 bitsize, bitpos, bitregion_start and bitregion_end fields from
1193 unsigned HOST_WIDE_INT to poly_uint64.
1194 (merged_store_group): Likewise load_align_base.
1195 (compatible_load_p, compatible_load_p): Update accordingly.
1196 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1197 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1198 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1199 bitregion_start and bitregion_end as poly_uint64s rather than
1200 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1201 (handled_load): Take the bitsize, bitpos,
1202 bitregion_start and bitregion_end as poly_uint64s rather than
1203 unsigned HOST_WIDE_INTs.
1204 (pass_store_merging::process_store): Update call to
1205 mem_valid_for_store_merging.
1207 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1208 Alan Hayward <alan.hayward@arm.com>
1209 David Sherwood <david.sherwood@arm.com>
1211 * builtins.c (get_object_alignment_2): Track polynomial offsets
1212 and sizes. Update the alignment handling.
1214 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1215 Alan Hayward <alan.hayward@arm.com>
1216 David Sherwood <david.sherwood@arm.com>
1218 * tree.h (get_inner_reference): Add a version that returns the
1219 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1220 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1221 the case in which bitpos is not associated with the first byte.
1223 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1224 Alan Hayward <alan.hayward@arm.com>
1225 David Sherwood <david.sherwood@arm.com>
1227 * tree-affine.h (get_inner_reference_aff): Return the size as a
1229 * tree-affine.c (get_inner_reference_aff): Likewise.
1230 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1231 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1233 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1234 Alan Hayward <alan.hayward@arm.com>
1235 David Sherwood <david.sherwood@arm.com>
1237 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1238 HOST_WIDE_INT rather than a poly_int64.
1240 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1241 Alan Hayward <alan.hayward@arm.com>
1242 David Sherwood <david.sherwood@arm.com>
1244 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1245 HOST_WIDE_INT to poly_int64_pod.
1246 (perform_symbolic_merge): Update accordingly.
1247 (bswap_replace): Likewise.
1249 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1250 Alan Hayward <alan.hayward@arm.com>
1251 David Sherwood <david.sherwood@arm.com>
1253 * tree-affine.h (aff_tree::offset): Change from widest_int
1255 (wide_int_ext_for_comb): Delete.
1256 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1257 constants as poly_widest_int rather than widest_int.
1258 (aff_combination_constant_multiple_p): Return the multiplier
1259 as a poly_widest_int.
1260 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1262 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1263 version static and add an overload for poly_widest_int.
1264 (aff_combination_const, aff_combination_add_cst)
1265 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1266 the constants as poly_widest_int rather than widest_int.
1267 (tree_to_aff_combination): Generalize INTEGER_CST case to
1269 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1270 (aff_combination_add_product, aff_combination_mult): Handle
1272 (aff_combination_constant_multiple_p): Return the multiplier
1273 as a poly_widest_int.
1274 * tree-predcom.c (determine_offset): Return the offset as a
1276 (split_data_refs_to_components, suitable_component_p): Update
1278 (valid_initializer_p): Update call to
1279 aff_combination_constant_multiple_p.
1280 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1281 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1282 as a poly_int64 rather than a HOST_WIDE_INT.
1283 (get_address_cost): Handle polynomial offsets.
1284 (iv_elimination_compare_lt): Likewise.
1285 (rewrite_use_nonlinear_expr): Likewise.
1287 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1288 Alan Hayward <alan.hayward@arm.com>
1289 David Sherwood <david.sherwood@arm.com>
1291 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1292 as a poly_int64_pod rather than a HOST_WIDE_INT.
1293 (get_addr_base_and_unit_offset): Likewise.
1294 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1295 (get_addr_base_and_unit_offset): Likewise.
1296 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1297 to poly_int64 in example.
1298 * fold-const.c (fold_binary_loc): Update call to
1299 get_addr_base_and_unit_offset.
1300 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1301 (maybe_canonicalize_mem_ref_addr): Likewise.
1302 (gimple_fold_stmt_to_constant_1): Likewise.
1303 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1305 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1306 * match.pd: Likewise.
1307 * omp-low.c (lower_omp_target): Likewise.
1308 * tree-sra.c (build_ref_for_offset): Likewise.
1309 (build_debug_ref_for_model): Likewise.
1310 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1311 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1312 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1313 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1314 (constant_pointer_difference): Likewise.
1315 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1316 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1317 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1318 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1319 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1320 (set_ssa_val_to): Likewise.
1321 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1322 (maybe_diag_stxncpy_trunc): Likewise.
1323 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1324 * tree.c (build_simple_mem_ref_loc): Likewise.
1325 (array_at_struct_end_p): Likewise.
1327 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1328 Alan Hayward <alan.hayward@arm.com>
1329 David Sherwood <david.sherwood@arm.com>
1331 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1332 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1333 (get_ref_base_and_extent_hwi): Declare.
1334 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1335 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1336 (get_ref_base_and_extent_hwi): New function.
1337 * cfgexpand.c (expand_debug_expr): Update call to
1338 get_ref_base_and_extent.
1339 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1340 * gimple-fold.c (get_base_constructor): Return the offset as a
1341 poly_int64_pod rather than a HOST_WIDE_INT.
1342 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1343 * ipa-polymorphic-call.c
1344 (ipa_polymorphic_call_context::set_by_invariant)
1345 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1346 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1347 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1348 rather than get_ref_base_and_extent.
1349 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1350 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1351 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1353 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1354 call to get_ref_base_and_extent.
1355 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1356 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1357 (stmt_kills_ref_p): Likewise.
1358 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1359 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1361 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1362 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1363 when calling native_encode_expr.
1364 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1365 call to get_ref_base_and_extent.
1366 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1367 get_ref_base_and_extent.
1368 * var-tracking.c (track_expr_p): Likewise.
1370 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1371 Alan Hayward <alan.hayward@arm.com>
1372 David Sherwood <david.sherwood@arm.com>
1374 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1375 HOST_WIDE_INT to poly_int64_pod.
1376 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1377 polynomail parameter offsets.
1379 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1380 Alan Hayward <alan.hayward@arm.com>
1381 David Sherwood <david.sherwood@arm.com>
1383 * gengtype.c (main): Handle poly_int64_pod.
1384 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1385 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1386 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1387 from HOST_WIDE_INT to poly_int64_pod.
1388 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1389 (copy_cfa): New function.
1390 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1391 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1392 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
1393 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
1394 (initial_return_save): Treat offsets as poly_ints.
1395 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
1396 offset is nonconstant.
1397 (reg_save): Take the offset as a poly_int64. Fall back to
1398 DW_CFA_expression for nonconstant offsets.
1399 (queue_reg_save): Take the offset as a poly_int64.
1400 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
1402 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1403 Alan Hayward <alan.hayward@arm.com>
1404 David Sherwood <david.sherwood@arm.com>
1406 * rtl.h (operand_subword, operand_subword_force): Take the offset
1407 as a poly_uint64 an unsigned int.
1408 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
1410 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1411 Alan Hayward <alan.hayward@arm.com>
1412 David Sherwood <david.sherwood@arm.com>
1414 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
1415 'p' format code. Use INT_LIST rather than SUBREG as the example of
1416 a code with an XINT and an XEXP. Remove the implication that
1417 accessing an rtx field using XINT is expected to work.
1418 * rtl.def (SUBREG): Change format from "ei" to "ep".
1419 * rtl.h (rtunion::rt_subreg): New field.
1420 (XCSUBREG): New macro.
1421 (SUBREG_BYTE): Use it.
1422 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
1423 Update constructor accordingly.
1424 (subreg_shape::operator ==): Update accordingly.
1425 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
1426 than an unsigned int.
1427 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
1428 a poly_uint64 rather than an unsigned int.
1429 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
1430 than an unsigned int.
1431 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
1432 (subreg_size_highpart_offset): Return a poly_uint64 rather than
1433 an unsigned int. Take the sizes as poly_uint64s.
1434 (subreg_offset_from_lsb): Return a poly_uint64 rather than
1435 an unsigned int. Take the shift as a poly_uint64 rather than
1437 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
1438 as a poly_uint64 rather than an unsigned int.
1439 (simplify_subreg_regno): Likewise.
1440 (byte_lowpart_offset): Return the memory offset as a poly_int64
1442 (subreg_memory_offset): Likewise. Take the subreg offset as a
1443 poly_uint64 rather than an unsigned int.
1444 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
1445 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
1446 poly_uint64 rather than an unsigned int.
1447 * rtl.c (rtx_format): Describe 'p' in comment.
1448 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
1449 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
1450 offset as a poly_uint64 rather than an unsigned int.
1451 (byte_lowpart_offset): Return the memory offset as a poly_int64
1453 (subreg_memory_offset): Likewise. Take the subreg offset as a
1454 poly_uint64 rather than an unsigned int.
1455 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
1456 mode sizes as poly_uint64s rather than unsigned ints. Return a
1457 poly_uint64 rather than an unsigned int.
1458 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
1459 (copy_insn_1): Handle 'p'.
1460 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
1461 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
1462 an unsigned int. Return the shift in the same way.
1463 (subreg_lsb): Return the shift as a poly_uint64 rather than an
1465 (subreg_size_offset_from_lsb): Take the sizes and shift as
1466 poly_uint64s rather than unsigned ints. Return the offset as
1468 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
1469 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
1471 * rtlhash.c (add_rtx): Handle 'p'.
1472 * genemit.c (gen_exp): Likewise.
1473 * gengenrtl.c (type_from_format, gendef): Likewise.
1474 * gensupport.c (subst_pattern_match, get_alternatives_number)
1475 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
1476 (subst_dup): Likewise.
1477 * gengtype.c (adjust_field_rtx_def): Likewise.
1478 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
1479 (match_pattern_2): Likewise.
1480 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
1481 (rtx_test::subreg_field): New function.
1482 (operator ==, safe_to_hoist_p, transition_parameter_type)
1483 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
1484 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
1485 * genpeep.c (match_rtx): Likewise.
1486 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
1487 (rtx_writer::print_rtx_operand): Handle 'p'.
1488 (print_value): Handle SUBREG.
1489 * read-rtl.c (apply_int_iterator): Likewise.
1490 (rtx_reader::read_rtx_operand): Handle 'p'.
1491 * alias.c (rtx_equal_for_memref_p): Likewise.
1492 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
1493 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
1495 * calls.c (expand_call): Likewise.
1496 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
1497 (make_extraction, gen_lowpart_for_combine): Likewise.
1498 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
1500 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
1501 rather than an unsigned int. Treat subreg offsets as poly_ints.
1502 (exp_equiv_p): Handle 'p'.
1503 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
1504 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
1505 * dse.c (find_shift_sequence): Likewise.
1506 * dwarf2out.c (rtl_for_decl_location): Likewise.
1507 * expmed.c (extract_low_bits): Likewise.
1508 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
1509 (expand_expr_real_2): Likewise.
1510 * final.c (alter_subreg): Likewise.
1511 (leaf_renumber_regs_insn): Handle 'p'.
1512 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
1513 Treat subreg offsets as poly_ints.
1514 * fwprop.c (forward_propagate_and_simplify): Likewise.
1515 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
1516 * ira.c (get_subreg_tracking_sizes): Likewise.
1517 * ira-conflicts.c (go_through_subreg): Likewise.
1518 * ira-lives.c (process_single_reg_class_operands): Likewise.
1519 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
1520 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
1521 as a poly_uint64 rather than an unsigned int.
1522 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
1523 subreg offsets as poly_ints.
1524 * lra-constraints.c (operands_match_p): Handle 'p'.
1525 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
1526 * lra-spills.c (assign_mem_slot): Likewise.
1527 * postreload.c (move2add_valid_value_p): Likewise.
1528 * recog.c (general_operand, indirect_operand): Likewise.
1529 * regcprop.c (copy_value, maybe_mode_change): Likewise.
1530 (copyprop_hardreg_forward_1): Likewise.
1531 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
1532 (record_subregs_of_mode): Likewise.
1533 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
1534 * reload.c (operands_match_p): Handle 'p'.
1535 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
1536 * reload1.c (alter_reg, choose_reload_regs): Likewise.
1537 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
1538 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
1539 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
1540 (simplify_const_poly_int_tests<N>::run): Likewise.
1541 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
1542 a poly_uint64 rather than an unsigned int.
1543 * valtrack.c (debug_lowpart_subreg): Likewise.
1544 * var-tracking.c (var_lowpart): Likewise.
1545 (loc_cmp): Handle 'p'.
1547 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1548 Alan Hayward <alan.hayward@arm.com>
1549 David Sherwood <david.sherwood@arm.com>
1551 * ira.c (get_subreg_tracking_sizes): New function.
1552 (init_live_subregs): Take an integer size rather than a register.
1553 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
1554 to init_live_subregs.
1556 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1557 Alan Hayward <alan.hayward@arm.com>
1558 David Sherwood <david.sherwood@arm.com>
1560 * expr.c (store_constructor_field): Change bitsize from a
1561 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
1562 HOST_WIDE_INT to a poly_int64.
1563 (store_constructor): Change size from a HOST_WIDE_INT to
1565 (store_field): Likewise bitsize and bitpos.
1567 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1568 Alan Hayward <alan.hayward@arm.com>
1569 David Sherwood <david.sherwood@arm.com>
1571 * expmed.h (store_bit_field): Change bitregion_start and
1572 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
1573 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
1574 (store_bit_field_1, store_integral_bit_field, store_bit_field)
1575 (store_fixed_bit_field, store_split_bit_field): Likewise.
1576 * expr.c (store_constructor_field, store_field): Likewise.
1577 (optimize_bitfield_assignment_op): Likewise. Make the same change
1578 to bitsize and bitpos.
1579 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
1580 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
1581 same change in the constructor arguments.
1582 (get_best_mode): Change bitregion_start and bitregion_end from
1583 unsigned HOST_WIDE_INT to poly_uint64.
1584 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
1585 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
1587 (bit_field_mode_iterator::next_mode): Update for new types
1588 of m_bitregion_start and m_bitregion_end.
1589 (get_best_mode): Change bitregion_start and bitregion_end from
1590 unsigned HOST_WIDE_INT to poly_uint64.
1592 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1593 Alan Hayward <alan.hayward@arm.com>
1594 David Sherwood <david.sherwood@arm.com>
1596 * rtl.h (simplify_gen_subreg): Add a temporary overload that
1597 accepts poly_uint64 offsets.
1598 * expmed.h (extract_bit_field): Take bitsize and bitnum as
1599 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1600 * expmed.c (lowpart_bit_field_p): Likewise.
1601 (extract_bit_field_as_subreg): New function, split out from...
1602 (extract_bit_field_1): ...here. Take bitsize and bitnum as
1603 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
1604 extractions, check that BITSIZE matches the size of the extracted
1605 value and that BITNUM is an exact multiple of that size.
1606 If all else fails, try forcing the value into memory if
1607 BITNUM is variable, and adjusting the address so that the
1608 offset is constant. Split the part that can only handle constant
1609 bitsize and bitnum out into...
1610 (extract_integral_bit_field): ...this new function.
1611 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
1612 rather than unsigned HOST_WIDE_INTs.
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 * expmed.h (store_bit_field): Take bitsize and bitnum as
1619 poly_uint64s rather than unsigned HOST_WIDE_INTs.
1620 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
1621 that returns the byte size.
1622 (store_bit_field_1): Take bitsize and bitnum as
1623 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
1624 to simple_mem_bitfield_p. Split the part that can only handle
1625 constant bitsize and bitnum out into...
1626 (store_integral_bit_field): ...this new function.
1627 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
1628 than unsigned HOST_WIDE_INTs.
1629 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
1631 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1632 Alan Hayward <alan.hayward@arm.com>
1633 David Sherwood <david.sherwood@arm.com>
1635 * lra-int.h (lra_reg): Change offset from int to poly_int64.
1636 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
1638 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1639 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
1640 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
1641 offset as a poly_int64 rather than an int.
1642 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
1643 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
1644 * lra-constraints.c (equiv_address_substitution): Track offsets
1646 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
1647 (curr_insn_transform): Handle the new form of sp_offset.
1648 * lra-eliminations.c (lra_elim_table): Change previous_offset
1649 and offset from HOST_WIDE_INT to poly_int64.
1650 (print_elim_table, update_reg_eliminate): Update accordingly.
1651 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
1652 (get_elimination): Update accordingly.
1653 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
1654 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
1655 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
1656 poly_int64 offsets generally.
1657 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
1658 (mark_not_eliminable, init_elimination): Update accordingly.
1659 (remove_reg_equal_offset_note): Return a bool and pass the new
1660 offset back by pointer as a poly_int64.
1661 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
1662 rather than a HOST_WIDE_INT.
1663 (do_remat): Track offsets poly_int64s.
1664 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
1666 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1667 Alan Hayward <alan.hayward@arm.com>
1668 David Sherwood <david.sherwood@arm.com>
1670 * rtl.h (mem_attrs): Add a default constructor. Change size and
1671 offset from HOST_WIDE_INT to poly_int64.
1672 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
1673 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
1674 (widen_memory_access): Take the sizes and offsets as poly_int64s
1675 rather than HOST_WIDE_INTs.
1676 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
1677 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
1679 (adjust_offset_for_component_ref): Change the offset from a
1680 HOST_WIDE_INT to a poly_int64.
1681 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
1682 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
1683 * dce.c (find_call_stack_args): Likewise.
1684 * dse.c (record_store): Likewise.
1685 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
1686 * print-rtl.c (rtx_writer::print_rtx): Likewise.
1687 * read-rtl-function.c (test_loading_mem): Likewise.
1688 * rtlanal.c (may_trap_p_1): Likewise.
1689 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1690 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
1691 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
1692 (mem_attrs::mem_attrs): New function.
1693 (set_mem_attributes_minus_bitpos): Change bitpos from a
1694 HOST_WIDE_INT to poly_int64.
1695 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
1696 (clear_mem_offset, clear_mem_size, change_address)
1697 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
1698 initialize mem_attrs.
1699 (set_mem_offset, set_mem_size, adjust_address_1)
1700 (adjust_automodify_address_1, offset_address, widen_memory_access):
1701 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
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 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
1708 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
1709 rather than 0 to represent an unknown size. Assert that the size
1710 is known when the mode isn't BLKmode.
1711 (may_trap_p_1): Use -1 for unknown sizes.
1712 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
1714 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1715 Alan Hayward <alan.hayward@arm.com>
1716 David Sherwood <david.sherwood@arm.com>
1718 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
1719 to poly_int64. Update commentary for positions_needed.large.
1720 (read_info_type): Change offset and width from HOST_WIDE_INT
1722 (set_usage_bits): Likewise.
1723 (canon_address): Return the offset as a poly_int64 rather than
1724 a HOST_WIDE_INT. Use strip_offset_and_add.
1725 (set_all_positions_unneeded, any_positions_needed_p): Use
1726 positions_needed.large to track stores with non-constant widths.
1727 (all_positions_needed_p): Likewise. Take the offset and width
1728 as poly_int64s rather than ints. Assert that rhs is nonnull.
1729 (record_store): Cope with non-constant offsets and widths.
1730 Nullify the rhs of an earlier store if we can't tell which bytes
1732 (find_shift_sequence): Take the access_size and shift as poly_int64s
1734 (get_stored_val): Take the read_offset and read_width as poly_int64s
1735 rather than HOST_WIDE_INTs.
1736 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
1737 non-constant offsets and widths.
1739 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1740 Alan Hayward <alan.hayward@arm.com>
1741 David Sherwood <david.sherwood@arm.com>
1743 * inchash.h (inchash::hash::add_poly_int): New function.
1744 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
1745 Use poly_int64 rather than HOST_WIDE_INT.
1746 (ao_ref::max_size_known_p): New function.
1747 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
1748 rather than HOST_WIDE_INT.
1749 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
1750 to temporaries until its interface is adjusted to match.
1751 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
1752 (aliasing_component_refs_p, decl_refs_may_alias_p)
1753 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
1754 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
1755 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
1757 * alias.c (ao_ref_from_mem): Likewise.
1758 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1759 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
1760 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
1761 (maybe_trim_complex_store, maybe_trim_constructor_store)
1762 (live_bytes_read, dse_classify_store): Likewise.
1763 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
1764 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
1765 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
1766 (vn_reference_lookup_3): Likewise.
1767 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
1769 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1770 Alan Hayward <alan.hayward@arm.com>
1771 David Sherwood <david.sherwood@arm.com>
1773 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
1774 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
1775 instead of ranges_overlap_p.
1777 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1778 Alan Hayward <alan.hayward@arm.com>
1779 David Sherwood <david.sherwood@arm.com>
1781 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
1782 sizes as poly_int64s rather than HOST_WIDE_INTs.
1784 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1785 Alan Hayward <alan.hayward@arm.com>
1786 David Sherwood <david.sherwood@arm.com>
1788 * gimple-fold.h (fold_ctor_reference): Take the offset and size
1789 as poly_uint64 rather than unsigned HOST_WIDE_INT.
1790 * gimple-fold.c (fold_ctor_reference): Likewise.
1792 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1793 Alan Hayward <alan.hayward@arm.com>
1794 David Sherwood <david.sherwood@arm.com>
1796 * target.def (dwarf_poly_indeterminate_value): New hook.
1797 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
1798 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
1799 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
1800 * doc/tm.texi: Regenerate.
1801 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
1802 offset as a poly_int64.
1803 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
1804 offset as a poly_int64.
1805 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
1806 Take the offset as a poly_int64.
1807 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
1808 (frame_pointer_fb_offset): Change to a poly_int64.
1809 (int_loc_descriptor): Take the offset as a poly_int64. Use
1810 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
1811 (based_loc_descr): Take the offset as a poly_int64.
1812 Use strip_offset_and_add to handle (plus X (const)).
1813 Use new_reg_loc_descr instead of an open-coded version of the
1814 previous implementation.
1815 (mem_loc_descriptor): Handle CONST_POLY_INT.
1816 (compute_frame_pointer_to_fb_displacement): Take the offset as a
1817 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
1819 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1820 Alan Hayward <alan.hayward@arm.com>
1821 David Sherwood <david.sherwood@arm.com>
1823 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
1824 (gen_rtx_REG_offset): Take the offset as a poly_int64.
1825 * inchash.h (inchash::hash::add_poly_hwi): New function.
1826 * gengtype.c (main): Register poly_int64.
1827 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
1828 offset as a poly_int.
1829 (reg_attr_hasher::equal): Use must_eq to compare offsets.
1830 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
1831 offset as a poly_int64.
1832 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
1833 * print-rtl.c (print_poly_int): New function.
1834 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
1836 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
1838 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
1839 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
1840 (vt_get_decl_and_offset): Return the offset as a poly_int64.
1841 Enforce track_offset_p for parts of a PARALLEL.
1842 (vt_add_function_parameter): Use const_offset for the final
1843 offset to track. Use get_tracked_reg_offset for the parts
1846 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1847 Alan Hayward <alan.hayward@arm.com>
1848 David Sherwood <david.sherwood@arm.com>
1850 * target.def (truly_noop_truncation): Take poly_uint64s instead of
1851 unsigned ints. Change default to hook_bool_puint64_puint64_true.
1852 * doc/tm.texi: Regenerate.
1853 * hooks.h (hook_bool_uint_uint_true): Delete.
1854 (hook_bool_puint64_puint64_true): Declare.
1855 * hooks.c (hook_bool_uint_uint_true): Delete.
1856 (hook_bool_puint64_puint64_true): New function.
1857 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
1858 instead of unsigned ints.
1859 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
1860 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
1862 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1863 Alan Hayward <alan.hayward@arm.com>
1864 David Sherwood <david.sherwood@arm.com>
1866 * optabs.h (expand_operand): Add an int_value field.
1867 (create_expand_operand): Add an int_value parameter and use it
1868 to initialize the new expand_operand field.
1869 (create_integer_operand): Replace with a declaration of a function
1870 that accepts poly_int64s. Move the implementation to...
1871 * optabs.c (create_integer_operand): ...here.
1872 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
1873 the mode preserves the value of int_value, instead of calling
1874 const_int_operand on the rtx. Use gen_int_mode to generate
1877 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1878 Alan Hayward <alan.hayward@arm.com>
1879 David Sherwood <david.sherwood@arm.com>
1881 * dumpfile.h (dump_dec): Declare.
1882 * dumpfile.c (dump_dec): New function.
1883 * pretty-print.h (pp_wide_integer): Turn into a function and
1884 declare a poly_int version.
1885 * pretty-print.c (pp_wide_integer): New function for poly_ints.
1887 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1888 Alan Hayward <alan.hayward@arm.com>
1889 David Sherwood <david.sherwood@arm.com>
1891 * doc/generic.texi (POLY_INT_CST): Document.
1892 * tree.def (POLY_INT_CST): New tree code.
1893 * treestruct.def (TS_POLY_INT_CST): New tree layout.
1894 * tree-core.h (tree_poly_int_cst): New struct.
1895 (tree_node): Add a poly_int_cst field.
1896 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
1897 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
1898 instead of a wide_int_ref.
1899 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
1901 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1902 instead of an unsigned HOST_WIDE_INT.
1903 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
1904 (ptrdiff_tree_p): Declare.
1905 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
1906 extern inline implementations if the target doesn't use POLY_INT_CST.
1907 (poly_int_tree_p): New function.
1908 (wi::unextended_tree): New class.
1909 (wi::int_traits <unextended_tree>): New override.
1910 (wi::extended_tree): Add a default constructor.
1911 (wi::extended_tree::get_tree): New function.
1912 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
1913 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
1914 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
1915 (wi::tree_to_poly_wide_ref): New typedefs.
1916 (wi::ints_for): Provide overloads for extended_tree and
1918 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
1919 (wi::to_wide): New functions.
1920 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
1921 * tree.c (poly_int_cst_hasher): New struct.
1922 (poly_int_cst_hash_table): New variable.
1923 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
1924 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
1926 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
1927 (init_ttree): Initialize poly_int_cst_hash_table.
1928 (build_int_cst, build_int_cst_type, build_invariant_address): Take
1929 a poly_int64 instead of a HOST_WIDE_INT.
1930 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
1931 instead of an unsigned HOST_WIDE_INT.
1932 (wide_int_to_tree): Rename to...
1933 (wide_int_to_tree_1): ...this.
1934 (build_new_poly_int_cst, build_poly_int_cst): New functions.
1935 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
1936 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
1937 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
1938 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
1939 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
1941 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
1942 (streamer_read_tree_body): Likewise.
1943 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
1944 (streamer_write_tree_body): Likewise.
1945 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
1946 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
1947 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
1948 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
1949 * gimple-expr.h (is_gimple_constant): Likewise.
1950 * gimplify.c (maybe_with_size_expr): Likewise.
1951 * print-tree.c (print_node): Likewise.
1952 * tree-data-ref.c (data_ref_compare_tree): Likewise.
1953 * tree-pretty-print.c (dump_generic_node): Likewise.
1954 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
1955 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
1956 * tree-vrp.c (compare_values_warnv): Likewise.
1957 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
1958 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
1959 (force_expr_to_var_cost): Likewise.
1960 * tree-ssa-loop.c (for_each_index): Likewise.
1961 * fold-const.h (build_invariant_address, size_int_kind): Take a
1962 poly_int64 instead of a HOST_WIDE_INT.
1963 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
1964 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
1966 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
1967 (int_const_binop_2): New function, split out from...
1968 (int_const_binop_1): ...here. Handle POLY_INT_CST.
1969 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
1970 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
1971 * gimple-ssa-strength-reduction.c (slsr_process_add)
1972 (slsr_process_mul): Check for INTEGER_CSTs before using them
1974 * stor-layout.c (bits_from_bytes): New function.
1975 (bit_from_pos): Use it.
1976 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
1977 by BITS_PER_UNIT to get the TYPE_SIZE.
1978 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
1979 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
1981 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1982 Alan Hayward <alan.hayward@arm.com>
1983 David Sherwood <david.sherwood@arm.com>
1985 * doc/rtl.texi (const_poly_int): Document. Also document the
1986 rtl sharing behavior.
1987 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
1988 * rtl.h (const_poly_int_def): New struct.
1989 (rtx_def::u): Add a cpi field.
1990 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
1991 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
1992 (wi::rtx_to_poly_wide_ref): New typedef
1993 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
1994 (poly_int_rtx_p): New functions.
1995 (trunc_int_for_mode): Declare a poly_int64 version.
1996 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
1997 (immed_wide_int_const): Take a poly_wide_int_ref rather than
1999 (strip_offset): Declare.
2000 (strip_offset_and_add): New function.
2001 * rtl.def (CONST_POLY_INT): New rtx code.
2002 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2003 (shared_const_p): Use poly_int_rtx_p.
2004 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2006 (gen_int_shift_amount): Likewise.
2007 * emit-rtl.c (const_poly_int_hasher): New class.
2008 (const_poly_int_htab): New variable.
2009 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2010 (const_poly_int_hasher::hash): New function.
2011 (const_poly_int_hasher::equal): Likewise.
2012 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2013 (immed_wide_int_const): Rename to...
2014 (immed_wide_int_const_1): ...this and make static.
2015 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2016 instead of a wide_int_ref.
2017 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2018 (gen_lowpart_common): Handle CONST_POLY_INT.
2019 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2020 * cselib.c (cselib_hash_rtx): Likewise.
2021 * dwarf2out.c (const_ok_for_output_1): Likewise.
2022 * expr.c (convert_modes): Likewise.
2023 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2024 * rtlhash.c (add_rtx): Likewise.
2025 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2026 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2027 Handle existing CONST_POLY_INT rtxes.
2028 * expmed.h (expand_shift): Take a poly_int64 instead of a
2030 * expmed.c (expand_shift): Likewise.
2031 * rtlanal.c (strip_offset): New function.
2032 (commutative_operand_precedence): Give CONST_POLY_INT the same
2033 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2035 * rtl-tests.c (const_poly_int_tests): New struct.
2036 (rtl_tests_c_tests): Use it.
2037 * simplify-rtx.c (simplify_const_unary_operation): Handle
2039 (simplify_const_binary_operation): Likewise.
2040 (simplify_binary_operation_1): Fold additions of symbolic constants
2041 and CONST_POLY_INTs.
2042 (simplify_subreg): Handle extensions and truncations of
2044 (simplify_const_poly_int_tests): New struct.
2045 (simplify_rtx_c_tests): Use it.
2046 * wide-int.h (storage_ref): Add default constructor.
2047 (wide_int_ref_storage): Likewise.
2048 (trailing_wide_ints): Use GTY((user)).
2049 (trailing_wide_ints::operator[]): Add a const version.
2050 (trailing_wide_ints::get_precision): New function.
2051 (trailing_wide_ints::extra_size): Likewise.
2053 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2054 Alan Hayward <alan.hayward@arm.com>
2055 David Sherwood <david.sherwood@arm.com>
2057 * emit-rtl.h (gen_int_shift_amount): Declare.
2058 * emit-rtl.c (gen_int_shift_amount): New function.
2059 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2061 * calls.c (shift_return_value): Likewise.
2062 * cse.c (fold_rtx): Likewise.
2063 * dse.c (find_shift_sequence): Likewise.
2064 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2065 (expand_shift, expand_smod_pow2): Likewise.
2066 * lower-subreg.c (shift_cost): Likewise.
2067 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2068 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2069 (expand_vec_perm_var): Likewise.
2070 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2071 (simplify_binary_operation_1): Likewise.
2072 * combine.c (try_combine, find_split_point, force_int_to_mode)
2073 (simplify_shift_const_1, simplify_shift_const): Likewise.
2074 (change_zero_ext): Likewise. Use simplify_gen_binary.
2076 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2078 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2080 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2082 * doc/invoke.texi (ARM Options): Document accepted extension options
2083 for -march=armv8.3-a.
2085 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2088 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2089 or --with-float=softfp, set the default CPU to arm10e.
2091 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2093 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2094 * config/visium/predicates.md (const_shift_operand): Likewise.
2095 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2096 (visium_legitimize_reload_address): Likewise.
2098 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2101 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2103 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2106 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2108 2017-12-20 Julia Koval <julia.koval@intel.com>
2110 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2111 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2112 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2113 * config.gcc: Include vpclmulqdqintrin.h.
2114 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2115 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2117 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2118 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2119 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2120 * config/i386/i386.c (isa2_opts): Add -mcx16.
2121 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2122 (ix86_option_override_internal): Move mcx16 to flags2.
2123 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2124 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2125 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2126 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2127 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2128 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2129 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2130 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2131 * doc/invoke.texi: Add -mvpclmulqdq.
2133 2017-12-20 Tom de Vries <tom@codesourcery.com>
2136 * config/i386/i386.c (ix86_static_chain): Move
2137 DECL_STATIC_CHAIN test ...
2138 * calls.c (rtx_for_static_chain): ... here. New function.
2139 * calls.h (rtx_for_static_chain): Declare.
2140 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2141 instead of targetm.calls.static_chain.
2142 * df-scan.c (df_get_entry_block_def_set): Same.
2144 2017-12-19 Tom de Vries <tom@codesourcery.com>
2146 PR tree-optimization/83493
2147 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2150 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2152 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2153 inform with hardcoded english plural handling.
2155 2017-12-18 Jeff Law <law@redhat.com>
2157 PR tree-optimization/83477
2158 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2159 a non-virtual PHI, always push a new range.
2161 2017-12-19 Martin Sebor <msebor@redhat.com>
2164 * builtins.c (compute_objsize): Handle non-constant offsets.
2166 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2168 PR tree-optimization/83444
2169 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2170 character load case, if get_stridx on MEM_REF's operand doesn't
2171 look usable, retry with get_addr_stridx.
2173 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2176 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2177 (variable_tracking_main_1): Keep markers even when VTA fails.
2180 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2181 even if there are markers before it.
2182 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2184 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2186 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2187 typical order conditions.
2188 * sel-sched.c (extract_new_fences_from): Likewise.
2189 * config/visium/constraints.md (J, K, L): Likewise.
2190 * config/visium/predicates.md (const_shift_operand): Likewise.
2191 * config/visium/visium.c (visium_legitimize_address,
2192 visium_legitimize_reload_address): Likewise.
2193 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2194 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2195 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2196 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2197 SET_DUMP_DETAIL): Likewise.
2198 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2199 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2200 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2201 avr_set_core_architecture, avr_set_current_function,
2202 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2203 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2204 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2205 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2206 avr_map_decompose, avr_fold_builtin): Likewise.
2207 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2208 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2209 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2210 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2211 * config/m32c/m32c.c (m32c_conditional_register_usage,
2212 m32c_address_cost): Likewise.
2213 * config/m32c/predicates.md (shiftcount_operand,
2214 longshiftcount_operand): Likewise.
2215 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2216 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2217 can_use_cdx_ldstw): Likewise.
2218 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2219 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2221 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2222 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2223 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2224 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2225 * config/vax/vax.c (adjacent_operands_p): Likewise.
2226 * config/ft32/constraints.md (L, b, KA): Likewise.
2227 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2229 * cfgexpand.c (expand_stack_alignment): Likewise.
2230 * gcse.c (insert_expr_in_table): Likewise.
2231 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2232 * cgraphunit.c (cgraph_node::expand): Likewise.
2233 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2234 * emit-rtl.c (add_insn): Likewise.
2235 * input.c (dump_location_info): Likewise.
2236 * passes.c (NEXT_PASS): Likewise.
2237 * read-rtl-function.c (parse_note_insn_name,
2238 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2240 * sched-rgn.c (sched_rgn_init): Likewise.
2241 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2242 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2243 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2245 * explow.c (eliminate_constant_term): Likewise.
2246 * final.c (leaf_renumber_regs_insn): Likewise.
2247 * cfgrtl.c (print_rtl_with_bb): Likewise.
2248 * genhooks.c (emit_init_macros): Likewise.
2249 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2250 * tree-data-ref.c (conflict_fn): Likewise.
2251 * selftest.c (assert_streq): Likewise.
2252 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2253 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2254 fold_binary_loc, multiple_of_p): Likewise.
2255 * reload.c (push_reload, find_equiv_reg): Likewise.
2256 * et-forest.c (et_nca, et_below): Likewise.
2257 * dbxout.c (dbxout_symbol_location): Likewise.
2258 * reorg.c (relax_delay_slots): Likewise.
2259 * dojump.c (do_compare_rtx_and_jump): Likewise.
2260 * gengtype-parse.c (type): Likewise.
2261 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2262 simplify_const_relational_operation): Likewise.
2263 * reload1.c (do_output_reload): Likewise.
2264 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2265 * gengtype.c (type_for_name): Likewise.
2266 * gimple-ssa-sprintf.c (format_directive): Likewise.
2268 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2271 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2272 accessing it. Adjust comment.
2274 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2277 * predict.c (zero_one_minusone): New function.
2278 (apply_return_prediction): Avoid return prediction for functions
2279 returning only -1, 0 and 1 values, unless they only return -1 and 0
2282 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2284 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2285 fall-through warning.
2286 (arc_compute_frame_size): Remove unused variables.
2287 (arc_print_operand): Fix fprintif format.
2288 (arc_can_follow_jump): Clean fall-through warning.
2290 2017-12-19 Marek Polacek <polacek@redhat.com>
2293 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2296 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2298 * config/arc/arc.c (overriderregs): New variable.
2299 (arc_override_options): Track fixed/call saved/call options.
2300 (arc_conditional_register_usage): Check against overrideregs
2301 variable whenever we change register properties.
2303 2017-12-19 Nathan Sidwell <nathan@acm.org>
2305 * opts.c (finish_options): Don't prefix dump_base_name if it
2306 already contains directories.
2308 2017-12-19 Martin Liska <mliska@suse.cz>
2310 PR rtl-optimization/82675
2311 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2312 more element in sbitmap.
2314 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2316 * gimplify.c (gimplify_expr): Use error_operand_p.
2318 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2321 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2324 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2326 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2327 the jump is to a label.
2329 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2331 PR tree-optimization/83444
2332 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2337 * ipa-inline.c (flatten_remove_node_hook): New function.
2338 (ipa_inline): Keep only nodes with flatten attribute at the end of
2339 the array in the order from ipa_reverse_postorder, only walk that
2340 portion of array for flattening, if there is more than one such
2341 node, temporarily register a removal hook and ignore removed nodes.
2343 PR tree-optimization/80631
2344 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2345 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2346 IFN_REDUC_MAX or IFN_REDUC_MIN.
2348 2017-12-18 Jeff Law <law@redhat.com>
2350 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2351 record symbolic equivalences from backedges in the CFG.
2354 2017-11-19 Jeff Law <law@redhat.com>
2356 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2357 of degenerates resulting from ignoring an edge.
2359 2017-12-18 Martin Sebor <msebor@redhat.com>
2362 PR tree-optimization/78450
2363 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2364 (handle_builtin_strlen): Call it.
2366 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2368 PR rtl-optimization/83424
2369 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2371 2017-12-18 Marek Polacek <polacek@redhat.com>
2374 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2375 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2378 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2380 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2381 (umaddsidi4, umaddsidi_split): Likewise.
2383 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2385 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2388 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2391 * config/s390/s390.c (s390_option_override): Avoid strncpy.
2393 2017-12-18 Richard Biener <rguenther@suse.de>
2395 PR tree-optimization/81877
2396 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
2397 (outermost_indep_loop): Adjust.
2398 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
2399 (can_sm_ref_p): Adjust.
2401 2017-12-18 Richard Biener <rguenther@suse.de>
2404 * tree.c (array_at_struct_end_p): Return true if the underlying
2405 object has space for at least one element in excess of what
2406 the array domain specifies.
2408 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2410 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
2411 example to avoid overfull hbox.
2412 * doc/invoke.texi (Option Summary): Add missing @gol.
2413 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
2416 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2417 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2419 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
2420 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
2421 (AArch64 Options): Document -mverbose-cost-dump.
2422 (ARM Options): Likewise, plus -mflip-thumb.
2424 2017-12-17 Martin Sebor <msebor@redhat.com>
2427 * gimple-ssa-warn-restrict.c
2428 (builtin_memref::offset_out_of_bounds): Correct the handling of
2431 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
2433 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
2435 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
2437 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
2439 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
2441 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
2444 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
2446 * doc/invoke.texi: Fix some typos.
2448 2017-12-16 Martin Sebor <msebor@redhat.com>
2450 PR tree-optimization/78918
2451 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
2452 * builtins.c (check_sizes): Rename...
2453 (check_access): ...to this. Rename function arguments for clarity.
2454 (check_memop_sizes): Adjust names.
2455 (expand_builtin_memchr, expand_builtin_memcpy): Same.
2456 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
2457 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
2458 (check_strncat_sizes, expand_builtin_strncat): Same.
2459 (expand_builtin_strncpy, expand_builtin_memset): Same.
2460 (expand_builtin_bzero, expand_builtin_memcmp): Same.
2461 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
2462 (maybe_emit_sprintf_chk_warning): Same.
2463 (expand_builtin_strcpy): Adjust.
2464 (expand_builtin_stpcpy): Same.
2465 (expand_builtin_with_bounds): Detect out-of-bounds accesses
2466 in pointer-checking forms of memcpy, memmove, and mempcpy.
2467 (gcall_to_tree_minimal, max_object_size): Define new functions.
2468 * builtins.h (max_object_size): Declare.
2469 * calls.c (alloc_max_size): Call max_object_size instead of
2470 hardcoding ssizetype limit.
2471 (get_size_range): Handle new argument.
2472 * calls.h (get_size_range): Add a new argument.
2473 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
2474 * doc/invoke.texi (-Wrestrict): Adjust, add example.
2475 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
2477 (gimple_fold_builtin_memory_chk): Same.
2478 (gimple_fold_builtin_stxcpy_chk): New function.
2479 * gimple-ssa-warn-restrict.c: New source.
2480 * gimple-ssa-warn-restrict.h: New header.
2481 * gimple.c (gimple_build_call_from_tree): Propagate location.
2482 * passes.def (pass_warn_restrict): Add new pass.
2483 * tree-pass.h (make_pass_warn_restrict): Declare.
2484 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
2486 (handle_builtin_strcat): Same.
2487 (strlen_optimize_stmt): Rename...
2488 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
2489 stpncpy, strncpy, and their checking forms.
2491 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
2493 PR rtl-optimization/82849
2494 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
2495 and get_max_loop_iterations_int.
2497 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2498 Alan Hayward <alan.hayward@arm.com>
2499 David Sherwood <david.sherwood@arm.com>
2501 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
2502 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
2503 size as a poly_uint64.
2504 (mode_for_vector, mode_for_int_vector): Take the number of vector
2505 elements as a poly_uint64.
2506 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
2507 size as a poly_uint64.
2508 (mode_for_vector, mode_for_int_vector): Take the number of vector
2509 elements as a poly_uint64.
2511 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2512 Alan Hayward <alan.hayward@arm.com>
2513 David Sherwood <david.sherwood@arm.com>
2515 * machmode.h (MACRO_MODE): New macro.
2516 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
2517 * targhooks.c (default_libcall_value, default_secondary_reload)
2518 (default_memory_move_cost, default_register_move_cost)
2519 (default_class_max_nregs): Likewise.
2521 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2522 Alan Hayward <alan.hayward@arm.com>
2523 David Sherwood <david.sherwood@arm.com>
2525 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
2526 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
2528 * genautomata.c (main): Likewise.
2529 * genconditions.c (write_header): Likewise.
2530 * genemit.c (main): Likewise.
2531 * genextract.c (print_header): Likewise.
2532 * genopinit.c (main): Likewise.
2533 * genoutput.c (output_prologue): Likewise.
2534 * genpeep.c (main): Likewise.
2535 * genpreds.c (write_insn_preds_c): Likewise.
2536 * genrecog.c (writer_header): Likewise.
2537 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
2538 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
2539 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
2540 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
2541 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
2542 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
2543 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
2544 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
2545 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
2546 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
2547 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
2548 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
2549 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
2550 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
2551 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
2552 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
2553 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
2554 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
2555 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
2556 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
2557 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
2558 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
2559 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
2560 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
2561 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
2562 * config/darwin.c (IN_TARGET_CODE): Likewise.
2563 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
2564 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
2565 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
2566 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
2567 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
2568 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
2569 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
2570 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
2571 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
2572 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
2573 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
2574 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
2575 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
2576 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
2577 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
2578 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
2579 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
2580 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
2581 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
2582 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
2583 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
2584 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
2585 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
2586 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
2587 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
2588 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
2589 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
2590 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
2591 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
2592 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
2593 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
2594 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
2595 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
2596 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
2597 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
2598 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
2599 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
2600 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
2601 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
2602 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
2603 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
2604 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
2605 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
2606 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
2607 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
2608 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
2609 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
2610 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
2611 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
2612 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
2613 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
2614 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
2615 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
2616 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
2617 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
2618 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
2619 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
2620 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
2621 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
2622 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
2623 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2624 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
2625 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
2626 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
2627 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
2628 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
2629 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
2630 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
2631 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
2632 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
2633 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
2634 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
2635 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
2636 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
2637 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
2638 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
2639 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
2640 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
2641 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
2642 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
2643 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
2644 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
2645 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
2646 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
2647 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
2648 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
2649 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
2650 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
2651 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
2652 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
2653 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
2654 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
2655 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
2656 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
2657 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
2658 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
2659 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
2660 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
2661 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
2662 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
2663 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
2664 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
2665 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
2666 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
2667 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
2668 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
2670 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2671 Alan Hayward <alan.hayward@arm.com>
2672 David Sherwood <david.sherwood@arm.com>
2674 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
2677 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2678 Alan Hayward <alan.hayward@arm.com>
2679 David Sherwood <david.sherwood@arm.com>
2681 * doc/generic.texi (VEC_SERIES_EXPR): Document.
2682 * doc/md.texi (vec_series@var{m}): Document.
2683 * tree.def (VEC_SERIES_EXPR): New tree code.
2684 * tree.h (build_vec_series): Declare.
2685 * tree.c (build_vec_series): New function.
2686 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
2687 * tree-pretty-print.c (dump_generic_node): Likewise.
2688 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
2689 * tree-inline.c (estimate_operator_cost): Likewise.
2690 * expr.c (expand_expr_real_2): Likewise.
2691 * optabs-tree.c (optab_for_tree_code): Likewise.
2692 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
2693 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
2694 * expmed.c (make_tree): Handle VEC_SERIES.
2695 * optabs.def (vec_series_optab): New optab.
2696 * optabs.h (expand_vec_series_expr): Declare.
2697 * optabs.c (expand_vec_series_expr): New function.
2698 * tree-vect-generic.c (expand_vector_operations_1): Check that
2699 the operands also have vector type.
2701 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
2702 Alan Hayward <alan.hawyard@arm.com>
2703 David Sherwood <david.sherwood@arm.com>
2705 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
2706 (VEC_COND_EXPR): Add missing @tindex.
2707 * doc/md.texi (vec_duplicate@var{m}): Document.
2708 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
2709 * tree.c (build_vector_from_val): Add stubbed-out handling of
2710 variable-length vectors, using VEC_DUPLICATE_EXPR.
2711 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
2712 * cfgexpand.c (expand_debug_expr): Likewise.
2713 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
2714 * tree-inline.c (estimate_operator_cost): Likewise.
2715 * tree-pretty-print.c (dump_generic_node): Likewise.
2716 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
2717 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
2718 (test_vec_duplicate_folding): New function.
2719 (fold_const_c_tests): Call it.
2720 * optabs.def (vec_duplicate_optab): New optab.
2721 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
2722 * optabs.h (expand_vector_broadcast): Declare.
2723 * optabs.c (expand_vector_broadcast): Make non-static. Try using
2724 vec_duplicate_optab.
2725 * expr.c (store_constructor): Try using vec_duplicate_optab for
2727 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
2729 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
2732 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
2733 div/mod latencies a bit.
2735 2017-12-15 Jeff Law <law@redhat.com>
2737 PR tree-optimization/36550
2738 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
2739 (mark_threaded_blocks): Rewrite code to avoid block copying when
2740 optimizing for size. Don't pessimize blocks which will be
2741 copied, but all the statements will be dead.
2743 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
2745 PR tree-optimization/81165
2746 * tree-ssa-threadupdate.c (uses_in_bb): New.
2747 (estimate_threading_killed_stmts): New.
2748 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
2749 * tree-ssa-threadedge.c
2750 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
2753 2017-12-15 Jeff Law <law@redhat.com>
2755 PR tree-optimization/83410
2756 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
2757 threads when parallelizing loops.
2759 2017-12-15 Jakub Jelinek <jakub@redhat.com>
2761 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
2763 * config/alpha/alpha.c (vms_attribute_table): Swap
2764 affects_type_identity and handler fields, adjust comments.
2765 * config/mips/mips.c (mips_attribute_table): Likewise.
2766 * config/visium/visium.c (visium_attribute_table): Likewise.
2767 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
2768 * config/microblaze/microblaze.c (microblaze_attribute_table):
2770 * config/spu/spu.c (spu_attribute_table): Likewise.
2771 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
2772 * config/arc/arc.c (arc_attribute_table): Likewise.
2773 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
2774 * config/v850/v850.c (v850_handle_interrupt_attribute,
2775 v850_handle_data_area_attribute): Formatting fixes.
2776 (v850_attribute_table): Swap affects_type_identity and handler
2777 fields, adjust comments.
2778 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
2779 * config/arm/arm.c (arm_attribute_table): Likewise.
2780 * config/avr/avr.c (avr_attribute_table): Likewise.
2781 * config/s390/s390.c (s390_attribute_table): Likewise.
2782 * config/sh/sh.c (sh_attribute_table): Likewise.
2783 * config/i386/i386.c (ix86_handle_cconv_attribute,
2784 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
2785 (ix86_attribute_table): Swap affects_type_identity and handler
2786 fields, adjust comments.
2787 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2788 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
2789 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
2790 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
2791 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
2792 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2793 * config/rx/rx.c (rx_attribute_table): Likewise.
2794 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
2795 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
2796 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
2797 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
2798 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2799 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
2800 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
2801 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
2802 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2803 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2804 * doc/plugins.texi (user_attr): Likewise. Add NULL for
2806 * attribs.c (empty_attribute_table): Swap affects_type_identity and
2808 (register_scoped_attributes, decl_attributes): Formatting fixes.
2810 PR tree-optimization/83269
2811 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
2812 subtraction in arg0's type if type is signed and arg0 is unsigned.
2816 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
2817 convert? on @0 instead of convert. Check type of @1, not @0.
2818 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
2819 convert? on @0 instead of convert. Check type of @1, not @0.
2820 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
2821 only optimize if either both @1 and @2 types are narrower
2822 precision, or both are wider or equal precision, and in the former
2823 case only if both have undefined overflow.
2825 2017-12-15 Richard Biener <rguenther@suse.de>
2828 * internal-fn.def (IFN_NOP): Add.
2829 * internal-fn.c (expand_NOP): Do nothing.
2830 * lto-streamer-in.c (input_function): Instead of removing
2831 sanitizer calls replace them with IFN_NOP calls.
2833 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
2834 Alan Hayward <alan.hayward@arm.com>
2835 David Sherwood <david.sherwood@arm.com>
2837 * dse.c (store_info, read_info_type): Replace begin and end with
2839 (print_range): New function.
2840 (set_all_positions_unneeded, any_positions_needed_p)
2841 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
2843 (record_store): Likewise. Optimize the case in which all positions
2845 (get_stored_val): Replace read_begin and read_end with read_offset
2847 (replace_read): Update call accordingly.
2849 2017-12-15 Bin Cheng <bin.cheng@arm.com>
2851 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
2852 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
2853 (loop_cand::loop_cand): Initialize above members.
2854 (loop_cand::supported_operations): Delete.
2855 (loop_cand::can_interchange_p): Inline above function.
2856 (loop_cand::classify_simple_reduction): Record number of constant
2857 initialized simple reductions.
2858 (should_interchange_loops): New parameters. Check stmt cost of loops
2860 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
2861 Update call to should_interchange_loops.
2862 (should_interchange_loop_nest): Update call to
2863 should_interchange_loops.
2865 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
2868 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
2869 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
2870 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
2872 2017-12-15 Julia Koval <julia.koval@intel.com>
2874 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
2875 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
2877 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
2878 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
2879 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
2881 2017-12-15 Julia Koval <julia.koval@intel.com>
2883 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
2884 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
2885 * config/i386/sse.md (vaesenc_<mode>): New pattern.
2886 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
2887 _mm_aesenc_epi128): New intrinsics.
2889 2017-12-15 Julia Koval <julia.koval@intel.com>
2891 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
2892 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
2894 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
2895 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
2896 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
2898 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2900 * gimple-ssa-strength-reduction.c (analyze_increments):
2901 Distinguish replacement costs for constant strides from those for
2904 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2906 * var-tracking.c (variable_tracking_main_1): Formatting fix.
2908 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
2910 * doc/invoke.texi: Document -Wcast-function-type.
2911 * recog.h (stored_funcptr): Change signature.
2912 * tree-dump.c (dump_node): Avoid warning.
2913 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
2915 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
2918 * gimple-fold.c (get_range_strlen): Add the handling of non-member
2921 2017-12-14 David Malcolm <dmalcolm@redhat.com>
2923 PR tree-optimization/83312
2924 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
2925 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
2926 "val" param, and to cope with arbitrary basic blocks.
2927 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
2928 handle NULL_TREE for "val", dropping "bb" param.
2929 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
2930 drop "bb" param. Handle NULL_TREE for "val".
2931 (find_case_label_for_value): Make "switch_stmt" param const.
2932 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
2934 (vrp_prop::check_all_array_refs): Reimplement as...
2935 (check_array_bounds_dom_walker::before_dom_children): ...this new
2936 vfunc. Replace linear search through BB block list, excluding
2937 those with non-executable in-edges via dominator walk.
2939 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2941 * config/arm/arm.opt (mverbose-cost-dump): New option.
2942 * config/arm/arm.c (arm_rtx_costs): Use it.
2944 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
2947 * reload1.c (emit_input_reload_insns): Skip debug markers.
2949 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
2951 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
2955 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
2956 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
2957 * config/sh/sh.c: Skip debug insns besides notes.
2958 * config/sh/sh.md: Likewise.
2959 * config/sh/sh_treg_combine.cc: Likewise.
2960 * config/sh/sync.md: Likewise.
2962 2017-12-14 Tom de Vries <tom@codesourcery.com>
2964 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2965 Add item for weak_undefined.
2967 2017-12-14 Richard Biener <rguenther@suse.de>
2969 PR tree-optimization/67842
2970 * sese.h (bb_in_region): Remove #if 0'ed code.
2972 2017-12-14 Richard Biener <rguenther@suse.de>
2974 PR tree-optimization/83326
2975 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2976 may_be_zero parameter and handle it by not marking the first
2977 peeled copy as not exiting the loop.
2978 (try_peel_loop): Likewise.
2979 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
2980 to handle the case of constant or zero iterations and perform
2981 loop header copying on-the-fly.
2983 2017-12-14 Richard Biener <rguenther@suse.de>
2985 PR tree-optimization/83418
2986 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
2987 Instead of asserting we don't get unfolded comparisons deal with them.
2989 2017-12-14 Jakub Jelinek <jakub@redhat.com>
2992 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
2993 basic blocks. Assert debug bind insns don't appear outside of bbs,
2994 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
2997 PR tree-optimization/83198
2998 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
2999 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3000 value if arg is a REAL_CST with incompatible type.
3002 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3003 Bin Cheng <bin.cheng@arm.com>
3006 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3008 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3011 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3013 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3014 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3016 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3017 Alan Hayward <alan.hayward@arm.com>
3018 David Sherwood <david.sherwood@arm.com>
3020 * poly-int.h: New file.
3021 * poly-int-types.h: Likewise.
3022 * coretypes.h: Include them.
3023 (POLY_INT_CONVERSION): Define.
3024 * target.def (estimated_poly_value): New hook.
3025 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3026 * doc/tm.texi: Regenerate.
3027 * doc/poly-int.texi: New file.
3028 * doc/gccint.texi: Include it.
3029 * doc/rtl.texi: Describe restrictions on subreg modes.
3030 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3031 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3032 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3033 * targhooks.h (default_estimated_poly_value): Declare.
3034 * targhooks.c (default_estimated_poly_value): New function.
3035 * target.h (estimated_poly_value): Likewise.
3036 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3037 (wi::unary_traits): Delete.
3038 (wi::binary_traits::signed_shift_result_type): Define for
3039 offset_int << HOST_WIDE_INT, etc.
3040 (generic_wide_int::operator <<=): Define for all types and use
3041 wi::lshift instead of <<.
3042 (wi::hwi_with_prec): Add a default constructor.
3043 (wi::ints_for): New class.
3044 (operator <<): Define for all wide-int types.
3045 (operator /): New function.
3046 (operator %): Likewise.
3047 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3048 (ASSERT_MAYBE_NE_AT): New macros.
3050 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3051 Dominik Vogt <vogt@linux.vnet.ibm.com>
3054 * emit-rtl.c (init_emit): Remove ??? comment.
3055 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3056 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3057 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3058 alignment of 3 virtual registers to BITS_PER_WORD.
3060 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3062 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3064 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3065 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3067 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3070 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3072 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3074 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3075 with the exception of debug begin stmt markers appear before
3079 * final.c (rest_of_handle_final): Call variable_tracking_main only
3080 if !flag_var_tracking.
3082 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3083 Jakub Jelinek <jakub@redhat.com>
3087 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3088 labels debug stmts that can only appear after labels.
3090 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3092 PR rtl-optimization/82398
3093 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3094 EXPR_USEFULNESS in priority comparison.
3096 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3098 PR rtl-optimization/83393
3099 * combine.c (move_deaths): If reg_stat points to a too new insn in
3100 last_death, do not use it: find the proper insn instead.
3102 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3105 * config/rs6000/rs6000-string.c (expand_block_compare,
3106 expand_strn_compare): Fix set_mem_size() calls.
3108 2017-12-12 Jeff Law <law@redhat.com>
3110 PR tree-optimization/83298
3111 PR tree-optimization/83362
3112 PR tree-optimization/83383
3113 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3114 push_value_range a public interface. Add new argument to
3115 record_ranges_from_stmt.
3116 * gimple-ssa-evrp-analyze.c
3117 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3118 Update comments. Handle recording temporary equivalences.
3119 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3120 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3121 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3122 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3123 gimple-ssa-evrp-analyze.h.
3124 (record_temporary_equivalences_from_phis): Add new argument. When
3125 the PHI arg is an SSA_NAME, set the result's range to the range
3127 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3128 from statements too.
3129 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3130 Pass it down to children as needed.
3131 (thread_outgoing_edges): Likewise.
3132 (thread_across_edge): Likewise. Push/pop range state as needed.
3133 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3135 2017-12-12 Julia Koval <julia.koval@intel.com>
3137 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3138 (PTA_CANNONLAKE): Remove PTA_CLWB.
3140 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3143 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3144 (vcondv2div2df): Likewise.
3145 (vconduv2dfv2di): Likewise.
3147 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3149 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3150 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3151 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3152 ATTR_MATHFN_FPROUNDING_ERRNO.
3154 2017-12-12 Richard Biener <rguenther@suse.de>
3156 PR tree-optimization/83385
3157 * graphite-scop-detection.c (get_order, order): Remove.
3158 (bb_to_rpo): New global.
3160 (build_scops): Sort pbbs in RPO order.
3162 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3164 * combine.c (simplify_set): Do not transform subregs to zero_extends
3165 if the destination mode is a vector mode.
3167 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3169 PR tree-optimization/83359
3170 * tree-cfg.h (fold_loop_internal_call): Declare.
3171 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3172 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3173 (find_loop_dist_alias): New function.
3174 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3175 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3178 PR tree-optimization/80631
3179 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3180 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3181 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3182 hardcoding zero as the value if COND_EXPR is never true. For
3183 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3184 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3185 hardcoding MAX_EXPR as the reduction operation.
3186 (is_nonwrapping_integer_induction): Allow negative step.
3187 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3188 vect_create_epilog_for_reduction, if no value is suitable, don't
3189 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3191 2017-12-12 Richard Biener <rguenther@suse.de>
3193 PR tree-optimization/81889
3194 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3195 range info from the non-wrapping IV instead of just the range
3198 2017-12-12 Julia Koval <julia.koval@intel.com>
3200 * config.gcc: Add vaesintrin.h.
3201 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3203 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3204 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3206 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3207 * config/i386/immintrin.h: Include vaesintrin.h.
3208 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3209 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3210 _mm_aesdec_epi128): New intrinsics.
3212 2017-12-12 Julia Koval <julia.koval@intel.com>
3214 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3215 OPTION_MASK_ISA_VAES_UNSET): New.
3216 (ix86_handle_option): Handle -mvaes.
3217 * config/i386/cpuid.h: Define bit_VAES.
3218 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3219 * config/i386/i386-c.c (__VAES__): New.
3220 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3221 (ix86_valid_target_attribute_inner_p): Ditto.
3222 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3223 * config/i386/i386.opt: Add -mvaes.
3224 * doc/invoke.texi: Ditto.
3226 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3228 * debug.h (gcc_debug_hooks): Add inline_entry.
3229 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3230 * debug.c (do_nothing_debug_hooks): Likewise.
3231 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3232 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3233 (dwarf2_lineno_debug_hooks): Likewise.
3235 * common.opt (gstatement-frontiers): New, setting
3236 debug_nonbind_markers_p.
3237 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3238 * toplev.c (process_options): Autodetect value for debug statement
3240 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3241 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3243 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3244 markers. Integrate source bind into debug stmt expand loop.
3245 (pass_expand::execute): Check debug marker limit. Avoid deep
3246 TER and expand debug locations for debug bind insns only.
3247 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3249 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3250 * final.c (reemit_insn_block_notes): Take current block from
3251 nonbind markers. Declare note where it's first set.
3252 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3253 begin stmt markers if enabled.
3254 (notice_source_line): Handle nonbind markers. Fail if their
3255 location is unknown or that of builtins.
3256 (rest_of_handle_final): Convert begin stmt markers to notes if
3257 var-tracking didn't run.
3258 (rest_of_clean_state): Skip begin stmt markers.
3259 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3261 * function.c (allocate_struct_function): Set begin_stmt_markers.
3262 * function.h (struct function): Add debug_marker_count counter
3263 and debug_nonbind_markers flag.
3264 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3265 * gimple-low.c (lower_function_body): Adjust
3266 debug_nonbind_markers.
3267 (lower_stmt): Drop or skip gimple debug stmts.
3268 (lower_try_catch): Skip debug stmts.
3269 * gimple.c (gimple_build_debug_begin_stmt): New.
3270 (gimple_copy): Increment debug_marker_count if copying one.
3271 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3272 * gimplify.c (rexpr_location): New.
3273 (rexpr_has_location): New.
3274 (warn_switch_unreachable_r): Handle gimple debug stmts.
3275 (shortcut_cond_r): Call expr_location.
3277 (find_goto_label): New.
3278 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3280 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3282 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3283 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3284 (LANG_HOOKS_INITIALIZER): ... this.
3285 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3286 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3288 (update_ebb_live_info): Skip debug insn markers.
3289 * lra.c (debug_insn_static_data): Rename to...
3290 (debug_bind_static_data): ... this.
3291 (debug_marker_static_data): New.
3292 (lra_set_insn_recog_data): Select one of the above depending
3294 (lra_update_isn_regno_info): Don't assume debug insns have
3296 (push_insns): Skip debug insns.
3297 * lto-streamer-in.c (input_function): Drop debug stmts
3298 depending on active options. Adjust debug_nonbind_markers.
3299 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3300 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3301 begin stmt marker notes.
3302 (print_insn): Likewise.
3303 * recog.c (extract_insn): Recognize rtl for debug markers.
3304 * rtl.def (DEBUG_MARKER): New.
3305 * tree-inline.c: Include params.h.
3306 (remap_gimple_stmt): Handle nonbind markers.
3307 (maybe_move_debug_stmts_to_successors): Likewise.
3308 (copy_debug_stmt): Likewise.
3309 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3310 markers regardless of no side effects.
3311 (tsi_link_before): Don't update container's side effects when adding
3312 a begin stmt marker.
3313 (tsi_link_after): Likewise.
3314 (expr_first): Skip begin stmt markers.
3315 (expr_last): Likewise.
3316 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3317 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3318 Disregard nonbind markers.
3319 * tree.c (make_node_stat): Don't set side effects for begin stmt
3321 (build1_stat): Likewise.
3322 * tree.def (DEBUG_BEGIN_STMT): New.
3323 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3324 * var-tracking.c (delete_debug_insns): Renamed to...
3325 (delete_vta_debug_insns): ... this.
3326 (reemit_marker_as_note): New.
3327 (vt_initialize): Reemit markers.
3328 (delete_vta_debug_insns): Likewise.
3329 (vt_debug_insns_local): Reemit or delete markers.
3330 (variable_tracking_main_1): Likewise.
3331 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3332 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3333 (gimple_debug_nonbind_marker_p): New.
3334 (gimple_build_debug_bind): Adjust.
3335 (gimple_build_debug_begin_stmt): New.
3336 * doc/invoke.texi (max-debug-marker-count): New param.
3337 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3338 (debug_parameter_ref, debug_marker): New.
3339 (NOTE_INSN_BEGIN_STMT): New.
3340 (DEBUG_INSN): Describe begin stmt markers.
3342 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3343 without debug insns, we wouldn't, but clean up debug insns
3344 after a control flow insn nevertheless.
3346 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3347 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3348 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3349 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3350 (rtl_tidy_fallthru_edge): Likewise.
3351 (rtl_verify_fallthru): Likewise.
3352 (rtl_verify_bb_layout): Likewise.
3353 (skip_insns_after_block): Likewise.
3354 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3355 * dwarf2out.c: Include print-rtl.h.
3356 (dwarf2out_next_real_insn): New.
3357 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3358 Dump debug binds in asm comments.
3359 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3360 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3361 callers to use gsi_start_nondebug_bb instead.
3362 (gsi_after_labels): Skip gimple debug stmts.
3363 (gsi_start_nondebug): New.
3364 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3365 (proper_loop_form_for_interchange): Adjust.
3366 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3367 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3368 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3369 (collect_fallthrough_labels): Likewise.
3370 (should_warn_for_implicit_fallthrough): Likewise.
3371 (warn_implicit_fallthrough_r): Likewise.
3372 (expand_FALLTHROUGH_r): Likewise.
3373 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3374 (graphite_copy_stmts_from_block): Skip nonbind markers.
3375 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3376 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3377 * jump.c (clean_barriers): Skip debug insns.
3378 * omp-expand.c (expand_parallel_call): Skip debug insns.
3379 (expand_task_call): Likewise.
3380 (remove_exit_barrier): Likewise.
3381 (expand_omp_taskreg): Likewise.
3382 (expand_omp_for_init_counts): Likewise.
3383 (expand_omp_for_generic): Likewise.
3384 (expand_omp_for_static_nochunk): Likewise.
3385 (expand_omp_for_static_chunk): Likewise.
3386 (expand_omp_simd): Likewise.
3387 (expand_omp_taskloop_for_outer): Likewise.
3388 (expand_omp_taskloop_for_inner): Likewise.
3389 (expand_oacc_for): Likewise.
3390 (expand_omp_sections): Likewise.
3391 (expand_omp_single): Likewise.
3392 (expand_omp_synch): Likewise.
3393 (expand_omp_atomic_load): Likewise.
3394 (expand_omp_atomic_store): Likewise.
3395 (expand_omp_atomic_fetch_op): Likewise.
3396 (expand_omp_atomic_pipeline): Likewise.
3397 (expand_omp_atomic_mutex): Likewise.
3398 (expand_omp_target): Likewise.
3399 (grid_expand_omp_for_loop): Likewise.
3400 (grid_expand_target_grid_body): Likewise.
3401 (build_omp_regions_1): Likewise.
3402 * omp-low.c (check_combined_parallel): Skip debug stmts.
3403 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
3404 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
3405 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
3407 * tree-cfg.c (make_blobs_1): Skip debug stmts.
3408 (make_edges): Likewise.
3409 (cleanup_dead_labels): Likewise.
3410 (gimple_can_merge_blocks_p): Likewise.
3411 (stmt_starts_bb_p): Likewise.
3412 (gimple_block_label): Likewise.
3413 (gimple_redirect_edge_and_branch): Likewise.
3414 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
3416 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
3418 * tree-pretty-print (print_declaration): Omit initializer in slim
3420 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
3422 (eliminate_unnecessary_stmts): Stabilize block removal order.
3423 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
3424 * var-tracking.c (get_first_insn): New.
3425 (vt_emit_notes): Call it.
3426 (vt_initialize): Walk any insns before the first BB.
3427 (delete_debug_insns): Likewise.
3429 * gimple.h (enum gimple_debug_subcode): Add
3430 GIMPLE_DEBUG_BEGIN_STMT.
3431 (gimple_debug_begin_stmt_p): New.
3432 (gimple_debug_nonbind_marker_p): New.
3433 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
3434 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
3435 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
3436 * insn-notes.def (BEGIN_STMT): New.
3437 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
3438 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
3439 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
3440 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
3441 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
3442 (INSN_DEBUG_MARKER_KIND): New.
3443 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
3444 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
3445 (INSN_VAR_LOCATION_PTR): New.
3446 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
3448 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
3449 deps only if debug bind insns are enabled.
3450 (pass_expand::execute): Avoid deep TER and expand
3451 debug locations for debug bind insns only.
3452 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
3453 debug stmts special handling down to debug bind stmts.
3454 * combine.c (try_combine): Narrow debug insns special handling
3455 down to debug bind insns.
3456 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
3457 Narrow debug insns preexisting special handling down to debug
3459 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
3460 handling down to debug bind insns.
3461 * function.c (instantiate_virtual_regs): Skip debug markers,
3462 adjust handling of debug binds.
3463 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
3464 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
3465 * haifa-sched.c (schedule_insn): Narrow special handling of debug
3466 insns to debug bind insns.
3467 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
3468 special handling of debug stmts to debug bind stmts.
3469 * ipa-split.c (split_function): Likewise.
3470 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
3471 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
3473 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
3474 * regrename.c (build_def_use): Likewise.
3475 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
3476 (pass_cprop_hardreg): Narrow special casing of debug insns to
3478 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
3479 * reload1.c (reload): Likewise.
3480 * sese.c (sese_insert_phis_for_liveouts): Narrow special
3481 casing of debug stmts to debug bind stmts.
3482 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
3483 * ssa-iterators.h (num_imm_uses): Likewise.
3484 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
3485 debug stmts to debug bind stmts.
3486 * tree-inline.c (tree_function_versioning): Narrow special casing
3487 of debug stmts to debug bind stmts.
3488 * tree-loop-distribution.c (generate_loops_for_partition):
3489 Narrow special casing of debug stmts to debug bind stmts.
3490 * tree-sra.c (analyze_access_subtree): Narrow special casing
3491 of debug stmts to debug bind stmts.
3492 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
3493 stmts to debug bind stmts.
3494 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
3495 casing of debug stmts to debug bind stmts.
3496 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
3497 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
3498 casing of debug stmts to debug bind stmts.
3499 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3501 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
3502 debug stmts to debug bind stmts.
3503 (gimple_replace_ssa_lhs): Likewise.
3504 (insert_debug_temp_for_var_def): Likewise.
3505 (insert_debug_temps_for_defs): Likewise.
3506 (reset_debug_uses): Likewise.
3507 * tree-ssanames.c (release_ssa_name_fn): Likewise.
3508 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
3509 (adjust_debug_stmts): Likewise.
3510 (adjust_phi_and_debug_stmts): Likewise.
3511 (vect_do_peeling): Likewise.
3512 * tree-vect-loop.c (vect_transform_loop): Likewise.
3513 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
3514 * var-tracking.c (adjust_mems): Narrow special casing of debug
3515 insns to debug bind insns.
3516 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
3517 (compute_bb_dataflow, vt_find_locations): Likewise.
3518 (vt_expand_loc, emit_notes_for_changes): Likewise.
3519 (vt_init_cfa_base): Likewise.
3520 (vt_emit_notes): Likewise.
3521 (vt_initialize): Likewise.
3522 (vt_finalize): Likewise.
3524 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
3525 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
3526 (next_nonnote_nondebug_insn_bb): New.
3527 (prev_nonnote_nondebug_insn_bb): New.
3528 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
3529 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
3530 (prev_nonnote_nondebug_insn_bb): Declare.
3531 (next_nonnote_nondebug_insn_bb): Declare.
3532 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
3533 * cfgrtl.c (get_last_bb_insn): Likewise.
3534 * lra.c (push_insns): Likewise.
3536 2017-12-11 David Malcolm <dmalcolm@redhat.com>
3539 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
3541 * selftest.c (selftest::test_runner::test_runner): New ctor.
3542 (selftest::test_runner::~test_runner): New dtor.
3543 * selftest.h (class selftest::test_runner): New class.
3545 2017-12-11 Carl Love <cel@us.ibm.com>
3547 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
3548 vec_extract_fp32_from_shortl]): Add #defines.
3549 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
3550 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
3551 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
3552 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
3553 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
3554 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
3555 * doc/extend.texi: Add documentation for the added builtins.
3557 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
3559 PR rtl-optimization/80693
3560 PR rtl-optimization/81019
3561 PR rtl-optimization/81020
3562 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
3563 are not mentioned in i3. Place the REG_UNUSED note on i2,
3564 possibly modified to REG_DEAD, if it did not originate in i3.
3566 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3568 * recog.c (store_data_bypass_p_1): New function.
3569 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
3570 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
3572 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
3574 PR rtl-optimization/83361
3575 * ifcvt.c (if_convert): Call fixup_partitions.
3577 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3579 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3580 early folding of splat_u{8,16,32}.
3582 2017-12-11 Jakub Jelinek <jakub@redhat.com>
3584 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
3585 output_operand_lossage first argument with capital letter.
3586 (aarch64_override_options): Don't start error and sorry first argument
3587 with capital letter.
3589 2017-12-11 Andi Kleen <ak@linux.intel.com>
3591 PR gcov-profile/83355
3592 * auto-profile.c (string_table::get_index_by_decl): Don't
3593 recurse when abstract origin points to itself.
3595 2017-12-11 Bin Cheng <bin.cheng@arm.com>
3597 PR tree-optimization/83320
3598 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
3599 (prune_datarefs_not_in_loop): Ditto.
3601 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
3603 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
3605 2017-12-10 Jakub Jelinek <jakub@redhat.com>
3607 PR tree-optimization/83337
3608 * gimple-loop-interchange.cc (compute_access_stride): Handle
3609 bitfield DRs properly.
3611 2017-12-09 Jakub Jelinek <jakub@redhat.com>
3613 PR tree-optimization/83338
3614 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
3615 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
3616 vector of unsigned integers to vector of signed integers.
3618 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
3620 PR rtl-optimization/83317
3621 * lra-constraints.c (process_address_1): Add insn code check.
3623 2017-12-08 Michael Matz <matz@suse.de>
3625 Fix PR tree-optimization/83323
3626 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
3627 head-controlled loops and loop BBs.
3628 * common.opt (funroll-and-jam): Remove, instead ...
3629 (floop-unroll-and-jam): ... reuse this option.
3630 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
3631 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
3632 (-floop-unroll-and-jam): ... this option.
3634 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3636 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
3637 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
3638 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
3640 2017-12-08 Richard Biener <rguenther@suse.de>
3643 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
3644 handle accesses outside of zero-sized vars.
3646 2017-12-08 Martin Jambor <mjambor@suse.cz>
3648 PR tree-optimization/83141
3649 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
3650 test for MEM_REFs implicitely changing types with padding. Remove
3652 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
3654 2017-12-08 Jakub Jelinek <jakub@redhat.com>
3656 * config/arc/arc.c (arc_attribute_table): Add exclusions to
3658 * config/avr/avr.c (avr_attribute_table): Likewise.
3659 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3660 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3661 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3662 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
3663 of struct attribute_spec.
3664 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3666 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
3669 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
3671 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
3674 * profile-count.c (profile_count::from_gcov_type): Move from
3675 profile-count.h; handle overflow.
3676 * profile-count.h (profile_count::from_gcov_type): Move offline.
3678 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
3680 PR rtl-optimization/83304
3681 * combine.c (move_deaths): If we do not know where a register died,
3684 2017-12-08 Richard Biener <rguenther@suse.de>
3686 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
3687 Provide -fopt-info-loop feedback when we interchange in a nest.
3689 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3691 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
3692 for armv6 ARM CPU IDs.
3694 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3696 * common/config/arm/arm-common.c: Include <algorithm>.
3697 (INCLUDE_VECTOR): Define.
3698 (compare_opt_names): New function.
3699 (arm_rewrite_selected_arch): Only strip out extensions that can be
3700 expressed through -mfpu. Sort the remaining extensions
3703 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
3705 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
3706 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
3707 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
3708 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
3710 2017-06-08 Tristan Gingold <gindold@adacore.com>
3713 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
3714 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
3716 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3718 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
3719 of struct attribute_spec.
3721 2017-12-08 Julia Koval <julia.koval@intel.com>
3723 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
3724 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
3725 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
3726 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
3727 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
3728 _mm_maskz_dpwssds_epi32): Ditto.
3730 2017-12-08 Richard Biener <rguenther@suse.de>
3732 PR tree-optimization/81303
3733 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
3734 conditions try to create a comparison vector type matching
3735 the data vector type.
3736 (vectorizable_condition): Adjust.
3737 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
3738 Leave invariant conditions alone in case we can vectorize those.
3740 2017-12-08 Julia Koval <julia.koval@intel.com>
3742 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
3743 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
3744 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
3745 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
3746 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
3748 2017-12-08 Julia Koval <julia.koval@intel.com>
3750 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
3751 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
3752 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
3753 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
3754 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
3755 _mm_maskz_dpbusds_epi32): New intrinsics.
3757 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3759 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
3761 (ld<bh>io_signed): Likewise.
3762 (st<bhw>io): Likewise.
3763 * config/nios2/predicates.md (ldstio_memory_operand): Allow
3764 SMALL_INT12 constant integer operand.
3766 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
3768 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
3770 (nios2_large_constant_allowed): ...this. Adjust uses.
3771 (nios2_plus_symbolic_constant_p): Rename to...
3772 (nios2_plus_large_constant_p): ...this. Adjust uses.
3773 (nios2_legitimate_address_p): Correct CONST_INT handling.
3774 (nios2_symbolic_memory_operand_p): Rename to...
3775 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
3776 (nios2_large_constant_p): Check for large constant integers too.
3777 (nios2_split_large_constant): Handle constant integers.
3778 (nios2_split_symbolic_memory_operand): Rename to...
3779 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
3780 (nios2_legitimize_constant_address): Handle constant integers.
3781 (r0rel_constant_p): Handle small constant integers.
3782 (nios2_print_operand_address): Handle r0-relative integer addresses.
3783 * config/nios2/nios2-protos.h: Adjust for renamed functions.
3784 * config/nios2/nios2.md: Adjust for renamed functions.
3786 2017-12-07 Andrew Waterman <andrew@sifive.com>
3788 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
3789 (TARGET_HAVE_SRODATA_SECTION): New define.
3790 (riscv_select_section): New function.
3792 2017-12-08 Joseph Myers <joseph@codesourcery.com>
3793 Alexander Monakov <amonakov@ispras.ru>
3794 Jakub Jelinek <jakub@redhat.com>
3797 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
3799 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3801 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
3802 a bool success value. Don't call output_operand_lossage here.
3803 (aarch64_print_ldpstp_address): Return a bool success value.
3804 (aarch64_print_operand_address): Call output_addr_const if
3805 aarch64_print_address_internal fails.
3806 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
3807 'y'; call output_operand_lossage instead. Call output_operand_lossage
3808 if aarch64_print_ldpstp_address fails.
3810 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3812 * tree-vector-builder.h
3813 (tree_vector_builder::binary_encoded_nelts): Declare.
3814 * tree-vector-builder.c
3815 (tree_vector_builder::binary_encoded_nelts): New function.
3816 * fold-const.c (negate_expr_p): Likewise.
3817 (operand_equal_p, fold_checksum_tree): Likewise.
3818 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
3819 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
3820 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
3821 (uniform_vector_p): Likewise.
3822 * varasm.c (const_hash_1, compare_constant): Likewise.
3823 * tree-ssa-ccp.c: Include tree-vector-builder.h.
3824 (valid_lattice_transition): Operate directly on the VECTOR_CST
3826 * ipa-icf.c: Include tree-vector-builder.h.
3827 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
3828 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
3830 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3832 * tree.c (build_vector): Delete.
3833 * tree.h (build_vector): Make static and move into the self-testing
3836 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3838 * vector-builder.h (vector_builder::derived): New const overload.
3839 (vector_builder::elt): New function.
3840 * tree-vector-builder.h (tree_vector_builder::type): New function.
3841 (tree_vector_builder::apply_step): Declare.
3842 * tree-vector-builder.c (tree_vector_builder::apply_step): New
3844 * gimple-fold.h (tree_vector_builder): Declare.
3845 (gimple_build_vector): Take a tree_vector_builder instead of a
3846 type and vector of elements.
3847 * gimple-fold.c (gimple_build_vector): Likewise.
3848 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
3850 (get_initial_defs_for_reduction): Likewise.
3851 (vectorizable_induction): Likewise.
3853 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3855 * tree-vector-builder.h
3856 (tree_vector_builder::new_binary_operation): Declare.
3857 * tree-vector-builder.c
3858 (tree_vector_builder::new_binary_operation): New function.
3859 * fold-const.c (fold_relational_const): Use it.
3860 (const_binop): Likewise. Check that both input vectors have
3861 the same number of elements, thus excluding things like WIDEN_SUM.
3862 Check whether it is possible to operate directly on the encodings
3865 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3867 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
3868 new_unary_operation, operating only on the encoded elements.
3869 (const_unop): Likewise.
3870 (exact_inverse): Likewise.
3871 (distributes_over_addition_p): New function.
3872 (const_binop): Use tree_vector_builder and new_unary_operation
3873 for combinations of VECTOR_CST and INTEGER_CST. Operate only
3874 on the encoded elements unless the encoding is strided and the
3875 operation does not distribute over addition.
3876 (fold_convert_const): Use tree_vector_builder and
3877 new_unary_operation. Operate only on the encoded elements
3878 for truncating integer conversions, or for non-stepped encodings.
3880 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3882 * config/sparc/sparc.c: Include tree-vector-builder.h.
3883 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
3884 * expmed.c: Include tree-vector-builder.h.
3885 (make_tree): Use tree_vector_builder instead of build_vector.
3886 * fold-const.c: Include tree-vector-builder.h.
3887 (const_binop): Use tree_vector_builder instead of build_vector.
3888 (const_unop): Likewise.
3889 (native_interpret_vector): Likewise.
3890 (fold_vec_perm): Likewise.
3891 (fold_ternary_loc): Likewise.
3892 * gimple-fold.c: Include tree-vector-builder.h.
3893 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
3895 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
3896 (simplify_vector_constructor): Use tree_vector_builder instead
3898 * tree-vect-generic.c: Include tree-vector-builder.h.
3899 (add_rshift): Use tree_vector_builder instead of build_vector.
3900 (expand_vector_divmod): Likewise.
3901 (optimize_vector_constructor): Likewise.
3902 * tree-vect-loop.c: Include tree-vector-builder.h.
3903 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
3904 of build_vector. Explicitly use a stepped encoding for
3906 * tree-vect-slp.c: Include tree-vector-builder.h.
3907 (vect_get_constant_vectors): Use tree_vector_builder instead
3909 (vect_transform_slp_perm_load): Likewise.
3910 (vect_schedule_slp_instance): Likewise.
3911 * tree-vect-stmts.c: Include tree-vector-builder.h.
3912 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
3913 (vect_gen_perm_mask_any): Likewise.
3914 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
3915 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
3917 (build_vector_from_val): Likewise. Explicitly use a duplicate
3920 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
3922 * doc/generic.texi (VECTOR_CST): Describe new representation of
3924 * vector-builder.h: New file.
3925 * tree-vector-builder.h: Likewise.
3926 * tree-vector-builder.c: Likewise.
3927 * Makefile.in (OBJS): Add tree-vector-builder.o.
3928 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
3929 * tree-core.h (tree_base): Add a vector_cst field to the u union.
3930 (tree_vector): Change the number of elements to
3931 vector_cst_encoded_nelts.
3932 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
3933 (VECTOR_CST_ELTS): Delete.
3934 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
3935 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
3936 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
3937 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
3938 (VECTOR_CST_ENCODED_ELT): Likewise.
3939 (vector_cst_encoded_nelts): New function.
3940 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3941 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3942 (vector_cst_int_elt, vector_cst_elt): Declare.
3943 * tree.c: Include tree-vector-builder.h.
3944 (tree_code_size): Abort if passed VECTOR_CST.
3945 (tree_size): Update for new VECTOR_CST layout.
3946 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
3947 VECTOR_CST_NELTS_PER_PATTERN as arguments.
3948 (build_vector): Use tree_vector_builder.
3949 (vector_cst_int_elt, vector_cst_elt): New functions.
3950 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
3951 encoded elements and then create the vector in the canonical form.
3952 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
3953 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
3954 (tree_c_tests): Call test_vector_cst_patterns.
3955 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
3957 (hash_tree): Likewise.
3958 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
3959 (streamer_write_tree_header): Likewise.
3960 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
3961 (streamer_alloc_tree): Likewise. Update call to make_vector.
3962 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
3964 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
3966 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
3967 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
3968 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
3970 2017-12-07 Bin Cheng <bin.cheng@arm.com>
3971 Richard Biener <rguenther@suse.de>
3973 PR tree-optimization/81303
3974 * Makefile.in (gimple-loop-interchange.o): New object file.
3975 * common.opt (floop-interchange): Reuse the option from graphite.
3976 * doc/invoke.texi (-floop-interchange): Ditto. New document for
3977 -floop-interchange and mention it for -O3.
3978 * opts.c (default_options_table): Enable -floop-interchange at -O3.
3979 * gimple-loop-interchange.cc: New file.
3980 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
3981 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
3982 * passes.def (pass_linterchange): New pass.
3983 * timevar.def (TV_LINTERCHANGE): New time var.
3984 * tree-pass.h (make_pass_linterchange): New declaration.
3985 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
3986 interchange. Record IV before/after increment in new parameters.
3987 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
3988 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
3990 (check_reduction_path): ...New function here.
3991 * tree-vectorizer.h (check_reduction_path): New declaration.
3993 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
3996 PR rtl-optimization/80818
3997 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
3998 always early clobbered.
3999 * lra-lives.c (process_bb_lives): Check input hard regs for early
4000 clobbered non-operand hard reg.
4002 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4005 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4006 types_compatible_p, just that TYPE_MODE is the same.
4008 2017-12-07 Martin Sebor <msebor@redhat.com>
4011 * attribs.c (empty_attribute_table): Initialize new member of
4012 struct attribute_spec.
4013 (decl_attributes): Add argument. Handle mutually exclusive
4014 combinations of attributes.
4015 (selftests::test_attribute_exclusions): New function.
4016 (selftests::attribute_c_tests): Ditto.
4017 * attribs.h (decl_attributes): Add default argument.
4018 * selftest.h (attribute_c_tests): Declare.
4019 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4020 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4022 * doc/extend.texi (Common Function Attributes): Update const and pure.
4023 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4024 of struct attribute_spec.
4025 * config/arc/arc.c (arc_attribute_table): Same.
4026 * config/arm/arm.c (arm_attribute_table): Same.
4027 * config/avr/avr.c ( avr_attribute_table): Same.
4028 * config/bfin/bfin.c (bfin_attribute_table): Same.
4029 * config/cr16/cr16.c (cr16_attribute_table): Same.
4030 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4031 * config/h8300/h8300.c (h8300_attribute_table): Same.
4032 * config/i386/i386.c (ix86_attribute_table): Same.
4033 * config/ia64/ia64.c (ia64_attribute_table): Same.
4034 * config/m32c/m32c.c (m32c_attribute_table): Same.
4035 * config/m32r/m32r.c (m32r_attribute_table): Same.
4036 * config/m68k/m68k.c (m68k_attribute_table): Same.
4037 * config/mcore/mcore.c (mcore_attribute_table): Same.
4038 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4039 * config/mips/mips.c (mips_attribute_table): Same.
4040 * config/msp430/msp430.c (msp430_attribute_table): Same.
4041 * config/nds32/nds32.c (nds32_attribute_table): Same.
4042 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4043 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4044 * config/rl78/rl78.c (rl78__attribute_table): Same.
4045 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4046 * onfig/rx/rx.c (rx_attribute_table): Same.
4047 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4048 * config/sh/sh.c (sh_attribute_table): Same.
4049 * config/sparc/sparc.c (sparc_attribute_table): Same.
4050 * config/spu/spu.c (spu_attribute_table): Same.
4051 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4052 * config/v850/v850.c (v850_attribute_table): Same.
4053 * config/visium/visium.c (visium_attribute_table): Same.
4055 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4058 * config/arm/arm.c (INCLUDE_STRING): Define.
4059 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4060 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4063 2017-12-07 Michael Matz <matz@suse.de>
4065 Add unroll and jam pass
4067 * gimple-loop-jam.c: New file.
4068 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4069 * common.opt (funroll-and-jam): New option.
4070 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4071 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4072 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4073 * passes.def: Add pass_loop_jam.
4074 * timevar.def (TV_LOOP_JAM): Add.
4075 * tree-pass.h (make_pass_loop_jam): Declare.
4076 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4077 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4078 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4079 to flow_loop_tree_node_add.
4080 (duplicate_subloops, copy_loops_to): Append to sibling list.
4081 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4082 * doc/invoke.texi (-funroll-and-jam): Document new option.
4083 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4085 2017-12-07 Richard Biener <rguenther@suse.de>
4087 PR tree-optimization/83296
4088 PR tree-optimization/67769
4089 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4090 flow sensitive info in an unrelated BB.
4091 (value_replacement): Use reset_flow_sensitive_info.
4092 (minmax_replacement): Reset flow sensitive info on the def
4093 we move. Do not reset flow sensitive info in the whole BB
4094 we move the stmt to.
4095 (abs_replacement): Likewise.
4097 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4100 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4101 rs6000_cpu to the given -mcpu=, or to the default processor.
4103 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4105 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4106 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4107 instead of rs6000_cpu_attr.
4108 (rs6000_adjust_cost): Ditto.
4109 (is_microcoded_insn): Ditto.
4110 (rs6000_adjust_priority): Ditto.
4111 (rs6000_issue_rate): Ditto.
4112 (rs6000_use_sched_lookahead): Ditto.
4113 (rs6000_use_sched_lookahead_guard): Ditto.
4114 (rs6000_sched_reorder): Ditto.
4115 (force_new_group): Ditto.
4116 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4117 (group_ending_nop): Ditto.
4119 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4121 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4122 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4123 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4124 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4126 (rs6000_reassociation_width): Ditto.
4127 (rs6000_emit_epilogue): Ditto.
4128 (rs6000_adjust_cost): Ditto.
4129 (is_microcoded_insn): Ditto.
4130 (is_cracked_insn): Ditto.
4131 (rs6000_adjust_priority): Ditto.
4132 (rs6000_sched_reorder): Ditto.
4133 (rs6000_sched_reorder2): Ditto.
4134 (insn_must_be_first_in_group): Ditto.
4135 (insn_must_be_last_in_group): Ditto.
4136 (rs6000_register_move_cost): Ditto.
4137 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4140 2017-12-07 Julia Koval <julia.koval@intel.com>
4142 * config.gcc: Add vaesintrin.h.
4143 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4145 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4146 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4148 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4149 * config/i386/immintrin.h: Include vaesintrin.h.
4150 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4151 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4152 _mm_aesdec_epi128): New intrinsics.
4154 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4156 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4157 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4158 c-family/c-spellcheck.cc.
4159 (best_macro_match::best_macro_match): Likewise.
4161 (struct edit_distance_traits<cpp_hashnode *>): Move to
4162 c-family/c-spellcheck.h.
4163 (class best_macro_match): Likewise.
4165 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4167 PR tree-optimization/83293
4168 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4169 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4170 might insert into empty bb.
4173 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4174 simplify for plus with :c added, and pointer_plus without that.
4175 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4176 with undefined overflow and the conversion is not widening,
4177 perform negation in utype and only convert to type afterwards.
4178 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4179 simplify for plus with :c added, and pointer_plus without that.
4180 If type is integral with undefined overflow and the conversion is
4181 not widening, perform minus in utype and only convert to type
4182 afterwards. Move the last pointer_diff_expr simplify into the
4185 2017-12-06 Martin Sebor <msebor@redhat.com>
4187 PR tree-optimization/82646
4188 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4189 strncpy, not maxlen.
4191 2017-12-06 Martin Sebor <msebor@redhat.com>
4193 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4196 PR tree-optimization/83075
4197 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4198 strncat/strncpy don't change length of source string.
4200 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4203 2017-11-29 Martin Aberg <maberg@gaisler.com>
4205 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4206 to prevent b2bst errata sequence.
4207 (sqrtdf2_fix): Likewise.
4209 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4211 PR tree-optimization/81945
4212 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4213 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4214 to dest_cfun has orig_loop_num set, either remap it to the new
4215 loop number if the loop got moved too, or clear it.
4217 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4219 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4221 (thunderx2t99_nothing): New insn reservation.
4222 (thunderx2t99_mrs): New insn reservation.
4223 (thunderx2t99_multiple): New insn reservation.
4224 (thunderx2t99_alu_basi): Add bfx to reservation.
4225 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4227 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4230 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4232 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4234 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4235 tree-ssa-pre.c::remove_dead_inserted_code.
4236 * tree-ssa-dce.h: New file.
4237 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4238 (remove_dead_inserted_code): Move and rename to function
4239 tree-ssa-dce.c::simple_dce_from_worklist.
4240 (pass_pre::execute): Update use.
4242 2017-12-05 Richard Biener <rguenther@suse.de>
4244 PR tree-optimization/83277
4245 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4246 to code-gen liveout vars.
4248 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4250 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4251 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4252 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4255 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4257 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4259 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4260 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4263 2017-12-05 Richard Biener <rguenther@suse.de>
4265 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4266 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4267 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4268 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4270 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4272 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4274 2017-12-05 Martin Liska <mliska@suse.cz>
4275 Jakub Jelinek <jakub@redhat.com>
4277 * doc/invoke.texi: Document the options.
4278 * flag-types.h (enum sanitize_code): Add
4279 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4280 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4281 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4282 * opts.c: Define new sanitizer options.
4283 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4284 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4286 2017-12-05 Julia Koval <julia.koval@intel.com>
4288 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4289 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4290 (ix86_handle_option): Handle -mavx512vnni.
4291 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4292 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4293 * config/i386/i386-c (__AVX512VNNI__): New.
4294 * config/i386/i386.c (ix86_target_string): Handle new option.
4295 (ix86_valid_target_attribute_inner_p): Handle new option.
4296 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4297 * config/i386/i386.opt (mavx512vnni): New option.
4299 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4302 * athlon.md: Disable for generic.
4303 * haswell.md: Enable for generic.
4304 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4305 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4307 (ix86_adjust_cost): Move generic to haswell path.
4309 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4311 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4312 instead of MEM_P in a couple more places. Fix formatting issues.
4314 2017-12-04 Jim Wilson <jimw@sifive.com>
4316 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4317 instead of GP_REG_LAST-1.
4318 (riscv_adjust_libcall_cfi_prologue): Likewise.
4319 (riscv_adjust_libcall_cri_epilogue): Likewise.
4320 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4323 2017-12-04 Luis Machado <luis.machado@linaro.org>
4325 * ipa-pure-const.c (check_decl): Add missing newline.
4326 (state_from_flags): Likewise.
4328 2017-12-04 Jeff Law <law@redhat.com>
4330 PR tree-optimizatin/78496
4331 * gimple-ssa-evrp-analyze.h
4332 (evrp_range_analyzer::get_vr_values): Simplify.
4333 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4334 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4335 and gimple-ssa-evrp-analyze.h.
4336 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4337 (simplify_stmt_for_jump_threading): Copy a blob of code from
4338 tree-vrp.c to use ranges to simplify statements.
4339 (dom_opt_dom_walker::before_dom_children): Call
4340 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4341 (dom_opt_dom_walker::after_dom_children): Similarly for
4342 evrp_range_analyzer::leave.
4343 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4346 * gimple-ssa-evrp-analyze.c
4347 (evrp_range_analyzer::extract_range_from_stmt): Always use
4348 vr_values::update_value_range so preexisting range info is
4349 medged with new range info, even if the new range is VR_VARYING.
4351 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4353 * combine.c: Adjust comment.
4354 (use_crosses_set_p): Delete.
4355 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4356 (try_combine): Ditto.
4358 2017-12-04 Richard Biener <rguenther@suse.de>
4360 PR tree-optimization/83255
4361 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4362 Re-add zero-iteration check.
4364 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4366 PR rtl-optimization/83245
4367 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4368 hard registers as earlyclobber, also if not in an asm.
4370 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4374 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4377 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4378 rs6000_cpu based on cpu_index, not tune_index.
4380 2017-12-04 Richard Biener <rguenther@suse.de>
4382 PR tree-optimization/83238
4383 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4384 code match comment, rejecting invalid SESE regions.
4386 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4388 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4389 require base operand is a REG_POINTER prior to reload on targets
4390 with non-equivalent space registers.
4392 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4394 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
4395 (update_specialized_profile): Fix updating of counts.
4396 (perhaps_add_new_callers): Likewise.
4398 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4401 * x86-tune.def: Remove obsolette FIXMEs.
4402 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
4403 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
4404 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
4406 (X86_TUNE_PAD_RETURNS): Disable for generic.
4408 2017-12-02 Jakub Jelinek <jakub@redhat.com>
4410 PR tree-optimization/83170
4411 PR tree-optimization/83241
4412 * gimple-ssa-store-merging.c
4413 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
4414 gimple_vuse (ins_stmt) in case it has changed.
4415 (imm_store_chain_info::output_merged_store): Likewise.
4417 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
4421 * tree-cfg.c (pass_warn_function_return::execute): Handle
4422 __builtin_ubsan_handle_missing_return like __builtin_unreachable
4423 with BUILTINS_LOCATION.
4427 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
4428 is BLKmode for vector field with vector raw mode, use TYPE_MODE
4429 instead of DECL_MODE.
4431 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
4432 last argument to rtx pointer.
4433 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
4434 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
4435 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
4436 depending on the chosen ISAs.
4437 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
4438 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
4439 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
4441 * config/i386/sse.md (mov<mode>_internal): Likewise.
4442 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
4444 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4446 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
4448 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4450 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
4451 parameters from prototype.
4452 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
4453 parameters. Don't print a comment.
4454 (emit_fusion_gpr_load): Adjust.
4455 (emit_fusion_load_store): Adjust.
4456 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
4457 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
4458 comment on the second line.
4460 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4463 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4464 rs6000_cpu based on cpu_index, not tune_index.
4466 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4468 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
4469 which_alternative instead of which_alternative + 1.
4470 (output_asm_insn): Print an extra tab if the template is short.
4472 2017-12-01 Jim Wilson <jimw@sifive.com>
4474 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
4476 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
4477 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
4479 * doc/tm.texi: Regenerate.
4481 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4484 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
4485 whether we can allocate pseudos before trying to fix an address.
4486 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
4487 memory address is indexed or indirect.
4488 (floatuns_<mode>si2_hw2): Likewise.
4490 2017-12-01 Jason Merrill <jason@redhat.com>
4492 * Makefile.in (TAGS): Add c-family/*.cc.
4494 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
4496 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
4497 (call_value_insn): Likewise.
4498 (sibcall_insn): Likewise.
4499 (sibcall_value_insn): Likewise.
4500 (movsi_aarch64): Likewise.
4501 (movdi_aarch64): Likewise.
4502 (add_losym_): Likewise.
4503 (ldr_got_small_): Likewise.
4504 (ldr_got_small_sidi): Likewise.
4505 (ldr_got_small_28k_): Likewise.
4506 (ldr_got_small_28k_sidi): Likewise.
4507 * config/aarch64/aarch64.c (aarch64_print_address_internal):
4508 Move output_addr_const to symbolic case. Add error check.
4510 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4512 * config/s390/predicates.md (plus16_Q_operand): New predicate.
4513 * config/s390/s390.md: Disable MVC merging peephole if it would
4514 disable operand forwarding.
4515 (new peephole2): Split MVCs if it would turn them into up to 2
4518 2017-12-01 Richard Biener <rguenther@suse.de>
4520 PR tree-optimization/83232
4521 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
4522 detection of same access. Instead of breaking the group here
4523 do not consider the duplicate. Add comment explaining real fix.
4525 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
4527 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
4529 2017-12-01 Sudakshina Das <sudi.das@arm.com>
4531 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
4533 2017-12-01 Jakub Jelinek <jakub@redhat.com>
4535 * function.h (struct function): Remove cilk_frame_decl,
4536 is_cilk_function and calls_cilk_spawn fields.
4537 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
4539 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
4540 cilk_elemental field.
4541 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
4542 * target.def: Adjust comment.
4543 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
4544 Don't test cilk_elemental.
4546 PR tree-optimization/83233
4547 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
4548 bswap_stat name for the struct.
4551 * tree.h (SWITCH_BREAK_LABEL_P): Define.
4552 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
4553 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
4554 SWITCH_BREAK_LABEL_P set on the label.
4555 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
4556 added for default case if it was missing and not all cases covered.
4557 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
4558 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
4560 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
4561 to avoid -Wimplicit-fallthrough warning.
4562 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
4563 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
4565 PR tree-optimization/83221
4566 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
4568 (init_reassoc): Formatting fix.
4571 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
4572 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
4576 * tree-cfg.c: Include asan.h.
4577 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
4579 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4581 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
4583 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4585 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
4586 vpcmpeqd instruction.
4588 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
4590 * config/i386/i386.c (standard_sse_constant_opcode): Fix
4591 registers type for 128bit mode.
4593 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4595 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
4597 * gengtype-state.c (read_state_pair): Likewise.
4598 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
4599 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
4600 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
4601 * ifcvt.c (noce_try_store_flag_constants): Likewise.
4602 * tree-ssa-ccp.c (ccp_finalize): Likewise.
4603 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
4604 * builtins.c (fold_builtin_3): Likewise.
4605 * graphite-scop-detection.c
4606 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
4607 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
4610 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
4613 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
4614 parameter. Return true if ignore_decl_rtl_set_p is true and other
4615 conditions are satisfied.
4616 * asan.h (asan_protect_global): Add new parameter.
4617 * varasm.c (categorize_decl_for_section): Pass true as second parameter
4618 to asan_protect_global calls.
4620 2017-11-30 Jim Wilson <jimw@sifive.com>
4622 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
4623 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
4624 -mstrict-align, add info on default value. Delete redundant lines for
4625 -mabi. Add missing -mexplicit-relocs docs.
4627 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4629 * config/arc/arc.md (trap): New pattern.
4631 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4633 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
4634 instruction to end into a delay slot.
4635 * config/arc/arc.md (cond_delay_insn): Check if the instruction
4636 can be placed into a delay slot against reg_note.
4638 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4640 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
4641 labels number of usages.
4643 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
4645 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
4647 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
4649 2017-11-30 Jakub Jelinek <jakub@redhat.com>
4652 * internal-fn.c (expand_mul_overflow): Optimize unsigned
4653 multiplication by power of 2 constant into two shifts + comparison.
4655 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
4658 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
4660 2017-11-30 Richard Biener <rguenther@suse.de>
4662 PR tree-optimization/83202
4663 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
4664 allow_peel argument and guard peeling.
4665 (canonicalize_loop_induction_variables): Likewise.
4666 (canonicalize_induction_variables): Pass false.
4667 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
4668 peeling from cunrolli.
4670 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4672 * combine.c (try_combine): Print a message to dump file whenever
4673 I0, I1, or I2 cannot be combined into I3.
4675 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4677 PR rtl-optimization/83156
4678 PR rtl-optimization/82621
4679 * combine.c (try_combine): Don't split an I2 if one of the dests is
4680 set again before I3. Allow unused dests.
4682 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
4684 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
4686 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
4688 PR rtl-optimization/80818
4689 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
4690 recursively. Use insn code for clobber.
4691 (lra_set_insn_recog_data): Pass the new arg to
4692 collect_non_operand_hard_regs.
4693 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
4695 (lra_update_insn_regno_info): Pass insn to
4696 add_regs_to_insn_regno_info.
4698 2017-11-29 Jim Wilson <jimw@sifive.com>
4699 Andrew Waterman <andrew@sifive.com>
4701 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
4702 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
4703 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
4704 use SINGLE_SHIFT_COST.
4705 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
4706 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
4708 2017-11-29 Julia Koval <julia.koval@intel.com>
4710 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
4711 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
4712 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
4713 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
4714 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
4715 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
4716 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
4717 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
4718 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
4719 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
4720 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
4721 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
4722 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
4723 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
4724 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
4725 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
4726 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
4727 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
4728 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
4729 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
4730 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
4731 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
4732 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
4733 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
4734 __builtin_ia32_vpshldv_v2di_mask,
4735 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
4736 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
4737 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
4739 2017-11-29 Julia Koval <julia.koval@intel.com>
4741 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
4742 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
4743 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
4744 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
4745 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
4746 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
4747 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
4748 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
4749 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
4750 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
4751 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
4752 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
4753 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
4754 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
4755 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
4756 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
4757 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
4758 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
4759 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4760 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
4761 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
4763 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4765 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
4766 movsi_pic_gotdata_op instruction as a load for the UT699 errata
4769 2017-11-29 Martin Aberg <maberg@gaisler.com>
4771 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4772 to prevent b2bst errata sequence.
4773 (sqrtdf2_fix): Likewise.
4775 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4777 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
4778 (div_sqrt_insn_p): New function.
4779 (sparc_do_work_around_errata): Insert NOP instructions to
4780 prevent sequences that could trigger the TN-0013 errata for
4781 certain LEON3 processors.
4782 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
4783 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
4784 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
4785 (in_branch_delay): Prevent div and sqrt in delay slot if
4787 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
4789 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4791 * config/sparc/sparc.c (atomic_insn_p): New function.
4792 (sparc_do_work_around_errata): Insert NOP instructions to
4793 prevent sequences that could trigger the TN-0010 errata for
4795 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
4796 instruction referable in atomic_insns_p.
4798 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4800 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
4801 (atomic_compare_and_swap_leon3_1): Likewise.
4804 2017-11-29 Daniel Cederman <cederman@gaisler.com>
4806 * config/sparc/sparc.c (fpop_insn_p): New function.
4807 (sparc_do_work_around_errata): Insert NOP instructions to
4808 prevent sequences that could trigger the TN-0012 errata for
4810 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
4811 * config/sparc/sparc.md (fix_gr712rc): New attribute.
4812 (in_branch_annul_delay): Prevent floating-point instructions
4813 in delay slot of annulled integer branch.
4815 2017-11-29 Richard Biener <rguenther@suse.de>
4817 PR tree-optimization/83202
4818 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
4820 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
4821 nodes vectorized to the same stmts multiple times.
4822 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
4823 (vect_analyze_slp_instance): Adjust.
4824 (scalar_stmts_to_slp_tree_map_t): New typedef.
4825 (vect_schedule_slp_instance): Add a map recording the SLP node
4826 representing the vectorized stmts for a set of scalar stmts.
4827 Avoid code-generating redundancies.
4828 (vect_schedule_slp): Allocate map and pass it down.
4830 2017-11-29 Nathan Sidwell <nathan@acm.org>
4833 * tree.c (build_complex_type): Fix canonicalization. Only fill in
4836 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
4838 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
4839 cases for printing LDP/STP memory addresses.
4840 (aarch64_print_address_internal): Renamed from
4841 aarch64_print_operand_address, added parameter, add Pmode check.
4842 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
4843 (aarch64_print_operand_address): Indirect to
4844 aarch64_print_address_internal.
4845 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
4846 'y' operand output specifier.
4848 2017-11-29 Jakub Jelinek <jakub@redhat.com>
4851 * tree.c (build_simple_mem_ref_loc): Handle
4852 get_addr_base_and_unit_offset returning a MEM_REF.
4855 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
4858 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
4861 2017-11-28 David Malcolm <dmalcolm@redhat.com>
4863 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
4865 (layout::move_to_column): Likewise.
4867 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4870 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
4871 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
4873 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4874 Martin Jambor <mjambor@suse.cz>
4877 * tree.h (expr_type_first_operand_type_p): Declare
4878 * tree.c (expr_type_first_operand_type_p): New function.
4879 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
4880 (ipa_value_from_jfunc): Adjust declaration.
4881 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
4882 Use it as result type for arithmetics, unless it is NULL in which case
4883 be more conservative.
4884 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
4885 ipa_get_jf_pass_through_result.
4886 (propagate_vals_across_pass_through): Likewise.
4887 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
4888 is to propagate_vals_across_pass_through.
4889 (propagate_constants_across_call): Pass PARM_TYPE to
4890 propagate_scalar_across_jump_function.
4891 (find_more_scalar_values_for_callers_subset): Pass parameter type to
4892 ipa_value_from_jfunc.
4893 (cgraph_edge_brings_all_scalars_for_node): Likewise.
4894 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
4895 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
4896 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
4897 target_type, pass it to ipa_value_from_jfunc.
4898 (update_indirect_edges_after_inlining): Pass parameter type to
4899 try_make_edge_direct_simple_call.
4901 2017-11-28 Jeff Law <law@redhat.com>
4903 * gimple-ssa-evrp-analyze.c
4904 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
4905 refine ranges if scev_initialized_p returns true.
4906 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
4908 2017-11-28 Julia Koval <julia.koval@intel.com>
4910 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
4911 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
4912 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
4913 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
4914 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
4915 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
4916 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
4917 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
4918 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
4919 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
4920 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
4921 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
4922 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
4923 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
4924 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
4925 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
4926 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
4927 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
4928 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
4929 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
4930 __builtin_ia32_vpshrd_v2di_mask): New builtins.
4931 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
4933 2017-11-28 Julia Koval <julia.koval@intel.com>
4935 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
4936 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
4937 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
4938 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
4939 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
4940 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
4941 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
4942 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
4943 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
4944 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
4945 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
4946 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
4947 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
4948 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
4949 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
4950 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
4951 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
4952 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
4953 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
4954 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
4955 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
4956 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
4957 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
4958 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
4959 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
4960 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
4961 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
4962 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
4963 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
4964 __builtin_ia32_vpshld_v2di_mask): New builtins.
4965 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
4966 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
4968 2017-11-28 Richard Biener <rguenther@suse.de>
4970 PR tree-optimization/80776
4971 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
4973 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
4975 (evrp_range_analyzer::record_ranges_from_incoming_edges):
4976 If the incoming edge is an effective fallthru because the other
4977 edge only reaches a __builtin_unreachable () then record ranges
4978 derived from the controlling condition in SSA info.
4979 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
4980 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4982 2017-11-28 Olivier Hainque <hainque@adacore.com>
4984 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
4985 on mingw build hosts.
4987 2017-11-28 Jakub Jelinek <jakub@redhat.com>
4990 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
4992 2017-11-28 Richard Biener <rguenther@suse.de>
4995 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
4996 copies generated from memcpy use a character array as reference
4999 2017-11-28 Julia Koval <julia.koval@intel.com>
5000 Sebastian Peryt <sebastian.peryt@intel.com>
5002 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5003 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5004 cilk-common.o, cilk.h, cilk-common.c): Remove.
5006 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5007 * builtins.c (is_builtin_name): Remove cilkplus condition.
5008 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5009 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5010 cilk-builtins.def, cilkplus.def): Remove.
5011 * cif-code.def (CILK_SPAWN): Remove.
5012 * cilk-builtins.def: Delete.
5013 * cilk-common.c: Ditto.
5015 * cilkplus.def: Ditto.
5016 * config/darwin.h (fcilkplus): Delete.
5017 * cppbuiltin.c: Ditto.
5018 * doc/extend.texi: Remove cilkplus doc.
5019 * doc/generic.texi: Ditto.
5020 * doc/invoke.texi: Ditto.
5021 * doc/passes.texi: Ditto.
5022 * gcc.c (fcilkplus): Remove.
5023 * gengtype.c (cilk.h): Remove.
5024 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5026 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5028 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5030 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5031 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5032 cilkplus conditions.
5033 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5034 inline_read_section): Ditto.
5035 * ipa-inline-analysis.c (cilk.h): Remove.
5036 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5037 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5038 append_linker_options): Remove condition for fcilkplus.
5039 * lto/lto-lang.c (cilk.h): Remove.
5040 (lto_init): Remove condition for fcilkplus.
5041 * omp-expand.c (expand_cilk_for_call): Delete.
5042 (expand_omp_taskreg, expand_omp_for_static_chunk,
5043 expand_omp_for): Remove cilkplus
5045 (expand_cilk_for): Delete.
5046 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5047 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5048 execute_lower_omp, diagnose_sb_0): Ditto.
5049 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5050 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5051 * tree-nested.c: Ditto.
5052 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5053 (dump_generic_node): Ditto.
5054 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5055 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5057 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5059 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5061 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5062 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5063 for vector mode and !TARGET_SIMD.
5065 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5067 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5069 * tree.h (SWITCH_LABELS): Remove.
5070 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5071 assert SWITCH_BODY is non-NULL.
5072 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5074 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5076 PR tree-optimization/80788
5077 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5078 has TREE_OVERFLOW set, call drop_tree_overflow.
5080 2017-11-28 Richard Biener <rguenther@suse.de>
5082 PR tree-optimization/83158
5083 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5085 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5088 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5089 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5091 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5093 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5094 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5096 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5099 * varasm.c (compare_constant): Take the mode of the constants into
5100 account when comparing floating point constants.
5102 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5104 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5105 from explicit instantiation of debug_helper.
5106 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5108 2017-11-27 Richard Biener <rguenther@suse.de>
5110 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5113 2017-11-27 Richard Biener <rguenther@suse.de>
5115 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5116 (type_hash_canon): Also clear int_cst_hash_table entry for
5118 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5120 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5122 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5123 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5124 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5126 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5128 * hash-map.h (gt_cleare_cache): Avoid UB.
5130 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5132 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5133 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5134 (decide_unroll_runtime_iterations): Likewise.
5135 (decide_unroll_stupid): Likewise.
5137 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5140 * config/i386/i386.md: Add a loop with incssp.
5142 2017-11-27 Martin Jambor <mjambor@suse.cz>
5144 PR tree-optimization/81248
5145 * tree-sra.c (splice_param_accesses): Remove size check.
5146 (decide_one_param_reduction): Fix size check.
5147 * gimple-pretty-print.c (dump_profile): Silence warning.
5148 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5150 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5153 * dbxout.c (lastlineno): New variable.
5154 (dbx_debug_hooks): Use dbxout_switch_text_section as
5155 switch_text_section debug hook.
5156 (dbxout_function_end): Switch to current_function_section
5157 rather than function_section. If crtl->has_bb_partition,
5158 output just one N_FUN, depending on in_cold_section_p.
5159 (dbxout_source_line): Remember last lineno in lastlineno.
5160 (dbxout_switch_text_section): New function.
5161 (dbxout_function_decl): Adjust dbxout_block caller.
5162 (dbx_block_with_cold_children): New function.
5163 (dbxout_block): Return true if any LBRAC/RBRAC have been
5164 emitted. Use dbx_block_with_cold_children at depth == 0
5165 in second partition. Add PARENT_BLOCKNUM argument, pass
5166 it optionally adjusted to children. Output LBRAC/RBRAC
5167 around recursive call only if the block is in the current
5168 partition, if not and anything was output, emit empty
5170 * final.c (final_scan_insn): Compute cold_function_name
5171 before calling switch_text_section debug hook. Call
5172 that hook even if dwarf2out_do_frame if not emitting
5176 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5177 TREE_READONLY decls.
5179 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5181 PR rtl-optimization/82488
5182 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5184 2017-11-26 Julia Koval <julia.koval@intel.com>
5186 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5188 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5191 2017-11-26 Julia Koval <julia.koval@intel.com>
5193 * config/i386/driver-i386.c (host_detect_local_cpu):
5194 Detect skylake-avx512.
5196 2017-11-26 Julia Koval <julia.koval@intel.com>
5198 * config.gcc: Add -march=cannonlake.
5199 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5200 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5201 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5202 (PTA_CANNONLAKE): New.
5203 (processor_target_table): Add cannonlake.
5204 (ix86_option_override_internal): Ditto.
5205 (fold_builtin_cpu): Ditto.
5206 (get_builtin_code_for_version): Handle cannonlake.
5207 (M_INTEL_COREI7_CANNONLAKE): New.
5208 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5209 * doc/invoke.texi: Add -march=cannonlake.
5211 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5213 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5214 (try_init_one_plugin): Alternative implementation for MinGW.
5215 * Makefile.in (plugin_implib): New.
5216 (gengtype-lex.c): Fix broken AIX workaround.
5217 * configure: Regenerate.
5218 * doc/plugins.texi: Document support for MinGW.
5220 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5222 PR rtl-optimization/81553
5223 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5224 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5225 is a shift where C1 has different mode than the whole shift, use C1's
5226 mode for MULT rather than the shift's mode.
5229 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5230 builtins not enabled in the currently selected ISA.
5232 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5234 PR tree-optimization/71026
5235 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5236 (insert_reciprocals): Change to insert reciprocals before a division
5237 by a square and to insert the square of a reciprocal.
5238 (execute_cse_reciprocals_1): Change to consider division by a square.
5239 (register_division_in): Add importance parameter.
5241 2017-11-24 Richard Biener <rguenther@suse.de>
5243 PR tree-optimization/82402
5244 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5245 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5247 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5249 * match.pd (0-ptr): New transformation.
5251 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5254 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5255 when skipping checking.
5257 2017-11-24 Richard Biener <rguenther@suse.de>
5259 PR tree-optimization/83128
5260 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5261 (vn_reference_lookup_3): Likewise.
5263 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5266 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5267 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5270 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5271 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5272 check for the latter with gimple_call_builtin_p. Do not handle
5273 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5275 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5277 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5280 2017-11-23 Julia Koval <julia.koval@intel.com>
5282 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5283 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5284 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5285 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5286 _mm512_maskz_expandloadu_epi16): New intrinsics.
5287 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5288 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5289 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5290 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5291 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5292 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5293 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5294 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5295 _mm256_maskz_expandloadu_epi8): New intrinsics.
5296 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5297 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5298 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5299 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5300 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5301 * config/i386/sse.md (VI248_VLBW): New iterator.
5302 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5304 2017-11-23 Julia Koval <julia.koval@intel.com>
5306 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5307 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5308 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5309 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5310 _mm512_mask_compressstoreu_epi16): New.
5311 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5312 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5313 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5314 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5315 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5316 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5317 _mm256_mask_compressstoreu_epi8): New.
5318 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5319 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5320 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5321 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5322 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5323 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5324 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5325 __builtin_ia32_compresshi128_mask,
5326 __builtin_ia32_compressstoreuqi512_mask,
5327 __builtin_ia32_compressstoreuhi512_mask,
5328 __builtin_ia32_compressstoreuqi256_mask,
5329 __builtin_ia32_compressstoreuqi128_mask,
5330 __builtin_ia32_compressstoreuhi256_mask,
5331 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5332 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5334 (ix86_expand_special_args_builtin): Handle new types.
5335 (s4fma_expand): Handle new builtin array.
5336 * config/i386/immintrin.h: Include new headers.
5337 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5338 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5340 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5342 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5343 fixes. Declare temp and g variables at the top in order to avoid
5344 {} in most of the cases.
5346 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5348 * match.pd (ptr-0): New transformation.
5350 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5352 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5353 (struct addr_mode_cost_table): New.
5354 (struct tune_params): Add field addr_mode_costs.
5355 * config/arm/arm.c (generic_addr_mode_costs): New.
5356 (arm_slowmul_tune): Initialise addr_mode_costs field.
5357 (arm_fastmul_tune): Likewise.
5358 (arm_strongarm_tune): Likewise.
5359 (arm_xscale_tune): Likewise.
5360 (arm_9e_tune): Likewise.
5361 (arm_marvell_pj4_tune): Likewise.
5362 (arm_v6t2_tune): Likewise.
5363 (arm_cortex_tune): Likewise.
5364 (arm_cortex_a8_tune): Likewise.
5365 (arm_cortex_a7_tune): Likewise.
5366 (arm_cortex_a15_tune): Likewise.
5367 (arm_cortex_a35_tune): Likewise.
5368 (arm_cortex_a53_tune): Likewise.
5369 (arm_cortex_a57_tune): Likewise.
5370 (arm_exynosm1_tune): Likewise.
5371 (arm_xgene1_tune): Likewise.
5372 (arm_cortex_a5_tune): Likewise.
5373 (arm_cortex_a9_tune): Likewise.
5374 (arm_cortex_a12_tune): Likewise.
5375 (arm_cortex_a73_tune): Likewise.
5376 (arm_v7m_tune): Likewise.
5377 (arm_cortex_m7_tune): Likewise.
5378 (arm_v6m_tune): Likewise.
5379 (arm_fa726te_tune): Likewise.
5380 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5382 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5384 * config/arm/arm.c (arm_mem_costs): New function.
5385 (arm_rtx_costs_internal): Use arm_mem_costs.
5387 2017-11-23 Mark Wielaard <mark@klomp.org>
5389 * dwarf2out.c (init_sections_and_labels): Use generation to create
5390 unique ranges_section_label and ranges_base_label. Return generation.
5391 (output_rnglists): Add generation argument. Use generation to create
5392 unique ranges labels.
5393 (dwarf2out_finish): Get generation from init_sections_and_labels
5394 and pass generation to output_rnglists.
5396 2017-11-23 Mike Stump <mikestump@comcast.net>
5397 Eric Botcazou <ebotcazou@adacore.com>
5399 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
5400 * cfgloop.h (struct loop): Add unroll field.
5401 * function.h (struct function): Add has_unroll bitfield.
5402 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
5403 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
5404 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
5405 (pass_rtl_unroll_loops::gate): Likewise.
5406 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
5407 for which loop->unroll==1.
5408 (decide_unroll_constant_iterations): Use note for consistency and
5409 take loop->unroll into account. Return early if loop->unroll is set.
5410 Fix thinko in existing test.
5411 (decide_unroll_runtime_iterations): Use note for consistency and
5412 take loop->unroll into account.
5413 (decide_unroll_stupid): Likewise.
5414 * lto-streamer-in.c (input_cfg): Read loop->unroll.
5415 * lto-streamer-out.c (output_cfg): Write loop->unroll.
5416 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
5418 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
5419 (print_loop): Print loop->unroll if set.
5420 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
5421 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
5422 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
5424 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
5425 loop->unroll is set and smaller than the trip count. Otherwise bypass
5426 entirely the heuristics if loop->unroll is set. Remove dead note.
5427 Fix off-by-one bug in other note.
5428 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
5429 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
5431 (tree_unroll_loops_completely): Make static.
5432 (pass_complete_unroll::execute): Use correct type for variable.
5433 (pass_complete_unrolli::execute): Fix formatting.
5434 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
5436 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
5438 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
5439 enable when TARGET_PREFER_AVX128 is set.
5441 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5443 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
5445 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
5446 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
5448 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5450 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
5451 frequencies. Fix estimation of aggregate parameters.
5453 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
5455 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
5458 2017-11-23 Tom de Vries <tom@codesourcery.com>
5460 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
5462 2017-11-23 Tom de Vries <tom@codesourcery.com>
5464 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
5467 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
5470 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
5471 sibcall_value_pcrel_fdpic): Use local variable instead of
5473 (calli_tbr_rel): Add missing operand 2.
5474 (call_valuei_tbr_rel): Add missing operand 3.
5476 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5479 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
5480 bitpos/bitsize covering the whole destination, use store_expr only if
5481 the complex mode is the same. Otherwise, use expand_normal and if
5482 it returns CONCAT, subreg each part separately instead of trying to
5483 subreg the whole result.
5485 2017-11-23 Richard Biener <rguenther@suse.de>
5487 PR tree-optimization/23094
5488 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
5489 come here from walking over backedges in the first iteration.
5490 (vn_reference_lookup_3): Skip clobbers that store the same value.
5492 2017-11-23 Richard Biener <rguenther@suse.de>
5494 PR tree-optimization/81403
5495 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
5496 a block we need a leader relative to.
5497 (phi_translate_1): For nary processing require a leader from
5498 get_representative_for given we run expression simplification
5499 using match-and-simplify. Remove previous fix.
5501 2017-11-22 Jeff Law <law@redhat.com>
5503 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
5504 Use new method allocate_value_range rather than accessing the
5505 vrp_value_range_pool data member directly.
5506 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
5507 to use extract_range_from_stmt method to avoid need for
5508 extract_range_from_assignment method.
5509 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
5510 method rather than setting values_propgated data member directly.
5511 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
5512 and values propagated data members and extract_range_from_assignment
5513 method. Reorder private data members to conform to standards.
5514 Add new methods set_lattice_propagation_complete and
5515 allocate_value_range.
5517 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
5519 PR rtl-optimization/83030
5520 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
5521 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
5522 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
5523 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
5524 CROSSING_JUMP_P flag is already set before setting it.
5525 * cfgrtl.c (fixup_partition_crossing): Likewise.
5526 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
5529 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5531 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
5532 Handle the case where both arguments are using gen_const_vec_series.
5534 2017-11-22 David Malcolm <dmalcolm@redhat.com>
5537 * pretty-print.c (pp_format): Move quoting implementation to
5538 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
5539 to pass address of "quote" local.
5540 (pp_begin_quote): New function.
5541 (pp_end_quote): New function.
5542 * pretty-print.h (printer_fn): Convert penultimate param from bool
5544 (pp_begin_quote): New decl.
5545 (pp_end_quote): New decl.
5546 * tree-diagnostic.c (default_tree_printer): Convert penultimate
5547 param from bool to bool *.
5548 * tree-diagnostic.h (default_tree_printer): Likewise.
5550 2017-11-22 Jeff Law <law@redhat.com>
5552 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
5553 Initialize vr_values.
5554 (evrp_range_analyzer::try_find_new_range): Call methods attached to
5555 vr_values via vr_values class instance rather than delegators.
5556 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
5557 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5558 (evrp_range_analyzer::push_value_range): Likewise.
5559 (evrp_range_analyzer::pop_value_range): Likewise.
5560 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
5561 most delegators. Those remaining are exposed as public interfaces.
5562 Make vr_values a pointer and private.
5563 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
5565 (evrp_range_analyzer::get_vr_value): New method.
5566 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
5567 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
5568 (class evrp_dom_walker): Attach evrp_folder class, initialize
5569 it in the ctor. Remove temporary delegators.
5570 (evrp_dom_walker::before_dom_children): Call methods in attached
5571 evrp_range_analyzer class via class instance pointer. Use
5572 free value_range_constant_singleton to remove need for
5573 op_with_constant_singleton_value delegator method. Do not
5574 create a vrp_prop class instance for every call! Narrow
5575 scope of a couple variables.
5576 (evrp_dom_walker::cleanup): Call methods in attached
5577 evrp_range_analyzer class via class instance pointer.
5578 * vr-values.h (class vr_values): Privatize many methods and
5581 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5583 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
5584 * cfgexpand.c (expand_debug_expr): Remove handling for them.
5585 * expr.c (expand_expr_real_2): Likewise.
5586 * fold-const.c (const_unop): Likewise.
5587 * optabs-tree.c (optab_for_tree_code): Likewise.
5588 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
5589 * tree-inline.c (estimate_operator_cost): Likewise.
5590 * tree-pretty-print.c (dump_generic_node): Likewise.
5591 (op_code_prio): Likewise.
5592 (op_symbol_code): Likewise.
5593 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
5594 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
5595 * internal-fn.c (direct_internal_fn_optab): New function.
5596 (direct_internal_fn_array, direct_internal_fn_supported_p
5597 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
5598 * fold-const-call.c (fold_const_reduction): New function.
5599 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
5601 * tree-vect-loop.c: Include internal-fn.h.
5602 (reduction_code_for_scalar_code): Rename to...
5603 (reduction_fn_for_scalar_code): ...this and return an internal
5605 (vect_model_reduction_cost): Take an internal_fn rather than
5607 (vect_create_epilog_for_reduction): Likewise. Build calls rather
5609 (vectorizable_reduction): Use internal functions rather than tree
5610 codes for the reduction operation. Update calls to the functions
5612 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
5613 Use calls to internal functions rather than REDUC tree codes.
5614 * config/aarch64/aarch64-simd.md: Update comment accordingly.
5616 2017-11-22 Olivier Hainque <hainque@adacore.com>
5618 * config/vxworks.c (vxworks_override_options): Pick default
5619 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
5620 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
5621 DWARF_GNAT_ENCODINGS_DEFAULT.
5622 * config/vxworksae.h: Likewise.
5624 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
5626 PR tree-optimization/83104
5627 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
5630 2017-11-22 Marek Polacek <polacek@redhat.com>
5631 H.J. Lu <hongjiu.lu@intel.com>
5632 Jason Merrill <jason@redhat.com>
5637 * calls.c (initialize_argument_information): Call
5638 warn_parameter_passing_abi target hook.
5639 (store_one_arg): Use 0 for empty record size. Don't push 0 size
5640 argument onto stack.
5641 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
5642 * common.opt: Update -fabi-version description.
5643 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
5644 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
5646 (ix86_is_empty_record): New function.
5647 (ix86_warn_parameter_passing_abi): New function.
5648 (TARGET_EMPTY_RECORD_P): Redefine.
5649 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
5650 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
5651 * doc/tm.texi: Regenerated.
5652 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
5653 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
5654 * dwarf2out.c (get_ultimate_context): Move to tree.c.
5655 * explow.c (hard_function_value): Call arg_int_size_in_bytes
5656 instead of int_size_in_bytes.
5657 * expr.c (copy_blkmode_to_reg): Likewise.
5658 * function.c (aggregate_value_p): Return 0 for empty types.
5659 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
5661 (locate_and_pad_parm): Call arg size_in_bytes instead
5663 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
5664 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
5665 * target.def (empty_record_p, warn_parameter_passing_abi): New target
5667 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
5668 (std_gimplify_va_arg_expr): Skip empty records. Call
5669 arg_size_in_bytes instead size_in_bytes.
5670 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
5671 * tree-core.h (tree_type_common): Add empty_flag.
5672 (tree_decl_common): Update comments.
5673 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
5675 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5676 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
5678 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
5679 * tree.c (default_is_empty_type): New function.
5680 (default_is_empty_record): New function.
5681 (arg_int_size_in_bytes): New function.
5682 (arg_size_in_bytes): New function.
5683 (get_ultimate_context): New function.
5684 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
5685 TRANSLATION_UNIT_WARN_EMPTY_P.
5686 (default_is_empty_record, arg_int_size_in_bytes,
5687 arg_size_in_bytes, get_ultimate_context): Declare.
5689 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5691 * config/arm/arm.c (cmse_clear_registers): New function.
5692 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
5693 code by call to cmse_clear_registers.
5694 (cmse_nonsecure_entry_clear_before_return): Likewise.
5696 2017-11-22 Tamar Christina <tamar.christina@arm.com>
5698 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
5699 (vdot_s32, vdotq_s32): New.
5700 (vdot_lane_u32, vdotq_lane_u32): New.
5701 (vdot_lane_s32, vdotq_lane_s32): New.
5704 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5707 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
5708 for unsigned values with fewer HWIs than the precision.
5709 (test_overflow): New function.
5710 (wide_int_cc_tests): Call it.
5712 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
5714 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
5715 has the same mode class as Pmode.
5717 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5719 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
5720 padding_bits_to_clear_ptr.
5721 (cmse_nonsecure_entry_clear_before_return): Likewise.
5723 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
5725 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
5726 auto_sbitap instead of integer bitfield to control register needing
5729 2017-11-22 Jakub Jelinek <jakub@redhat.com>
5731 PR tree-optimization/83044
5732 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
5733 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
5734 that need the upper bound. Subtract offset from
5735 get_addr_base_and_unit_offset only if positive and subtract it
5736 before division by eltsize rather than after it.
5739 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
5740 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
5741 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
5745 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
5746 expand_binop with *mul_widen_optab, make sure at least one of the
5747 operands doesn't have VOIDmode.
5750 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
5752 PR rtl-optimization/82044
5753 PR tree-optimization/82042
5754 * dse.c (record_store): Check for overflow.
5755 (check_mem_read_rtx): Properly check for overflow if width == -1, call
5756 add_wild_read instead of clear_rhs_from_active_local_stores on
5757 overflow and log it into dump_file.
5759 2017-11-22 Richard Biener <rguenther@suse.de>
5761 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
5762 fake edges to exit when looking for a place to insert.
5763 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
5765 (insert_into_preds_of_block): Commit edge insertion immediately,
5766 assert that doesn't require new BBs.
5767 (fini_pre): Release expressions.
5768 (pass_pre::execute): Shuffle things around a bit, if the fn
5769 is too large do not compute AVAIL either as this is really the
5772 2017-11-22 Richard Biener <rguenther@suse.de>
5774 PR tree-optimization/83089
5775 * tree-if-conv.c (pass_if_conversion::execute): If anything
5776 changed reset SCEV and free the number of iteration estimates.
5778 2017-11-21 Martin Sebor <msebor@redhat.com>
5780 PR tree-optimization/82945
5781 * calls.h (warn_nonstring_bound): Remove unused function.
5783 2017-11-21 Martin Sebor <msebor@redhat.com>
5785 PR tree-optimization/82945
5786 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
5787 * calls.h (maybe_warn_nonstring_arg): Declare new function.
5788 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
5790 (initialize_argument_information): Call maybe_warn_nonstring_arg.
5791 * calls.h (get_attr_nonstring_decl): Declare new function.
5792 * doc/extend.texi (attribute nonstring): Update.
5793 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
5794 get_attr_nonstring_decl and handle it.
5795 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
5796 detection of nul-termination.
5797 (strlen_to_stridx): Change to a pointer.
5798 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
5799 (pass_strlen::execute): Same.
5801 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
5803 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
5804 for the new option -mprefer-vector-width=[none|128|256|512].
5805 * config/i386/i386.c (ix86_target_string): remove old style options
5806 -mprefer-avx256 and make -mprefer-avx128 as alias.
5807 (ix86_option_override_internal): Apply defaults for the
5808 -mprefer-vector-width=[128|256] option.
5809 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
5810 Implement macros to work with -mprefer-vector-width=.
5811 * config/i386/i386.opt: Implemented option
5812 -mprefer-vector-width=[none|128|256|512].
5813 * doc/invoke.texi: Documentation for
5814 -mprefer-vector-width=[none|128|256|512].
5816 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
5818 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
5820 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
5822 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
5824 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
5826 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
5828 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
5829 * expr.c (expand_expr_real_2): Likewise.
5830 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
5831 fold_binary_loc): Likewise.
5832 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
5833 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
5834 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
5835 MINUS_EXPR transformations.
5836 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
5837 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
5838 * tree-inline.c (estimate_operator_cost): Likewise.
5839 * tree-pretty-print.c (dump_generic_node, op_code_prio,
5840 op_symbol_code): Likewise.
5841 * tree-vect-stmts.c (vectorizable_operation): Likewise.
5842 * vr-values.c (extract_range_from_binary_expr): Likewise.
5843 * varasm.c (initializer_constant_valid_p_1): Likewise.
5844 * tree.def: New tree code POINTER_DIFF_EXPR.
5846 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
5848 * config/i386/i386.md (*bswap<mode>2_movbe): Add
5849 integer suffix to movbe mnemonic.
5850 (*bswaphi2_movbe): Ditto.
5851 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
5853 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5856 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
5857 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
5858 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
5860 2017-11-21 Martin Liska <mliska@suse.cz>
5862 * tree-inline.c (expand_call_inline): Remove not needed
5865 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
5866 Jakub Jelinek <jakub@redhat.com>
5869 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
5870 Remove static keyword from f variable.
5872 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5874 PR tree-optimization/83086
5875 * gimple-ssa-store-merging.c
5876 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
5877 rather than n.base_addr.
5879 2017-11-21 Martin Liska <mliska@suse.cz>
5881 PR rtl-optimization/82044
5882 PR tree-optimization/82042
5883 * dse.c (check_mem_read_rtx): Check for overflow.
5885 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5887 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
5890 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5892 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
5893 and its corresponding call arg location note.
5895 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
5896 Andrew Burgess <andrew.burgess@embecosm.com>
5898 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
5900 (arc_return_slot_offset): Likewise.
5901 (arc_eh_return_address_location): New declaration.
5902 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
5903 (MUST_SAVE_REGISTER): Add exception handler case.
5904 (MUST_SAVE_RETURN_ADDR): Likewise.
5905 (arc_frame_pointer_required): Likewise.
5906 (arc_frame_pointer_needed): New function.
5907 (arc_compute_frame_size): Changed.
5908 (arc_expand_prologue): Likewise.
5909 (arc_expand_epilogue): Likewise.
5910 (arc_initial_elimination_offset): Likewise.
5911 (arc_return_slot_offset): Delete.
5912 (arc_eh_return_address_location): New function.
5913 (arc_builtin_setjmp_frame_value): Likewise.
5914 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
5915 (EH_RETURN_STACKADJ_RTX): Define.
5916 (EH_RETURN_HANDLER_RTX): Likewise.
5917 * config/arc/arc.md (eh_return): Delete.
5919 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
5921 * print-tree.h (debug_vec_tree): Remove prototype.
5922 * gdbinit.in (pvt): Remove macro.
5924 2017-11-21 Jakub Jelinek <jakub@redhat.com>
5926 PR tree-optimization/83047
5927 * gimple-ssa-store-merging.c
5928 (imm_store_chain_info::output_merged_store): If the loads with the
5929 same vuse are in different basic blocks, for load_gsi pick a load
5930 location that is dominated by the other loads.
5933 * config/i386/i386.c (ix86_memmodel_check): Start
5934 -Winvalid-memory-model diagnostics with lowercase letter.
5937 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
5938 set in_cold_section_p to the partition containing loc_list->first.
5939 When seeing loc_list->last_before_switch node, update secname and
5940 perform range_across_switch second partition handling only after that.
5943 * run-rtl-passes.c: Include debug.h.
5944 (run_rtl_passes): Call debug_hooks->assembly_start.
5945 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
5949 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
5950 OPTAB_DIRECT in calls to expand_simple_binop.
5952 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5955 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
5957 2017-11-20 David Malcolm <dmalcolm@redhat.com>
5960 * spellcheck.h (best_match::blithely_get_best_candidate): New
5963 2017-11-20 Jakub Jelinek <jakub@redhat.com>
5965 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
5966 with lower case letter.
5968 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
5970 * config/i386/i386.md (bswaphi2): New expander.
5971 (*bswaphi2_movbe): New insn pattern.
5972 (bswaphi -> rorhi pepehole2): New peephole pattern.
5974 2017-11-20 Jeff Law <law@redhat.com>
5976 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
5977 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
5978 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
5979 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
5980 gimple-ssa-evrp-analyze.h.
5982 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
5983 set BB_VISITED here.
5984 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
5986 2017-11-20 Bin Cheng <bin.cheng@arm.com>
5988 * tree-predcom.c: Add general comment on Store-Store chains.
5989 (split_data_refs_to_components): Postpone clearing eliminate_store_p
5991 (get_chain_last_ref_at): Rename into...
5992 (get_chain_last_write_at): ...this.
5993 (get_chain_last_write_before_load): New function.
5994 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
5995 CT_STORE_STORE when write reference is added.
5996 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
5997 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
5998 (initialize_root_vars_store_elim_1): Ditto.
5999 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6000 definition is created.
6001 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6002 chain by replacing it with dominant stored value.
6004 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6006 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6008 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6010 * vr-values.c (extract_range_from_binary_expr): Use a full range
6013 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6015 * config/arm/arm.md (R4_REGNUM): Define constant.
6016 (nonsecure_call_internal): Remove r4 clobber.
6017 (nonsecure_call_value_internal): Likewise.
6018 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6019 clobber and resequence match_operands.
6020 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6021 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6022 (nonsecure_call_value_reg_thumb2): Likewise.
6024 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6026 PR tree-optimization/78821
6027 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6028 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6029 to the address of the base rather than the base itself.
6030 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6031 (find_bswap_or_nop_finalize): New function.
6032 (find_bswap_or_nop): Use it.
6033 (bswap_replace): Return a tree rather than bool, change first
6034 argument from gimple * to gimple_stmt_iterator, allow inserting
6035 into an empty sequence, allow ins_stmt to be NULL - then emit
6036 all stmts into gsi. Fix up MEM_REF address gimplification.
6037 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6038 (struct store_immediate_info): Add N and INS_STMT non-static
6040 (store_immediate_info::store_immediate_info): Initialize them
6041 from newly added ctor args.
6042 (merged_store_group::apply_stores): Formatting fixes. Sort by
6044 (stmts_may_clobber_ref_p): For stores call also
6045 refs_anti_dependent_p.
6046 (gather_bswap_load_refs): New function.
6047 (imm_store_chain_info::try_coalesce_bswap): New method.
6048 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6049 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6050 (imm_store_chain_info::output_merged_store): Fail if number of
6051 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6052 and NOP_EXPR rhs_code.
6053 (pass_store_merging::process_store): Compute n and ins_stmt, if
6054 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6055 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6057 (pass_store_merging::execute): Calculate dominators.
6059 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6060 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6061 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6062 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6063 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6064 class pass_optimize_bswap, bswap_replace,
6065 pass_optimize_bswap::execute): Moved to ...
6066 * gimple-ssa-store-merging.c: ... this file.
6067 Include optabs-tree.h.
6068 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6069 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6070 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6071 anonymous namespace, remove static keywords.
6072 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6073 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6075 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6078 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6080 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6082 * vec.h (debug_helper): New function.
6083 (DEFINE_DEBUG_VEC): New macro.
6084 * hash-set.h (debug_helper): New function.
6085 (DEFINE_DEBUG_HASH_SET): New macro.
6086 * cfg.c (debug_slim (edge)): New function.
6087 Call DEFINE_DEBUG_VEC for edges.
6088 Call DEFINE_DEBUG_HASH_SET for edges.
6089 * cfghooks.c (debug_slim (basic_block)): New function.
6090 Call DEFINE_DEBUG_VEC for basic blocks.
6091 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6092 * print-tree.c (debug_slim): New function to handle trees.
6093 Call DEFINE_DEBUG_VEC for trees.
6094 Call DEFINE_DEBUG_HASH_SET for trees.
6095 (debug (vec<tree, va_gc>) &): Remove.
6096 (debug (<vec<tree, va_gc>) *): Remove.
6097 * print-rtl.c (debug_slim): New function to handle const_rtx.
6098 Call DEFINE_DEBUG_VEC for rtx_def.
6099 Call DEFINE_DEBUG_VEC for rtx_insn.
6100 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6101 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6102 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6103 (debug (vec<rtx_insn *> *ptr): Remove.
6104 (debug_insn_vector): Remove.
6105 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6107 2017-11-20 Tom de Vries <tom@codesourcery.com>
6109 PR rtl-optimization/82020
6110 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6111 IF_THEN_ELSE condition.
6113 2017-11-19 Jeff Law <law@redhat.com>
6115 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6116 of degenerates resulting from ignoring an edge.
6118 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6121 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6123 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6126 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6127 for uninitialied counts.
6129 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6132 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6134 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6137 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6139 2017-11-19 Tom de Vries <tom@codesourcery.com>
6141 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6144 2017-11-19 Tom de Vries <tom@codesourcery.com>
6146 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6149 2017-11-19 Tom de Vries <tom@codesourcery.com>
6151 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6154 2017-11-19 Tom de Vries <tom@codesourcery.com>
6156 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6158 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6160 2017-11-19 Tom de Vries <tom@codesourcery.com>
6162 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6164 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6165 semicolon after MCORE_EXPORT_NAME call.
6167 2017-11-19 Tom de Vries <tom@codesourcery.com>
6170 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6171 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6173 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6175 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6176 (cgraph_edge::clone): Cleanup updating of profile.
6177 * ipa-cp.c (update_profiling_info): Likewise.
6178 * ipa-inline-transform.c (inline_transform): Likewise.
6179 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6180 * ipa-split.c (execute_split_functions): Do not split when function
6182 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6183 * profile-count.c (profile_count::dump): Add global0.
6184 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6186 (profile_count::to_sreal_scale): Likewise.
6187 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6188 (profile_count::combine_with_ipa_count): New function.
6189 * profile-count.h (profile_guessed_global0adjusted): New.
6190 (profile_count::adjusted_zero): New.
6191 (profile_count::global0adjusted): New.
6192 (profile_count::combine_with_ipa_count): New.
6193 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6194 correct profile of return block of split functions.
6195 (copy_cfg_body): Remove unused profile_count.
6196 (copy_body): Likewise.
6197 (expand_call_inline): Update.
6198 (tree_function_versioning): Update.
6200 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6202 * hash-set.h (hash_set::empty): New.
6203 * tree-ssa-threadbackward.h: Delete.
6204 * tree-ssa-threadbackward.c (class thread_jumps): New.
6205 Move max_threaded_paths into class.
6206 (fsm_find_thread_path): Remove arguments that are now in class.
6207 (profitable_jump_thread_path): Rename to...
6208 (thread_jumps::profitable_jump_thread_path): ...this.
6209 (convert_and_register_jump_thread_path): Rename to...
6210 (thread_jumps::convert_and_register_current_path): ...this.
6211 (check_subpath_and_update_thread_path): Rename to...
6212 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6213 (register_jump_thread_path_if_profitable): Rename to...
6214 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6215 (handle_phi): Rename to...
6216 (thread_jumps::handle_phi): ...this.
6217 (handle_assignment): Rename to...
6218 (thread_jumps::handle_assignment): ...this.
6219 (fsm_find_control_statement_thread_paths): Rename to...
6220 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6221 (find_jump_threads_backwards): Rename to...
6222 (thread_jumps::find_jump_threads_backwards): ...this.
6223 Initialize path local data.
6224 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6225 from within thread_jumps class.
6226 (pass_early_thread_jumps::execute): Same.
6228 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6230 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6232 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6234 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6235 to canonical location.
6237 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6240 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6242 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6244 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6246 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6247 rcrt1.o%s/grcrt1.o%s for -static-pie.
6249 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6251 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6252 ix86_shift_rotate_cost): Break out from ...
6253 (ix86_rtx_costs): ... here.
6254 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6257 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6259 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6260 when entry block was promoted unlikely.
6261 (estimate_bb_frequencies): Increase frequency scale.
6262 * profile-count.h (profile_count): Export precision info.
6264 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6266 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6267 disturbin profile of entry block.
6269 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6271 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6274 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6276 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6277 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6279 2017-11-17 Jeff Law <law@redhat.com>
6281 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6282 from evrp_dom_walker class. Various methods moved into new class.
6283 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6284 (evrp_range_analyzer::enter): New method.
6285 (evrp_range_analyzer::leave): New method.
6286 (evrp_dom_walker): Remove delegators no longer needed by this class.
6287 Replace vr_values data member with evrp_range_analyzer
6289 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6290 method extracted from evrp_dom_walker::before_dom_children.
6291 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6292 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6294 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6295 Add private copy constructor and move assignment operators.
6296 Privatize methods and class data where trivially possible.
6297 (evrp_dom_walker::cleanup): New function, extracted from
6298 execute_early_vrp. Simplify access to class data.
6300 * vr-values.h (get_output_for_vrp): Prototype.
6301 * vr-values.c (get_output_for_vrp): New function extracted from
6302 vrp_visit_assignment_or_call and extract_range_from_stmt.
6303 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6305 2017-11-17 Luis Machado <luis.machado@linaro.org>
6307 * config/aarch64/aarch64.c
6308 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6309 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6310 tune_params::AUTOPREFETCHER_WEAK.
6312 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6315 * config/arm/arm.c (arm_valid_target_attribute_rec):
6316 Parse "arch=" and "+<ext>".
6317 (arm_valid_target_attribute_tree): Re-init global options.
6318 (arm_option_override): Make non-static.
6319 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6320 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6321 (__ARM_FEATURE_CRC32): Support undef.
6322 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6323 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6325 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6327 * gdbinit.in (break-on-diagnostic): New command.
6329 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6331 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6332 return to indirect jump for EH return if control-flow
6333 protection is enabled. Change explicit 'false' argument in
6334 pro_epilogue_adjust_stack with a value of
6336 * config/i386/i386.md (simple_return_indirect_internal):
6337 Remove SImode restriction to support 64-bit.
6339 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6341 * combine.c (added_notes_insn): New.
6342 (try_combine): Handle added_notes_insn like added_links_insn.
6343 Rewrite return value code.
6344 (distribute_notes): Set added_notes_insn to the earliest insn we added
6347 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6349 PR rtl-optimization/82621
6350 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6351 dest of one of those SETs is unused.
6353 2017-11-17 Richard Biener <rguenther@suse.de>
6356 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6357 * tree-pretty-print.c (dump_generic_node): Handle
6358 annot_expr_parallel_kind.
6359 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6360 * gimplify.c (gimple_boolify): Likewise.
6362 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6364 * config.gcc (extra_headers): Add cet.h for x86 targets.
6365 * config/i386/cet.h: New file.
6366 * doc/install.texi: Add --enable-cet/--disable-cet.
6368 2017-11-17 Richard Biener <rguenther@suse.de>
6370 PR tree-optimization/83017
6371 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6372 (gen_parallel_loop): Properly count iterations.
6373 (parallelize_loops): Handle loop->can_be_parallel independent
6374 of flag_loop_parallelize_all. Make static profitability test match
6376 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6377 * invoke.texi (parloops-min-per-thread): Document.
6379 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6381 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6382 upstreaming review comments.
6384 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6386 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6387 with fast unaligned access.
6388 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6390 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6392 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
6393 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
6395 2017-11-17 Richard Biener <rguenther@suse.de>
6397 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
6398 folding of references.
6400 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
6403 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
6404 of replacing call to strncmp with corresponding call to strcmp when
6407 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
6409 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
6410 option prefer-avx256 for skylake-avx512 configuration.
6411 * config/i386/i386.c (ix86_option_override_internal): Ditto.
6412 (get_builtin_code_for_version): Ditto.
6414 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6415 Monk Chiang <sh.chiang04@gmail.com>
6417 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
6418 (FIXED_REGISTERS): Reserve more register numbers.
6419 (CALL_USED_REGISTERS): Likewise.
6420 (REG_ALLOC_ORDER): Likewise.
6421 (REG_CLASS_CONTENTS): Likewise.
6422 (REGISTER_NAMES): Likewise.
6424 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
6425 Kito Cheng <kito.cheng@gmail.com>
6427 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
6429 * config/nds32/iterators.md: Add vector mode iterators and attributes.
6431 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
6433 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
6434 parameter list for vec_splats.
6436 2017-11-16 Joseph Myers <joseph@codesourcery.com>
6438 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
6440 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
6442 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
6445 * doc/install.texi: Document incompatibility of Perl >=5.6.26
6446 with the required version of automake 1.11.6.
6448 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
6450 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
6452 (DU_C2_3_power9): Correct reservation combinations.
6453 (FP_DIV_power9, VEC_DIV_power9): New.
6454 (power9-alu): Split out rotate/shift...
6455 (power9-rot): ...to here, correct dispatch resource.
6456 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
6458 (power9-fp): Correct latency.
6459 (power9-sdiv): Add div/sqrt resource.
6460 (power9-ddiv): Correct latency, add div/sqrt resource.
6461 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
6462 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
6464 (power9-qpdiv, power9-qpmul): Adjust resource usage.
6466 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
6468 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
6469 switch statement mapping KF built-ins to TF built-ins if we don't
6470 have the proper ISA 3.0 assembler support.
6472 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6474 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
6475 (lower_emutls_function_body): Do not compute it.
6477 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6479 * ipa-split.c (split_bb_info): Turn time to sreal.
6480 (split_point): Likewise.
6481 (dump_split_point): Likewise.
6482 (fine_split_points): Likewise.
6483 (execute_split_functions): Only zero split_bbs; turn time to sreals.
6485 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6487 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
6490 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6492 * predict.c (combine_predictions_for_bb): Preserve zero predicted
6494 (expensive_function_p): Remove useless assert.
6495 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
6497 2017-11-16 Martin Sebor <msebor@redhat.com>
6499 PR tree-optimization/82588
6500 PR tree-optimization/82583
6501 * tree-vrp.c (check_array_ref): Handle flexible array members,
6502 string literals, and inner indices.
6503 (search_for_addr_array): Add detail to diagnostics.
6505 2017-11-16 Nathan Sidwell <nathan@acm.org>
6509 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
6510 (SET_DECL_ASSEMBLER_NAME): Forward to
6511 overwrite_decl_assembler_name.
6512 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
6513 (overwrite_decl_assembler_name): Declare.
6514 * tree.c (overwrite_decl_assembler_name): New.
6515 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
6516 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
6517 (LANG_HOOKS_INITIALIZER): Add it.
6518 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
6519 * langhooks.c (lhd_set_decl_assembler_name): Use
6520 SET_DECL_ASSEMBLER_NAME.
6521 (lhd_overwrite_decl_assembler_name): Default implementation.
6523 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6524 Jackson Woodruff <jackson.woodruff@arm.com>
6526 PR tree-optimization/71026
6527 * match.pd: Canonicalize constant multiplies in division.
6529 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
6531 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
6532 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
6533 Remove OPT_fomit_frame_pointer entry.
6534 * common/config/arc/arc-common.c: Likewise.
6535 * common/config/arm/arm-common.c: Likewise.
6536 * common/config/avr/avr-common.c: Likewise.
6537 * common/config/c6x/c6x-common.c: Likewise.
6538 * common/config/cr16/cr16-common.c: Likewise.
6539 * common/config/cris/cris-common.c: Likewise.
6540 * common/config/epiphany/epiphany-common.c: Likewise.
6541 * common/config/fr30/fr30-common.c: Likewise.
6542 * common/config/frv/frv-common.c: Likewise.
6543 * common/config/ia64/ia64-common.c: Likewise.
6544 * common/config/iq2000/iq2000-common.c: Likewise.
6545 * common/config/lm32/lm32-common.c: Likewise.
6546 * common/config/m32r/m32r-common.c: Likewise.
6547 * common/config/mcore/mcore-common.c: Likewise.
6548 * common/config/microblaze/microblaze-common.c: Likewise.
6549 * common/config/mips/mips-common.c: Likewise.
6550 * common/config/mmix/mmix-common.c: Likewise.
6551 * common/config/mn10300/mn10300-common.c: Likewise.
6552 * common/config/nios2/nios2-common.c: Likewise.
6553 * common/config/pa/pa-common.c: Likewise.
6554 * common/config/pdp11/pdp11-common.c: Likewise.
6555 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
6556 * common/config/riscv/riscv-common.c: Likewise.
6557 * common/config/rs6000/rs6000-common.c: Likewise.
6558 * common/config/rx/rx-common.c: Likewise.
6559 * common/config/s390/s390-common.c: Likewise.
6560 * common/config/sh/sh-common.c: Likewise.
6561 * common/config/sparc/sparc-common.c: Likewise.
6562 * common/config/tilegx/tilegx-common.c: Likewise.
6563 * common/config/tilepro/tilepro-common.c: Likewise.
6564 * common/config/v850/v850-common.c: Likewise.
6565 * common/config/visium/visium-common.c: Likewise.
6566 * common/config/xstormy16/xstormy16-common.c: Likewise.
6567 * common/config/xtensa/xtensa-common.c: Likewise.
6568 * invoke.texi (-fomit-frame-pointer): Update documentation.
6570 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
6572 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
6574 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6576 * cfg.c (scale_bbs_frequencies_int,
6577 cale_bbs_frequencies_gcov_type): Remove.
6578 * cfg.h (scale_bbs_frequencies_int,
6579 cale_bbs_frequencies_gcov_type): Remove.
6581 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6583 * tree-ssa-loop-manip.c
6584 (scale_dominated_blocks_in_loop): Update to profile counts.
6585 (tree_transform_and_unroll_loop): Likewise.
6587 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6589 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
6590 scale_bbs_frequencies_int.
6592 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6594 * final.c (compute_alignments): Use counts rather than frequencies.
6596 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6598 * cfgloopanal.c: Include sreal.h
6599 (average_num_loop_insns): Use counts and sreal for accounting.
6601 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
6603 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
6606 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6607 Kito Cheng <kito.cheng@gmail.com>
6609 * config/nds32/constraints.md: Provide more constraints.
6610 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
6611 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
6612 support constraints usage.
6614 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6616 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
6618 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
6619 Kito Cheng <kito.cheng@gmail.com>
6621 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
6622 * config/nds32/nds32.opt: Refine the layout.
6623 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
6624 TARGET_EXT_STRING): Support new options.
6625 * config/nds32/nds32.h: Likewise.
6626 * config/nds32/nds32.md: Likewise.
6627 * config/nds32/nds32-predicates.c: Likewise.
6628 * config/nds32/constraints.md: Likewise.
6629 * common/config/nds32/nds32-common.c: Likewise.
6631 2017-11-16 Julia Koval <julia.koval@intel.com>
6634 * config/i386/gfniintrin.h: Add sse check.
6635 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
6637 2017-11-16 Julia Koval <julia.koval@intel.com>
6639 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
6640 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
6641 (ix86_handle_option): Handle -mavx512vbmi2.
6642 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
6643 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
6644 * config/i386/i386-c.c (__AVX512VBMI2__): New.
6645 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
6646 (ix86_valid_target_attribute_inner_p): Ditto.
6647 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
6648 * config/i386/i386.opt (mavx512vbmi2): New option.
6649 * doc/invoke.texi: Add new option.
6651 2017-11-16 Julia Koval <julia.koval@intel.com>
6653 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
6654 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
6655 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
6656 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
6657 _mm512_gf2p8mul_epi8): New intrinsics.
6658 * config/i386/i386-builtin-types.def
6659 (V64QI_FTYPE_V64QI_V64QI): New type.
6660 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
6661 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
6662 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
6663 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
6664 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
6665 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
6667 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
6669 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
6670 explicitly as a stream of bytes.
6672 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6674 * config/rs6000/altivec.h (vec_xst_be): New #define.
6675 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
6676 and externalize from *altivec_vperm_<mode>_internal.
6677 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
6679 (XL_BE_V8HI): Likewise.
6680 (XL_BE_V4SI): Likewise.
6681 (XL_BE_V4SI): Likewise.
6682 (XL_BE_V2DI): Likewise.
6683 (XL_BE_V4SF): Likewise.
6684 (XL_BE_V2DF): Likewise.
6685 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
6686 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
6687 all array entries with these keys: VSX_BUILTIN_VEC_XL,
6688 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
6689 VSX_BUILTIN_VEC_XST_BE.
6690 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
6691 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
6693 (altivec_init_builtins): Replace conditional calls to def_builtin
6694 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
6695 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
6696 with unconditional calls. Remove calls to def_builtin for
6697 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
6698 __builtin_vec_xst_be.
6699 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
6700 to define_expand, and add alternate RTL generation for P8.
6701 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
6702 vsx_ld_elemrev_v8hi.
6703 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
6704 add alternate RTL generation for P8.
6705 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
6706 vsx_ld_elemrev_v16qi.
6707 (vsx_st_elemrev_v8hi): Convert define_insn
6708 to define_expand, and add alternate RTL generation for P8.
6709 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
6710 vsx_st_elemrev_v8hi.
6711 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
6712 add alternate RTL generation for P8.
6713 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
6714 vsx_st_elemrev_v16qi.
6716 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
6719 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
6720 TARGET_AVX512ER check.
6721 (ix86_option_override_internal): Set MASK_VZEROUPPER if
6722 neither -mzeroupper nor -mno-zeroupper is used and
6723 TARGET_EMIT_VZEROUPPER is set.
6724 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
6725 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
6727 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6729 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6730 folding of vector compares.
6731 (fold_build_vec_cmp): New helper function.
6732 (fold_compare_helper): New helper function.
6733 (builtin_function_type): Add compare builtins to the list of functions
6734 having unsigned arguments. Cosmetic updates to comment indentation.
6735 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
6736 the not+eq combination.
6738 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6740 PR tree-optimization/82726
6741 PR tree-optimization/70754
6742 * tree-predcom.c (order_drefs_by_pos): New function.
6743 (combine_chains): Move code setting has_max_use_after to...
6744 (try_combine_chains): ...here. New parameter. Sort combined chains
6745 according to position information.
6746 (tree_predictive_commoning_loop): Update call to above function.
6747 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
6749 2017-11-15 Bin Cheng <bin.cheng@arm.com>
6751 PR tree-optimization/82726
6753 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6755 PR tree-optimization/70754
6756 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6757 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6758 combined stmt before it if not NULL.
6759 (combine_chains): Process refs reversely and compute dominance point
6763 2017-02-23 Bin Cheng <bin.cheng@arm.com>
6765 PR tree-optimization/79663
6766 * tree-predcom.c (combine_chains): Process refs in reverse order
6767 only for ZERO length chains, and add explaining comment.
6769 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6771 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
6772 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
6773 * doc/invoke.texi (ARM Options): Add armv8.3-a.
6775 2017-11-15 Tamar Christina <tamar.christina@arm.com>
6777 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
6779 2017-11-15 Martin Liska <mliska@suse.cz>
6781 * tree-cfg.c (pass_warn_function_return::execute):
6782 Compare warn_return_type for greater than zero.
6784 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
6788 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
6789 to return true on Xeon and not on Xeon Phi.
6790 (ix86_check_avx256_register): Changed to ...
6791 (ix86_check_avx_upper_register): ... this. Add extra check for
6792 VALID_AVX512F_REG_OR_XI_MODE.
6793 (ix86_avx_u128_mode_needed): Changed
6794 ix86_check_avx256_register to ix86_check_avx_upper_register.
6795 (ix86_check_avx256_stores): Changed to ...
6796 (ix86_check_avx_upper_stores): ... this. Changed
6797 ix86_check_avx256_register to ix86_check_avx_upper_register.
6798 (ix86_avx_u128_mode_after): Changed
6799 avx_reg256_found to avx_upper_reg_found. Changed
6800 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
6801 (ix86_avx_u128_mode_entry): Changed
6802 ix86_check_avx256_register to ix86_check_avx_upper_register.
6803 (ix86_avx_u128_mode_exit): Ditto.
6804 * config/i386/i386.h: (host_detect_local_cpu): New define.
6806 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
6808 * config/arm/xgene1.md (xgene1): Split into automatons
6809 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
6810 (xgene1_f_load): Adjust reservations and/or types.
6811 (xgene1_f_store): Likewise.
6812 (xgene1_load_pair): Likewise.
6813 (xgene1_store_pair): Likewise.
6814 (xgene1_fp_load1): Likewise.
6815 (xgene1_load1): Likewise.
6816 (xgene1_store1): Likewise.
6817 (xgene1_move): Likewise.
6818 (xgene1_alu): Likewise.
6819 (xgene1_simd): Likewise.
6820 (xgene1_bfm): Likewise.
6821 (xgene1_neon_load1): Likewise.
6822 (xgene1_neon_store1): Likewise.
6823 (xgene1_neon_logic): Likewise.
6824 (xgene1_neon_st1): Likewise.
6825 (xgene1_neon_ld1r): Likewise.
6826 (xgene1_alu_cond): Added.
6827 (xgene1_shift_reg): Likwise.
6828 (xgene1_bfx): Likewise.
6829 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
6831 2017-11-15 Jakub Jelinek <jakub@redhat.com>
6834 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
6836 (can_widen_mult_without_libcall): New function.
6837 (expand_mul_overflow): If only checking unsigned mul overflow,
6838 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
6839 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
6840 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
6841 (expand_DIVMOD): Formatting fix.
6842 * expmed.h (expand_mult): Add NO_LIBCALL argument.
6843 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
6844 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
6846 PR tree-optimization/82977
6847 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
6848 constructed temporary to strlen_to_stridx.put.
6850 2017-11-15 Martin Liska <mliska@suse.cz>
6852 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
6853 * configure: Regenerate.
6855 2017-11-15 Martin Liska <mliska@suse.cz>
6858 * config/sh/sh-mem.cc: Use proper probability for
6861 2017-11-14 Jeff Law <law@redhat.com>
6863 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
6864 the red zone for stack_clash_protection_final_dynamic_probe targets
6865 when the total dynamic stack size is zero bytes.
6867 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
6868 blocks is post order.
6870 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
6872 * dumpfile.h (TDF_COMPARE_DEBUG): New.
6873 * final.c (rest_of_clean_state): Set it for the
6874 -fcompare-debug dump.
6875 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
6876 class when TDF_COMPARE_DEBUG is set.
6878 * dwarf2out.c (gen_producer_string): Discard
6881 2017-11-15 Joseph Myers <joseph@codesourcery.com>
6884 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
6885 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
6886 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
6887 (__TGMATH_REAL_2_3): Remove macros.
6888 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
6889 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
6891 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
6893 (remquo): Define using __TGMATH_REAL_3.
6895 2017-11-14 Jeff Law <law@redhat.com>
6897 * vr-values.c: New file with contents extracted from tree-vrp.c.
6898 * Makefile.in (OBJS): Add vr-values.o
6899 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
6900 (set_value_range, set_and_canonicalize_value_range): Likewise.
6901 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
6902 (value_range_constant_singleton, symbolic_range_p): Likewise.
6903 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
6904 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
6905 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
6906 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
6907 (find_case_label_range, find_case_label_index): Likewise.
6908 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
6909 (range_int_cst_singleton_p, value_inside_range): Likewise.
6910 (get_single_symbol): Likewise.
6911 (switch_update): Move structure definition here.
6912 (to_remove_edges, to_update_switch_stmts): Provide externs.
6913 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
6914 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
6915 (vrp_val_is_min, set_value_range): Likewise.
6916 (set_and_canonicalize_value_range, copy_value_range): Likewise.
6917 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
6918 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
6919 (range_is_nonnull, range_int_cst_p): Likewwise.
6920 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
6921 (get_single_symbol, operand_less_p): Likewise
6922 (compare_values_warnv, compare_values): Likewise.
6923 (value_inside_range, value_range_constant_singleton): Likewise.
6924 (zero_nonzero_bitgs_from_vr): Likewise.
6925 (extract_range_from_binary_expr_1): Likewise.
6926 (overflow_comparison_p): Likewise.
6927 (to_remove_edges, to_update_switch_stmts): Likewise.
6928 (find_case_label-index, find_case_label_range): Likewise.
6929 (switch_update, set_value_range_to_nonnegative): Remove.
6930 (set_value_range_to_truthvalue): Likewise.
6931 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
6932 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
6933 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
6934 (find_case_label_ranges, test_for_singularity): Likewise.
6935 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
6936 (x_vr_values): Move to its remaining use site.
6938 2017-11-10 Jeff Law <law@redhat.com>
6940 * vr-values.h (VR_INITIALIZER): Move #define here.
6941 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
6942 * Makefile.in (OBJS): Add tree-evrp.o
6943 * tree-vrp.h (assert_info): Move structure definition here.
6944 (set_value_range_to_varying): Prototype.
6945 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6946 (infer_value_range, register_edge_assert_for): Likewise.
6947 (stmt_interesting_for_vrp): Likewise.
6948 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
6949 (set_value_range_to_varying): No longer static.
6950 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
6951 (infer_value_range, register_edge_assert_for): Likewise.
6953 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6955 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
6956 generate the XXBRD instruction.
6958 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
6959 (rs6000_builtin_type_compatible): Treat _Float128 and long double
6960 as being compatible if -mabi=ieeelongdouble.
6961 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
6962 to setup float128 built-ins with hardware support.
6963 (BU_FLOAT128_HW_2): Likewise.
6964 (BU_FLOAT128_HW_3): Likewise.
6965 (BU_FLOAT128_HW_VSX_1): Likewise.
6966 (BU_FLOAT128_HW_VSX_2): Likewise.
6967 (scalar_extract_expq): Change float128 built-in functions to
6968 accommodate having both KFmode and TFmode functions. Use the
6969 KFmode variant as the default.
6970 (scalar_extract_sigq): Likewise.
6971 (scalar_test_neg_qp): Likewise.
6972 (scalar_insert_exp_q): Likewise.
6973 (scalar_insert_exp_qp): Likewise.
6974 (scalar_test_data_class_qp): Likewise.
6975 (sqrtf128_round_to_odd): Delete processing the round to odd
6976 built-in functions as special built-in functions, and define them
6977 as float128 built-ins. Use the KFmode variant as the default.
6978 (truncf128_round_to_odd): Likewise.
6979 (addf128_round_to_odd): Likewise.
6980 (subf128_round_to_odd): Likewise.
6981 (mulf128_round_to_odd): Likewise.
6982 (divf128_round_to_odd): Likewise.
6983 (fmaf128_round_to_odd): Likewise.
6984 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6985 support for KFmode and TFmode xststdcqp calls.
6986 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
6987 point, switch the built-in handlers for the get/set float128
6988 exponent, get float128 mantissa, float128 test built-ins, and the
6989 float128 round to odd built-in functions. Eliminate creating the
6990 float128 round to odd built-in functions as special built-ins.
6991 (rs6000_init_builtins): Eliminate special creation of the float128
6992 round to odd built-in functions.
6993 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
6994 function insns to support both TFmode and KFmode variants.
6995 (xsxsigqp_<mode>): Likewise.
6996 (xsiexpqpf_<mode>): Likewise.
6997 (xsiexpqp_<mode>): Likewise.
6998 (xststdcqp_<mode>): Likewise.
6999 (xststdcnegqp_<mode>): Likewise.
7000 (xststdcqp_<mode>): Likewise.
7002 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7004 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7005 unused path_in_freq_ptr parameter.
7006 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7008 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7010 * ipa-inline.c (edge_badness): Dump sreal frequency.
7011 (compute_inlined_call_time): Match natural implementaiton ...
7012 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7013 forgotten division by CGRAPH_FREQ_BASE.
7015 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7017 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7018 Solaris 11. Update comment.
7019 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7021 * config/sol2.h (STARTFILE_SPEC): Likewise.
7022 * configure: Regenerate.
7024 2017-11-14 Carl Love <cel@us.ibm.com>
7026 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7027 le_ and be_ prefixes to swap* variables. Remove
7028 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7031 2017-11-14 Jason Merrill <jason@redhat.com>
7033 Support GTY((cache)) on hash_map.
7034 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7035 (ggc_cache_remove): Override it instead of ggc_mx.
7036 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7037 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7038 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7039 (simple_cache_map_traits): Override maybe_mx.
7040 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7041 (hash_map): Friend gt_cleare_cache.
7042 (gt_cleare_cache): New.
7043 * tree.h (tree_cache_traits): New hash_map traits class.
7044 (tree_cache_map): New typedef.
7046 2017-11-14 Richard Biener <rguenther@suse.de>
7048 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7049 paramter and handling.
7050 (cleanup_control_flow_bb): Likewise.
7051 (cleanup_control_flow_pre): New helper performing a DFS walk
7052 to call cleanup_control_flow_bb in PRE order.
7053 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7054 via cleanup_control_flow_pre.
7056 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7058 * config/aarch64/aarch64-simd.md
7059 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7060 (*aarch64_simd_bsl<mode>_alt): Likewise.
7061 (aarch64_simd_bsldi_internal): New.
7062 (aarch64_simd_bsldi_alt): Likewise.
7064 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7066 * tracer.c (better_p): Do not compare frequencies.
7067 * reg-stack.c (better_edge): Likewise.
7068 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7071 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7073 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7074 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7075 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7076 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7077 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7078 * predict.c (maybe_hot_frequency_p): Inline to ...
7079 (maybe_hot_count_p): ... here; rewrite to counts.
7080 (counts_to_freqs): Rename to ...
7081 (update_max_bb_count): ... this one.
7082 (expensive_function_p): Use counts.
7083 (estimate_bb_frequencies): Update.
7084 (rebuild_frequencies): Update.
7085 * predict.h (counts_to_freqs): Rename to ...
7086 (update_max_bb_count): ... this one.
7087 * profile.c (compute_branch_probabilities): Add debug info
7088 * tree-inline.c (expand_call_inline): Update debug info.
7089 (optimize_inline_calls): Use update_max_bb_count..
7090 (tree_function_versioning): Use update_max_bb_count..
7091 * value-prof.c (gimple_value_profile_transformations):
7092 Do not use update_max_bb_count.
7094 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7096 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7097 always use frequencies.
7099 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7101 * bb-reorder.c: Remove frequencies from comments.
7102 (better_edge_p): Use profile counts.
7103 (find_traces): Dump profile counts.
7104 (rotate_loop): Use profile counts.
7105 (find_traces_1_round): Likewise.
7106 (connect_better_edge_p): Use counts instead of probabilities for
7108 (copy_bb_p): Drop early check for non-0 frequency.
7109 (sanitize_hot_paths): Update comments.
7111 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7113 * ipa-split.c (struct split_point): Add count.
7114 (consider_split): Do not compute incoming frequency; compute incoming
7115 count and store it to split_point.
7116 (split_function): Set count of the call to split part correctly.
7118 2017-11-13 Carl Love <cel@us.ibm.com>
7120 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7122 2017-11-13 Tom Tromey <tom@tromey.com>
7124 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7126 2017-11-13 Carl Love <cel@us.ibm.com>
7128 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7129 Add support for builtins:
7130 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7131 vector {un,}signed {char,int,short},
7132 vector {un,}signed {char,int,short}) arguments.
7133 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7134 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7135 Add BU_P9V_AV_2 expansions for the builtins.
7136 * config/rs6000/altivec.h (vec_first_match_index,
7137 vec_first_mismatch_index, vec_first_match_or_eos_index,
7138 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7139 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7140 new extern declaration.
7141 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7142 * config/rs6000/vsx.md (first_match_index_<mode>,
7143 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7144 first_mismatch_or_eos_index_<mode>): Add define expand.
7145 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7146 * doc/extend.texi: Update the built-in documenation file for the new
7149 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7151 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7152 into the min/max operations for _Float<N> and _Float<N>X types.
7154 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7157 * dwarf2out.c (do_eh_frame): New static variable.
7158 (dwarf2out_begin_prologue): Set it.
7159 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7161 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7163 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7165 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7167 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7170 * debug.h (dwarf2out_do_eh_frame): Declare.
7171 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7172 (dwarf2out_do_frame): Use it.
7173 (dwarf2out_do_cfi_asm): Likewise.
7174 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7175 (dwarf2out_assembly_start): Likewise.
7176 (dwarf2out_begin_prologue): Fix comment.
7177 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7178 if the target needs either debug or unwind DWARF2 info.
7179 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7180 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7181 -ftrapping-math, -ftrapv and -fwrapv.
7183 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7185 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7186 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7187 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7188 (estimate_edge_size_and_time): Likewise.
7189 (ipa_merge_fn_summary_after_inlining): Likewise.
7190 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7191 (compute_uninlined_call_time): Use sreal_frequency.
7192 (compute_inlined_call_time): Likewise.
7193 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7194 * profile-count.c: Include sreal.h.
7195 (profile_count::to_sreal_scale): New.
7196 * profile-count.h: Forward declare sreal.
7197 (profile_count::to_sreal_scale): Declare.
7199 2017-11-13 Nathan Sidwell <nathan@acm.org>
7201 * diagnostic.c (maybe_line_and_column): New.
7202 (diagnostic_get_location_text): Use it.
7203 (diagnostic_report_current_module): Likewise.
7204 (test_diagnostic_get_location_text): Add tests.
7206 2017-11-13 Luis Machado <luis.machado@linaro.org>
7208 * doc/md.texi (Specifying processor pipeline description): Fix
7209 incorrect latency for the div instruction example.
7211 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7213 PR tree-optimization/78821
7214 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7215 that bit_not_p is the same.
7216 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7217 (split_group): Count precisely bit_not_p bits in each statement.
7218 (invert_op): New function.
7219 (imm_store_chain_info::output_merged_store): Use invert_op to
7220 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7221 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7223 2017-11-13 Martin Liska <mliska@suse.cz>
7225 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7226 (struct source_info): Likewise.
7227 (add_branch_counts): Likewise.
7228 (add_line_counts): Likewise.
7229 (function_summary): Likewise.
7230 (output_intermediate_line): Likewise.
7231 (generate_results): Likewise.
7233 2017-11-13 Martin Liska <mliska@suse.cz>
7235 * gcov.c (struct block_info): Remove typedef for block_t.
7236 (struct line_info): Likewise.
7237 (line_info::has_block): Likewise.
7238 (EXIT_BLOCK): Likewise.
7239 (unblock): Likewise.
7240 (circuit): Likewise.
7241 (get_cycles_count): Likewise.
7242 (process_file): Likewise.
7243 (read_graph_file): Likewise.
7244 (solve_flow_graph): Likewise.
7245 (find_exception_blocks): Likewise.
7246 (add_line_counts): Likewise.
7247 (accumulate_line_info): Likewise.
7248 (output_line_details): Likewise.
7250 2017-11-13 Martin Liska <mliska@suse.cz>
7252 * gcov.c (struct arc_info): Remove typedef for arc_t.
7253 (struct line_info): Likewise.
7254 (add_branch_counts): Likewise.
7255 (output_branch_count): Likewise.
7256 (function_info::~function_info): Likewise.
7257 (circuit): Likewise.
7258 (output_intermediate_line): Likewise.
7259 (read_graph_file): Likewise.
7260 (solve_flow_graph): Likewise.
7261 (find_exception_blocks): Likewise.
7262 (add_line_counts): Likewise.
7263 (accumulate_line_info): Likewise.
7264 (output_line_details): Likewise.
7265 (output_function_details): Likewise.
7267 2017-11-13 Martin Liska <mliska@suse.cz>
7269 * gcov.c (struct function_info): Remove typedef for function_t.
7270 (struct source_info): Likewise.
7271 (source_info::get_functions_at_location): Likewise.
7272 (solve_flow_graph): Likewise.
7273 (find_exception_blocks): Likewise.
7274 (add_line_counts): Likewise.
7275 (output_intermediate_file): Likewise.
7276 (process_file): Likewise.
7277 (generate_results): Likewise.
7278 (release_structures): Likewise.
7279 (read_graph_file): Likewise.
7280 (read_count_file): Likewise.
7281 (accumulate_line_counts): Likewise.
7282 (output_lines): Likewise.
7284 2017-11-13 Martin Liska <mliska@suse.cz>
7286 * gcov.c (function_info::function_info): Remove num_counts
7287 and add vector<gcov_type>.
7288 (function_info::~function_info): Use the vector.
7289 (process_file): Likewise.
7290 (read_graph_file): Likewise.
7291 (read_count_file): Likewise.
7292 (solve_flow_graph): Likewise.
7294 2017-11-13 Martin Liska <mliska@suse.cz>
7296 * gcov.c (function_info::is_artificial): New function.
7297 (process_file): Erase all artificial early.
7298 (generate_results): Skip as all artificial are already
7301 2017-11-13 Martin Liska <mliska@suse.cz>
7303 * gcov.c (read_graph_file): Store to global vector of functions.
7304 (read_count_file): Iterate the vector.
7305 (process_file): Likewise.
7306 (generate_results): Likewise.
7307 (release_structures): Likewise.
7309 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7311 PR tree-optimization/82954
7312 * gimple-ssa-store-merging.c
7313 (imm_store_chain_info::coalesce_immediate_stores): If
7314 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7316 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7318 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7319 Upddate call to ENDIAN_LANE_N.
7320 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7321 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7322 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7323 and use aarch64_endian_lane_rtx.
7324 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7326 2017-11-12 Tom de Vries <tom@codesourcery.com>
7328 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7330 2017-11-12 Tom de Vries <tom@codesourcery.com>
7332 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7334 2017-11-12 Tom de Vries <tom@codesourcery.com>
7336 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7338 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7339 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7340 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7342 2017-11-11 Martin Sebor <msebor@redhat.com>
7345 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7348 * prefic.c (translate_name): Replace strncpy with memcpy to
7349 avoid -Wstringop-truncation.
7351 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7353 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7355 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7357 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7359 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7361 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7364 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7366 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7367 enable generating XXBRH if the value is in a vector register.
7368 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7369 value is in a vector register.
7370 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7371 register to register bswap64's instead of doing the GPR sequence
7372 used on previous machines.
7373 (bswapdi2_xxbrd): New insn.
7374 (bswapdi2_reg): Disallow on ISA 3.0.
7375 (register to register bswap64 splitter): Do not split the insn on
7376 ISA 3.0 systems that use XXBRD.
7378 2017-11-10 Martin Sebor <msebor@redhat.com>
7381 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7383 (find_subframework_file): Same.
7385 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7387 * auto-profile.c (afdo_indirect_call): Drop frequency.
7388 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7389 (cgraph_node::create_edge): Drop frequency argument.
7390 (cgraph_node::create_indirect_edge): Drop frequency argument.
7391 (cgraph_edge::make_speculative): Drop frequency arguments.
7392 (cgraph_edge::resolve_speculation): Do not update frequencies
7393 (cgraph_edge::dump_edge_flags): Do not dump frequency.
7394 (cgraph_node::dump): Check consistency in IPA mode.
7395 (cgraph_edge::maybe_hot_p): Use IPA counter.
7396 (cgraph_edge::verify_count_and_frequency): Rename to ...
7397 (cgraph_edge::verify_count): ... this one; drop frequency checking.
7398 (cgraph_node::verify_node): Update.
7399 * cgraph.h (struct cgraph_edge): Drop frequency.
7400 (cgraph_edge::frequency): New function.
7401 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
7403 (cgraph_edge::rebuild_edges): Likewise.
7404 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
7405 (duplicate_thunk_for_node): Do not pass frequency.
7406 (cgraph_node::create_clone): Scale only counts.
7407 (cgraph_node::create_virtual_clone): Do not pass frequency.
7408 (cgraph_node::create_edge_including_clones): Do not pass frequency.
7409 (cgraph_node::create_version_clone): Do not pass frequency.
7410 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
7411 (cgraph_node::expand_thunk): Do not pass frequency.
7412 (cgraph_node::create_wrapper): Do not pass frequency.
7413 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
7415 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
7416 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
7417 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
7418 (gather_caller_stats): Use frequency function.
7419 (ipcp_cloning_candidate_p): Use frequency function.
7420 (ipcp_propagate_stage): Use frequency function.
7421 (get_info_about_necessary_edges): Use frequency function.
7422 (update_profiling_info): Update only IPA profile.
7423 (update_specialized_profile): Use frequency functoin.
7424 (perhaps_add_new_callers): Update only IPA profile.
7425 * ipa-devirt.c (ipa_devirt): Use IPA profile.
7426 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
7427 (dump_ipa_call_summary): Use frequency function.
7428 (estimate_edge_size_and_time): Use frequency function.
7429 (ipa_merge_fn_summary_after_inlining): Use frequency function.
7430 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
7431 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
7432 (update_noncloned_counts): ... ths one; scale counts only.
7433 (clone_inlined_nodes): Do not scale frequency.
7434 (inline_call): Do not pass frequency.
7435 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
7436 (compute_inlined_call_time): Use IPA profile.
7437 (want_inline_small_function_p): Use IPA profile.
7438 (want_inline_self_recursive_call_p): Use IPA profile.
7439 (edge_badness): Use IPA profile.
7440 (lookup_recursive_calls): Use IPA profile.
7441 (recursive_inlining): Do not pass frequency.
7442 (resolve_noninline_speculation): Do not update frequency.
7443 (inline_small_functions): Collect max of IPA profile.
7444 (dump_overall_stats): Dump IPA porfile.
7445 (dump_inline_stats): Dump IPA porfile.
7446 (ipa_inline): Collect IPA stats.
7447 * ipa-inline.h (clone_inlined_nodes): Update prototype.
7448 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
7449 (ipa_propagate_frequency): Use frequency function.
7450 (ipa_profile): Cleanup.
7451 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
7452 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
7453 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
7454 (input_node): Do not stream frequency.
7455 (input_edge): Do not stream frequency.
7456 (merge_profile_summaries): Scale only IPA profiles.
7457 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
7458 * predict.c (drop_profile): Do not recompute frequency.
7459 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
7460 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
7461 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
7462 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
7463 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
7464 * tree-inline.c (copy_bb): Do not scale frequency.
7465 (expand_call_inline): Do not scale frequency.
7466 (tree_function_versioning): Do not scale frequency.
7467 * ubsan.c (ubsan_create_edge): Do not pass frequency.
7469 2017-11-10 Julia Koval <julia.koval@intel.com>
7471 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
7472 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
7473 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
7474 (_mm256_maskz_gf2p8affine_epi64_epi8)
7475 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
7476 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
7477 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
7478 (__builtin_ia32_vgf2p8affineqb_v32qi)
7479 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
7480 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
7482 2017-11-10 Tamar Christina <tamar.christina@arm.com>
7486 (arm_option_override): Refactor.
7487 (arm_option_reconfigure_globals): New.
7488 (arm_options_perform_arch_sanity_checks): New.
7489 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
7491 (arm_options_perform_arch_sanity_checks): Likewise
7493 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
7495 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
7496 (power9-qpmul): New.
7497 * rs6000/rs6000.md ("type" attr): Add qmul.
7498 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
7499 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
7500 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
7502 2017-11-10 Martin Sebor <msebor@redhat.com>
7505 * builtins.c (compute_objsize): Handle arrays that
7506 compute_builtin_object_size likes to fail for. Make extern.
7507 * builtins.h (compute_objsize): Declare.
7508 (check_strncpy_sizes): New function.
7509 (expand_builtin_strncpy): Call check_strncpy_sizes.
7510 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
7511 -Wstringop-truncation.
7512 (gimple_fold_builtin_strncat): Same.
7513 * gimple.c (gimple_build_call_from_tree): Set call location.
7514 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
7515 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
7516 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
7517 (handle_builtin_strlen): Use strlen_to_stridx.
7518 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
7520 Use strlen_to_stridx.
7521 (pass_strlen::execute): Release strlen_to_stridx.
7522 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
7523 (-Wstringop-truncation): Document new option.
7525 2017-11-10 Martin Liska <mliska@suse.cz>
7527 PR gcov-profile/82702
7528 * gcov.c (main): Handle intermediate files in a different
7530 (get_gcov_intermediate_filename): New function.
7531 (output_gcov_file): Remove support of intermediate files.
7532 (generate_results): Allocate intermediate file.
7533 (release_structures): Clean-up properly fn_end.
7534 (output_intermediate_file): Start iterating with line 1.
7536 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7538 PR tree-optimization/82929
7539 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7540 ops_swapped_p non-static data member.
7541 (store_immediate_info::store_immediate_info): Clear it.
7542 (imm_store_chain_info::coalesce_immediate_stores): If swapping
7543 ops set ops_swapped_p.
7544 (count_multiple_uses): Handle ops_swapped_p.
7546 2017-11-10 Martin Liska <mliska@suse.cz>
7548 * coverage.c (coverage_init): Stream information about
7549 support of has_unexecuted_blocks.
7550 * doc/gcov.texi: Document that.
7551 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
7552 * gcov.c (read_graph_file): Likewise.
7553 (output_line_beginning): Fix a small issue with
7556 2017-11-10 Bin Cheng <bin.cheng@arm.com>
7558 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
7559 reference of trivial component.
7561 2017-11-10 Jakub Jelinek <jakub@redhat.com>
7564 * gimple-ssa-store-merging.c
7565 (pass_store_merging::terminate_all_aliasing_chains): For
7566 gimple_store_p stmts also call refs_output_dependent_p.
7568 PR rtl-optimization/82913
7569 * compare-elim.c (try_merge_compare): Punt if def_insn is not
7572 2017-11-09 Jeff Law <law@redhat.com>
7574 * vr-values.h: New file with vr_values class.
7575 * tree-vrp.c: Include vr-values.h
7576 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
7577 data objects into the vr_values class.
7578 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
7579 (get_value_range): Make it a member function within vr_values class.
7580 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
7581 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
7582 (op_with_constant_singleton_value_range): Likewise.
7583 (extract_range_for_var_from_comparison_expr): Likewise.
7584 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
7585 (extract_range_from_binary_expr): Likewise.
7586 (extract_range_from_unary_expr): Likewise.
7587 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
7588 (check_for_binary_op_overflow, extract_range_basic): Likewise.
7589 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
7590 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
7591 (compare_name_with_value, compare_names): Likewise.
7592 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
7593 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
7594 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
7595 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
7596 (extract_range_from_phi_node): Likewise.
7597 (simplify_truth_ops_using_ranges): Likewise.
7598 (simplify_div_or_mod_using_ranges): Likewise.
7599 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
7600 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
7601 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
7602 (simplify_float_conversion_using_ranges): Likewise.
7603 (simplify_internal_call_using_ranges): Likewise.
7604 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
7605 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
7606 poitner via x_vr_values for calls into gimple folder.
7607 (vrp_initialize_lattice): Make this the vr_values ctor.
7608 (vrp_free_lattice): Make this the vr_values dtor.
7609 (set_vr_value): New function.
7610 (class vrp_prop): Add vr_values data member. Add various member
7611 functions as well as member functions that delegate to vr_values.
7612 (check_array_ref): Make a member function within vrp_prop class.
7613 (search_for_addr_array, vrp_initialize): Likewise.
7614 (vrp_finalize): Likewise. Revamp to avoid direct access to
7615 vr_value, values_propagated, etc.
7616 (check_array_bounds): Extract vrp_prop class instance pointer from
7617 walk info structure. Use it to call member functions.
7618 (check_all_array_refs): Make a member function within vrp_prop class.
7619 Smuggle class instance pointer via walk info structure.
7620 (x_vr_values): New local static.
7621 (vrp_valueize): Use x_vr_values to get class instance.
7622 (vr_valueize_1): Likewise.
7623 (class vrp_folder): Add vr_values data member. Add various member
7624 functions as well as member functions that delegate to vr_values.
7625 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
7626 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
7627 class instance from vr_values. Use it to call member functions.
7628 (vrp_dom_walker): Add vr_values data member.
7629 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
7630 instance via x_vr_values.
7631 (identify_jump_threads): Accept vr_values as argument. Store
7632 it into the walker structure.
7633 (evrp_dom_walker): Add vr_values class data member. Add various
7635 (evrp_dom_walker::try_find_new_range): Use vr_values data
7636 member to access the memory allocator.
7637 (evrp_dom_walker::before_dom_children): Store vr_values class
7638 instance into the vrp_folder class.
7639 (evrp_dom_walker::push_value_range): Rework to avoid direct
7640 access to num_vr_values and vr_value.
7641 (evrp_dom_walker::pop_value_range): Likewise.
7642 (execute_early_vrp): Remove call to vrp_initialize_lattice.
7643 Use vr_values to get to dump_all_value_ranges member function.
7644 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
7645 and simplify_cond_using_ranges_2 via vrp_prop class instance.
7646 Pass vr_values class instance down to identify_jump_threads.
7647 Remove call to vrp_free_lattice.
7648 (debug_all_value_ranges): Remove.
7650 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
7651 (vrp_folder): Likewise.
7653 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
7654 Get it from the existing bitmap instead.
7655 (vrp_intersect_ranges_1): Likewise.
7657 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7659 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
7661 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7662 set bit_not_p to it.
7663 (imm_store_chain_info::coalesce_immediate_stores): Break group
7664 if bit_not_p is different.
7665 (count_multiple_uses, split_group,
7666 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
7667 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
7668 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
7669 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
7670 to store_immediate_info ctor.
7672 2017-11-09 Jim Wilson <jimw@sifive.com>
7674 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
7675 (scan_prog_file): Likewise.
7677 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7679 * bb-reorder.c (max_entry_frequency): Remove.
7680 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
7681 connect_traces, push_to_next_round_p): Remove prototypes.
7682 (find_traces_1_round): Use counts only.
7683 (push_to_next_round_p): Likewise.
7684 (find_traces): Likewise.
7685 (rotate_loop): Likewise.
7686 (find_traces_1_round): Likewise.
7687 (connect_traces): Likewise.
7688 (edge_order): Likewise.
7690 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
7692 * config/arm/arm.c (output_return_instruction): Add comments to
7693 indicate requirement for cmse_nonsecure_entry return to account
7694 for the size of clearing instruction output here.
7695 (thumb_exit): Likewise.
7696 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
7697 return in hardfloat mode.
7699 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
7701 * config/rs6000/rs6000.c (machine_function): Add a bool,
7702 "toc_is_wrapped_separately".
7703 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
7704 if it wasn't explicitly set or unset, we are optimizing for speed, and
7705 doing separate shrink-wrapping.
7706 (rs6000_get_separate_components): Enable the TOC component if
7707 saving the TOC register in the prologue.
7708 (rs6000_components_for_bb): Handle the TOC component.
7709 (rs6000_emit_prologue_components): Store the TOC register where needed.
7710 (rs6000_set_handled_components): Mark TOC as handled, if handled.
7711 (rs6000_emit_prologue): Don't save the TOC if that is already done.
7713 2017-11-09 Martin Jambor <mjambor@suse.cz>
7715 * ipa-param-manipulation.c: New file.
7716 * ipa-param-manipulation.h: Likewise.
7717 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
7718 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
7719 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
7720 (ipa_parm_adjustment): Likewise.
7721 (ipa_parm_adjustment_vec): Likewise.
7722 (ipa_get_vector_of_formal_parms): Moved declaration to
7723 ipa-param-manipulation.h.
7724 (ipa_get_vector_of_formal_parm_types): Likewise.
7725 (ipa_modify_formal_parameters): Likewise.
7726 (ipa_modify_call_arguments): Likewise.
7727 (ipa_combine_adjustments): Likewise.
7728 (ipa_dump_param_adjustments): Likewise.
7729 (ipa_modify_expr): Likewise.
7730 (ipa_get_adjustment_candidate): Likewise.
7731 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
7732 ipa-param-manipulation.c.
7733 (ipa_get_vector_of_formal_parm_types): Likewise.
7734 (ipa_modify_formal_parameters): Likewise.
7735 (ipa_modify_call_arguments): Likewise.
7736 (ipa_modify_expr): Likewise.
7737 (get_ssa_base_param): Likewise.
7738 (ipa_get_adjustment_candidate): Likewise.
7739 (index_in_adjustments_multiple_times_p): Likewise.
7740 (ipa_combine_adjustments): Likewise.
7741 (ipa_dump_param_adjustments): Likewise.
7742 * tree-sra.c: Also include ipa-param-manipulation.h
7743 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
7746 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7747 Alan Hayward <alan.hayward@arm.com>
7748 David Sherwood <david.sherwood@arm.com>
7750 * doc/sourcebuild.texi (vect_masked_store): Document.
7752 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7753 Alan Hayward <alan.hayward@arm.com>
7754 David Sherwood <david.sherwood@arm.com>
7756 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
7758 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7759 Alan Hayward <alan.hayward@arm.com>
7760 David Sherwood <david.sherwood@arm.com>
7762 * doc/sourcebuild.texi (vect_variable_length): Document.
7764 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7765 Alan Hayward <alan.hayward@arm.com>
7766 David Sherwood <david.sherwood@arm.com>
7768 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
7770 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7771 Alan Hayward <alan.hayward@arm.com>
7772 David Sherwood <david.sherwood@arm.com>
7774 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
7776 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7777 Alan Hayward <alan.hayward@arm.com>
7778 David Sherwood <david.sherwood@arm.com>
7780 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
7781 previously undocumented selectors.
7782 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
7784 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7786 * doc/rtl.texi (const_vector): Say that elements can be
7787 const_wide_ints too.
7788 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
7789 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
7790 (gen_vec_duplicate): Use it instead of CONSTANT_P.
7791 * optabs.c (expand_vector_broadcast): Likewise.
7793 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7794 Alan Hayward <alan.hayward@arm.com>
7795 David Sherwood <david.sherwood@arm.com>
7797 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
7798 scaled index even if the unscaled address was invalid.
7799 Don't increase the complexity of using a scale in that case.
7801 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
7802 Alan Hayward <alan.hayward@arm.com>
7803 David Sherwood <david.sherwood@arm.com>
7805 * doc/rtl.texi: Rewrite the subreg rules so that they partition
7806 the inner register into REGMODE_NATURAL_SIZE bytes rather than
7807 UNITS_PER_WORD bytes.
7808 * emit-rtl.c (validate_subreg): Divide subregs into blocks
7809 based on REGMODE_NATURAL_SIZE of the inner mode.
7810 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
7811 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
7812 * expmed.c (lowpart_bit_field_p): Divide the value up into
7813 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
7814 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
7815 whether something is likely to occupy more than one register.
7817 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
7820 * ipa-inline-transform.c (update_noncloned_frequencies): Use
7821 profile_count::adjust_for_ipa_scaling.
7822 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
7823 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
7825 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
7827 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7829 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
7830 (split_group): Add total_orig and total_new arguments, estimate the
7831 number of statements related to the store group without store merging
7832 and with store merging.
7833 (imm_store_chain_info::output_merged_store): Adjust split_group
7834 callers, punt if estimated number of statements with store merging
7835 is not smaller than estimated number of statements without it.
7837 (handled_load): Remove has_single_use checks.
7838 (pass_store_merging::process_store): Likewise.
7840 2017-11-09 Richard Biener <rguenther@suse.de>
7842 PR tree-optimization/82902
7843 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
7845 2017-11-09 Martin Liska <mliska@suse.cz>
7848 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
7851 2017-11-09 Martin Liska <mliska@suse.cz>
7853 PR tree-optimization/82669
7854 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
7856 2017-11-09 Martin Liska <mliska@suse.cz>
7858 PR gcov-profile/48463
7859 * coverage.c (coverage_begin_function): Output also end locus
7860 of a function and information whether the function is
7862 * gcov-dump.c (tag_function): Parse and print the information.
7863 * gcov.c (INCLUDE_MAP): Add include.
7864 (INCLUDE_SET): Likewise.
7865 (struct line_info): Move earlier in the source file because
7866 of vector<line_info> in function_info structure.
7867 (line_info::line_info): Likewise.
7868 (line_info::has_block): Likewise.
7869 (struct source_info): Add new member index.
7870 (source_info::get_functions_at_location): New function.
7871 (function_info::group_line_p): New function.
7872 (output_intermediate_line): New function.
7873 (output_intermediate_file): Use the mentioned function.
7874 (struct function_start): New.
7875 (struct function_start_pair_hash): Likewise.
7876 (process_file): Add code that identifies group functions.
7877 Assign lines either to global or function scope.
7878 (generate_results): Skip artificial functions.
7879 (find_source): Assign index for each source file.
7880 (read_graph_file): Read new flag artificial and end_line.
7881 (add_line_counts): Assign it either to global of function scope.
7882 (accumulate_line_counts): Isolate core of the function to
7883 accumulate_line_info and call it for both function and global
7885 (accumulate_line_info): New function.
7886 (output_line_beginning): Fix GNU coding style.
7887 (print_source_line): New function.
7888 (output_line_details): Likewise.
7889 (output_function_details): Likewise.
7890 (output_lines): Iterate both source (global) scope and function
7892 (struct function_line_start_cmp): New class.
7893 * doc/gcov.texi: Reflect changes in documentation.
7895 2017-11-09 Jakub Jelinek <jakub@redhat.com>
7898 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
7899 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
7900 and similarly for not instead of neg.
7902 2017-11-08 Andi Kleen <ak@linux.intel.com>
7904 * config/i386/i386.opt: Add -mforce-indirect-call.
7905 * config/i386/predicates.md: Check for flag_force_indirect_call.
7906 * doc/invoke.texi: Document -mforce-indirect-call
7908 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
7910 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
7912 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
7913 config/riscv/riscv.c (predict.h): New include.
7914 (riscv_slow_unaligned_access_p): No longer static.
7915 (riscv_block_move_straight): Add require.
7916 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
7918 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7921 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
7922 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
7923 nonimmediate_operand predicate for operand 1 instead of
7926 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7928 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
7930 * config/aarch64/constraints.md (Uml): New constraint.
7931 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
7934 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7936 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
7937 of two vec_duplicates into a vec_concat.
7939 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7941 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7942 Simplify vec_merge of vec_duplicate and vec_concat.
7943 * config/aarch64/constraints.md (Utq): New constraint.
7944 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
7947 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7949 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
7950 Simplify vec_merge of vec_duplicate and const_vector.
7951 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
7953 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
7954 mode iterator. Update predicate on operand 1 to
7955 handle non-const_vec constants. Delete constraints.
7956 (*aarch64_combinez_be<mode>): Likewise for operand 2.
7958 2017-11-08 Jakub Jelinek <jakub@redhat.com>
7960 PR tree-optimization/78821
7961 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
7963 (store_operand_info::store_operand_info): Initialize it to false.
7964 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
7965 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
7966 store in the group, and if chain_info is non-NULL, to ignore altogether
7968 (compatible_load_p): Fail if bit_not_p does not match.
7969 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
7970 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
7971 (pass_store_merging::process_store): Adjust
7972 terminate_all_aliasing_chains calls to pass NULL in all current spots,
7973 call terminate_all_aliasing_chains newly when adding a store into
7974 a chain with non-NULL chain_info.
7976 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7978 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
7980 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
7982 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7984 (aarch64_layout_frame): Initialise emit_frame_chain.
7985 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
7986 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
7988 2017-11-08 Martin Liska <mliska@suse.cz>
7990 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
7991 of gimple_call_internal_p.
7993 2017-11-07 Tom de Vries <tom@codesourcery.com>
7995 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
7997 2017-11-07 Tom de Vries <tom@codesourcery.com>
7999 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8002 2017-11-08 Martin Liska <mliska@suse.cz>
8005 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8007 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8009 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8011 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8014 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8015 (*cmp<mode>_ccz_1): New insn with $k alternative.
8018 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8019 RTX_COMM_COMPARE as commutative as well.
8020 (ix86_binary_operator_ok): Formatting fix.
8021 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8022 *<code><mode>3<mask_name><round_saeonly_name>,
8023 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8024 *<s>mul<mode>3_highpart<mask_name>,
8025 *vec_widen_umult_even_v16si<mask_name>,
8026 *vec_widen_umult_even_v8si<mask_name>,
8027 *vec_widen_umult_even_v4si<mask_name>,
8028 *vec_widen_smult_even_v16si<mask_name>,
8029 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8030 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8031 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8032 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8033 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8034 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8035 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8036 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8037 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8038 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8039 ix86_binary_operator_ok. Formatting fixes.
8040 (*<plusminus_insn><mode>3<mask_name><round_name>,
8041 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8044 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8046 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8047 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8048 eq and ne if TARGET_ISEL.
8049 (cmp): New code_iterator.
8050 (UNS, UNSU_, UNSIK): New code_attrs.
8051 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8052 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8054 ("eq<mode>3"): ... this.
8055 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8057 ("ne<mode>3"): ... this.
8059 2017-11-07 Julia Koval <julia.koval@intel.com>
8062 * common/config/i386/i386-common.c
8063 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8064 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8065 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8066 * config/i386/i386.opt: Ditto.
8067 * config/i386/i386.c (ix86_target_string): Ditto.
8068 (ix86_option_override_internal): Ditto.
8069 (ix86_init_mpx_builtins): Move MPX to args2.
8070 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8071 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8072 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8073 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8074 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8075 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8076 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8078 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8081 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8082 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8083 (zero-extendsidi peephole2): Remove peephole.
8085 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8088 * stor-layout.c: Include attribs.h.
8089 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8090 explicit lookup of "aligned" attribute.
8092 2017-11-07 Andrew Waterman <andrew@sifive.com>
8094 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8095 (riscv_expand_block_move): Likewise.
8096 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8098 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8099 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8100 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8101 (riscv_adjust_block_mem): Likewise.
8102 (riscv_block_move_loop): Likewise.
8103 (riscv_expand_block_move): Likewise.
8104 * config/riscv/riscv.md (movmemsi): New pattern.
8106 2017-11-07 Michael Clark <michaeljclark@mac.com>
8108 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8109 (MUSL_DYNAMIC_LINKER): Likewise.
8111 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8113 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8114 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8117 2017-11-07 Richard Biener <rguenther@suse.de>
8119 * match.pd: Fix build.
8121 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8122 Jackson Woodruff <jackson.woodruff@arm.com>
8124 PR tree-optimization/71026
8125 * match.pd: Canonicalize negate in division.
8127 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8130 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8132 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8134 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8135 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8137 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8139 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8140 non-scalar integral types.
8141 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8142 (-(A-B), -(~A)): New transformations.
8144 2017-11-07 Tom de Vries <tom@codesourcery.com>
8146 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8147 semicolon after "do {} while (0)".
8148 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8149 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8150 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8151 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8152 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8154 2017-11-07 Tom de Vries <tom@codesourcery.com>
8156 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8157 after "do {} while (0)".
8158 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8159 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8160 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8161 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8162 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8164 2017-11-07 Tom de Vries <tom@codesourcery.com>
8167 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8169 (arm_rtx_costs_internal): Add missing semicolon after
8170 HANDLE_NARROW_SHIFT_ARITH call.
8172 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8174 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8175 disable isel if it was not set explicitly.
8177 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8179 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8180 (add_type_attribute) likewise.
8182 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8184 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8185 of struct ix86_frame.
8186 (ix86_initial_elimination_offset): Likewise.
8187 (ix86_expand_split_stack_prologue): Likewise.
8189 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8191 * tree-vrp.h (enum value_range_type): Update stale comment.
8193 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8194 Alan Hayward <alan.hayward@arm.com>
8195 David Sherwood <david.sherwood@arm.com>
8197 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8198 (aarch64_expand_vec_perm_const): Take the number of units too.
8199 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8200 (aarch64_expand_vec_perm_const): Likewise.
8201 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8202 (vec_perm<mode>): Update accordingly.
8204 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8205 Alan Hayward <alan.hayward@arm.com>
8206 David Sherwood <david.sherwood@arm.com>
8208 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8209 Take the number of units too.
8210 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8211 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8212 but check for a vector mode before rather than after the call.
8213 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8214 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8215 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8216 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8217 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8218 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8219 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8220 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8221 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8222 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8223 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8224 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8225 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8226 (aarch64_sqdmull2_n<mode>): Update accordingly.
8228 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8229 Alan Hayward <alan.hayward@arm.com>
8230 David Sherwood <david.sherwood@arm.com>
8232 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8233 the number of units too.
8234 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8235 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8236 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8237 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8238 (vec_store_lanesxi<mode>): Update accordingly.
8240 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8241 Alan Hayward <alan.hayward@arm.com>
8242 David Sherwood <david.sherwood@arm.com>
8244 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8245 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8246 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8247 of units rather than the mode.
8248 * config/aarch64/iterators.md (nunits): New mode attribute.
8249 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8250 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8251 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8252 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8253 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8254 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8255 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8256 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8257 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8258 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8259 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8260 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8261 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8262 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8263 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8264 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8265 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8266 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8267 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8268 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8269 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8270 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8271 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8272 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8273 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8274 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8275 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8276 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8277 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8278 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8279 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8280 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8281 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8282 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8283 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8284 (aarch64_simd_vec_setv2di): Likewise.
8286 2017-11-06 Carl Love <cel@us.ibm.com>
8288 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8290 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8291 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8293 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8295 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8296 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8297 Add power 8 macro expansions.
8298 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8299 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8300 power 8 instructions. (VSX_XXBR): Add iterator.
8302 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8304 * config/arm/arm.md (predicable_short_it): Change default to "no",
8305 improve documentation, remove uses that are identical to the default.
8306 (enabled_for_depr_it): Rename to enabled_for_short_it.
8307 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8308 Remove default uses.
8309 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8310 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8311 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8312 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8314 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8317 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8318 float128 helper macros, which are no longer used after deleting
8319 the old 'q' built-in functions, and moving the round to odd
8320 built-in functions to being special built-in functions.
8321 (BU_FLOAT128_2): Likewise.
8322 (BU_FLOAT128_1_HW): Likewise.
8323 (BU_FLOAT128_2_HW): Likewise.
8324 (BU_FLOAT128_3_HW): Likewise.
8325 (FABSQ): Delete old 'q' built-in functions.
8326 (COPYSIGNQ): Likewise.
8327 (SQRTF128_ODD): Move round to odd built-in functions to be
8328 special built-in functions, so that we can handle
8329 -mabi=ieeelongdouble.
8330 (TRUNCF128_ODD): Likewise.
8331 (ADDF128_ODD): Likewise.
8332 (SUBF128_ODD): Likewise.
8333 (MULF128_ODD): Likewise.
8334 (DIVF128_ODD): Likewise.
8335 (FMAF128_ODD): Likewise.
8336 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8337 built-in names to 'f128'.
8338 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8339 old 'q' built-in functions, as the machine independent code for
8340 'f128' built-in functions handles this.
8341 (rs6000_expand_builtin): Add expansion for float128 round to odd
8342 functions, keying off on -mabi=ieeelongdouble of whether to use
8343 the KFmode or TFmode variant.
8344 (rs6000_init_builtins): Initialize the _Float128 round to odd
8346 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8347 _Float128 'q' built-in functions are now mapped into the new
8348 'f128' built-in functions.
8350 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8352 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8353 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8355 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8358 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8359 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8360 * toplev.c: Include "ipa-fnsummary.h".
8361 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8363 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8365 PR tree-optimization/82838
8366 * gimple-ssa-store-merging.c
8367 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8368 on a separate gimple_seq which is then appended to seq.
8370 2017-11-06 Jeff Law <law@redhat.com>
8373 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8374 (get_probe_interval): New functions.
8375 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8376 (ix86_adjust_stack_and_probe): Likewise.
8377 (output_adjust_stack_and_probe): Likewise.
8378 (ix86_emit_probe_stack_range): Likewise.
8379 (ix86_expand_prologue): Likewise.
8381 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8383 PR tree-optimization/82816
8384 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8385 if the modes of the two types are the same.
8386 (convert_plusminus_to_widen): Likewise.
8388 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8390 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8392 (usadv16qi): New define_expand.
8393 (usadv8hi): New define_expand.
8395 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
8398 * ipa-inline-transform.c (update_noncloned_frequencies): Always
8400 (inline_transform): Likewise.
8401 * predict.c (counts_to_freqs): Remove useless conditional.
8402 * profile-count.h (profile_count::apply_scale): Move sanity check.
8403 * tree-inline.c (copy_bb): Always scale.
8404 (copy_cfg_body): Likewise.
8406 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
8409 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
8411 (*cmp_ite0): Add enabled_for_depr_it attribute.
8412 (*cmp_ite1): Likewise.
8414 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8416 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
8419 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8421 * tree-vrp.c (vrp_int_const_binop): Return true on success and
8422 return the value by pointer.
8423 (extract_range_from_multiplicative_op_1): Update accordingly.
8424 Return as soon as an operation fails.
8426 2017-11-05 Tom de Vries <tom@codesourcery.com>
8429 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
8430 (DEF_SANITIZER_BUILTIN): ... here.
8431 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
8432 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
8434 2017-11-05 Tom de Vries <tom@codesourcery.com>
8437 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
8439 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
8440 ASM_OUTPUT_BEFORE_CASE_LABEL call.
8441 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
8443 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8444 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8445 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
8447 2017-11-05 Tom de Vries <tom@codesourcery.com>
8450 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
8453 2017-11-04 Michael Clark <michaeljclark@mac.com>
8455 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
8456 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
8458 (*addsi3_extended): Likewise.
8459 (*addsi3_extended2): Likewise.
8460 (<optab>si3): Likewise.
8461 (<optab>di3): Likewise.
8462 (<optab><mode>3): Likewise.
8463 (<*optabe>si3_internal): Likewise.
8464 (zero_extendqi<SUPERQI:mode>2): Likewise.
8465 (*add<mode>hi3): Likewise.
8466 (*xor<mode>hi3): Likewise.
8467 (<optab>di3): Likewise.
8468 (*<optab>si3_extend): Likewise.
8469 (*sge<u>_<X:mode><GPR:mode>): Likewise.
8470 (*slt<u>_<X:mode><GPR:mode>): Likewise.
8471 (*sle<u>_<X:mode><GPR:mode>): Likewise.
8473 2017-11-04 Andrew Waterman <andrew@sifive.com>
8475 * config/riscv/riscv.c (riscv_option_override): Conditionally set
8476 TARGET_STRICT_ALIGN based upon -mtune argument.
8478 2017-11-04 Andrew Waterman <andrew@sifive.com>
8480 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
8482 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
8484 * config/i386/i386.c (choose_basereg): Use optional scratch
8485 register and add assertion.
8486 (x86_emit_outlined_ms2sysv_save): Use scratch register when
8487 needed, and don't allocate stack.
8488 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
8489 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
8490 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
8492 2017-11-03 Jeff Law <law@redhat.com>
8494 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
8495 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
8496 to probe at the start of a noreturn function.
8498 2017-11-03 Jakub Jelinek <jakub@redhat.com>
8500 PR tree-optimization/78821
8501 * gimple-ssa-store-merging.c: Update the file comment.
8502 (MAX_STORE_ALIAS_CHECKS): Define.
8503 (struct store_operand_info): New type.
8504 (store_operand_info::store_operand_info): New constructor.
8505 (struct store_immediate_info): Add rhs_code and ops data members.
8506 (store_immediate_info::store_immediate_info): Add rhscode, op0r
8507 and op1r arguments to the ctor, initialize corresponding data members.
8508 (struct merged_store_group): Add load_align_base and load_align
8510 (merged_store_group::merged_store_group): Initialize them.
8511 (merged_store_group::do_merge): Update them.
8512 (merged_store_group::apply_stores): Pick the constant for
8513 encode_tree_to_bitpos from one of the two operands, or skip
8514 encode_tree_to_bitpos if neither operand is a constant.
8515 (class pass_store_merging): Add process_store method decl. Remove
8516 bool argument from terminate_all_aliasing_chains method decl.
8517 (pass_store_merging::terminate_all_aliasing_chains): Remove
8518 var_offset_p argument and corresponding handling.
8519 (stmts_may_clobber_ref_p): New function.
8520 (compatible_load_p): New function.
8521 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
8522 if there is overlap and rhs_code is not INTEGER_CST. For
8523 non-overlapping stores terminate group if rhs is not mergeable.
8524 (get_alias_type_for_stmts): Change first argument from
8525 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
8526 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
8527 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
8529 (get_location_for_stmts): Change first argument from
8530 auto_vec<gimple *> & to vec<gimple *> &.
8531 (struct split_store): Remove orig_stmts data member, add orig_stores.
8532 (split_store::split_store): Create orig_stores rather than orig_stmts.
8533 (find_constituent_stmts): Renamed to ...
8534 (find_constituent_stores): ... this. Change second argument from
8535 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
8536 to info structures rather than the statements.
8537 (split_group): Rename ALLOW_UNALIGNED argument to
8538 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
8539 it. Adjust find_constituent_stores caller.
8540 (imm_store_chain_info::output_merged_store): Handle rhs_code other
8541 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
8542 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
8543 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
8544 (mem_valid_for_store_merging): New function.
8545 (handled_load): New function.
8546 (pass_store_merging::process_store): New method.
8547 (pass_store_merging::execute): Use process_store method. Adjust
8548 terminate_all_aliasing_chains caller.
8550 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8552 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8553 Return true for more constants, symbols and label references.
8554 (aarch64_valid_floating_const): Remove unused function.
8556 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8559 * config/aarch64/aarch64.c (aarch64_layout_frame):
8560 Undo forcing of LR at bottom of frame.
8562 2017-11-03 Jeff Law <law@redhat.com>
8565 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
8566 for int_registers_saved.
8568 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
8569 extracted from tree-ssa-dom.c.
8570 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
8571 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
8572 (record_equivalences_from_incoming_edge): Add additional argument
8573 to single_pred_edge_ignoring_loop_edges call.
8574 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
8575 (uncprop_dom_walker::before_dom_children): Add additional argument
8576 to single_pred_edge_ignoring_loop_edges call.
8577 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
8578 single_pred_edge_ignoring_loop_edges rather than open coding.
8579 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
8581 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
8583 * match.pd (-(-A)): Rewrite.
8585 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8587 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
8588 (rs6000_emit_int_cmove): New declaration.
8589 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
8590 (rs6000_emit_sISEL): Delete.
8591 (rs6000_emit_int_cmove): Make non-static.
8592 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
8593 instead of rs6000_emit_sISEL.
8595 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
8597 * asan.c (create_cond_insert_point): Maintain profile.
8598 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
8600 * basic-block.h (struct basic_block_def): Remove frequency.
8601 (EDGE_FREQUENCY): Use to_frequency
8602 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
8604 (find_traces): Update to use to_frequency.
8605 (find_traces_1_round): Likewise; use only IPA counts.
8606 (bb_to_key): Likewise.
8607 (connect_traces): Use IPA counts only.
8608 (copy_bb_p): Update to use to_frequency.
8609 (fix_up_crossing_landing_pad): Likewise.
8610 (sanitize_hot_paths): Likewise.
8611 * bt-load.c (basic_block_freq): Likewise.
8612 * cfg.c (init_flow): Set count_max to uninitialized.
8613 (check_bb_profile): Remove frequencies; check counts.
8614 (dump_bb_info): Do not dump frequencies.
8615 (update_bb_profile_for_threading): Update counts only.
8616 (scale_bbs_frequencies_int): Likewise.
8617 (MAX_SAFE_MULTIPLIER): Remove.
8618 (scale_bbs_frequencies_gcov_type): Update counts only.
8619 (scale_bbs_frequencies_profile_count): Update counts only.
8620 (scale_bbs_frequencies): Update counts only.
8621 * cfg.h (struct control_flow_graph): Add count-max.
8622 (update_bb_profile_for_threading): Update prototype.
8623 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
8624 (find_many_sub_basic_blocks): Likewise.
8625 * cfgcleanup.c (try_forward_edges): Likewise.
8626 (try_crossjump_to_edge): Likewise.
8627 * cfgexpand.c (expand_gimple_cond): Likewise.
8628 (expand_gimple_tailcall): Likewise.
8629 (construct_init_block): Likewise.
8630 (construct_exit_block): Likewise.
8631 * cfghooks.c (verify_flow_info): Check consistency of counts.
8632 (dump_bb_for_graph): Do not dump frequencies.
8633 (split_block_1): Do not update frequencies.
8634 (split_edge): Do not update frequencies.
8635 (make_forwarder_block): Do not update frequencies.
8636 (duplicate_block): Do not update frequencies.
8637 (account_profile_record): Do not update frequencies.
8638 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
8639 for global heuristics.
8640 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
8641 (expected_loop_iterations_unbounded): Use counts only.
8642 * cfgloopmanip.c (scale_loop_profile): Simplify.
8643 (create_empty_loop_on_edge): Simplify
8645 (duplicate_loop_to_header_edge): Simplify
8646 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
8647 (update_br_prob_note): Take care of removing note when profile
8649 (relink_block_chain): Do not dump frequency.
8650 (rtl_account_profile_record): Use to_frequency.
8651 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
8652 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
8653 (cgraph_update_edges_for_call_stmt_node): Likewise.
8654 (cgraph_edge::verify_count_and_frequency): Update.
8655 (cgraph_node::verify_node): Temporarily disable frequency verification.
8656 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
8657 to_cgraph_frequency.
8658 (cgraph_edge::rebuild_edges): Convert to ipa counts.
8659 * cgraphunit.c (init_lowered_empty_function): Do not initialize
8661 (cgraph_node::expand_thunk): Update profile.
8662 * except.c (dw2_build_landing_pads): Do not update frequency.
8663 * final.c (compute_alignments): Use to_frequency.
8664 (dump_basic_block_info): Do not dump frequency.
8665 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
8666 (dump_gimple_bb_header): Do not dump frequency.
8667 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
8669 * gimple-streamer-in.c (input_bb): Do not stream frequency.
8670 * gimple-streamer-out.c (output_bb): Do not stream frequency.
8671 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
8672 (init_before_recovery): Do not update frequency.
8673 (sched_create_recovery_edges): Do not update frequency.
8674 * hsa-gen.c (convert_switch_statements): Do not update frequency.
8675 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
8676 (ipa_cp_c_finalize): Set max_count to uninitialized.
8677 * ipa-fnsummary.c (get_minimal_bb): Use counts.
8678 (param_change_prob): Use counts.
8679 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
8681 * ipa-split.c (consider_split): Use to_frequency.
8682 (split_function): Use to_frequency.
8683 * ira-build.c (loop_compare_func): Likewise.
8684 (mark_loops_for_removal): Likewise.
8685 (mark_all_loops_for_removal): Likewise.
8686 * loop-doloop.c (doloop_modify): Do not update frequency.
8687 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
8689 * lto-streamer-in.c (input_function): Update count_max.
8690 * omp-expand.c (expand_omp_taskreg): Update count_max.
8691 * omp-simd-clone.c (simd_clone_adjust): Update profile.
8692 * predict.c (maybe_hot_frequency_p): Use to_frequency.
8693 (maybe_hot_count_p): Use ipa counts only.
8694 (maybe_hot_bb_p): Simplify.
8695 (maybe_hot_edge_p): Simplify.
8696 (probably_never_executed): Do not take frequency argument.
8697 (probably_never_executed_bb_p): Do not pass frequency.
8698 (probably_never_executed_edge_p): Likewise.
8699 (combine_predictions_for_bb): Check that profile is nonzero.
8700 (propagate_freq): Do not set frequency.
8701 (drop_profile): Simplify.
8702 (counts_to_freqs): Simplify.
8703 (expensive_function_p): Use to_frequency.
8704 (propagate_unlikely_bbs_forward): Simplify.
8705 (determine_unlikely_bbs): Simplify.
8706 (estimate_bb_frequencies): Add hack to silence graphite issues.
8707 (compute_function_frequency): Use ipa counts.
8708 (pass_profile::execute): Update.
8709 (rebuild_frequencies): Use counts only.
8710 (force_edge_cold): Use counts only.
8711 * profile-count.c (profile_count::dump): Dump new count types.
8712 (profile_count::differs_from_p): Check compatiblity.
8713 (profile_count::to_frequency): New function.
8714 (profile_count::to_cgraph_frequency): New function.
8715 * profile-count.h (struct function): Declare.
8716 (enum profile_quality): Add profile_guessed_local and
8717 profile_guessed_global0.
8718 (class profile_proability): Decrease number of bits to 29;
8719 update from_reg_br_prob_note and to_reg_br_prob_note.
8720 (class profile_count: Update comment; decrease number of bits
8721 to 61. Check compatibility.
8722 (profile_count::compatible_p): New private member function.
8723 (profile_count::ipa_p): New member function.
8724 (profile_count::operator<): Handle global zero correctly.
8725 (profile_count::operator>): Handle global zero correctly.
8726 (profile_count::operator<=): Handle global zero correctly.
8727 (profile_count::operator>=): Handle global zero correctly.
8728 (profile_count::nonzero_p): New member function.
8729 (profile_count::force_nonzero): New member function.
8730 (profile_count::max): New member function.
8731 (profile_count::apply_scale): Handle IPA scalling.
8732 (profile_count::guessed_local): New member function.
8733 (profile_count::global0): New member function.
8734 (profile_count::ipa): New member function.
8735 (profile_count::to_frequency): Declare.
8736 (profile_count::to_cgraph_frequency): Declare.
8737 * profile.c (OVERLAP_BASE): Delete.
8738 (compute_frequency_overlap): Delete.
8739 (compute_branch_probabilities): Do not use compute_frequency_overlap.
8740 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
8741 * sched-ebb.c (rank): Use counts only.
8742 * shrink-wrap.c (handle_simple_exit): Use counts only.
8743 (try_shrink_wrapping): Use counts only.
8744 (place_prologue_for_one_component): Use counts only.
8745 * tracer.c (find_best_predecessor): Use to_frequency.
8746 (find_trace): Use to_frequency.
8747 (tail_duplicate): Use to_frequency.
8748 * trans-mem.c (expand_transaction): Do not update frequency.
8749 * tree-call-cdce.c: Do not update frequency.
8750 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
8751 (gimple_merge_blocks): Likewise.
8752 (gimple_split_edge): Likewise.
8753 (gimple_duplicate_sese_region): Likewise.
8754 (gimple_duplicate_sese_tail): Likewise.
8755 (move_sese_region_to_fn): Likewise.
8756 (gimple_account_profile_record): Likewise.
8757 (insert_cond_bb): Likewise.
8758 * tree-complex.c (expand_complex_div_wide): Likewise.
8759 * tree-eh.c (lower_resx): Update profile.
8760 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
8762 (initialize_cfun): Do not initialize frequencies
8763 (freqs_to_counts): Delete.
8764 (copy_cfg_body): Ignore count parameter.
8765 (copy_body): Update.
8766 (expand_call_inline): Update count_max.
8767 (optimize_inline_calls): Update count_max.
8768 (tree_function_versioning): Update count_max.
8769 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
8770 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
8772 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
8773 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
8774 (try_peel_loop): Likewise.
8775 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
8777 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
8778 (tree_transform_and_unroll_loop): Do not use frequencies
8779 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
8780 Use reliable prediction only.
8781 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
8782 * tree-ssa-sink.c (select_best_block): Use to_frequency.
8783 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
8784 probability scaling.
8785 * tree-ssa-threadupdate.c (create_block_for_threading): Do
8786 not update frequency
8787 (any_remaining_duplicated_blocks): Likewise.
8788 (update_profile): Likewise.
8789 (estimated_freqs_path): Delete.
8790 (freqs_to_counts_path): Delete.
8791 (clear_counts_path): Delete.
8792 (ssa_fix_duplicate_block_edges): Likewise.
8793 (duplicate_thread_path): Likewise.
8794 * tree-switch-conversion.c (gen_inbound_check): Use counts.
8795 * tree-tailcall.c (decrease_profile): Do not update frequency.
8796 (eliminate_tail_call): Likewise.
8797 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
8798 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
8799 (optimize_mask_stores): Likewise.
8800 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
8801 * ubsan.c (ubsan_expand_null_ifn): Update profile.
8802 (ubsan_expand_ptr_ifn): Update profile.
8803 * value-prof.c (gimple_ic): Simplify.
8804 * value-prof.h (gimple_ic): Update prototype.
8805 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
8806 * ipa-inline.c (compute_uninlined_call_time): Be sure that
8808 (want_inline_self_recursive_call_p): Likewise.
8809 (resolve_noninline_speculation): Only cummulate defined counts.
8810 (inline_small_functions): Use nonzero_p.
8811 (ipa_inline): Do not access freed node.
8813 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
8815 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8816 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
8818 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8820 * config/riscv/riscv.c (riscv_legitimize_move): Handle
8821 non-legitimate address.
8823 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8825 * config/rs6000/rs6000.md (*lt0_disi): Delete.
8826 (*lt0_<mode>di, *lt0_<mode>si): New.
8828 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
8830 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
8831 TARGET_PAIRED_FLOAT.
8833 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
8834 Jim Wilson <jim.wilson@linaro.org>
8836 * config/aarch64/aarch64-cores.def (saphira): New CPU.
8837 * config/aarch64/aarch64-tune.md: Regenerated.
8838 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
8839 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
8841 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
8843 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
8844 (arc_expand_prologue): Restore blink for millicode.
8845 * config/arc/linux.h (LINK_EH_SPEC): Defined.
8847 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
8850 * config/i386/i386.c (ix86_vector_duplicate_value): Use
8851 gen_vec_duplicate after forcing the scalar into a register.
8853 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
8855 * combine (try_combine): Print the insns input to try_combine to the
8858 2017-11-02 Steve Ellcey <sellcey@cavium.com>
8861 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
8862 Remove second argument from aarch64_process_target_attr call.
8863 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
8865 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
8867 (aarch64_handle_attr_arch): Remove second argument.
8868 (aarch64_handle_attr_cpu): Ditto.
8869 (aarch64_handle_attr_tune): Ditto.
8870 (aarch64_handle_attr_isa_flags): Ditto.
8871 (aarch64_process_one_target_attr): Ditto.
8872 (aarch64_process_target_attr): Ditto.
8873 (aarch64_option_valid_attribute_p): Remove second argument.
8874 on aarch64_process_target_attr call.
8876 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8878 * diagnostic.c: Include "selftest-diagnostic.h".
8879 (selftest::assert_location_text): New function.
8880 (selftest::test_diagnostic_get_location_text): New function.
8881 (selftest::diagnostic_c_tests): Call it.
8883 2017-11-02 David Malcolm <dmalcolm@redhat.com>
8885 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
8886 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
8887 (class selftest::test_diagnostic_context): Move to...
8888 * selftest-diagnostic.c: New file.
8889 * selftest-diagnostic.h: New file.
8891 2017-11-02 James Bowman <james.bowman@ftdichip.com>
8893 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
8894 offset range for FT32B.
8895 * config/ft32/ft32.h: option "mcompress" enables relaxation.
8896 * config/ft32/ft32.md: Add TARGET_NOPM.
8897 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
8898 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
8900 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
8902 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
8904 2017-11-02 Jeff Law <law@redhat.com>
8906 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
8907 virtual keyword on FINAL OVERRIDE members.
8909 * tree-ssa-propagate.h (ssa_propagation_engine): Group
8910 virtuals together. Add virtual destructor.
8911 (substitute_and_fold_engine): Similarly.
8913 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
8915 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
8917 2017-11-02 Richard Biener <rguenther@suse.de>
8919 PR tree-optimization/82795
8920 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
8922 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8924 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
8926 (gcc_GAS_CHECK_FEATURE): Remove.
8927 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
8928 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
8929 * configure: Regenerate.
8931 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
8933 * config/arc/arc.c (hwloop_optimize): Account for empty
8936 2017-11-02 Richard Biener <rguenther@suse.de>
8939 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
8940 Truncate ARRAY_REF index and element size.
8942 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
8944 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
8946 2017-11-01 Jeff Law <law@redhat.com>
8948 * tree-ssa-ccp.c (ccp_folder): New class derived from
8949 substitute_and_fold_engine.
8950 (ccp_folder::get_value): New member function.
8951 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
8952 (ccp_fold_stmt): Remove prototype.
8953 (ccp_finalize): Call substitute_and_fold from the ccp_class.
8954 * tree-ssa-copy.c (copy_folder): New class derived from
8955 substitute_and_fold_engine.
8956 (copy_folder::get_value): Renamed from get_value.
8957 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
8958 * tree-vrp.c (vrp_folder): New class derived from
8959 substitute_and_fold_engine.
8960 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
8961 (vrp_folder::get_value): New member function.
8962 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
8963 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
8964 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
8965 provide a class interface to folder/substitute routines.
8966 (ssa_prop_fold_stmt_fn): Remove typedef.
8967 (ssa_prop_get_value_fn): Likewise.
8968 (subsitute_and_fold): Remove prototype.
8969 (replace_uses_in): Likewise.
8970 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
8971 Renamed from replace_uses_in. Call the virtual member function
8972 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
8973 (substitute_and_fold_dom_walker): Remove initialization of
8974 data member entries for calbacks. Add substitute_and_fold_engine
8975 member and initialize it.
8976 (substitute_and_fold_dom_walker::before_dom_children0: Use the
8977 member functions for get_value, replace_phi_args_in c
8978 replace_uses_in, and fold_stmt calls.
8979 (substitute_and_fold_engine::substitute_and_fold): Renamed from
8980 substitute_and_fold. Remove assert. Update ctor call.
8982 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
8983 (ssa_prop_visit_phi_fn): Likewise.
8984 (class ssa_propagation_engine): New class to provide an interface
8986 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
8988 (ssa_prop_visit_phi): Likewise.
8989 (ssa_propagation_engine::simulate_stmt): Moved into class.
8990 Call visit_phi/visit_stmt from the class rather than via
8991 file scoped static variables.
8992 (ssa_propagation_engine::simulate_block): Moved into class.
8993 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
8994 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
8995 set file scoped statics for the visit_stmt/visit_phi callbacks.
8996 * tree-complex.c (complex_propagate): New class derived from
8997 ssa_propagation_engine.
8998 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
8999 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9000 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9002 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9003 ssa_propagation_engine.
9004 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9005 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9006 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9007 * tree-ssa-copy.c (copy_prop): New class derived from
9008 ssa_propagation_engine.
9009 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9010 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9011 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9012 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9013 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9014 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9015 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9017 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9019 PR rtl-optimization/82778
9020 PR rtl-optimization/82597
9021 * compare-elim.c (struct comparison): Add in_a_setter field.
9022 (find_comparison_dom_walker::before_dom_children): Remove killed
9023 bitmap and df_simulate_find_defs call, instead walk the defs.
9024 Compute last_setter and initialize in_a_setter. Merge definitions
9025 with first initialization for a few variables.
9026 (try_validate_parallel): Use insn_invalid_p instead of
9027 recog_memoized. Return insn rather than just the pattern.
9028 (try_merge_compare): Fix up comment. Don't uselessly test if
9029 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9031 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9034 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9035 Alan Hayward <alan.hayward@arm.com>
9036 David Sherwood <david.sherwood@arm.com>
9038 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9039 aarch64_hard_regno_nregs to get the number of registers
9042 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9043 Alan Hayward <alan.hayward@arm.com>
9044 David Sherwood <david.sherwood@arm.com>
9046 * config/aarch64/constraints.md (Upl): Rename to...
9048 * config/aarch64/aarch64.md
9049 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9052 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9053 Alan Hayward <alan.hayward@arm.com>
9054 David Sherwood <david.sherwood@arm.com>
9056 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9057 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9060 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9061 Alan Hayward <alan.hayward@arm.com>
9062 David Sherwood <david.sherwood@arm.com>
9064 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9065 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9066 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9068 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9069 of a single element.
9070 * config/aarch64/iterators.md: Add a comment above the permute
9071 unspecs to say that they are generated directly by
9072 aarch64_expand_vec_perm_const.
9073 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9075 2017-11-01 Nathan Sidwell <nathan@acm.org>
9077 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9079 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9081 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9082 and medany code models, and describe what they do.
9084 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9086 Revert accidental duplicate:
9088 * combine.c (can_change_dest_mode): Reject changes in
9089 REGMODE_NATURAL_SIZE.
9091 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9093 PR rtl-optimization/64682
9094 PR rtl-optimization/69567
9095 PR rtl-optimization/69737
9096 PR rtl-optimization/82683
9097 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9098 register mentioned in the note, drop the note, unless it came from I3,
9099 in which case it should go to I3 again.
9101 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9103 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9104 and return false if not.
9105 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9107 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9109 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9110 range is known to be empty.
9112 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9113 Alan Hayward <alan.hayward@arm.com>
9114 David Sherwood <david.sherwood@arm.com>
9116 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9117 and CONST_VECTOR_NUNITS instead of computing the number of units from
9118 the byte sizes of the vector and element.
9119 (simplify_binary_operation_1): Likewise.
9120 (simplify_const_binary_operation): Likewise.
9121 (simplify_ternary_operation): Likewise.
9123 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9124 Alan Hayward <alan.hayward@arm.com>
9125 David Sherwood <david.sherwood@arm.com>
9127 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9128 (int_mem_offset): ...this new function.
9129 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9130 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9131 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
9134 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9135 Alan Hayward <alan.hayward@arm.com>
9136 David Sherwood <david.sherwood@arm.com>
9138 * lower-subreg.c (interesting_mode_p): New function.
9139 (compute_costs, find_decomposable_subregs, decompose_register)
9140 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9141 (resolve_clobber, dump_choices): Use it.
9143 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9144 Alan Hayward <alan.hayward@arm.com>
9145 David Sherwood <david.sherwood@arm.com>
9147 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9149 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9150 Alan Hayward <alan.hayward@arm.com>
9151 David Sherwood <david.sherwood@arm.com>
9153 * alias.c (find_base_value, find_base_term): Only process integer
9154 truncations. Check the precision rather than the size.
9156 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9157 Alan Hayward <alan.hayward@arm.com>
9158 David Sherwood <david.sherwood@arm.com>
9160 * machmode.h (is_narrower_int_mode): New function
9161 * optabs.c (expand_float, expand_fix): Use it.
9162 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9164 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9165 Alan Hayward <alan.hayward@arm.com>
9166 David Sherwood <david.sherwood@arm.com>
9168 * rtl.h (narrower_subreg_mode): New function.
9169 * ira-color.c (update_costs_from_allocno): Use it.
9171 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9172 Alan Hayward <alan.hayward@arm.com>
9173 David Sherwood <david.sherwood@arm.com>
9175 * optabs-query.h (convert_optab_p): New function, split out from...
9176 (convert_optab_handler): ...here.
9177 (widening_optab_handler): Delete.
9178 (find_widening_optab_handler): Remove permit_non_widening parameter.
9179 (find_widening_optab_handler_and_mode): Likewise. Provide an
9180 override that operates on mode class wrappers.
9181 * optabs-query.c (widening_optab_handler): Delete.
9182 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9183 parameter. Assert that the two modes are the same class and that
9184 the "from" mode is narrower than the "to" mode. Use
9185 convert_optab_handler instead of widening_optab_handler.
9186 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9187 instead of widening_optab_handler.
9188 * expr.c (expand_expr_real_2): Update calls to
9189 find_widening_optab_handler.
9190 * optabs.c (expand_widen_pattern_expr): Likewise.
9191 (expand_binop_directly): Take the insn_code as a parameter.
9192 (expand_binop): Only call find_widening_optab_handler for
9193 conversion optabs; use optab_handler otherwise. Update calls
9194 to find_widening_optab_handler and expand_binop_directly.
9195 Use convert_optab_handler instead of widening_optab_handler.
9196 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9197 find_widening_optab_handler and use scalar_mode rather than
9199 (convert_plusminus_to_widen): Likewise.
9201 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9202 Alan Hayward <alan.hayward@arm.com>
9203 David Sherwood <david.sherwood@arm.com>
9205 * machmode.h (fixed_size_mode): New class.
9206 * rtl.h (get_pool_mode): Return fixed_size_mode.
9207 * gengtype.c (main): Add fixed_size_mode.
9208 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9209 (get_raw_arg_mode): Likewise.
9210 * doc/tm.texi: Regenerate.
9211 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9212 * targhooks.c (default_get_reg_raw_mode): Likewise.
9213 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9214 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9215 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9216 (msp430_get_raw_result_mode): Likewise.
9217 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9218 * dbxout.c (dbxout_parms): Require fixed-size modes.
9219 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9220 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9221 * omp-low.c (lower_oacc_reductions): Likewise.
9222 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9223 (simplify_subreg): Update accordingly.
9224 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9225 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9226 that aren't fixed-size.
9227 (get_pool_mode): Return a fixed_size_mode.
9228 (output_constant_pool_2): Take a fixed_size_mode.
9230 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9231 Alan Hayward <alan.hayward@arm.com>
9232 David Sherwood <david.sherwood@arm.com>
9234 * doc/rtl.texi (vec_series): Document.
9235 (const): Say that the operand can be a vec_series.
9236 * rtl.def (VEC_SERIES): New rtx code.
9237 * rtl.h (const_vec_series_p_1): Declare.
9238 (const_vec_series_p): New function.
9239 * emit-rtl.h (gen_const_vec_series): Declare.
9240 (gen_vec_series): Likewise.
9241 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9242 (gen_vec_series): Likewise.
9243 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9244 * simplify-rtx.c (simplify_unary_operation): Handle negations
9246 (simplify_binary_operation_series): New function.
9247 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9248 (test_vector_ops_series): New function.
9249 (test_vector_ops): Call it.
9250 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9251 gen_const_vec_series.
9252 (altivec_lvsr): Likewise.
9253 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9255 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9256 Alan Hayward <alan.hayward@arm.com>
9257 David Sherwood <david.sherwood@arm.com>
9259 * doc/rtl.texi (const): Update description of address constants.
9260 Say that vector constants are allowed too.
9261 * common.md (E, F): Use CONSTANT_P instead of checking for
9263 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9264 checking for CONST_VECTOR.
9265 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9267 * expr.c (expand_expr_real_2): Check for vector modes instead
9268 of checking for CONST_VECTOR.
9269 * rtl.h (const_vec_p): New function.
9270 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9271 (unwrap_const_vec_duplicate): Handle them here too.
9273 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9274 David Malcolm <dmalcolm@redhat.com>
9275 Alan Hayward <alan.hayward@arm.com>
9276 David Sherwood <david.sherwood@arm.com>
9278 * rtl.h (vec_duplicate_p): New function.
9279 * selftest-rtl.c (assert_rtx_eq_at): New function.
9280 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9281 (assert_rtx_eq_at): Declare.
9282 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9283 * selftest-run-tests.c (selftest::run_tests): Call it.
9284 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9285 (simplify_unary_operation_1): Recursively handle vector duplicates.
9286 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9288 (simplify_subreg): Handle subregs of vector duplicates.
9289 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9290 (selftest::simplify_rtx_c_tests): New functions.
9292 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9293 Alan Hayward <alan.hayward@arm.com>
9294 David Sherwood <david.sherwood@arm.com>
9296 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9297 (gen_vec_duplicate): Likewise.
9298 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9300 (gen_const_vector): ...here.
9301 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9302 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9303 whose elements are all equal.
9304 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9305 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9306 (simplify_relational_operation): Likewise.
9307 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9309 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9310 (aarch64_expand_vector_init): Likewise.
9311 * config/arm/arm.c (neon_vdup_constant): Likewise.
9312 (neon_expand_vector_init): Likewise.
9313 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9314 (arm_block_set_unaligned_vect): Likewise.
9315 (arm_block_set_aligned_vect): Likewise.
9316 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9317 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9318 (expand_vec_perm_even_odd_pack): Likewise.
9319 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9320 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9321 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9322 gen_const_vec_duplicate.
9323 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9324 * config/mips/mips.c (mips_gen_const_int_vector): Use
9325 gen_const_vec_duplicate.
9326 (mips_expand_vector_init): Use CONST0_RTX.
9327 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9328 (define_split): Use gen_const_vec_duplicate.
9329 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9330 (define_split): Use gen_const_vec_duplicate.
9331 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9332 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9333 * config/spu/spu.c (spu_const): Likewise.
9335 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9336 Alan Hayward <alan.hayward@arm.com>
9337 David Sherwood <david.sherwood@arm.com>
9339 * combine.c (can_change_dest_mode): Reject changes in
9340 REGMODE_NATURAL_SIZE.
9342 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9344 * configure.ac (--enable-libssp): New.
9345 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9346 trying to determine target-specific default. Adjust indentation.
9347 * configure: Regenerated.
9348 * doc/install.texi (Configuration): Expand --disable-libssp
9351 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9353 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9356 2017-10-31 Martin Jambor <mjambor@suse.cz>
9359 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9361 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9363 * auto-profile.c (autofdo_source_profile::read): Use
9364 UNKNOWN_LOCATION rather than 0.
9365 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9366 (warning_at): ...this overload.
9367 (warning_at_rich_loc_n): Rename to...
9368 (warning_n): ...this overload.
9369 (error_at_rich_loc): Rename to...
9370 (error_at): ...this overload.
9371 (pedwarn_at_rich_loc): Rename to...
9372 (pedwarn): ...this overload.
9373 (permerror_at_rich_loc): Rename to...
9374 (permerror): ...this overload.
9375 (inform_at_rich_loc): Rename to...
9376 (inform): ...this overload.
9377 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9378 (diagnostic_n_impl_richloc): Rename to...
9379 (diagnostic_n_impl): ...this rich_location *-based decl.
9380 (inform_at_rich_loc): Rename to...
9381 (inform): ...this, and add an assertion.
9382 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9383 (warning_at_rich_loc): Rename to...
9384 (warning_at): ...this, and add an assertion.
9385 (warning_at_rich_loc_n): Rename to...
9386 (warning_n): ...this, and add an assertion.
9387 (warning_n): Update location_t-based implementation for removal of
9388 location_t-based diagnostic_n_impl.
9389 (pedwarn_at_rich_loc): Rename to...
9390 (pedwarn): ...this, and add an assertion.
9391 (permerror_at_rich_loc): Rename to...
9392 (permerror): ...this, and add an assertion.
9393 (error_n): Update for removal of location_t-based diagnostic_n_impl.
9394 (error_at_rich_loc): Rename to...
9395 (error_at): ...this, and add an assertion.
9396 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
9397 (driver::do_spec_on_infiles): Likewise.
9398 * substring-locations.c (format_warning_va): Update for renaming
9399 of inform_at_rich_loc.
9401 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
9403 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
9404 _Float<N>X built-in functions so that the variant without the
9405 "__builtin_" prefix is only enabled for the GNU C and Objective C
9406 languages when they are in non-strict ANSI/ISO mode.
9407 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
9408 * target.def (floatn_builtin_p): Add a target hook to control
9409 whether _Float<N> and _Float<N>X built-in functions without the
9410 "__builtin_" prefix are enabled, and return true for C and
9411 Objective C in the default hook. Include langhooks.h in
9413 * targhooks.h (default_floatn_builtin_p): Likewise.
9414 * targhooks.c (default_floatn_builtin_p): Likewise.
9415 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
9416 floatn_builtin_p target hook.
9417 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
9419 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
9420 Eric Botcazou <ebotcazou@adacore.com>
9422 PR rtl-optimization/81803
9423 * lra-constraints.c (curr_insn_transform): Also reload the whole
9424 register for a strict subreg no wider than a word if this is for
9425 a WORD_REGISTER_OPERATIONS target.
9427 2017-10-31 Jason Merrill <jason@redhat.com>
9429 * gdbinit.in: Skip over inlines from timevar.h.
9431 2017-10-31 Martin Liska <mliska@suse.cz>
9433 * doc/gcov.texi: Document new option.
9434 * gcov.c (print_usage): Likewise print it.
9435 (process_args): Support the argument.
9436 (format_count): New function.
9437 (format_gcov): Use the function.
9439 2017-10-31 Martin Liska <mliska@suse.cz>
9441 * gcov.c (struct name_map): do not use typedef.
9442 Define operator== and operator<.
9443 (name_search): Remove.
9444 (name_sort): Remove.
9445 (main): Do not allocate names.
9446 (process_file): Add vertical space.
9447 (generate_results): Use std::find.
9448 (release_structures): Do not release memory.
9449 (find_source): Use std::find.
9451 2017-10-31 Martin Liska <mliska@suse.cz>
9453 * gcov.c (struct line_info): Remove it's typedef.
9454 (line_info::line_info): Add proper ctor.
9455 (line_info::has_block): Do not use a typedef.
9456 (struct source_info): Do not use typedef.
9457 (circuit): Likewise.
9458 (get_cycles_count): Likewise.
9459 (output_intermediate_file): Iterate via vector iterator.
9460 (add_line_counts): Use std::vector methods.
9461 (accumulate_line_counts): Likewise.
9462 (output_lines): Likewise.
9464 2017-10-31 Martin Liska <mliska@suse.cz>
9466 * gcov.c (struct source_info): Remove typedef.
9467 (source_info::source_info): Add proper ctor.
9468 (accumulate_line_counts): Use struct, not it's typedef.
9469 (output_gcov_file): Likewise.
9470 (output_lines): Likewise.
9471 (main): Do not allocate an array.
9472 (output_intermediate_file): Use size of vector container.
9473 (process_file): Resize the vector.
9474 (generate_results): Do not preallocate, use newly added vector
9476 (release_structures): Do not release sources.
9477 (find_source): Use vector methods.
9478 (add_line_counts): Do not use typedef.
9480 2017-10-31 Martin Liska <mliska@suse.cz>
9482 * doc/gcov.texi: Document that.
9483 * gcov.c (add_line_counts): Mark lines with a non-executed
9485 (output_line_beginning): Handle such lines.
9486 (output_lines): Pass new argument.
9487 (output_intermediate_file): Print it in intermediate format.
9489 2017-10-31 Martin Liska <mliska@suse.cz>
9491 * color-macros.h: New file.
9492 * diagnostic-color.c: Factor out color related to macros to
9494 * doc/gcov.texi: Document -k option.
9495 * gcov.c (INCLUDE_STRING): Include string.h.
9496 (print_usage): Add -k option.
9497 (process_args): Parse it.
9498 (pad_count_string): New function.
9499 (output_line_beginning): Likewise.
9500 (DEFAULT_LINE_START): New macro.
9501 (output_lines): Support color output.
9503 2017-10-31 Martin Liska <mliska@suse.cz>
9505 PR gcov-profile/82633
9506 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
9507 their interaction with GCOV infrastructure.
9508 * configure.ac: Add -fkeep-{inline,static}-functions to
9510 * configure: Regenerate.
9512 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
9515 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
9517 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
9520 * config/rs6000/rs6000.md (allocate_stack): Force update interval
9521 into a register if it does not fit into an immediate offset field.
9523 2017-10-31 Olivier Hainque <hainque@adacore.com>
9525 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
9527 2017-10-31 Julia Koval <julia.koval@intel.com>
9529 * config.gcc: Add gfniintrin.h.
9530 * config/i386/gfniintrin.h: New.
9531 * config/i386/i386-builtin-types.def
9532 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
9533 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
9534 __builtin_ia32_vgf2p8affineinvqb_v32qi,
9535 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
9536 __builtin_ia32_vgf2p8affineinvqb_v16qi,
9537 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
9538 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
9539 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
9540 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
9541 V64QI_FTYPE_V64QI_V64QI_INT): New types.
9542 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
9543 * config/i386/immintrin.h: Include gfniintrin.h.
9544 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
9546 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
9548 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
9550 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9552 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
9553 (arm_ashldi3_1bit): Remove pattern.
9554 (ashrdi3): Remove shift by 1 expansion.
9555 (arm_ashrdi3_1bit): Remove pattern.
9556 (lshrdi3): Remove shift by 1 expansion.
9557 (arm_lshrdi3_1bit): Remove pattern.
9558 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
9559 cost of ashldi3 by 1.
9560 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
9561 (<shift>di3_neon): Likewise.
9563 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
9565 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
9566 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
9567 and (*aarch64_simd_mov<VQ:mode>).
9568 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
9569 pattern alternative.
9570 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
9571 attributes to match pattern alternative.
9573 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
9575 * config.gcc (powerpc*-*-*): Add emmintrin.h.
9576 * config/rs6000/emmintrin.h: New file.
9577 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
9579 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
9581 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
9582 * (movdi_vfp_cortexa8): Remove pattern.
9584 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9586 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
9588 (Specific, alpha*-dec-osf5.1): Remove.
9589 (Specific, mips-sgi-irix5): Remove.
9590 (Specific, mips-sgi-irix6): Remove.
9592 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9595 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
9596 arguments to clear_bit_region_be.
9598 2017-10-30 Jim Wilson <wilson@tuliptree.org>
9600 * gimplify.c: Include memmodel.h.
9602 2017-10-30 Martin Jambor <mjambor@suse.cz>
9604 * omp-grid.c (grid_attempt_target_gridification): Also insert a
9605 condition whether loop should be executed at all.
9607 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
9609 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9610 gimple folding of vec_madd() intrinsics.
9611 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
9612 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
9613 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
9615 2017-10-30 Richard Biener <rguenther@suse.de>
9617 PR tree-optimization/82762
9619 2017-10-23 Richard Biener <rguenther@suse.de>
9621 PR tree-optimization/82129
9623 2017-08-01 Richard Biener <rguenther@suse.de>
9625 PR tree-optimization/81181
9626 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9627 (compute_antic): ... end of iteration here.
9629 2017-10-30 Joseph Myers <joseph@codesourcery.com>
9631 * doc/invoke.texi (C Dialect Options): Document -std=c17,
9632 -std=iso9899:2017 and -std=gnu17.
9633 * doc/standards.texi (C Language): Document C17 support.
9634 * doc/cpp.texi (Overview): Mention -std=c17.
9635 (Standard Predefined Macros): Document C11 and C17 values of
9636 __STDC_VERSION__. Do not refer to C99 support as incomplete.
9637 * doc/extend.texi (Inline): Do not list individual options for
9638 standards newer than C99.
9639 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
9641 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
9644 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
9646 * asan.c (asan_finish_file): Align asan globals array by shadow
9649 2017-10-30 Jakub Jelinek <jakub@redhat.com>
9652 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
9653 (struct store_immediate_info): Add bitregion_start and bitregion_end
9655 (store_immediate_info::store_immediate_info): Add brs and bre
9656 arguments and initialize bitregion_{start,end} from those.
9657 (struct merged_store_group): Add bitregion_start, bitregion_end,
9658 align_base and mask fields. Drop unnecessary struct keyword from
9659 struct store_immediate_info. Add do_merge method.
9660 (clear_bit_region_be): Use memset instead of loop storing zeros.
9661 (merged_store_group::do_merge): New method.
9662 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
9663 stores as long as the surrounding bitregions have no gaps.
9664 (merged_store_group::merge_overlapping): Use do_merge.
9665 (merged_store_group::apply_stores): Test that bitregion_{start,end}
9666 is byte aligned, rather than requiring that start and width are
9667 byte aligned. Drop unnecessary struct keyword from
9668 struct store_immediate_info. Allocate and populate also mask array.
9669 Make start of the arrays relative to bitregion_start rather than
9670 start and size them according to bitregion_{end,start} difference.
9671 (struct imm_store_chain_info): Drop unnecessary struct keyword from
9672 struct store_immediate_info.
9673 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
9674 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
9675 struct keyword from struct store_immediate_info.
9676 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
9677 between stores as long as the surrounding bitregions have no gaps.
9679 (struct split_store): Add orig non-static data member.
9680 (split_store::split_store): Initialize orig to false.
9681 (find_constituent_stmts): Return store_immediate_info *, non-NULL
9682 if there is exactly a single original stmt. Change stmts argument
9683 to pointer from reference, if NULL, don't push anything to it. Add
9684 first argument, use it to optimize skipping over orig stmts that
9685 are known to be before bitpos already. Simplify.
9686 (split_group): Return unsigned int count how many stores are or
9687 would be needed rather than a bool. Add allow_unaligned argument.
9688 Change split_stores argument from reference to pointer, if NULL,
9689 only do a dry run computing how many stores would be produced.
9690 Rewritten algorithm to use both alignment and misalign if
9691 !allow_unaligned and handle bitfield stores with gaps.
9692 (imm_store_chain_info::output_merged_store): Set start_byte_pos
9693 from bitregion_start instead of start. Compute allow_unaligned
9694 here, if true, do 2 split_group dry runs to compute which one
9695 produces fewer stores and prefer aligned if equal. Punt if
9696 new count is bigger or equal than original before emitting any
9697 statements, rather than during that. Remove no longer needed
9698 new_ssa_names tracking. Replace num_stmts with
9699 split_stores.length (). Use 32-bit stack allocated entries
9700 in split_stores auto_vec. Try to reuse original store lhs/rhs1
9701 if possible. Handle bitfields with gaps.
9702 (pass_store_merging::execute): Ignore bitsize == 0 stores.
9703 Compute bitregion_{start,end} for the stores and construct
9704 store_immediate_info with that. Formatting fixes.
9706 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
9709 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
9710 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
9712 2017-10-29 Jim Wilson <wilson@tuliptree.org>
9714 * gimplify.c: Include tm_p.h.
9716 * common.opt (gcoff): Re-add as ignored option.
9717 (gcoff1, gcoff2, gcoff3): Likewise.
9719 * Makefile.in (OBJS): Delete sdbout.o.
9720 (GTFILES): Delete $(srcdir)/sdbout.c.
9721 * debug.h: Delete sdb_debug_hooks.
9722 * final.c: Delete sdbout.h include.
9723 (final_scan_insn): Delete SDB_DEBUG check.
9724 (rest_of_clean_state): Likewise.
9725 * output.h: Delete sdb_begin_function_line.
9728 * toplev.c: Delete sdbout.h include.
9729 (process_options): Delete SDB_DEBUG check.
9730 * tree-core.h (tree_type_common): Delete pointer field of
9732 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
9733 TYPE_SYMTAB_POINTER.
9734 * tree.h (TYPE_SYMTAB_POINTER): Delete.
9735 (TYPE_SYMTAB_IS_POINTER): Delete.
9736 (TYPE_SYMTAB_IS_DIE): Renumber.
9737 * xcoffout.c: Refer to former sdbout.c file.
9738 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
9740 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
9741 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
9742 (Debugging Options): Delete -gcoff.
9743 (-gstabs): Delete SDB reference.
9745 (-gcoff@var{level}): Delete.
9746 * doc/passes.texi (Debugging information output): Delete SDB and
9747 sdbout.c references.
9748 * doc/tm.texi: Regenerate.
9749 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
9750 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
9752 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
9753 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
9754 (SDB_DEBUGGING_INFO): Delete.
9755 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
9756 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
9757 * target.def (output_source_filename): Delete COFF reference.
9759 * common.opt (gcoff): Delete.
9760 (gxcoff+): Update Negative chain.
9761 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
9763 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
9764 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
9765 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
9767 (expand_function_start): Change sdb reference to past tense.
9768 (expand_function_end): Change sdb reference to past tense.
9769 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
9770 * opts.c (debug_type_names): Delete coff entry.
9771 (common_handle_option): Delete OPT_gcoff case.
9772 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
9774 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
9775 * config/cris/cris.h: Delete SDB reference in comment.
9776 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
9777 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
9778 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
9779 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
9781 (ix86_expand_prologue): Likewise.
9782 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
9783 * config/ia64/ia64.h: Likewise.
9784 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
9785 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
9787 * config/mmix/mmix.h: Likewise.
9788 * config/nds32/nds32.c: Likewise.
9789 * config/stormy/storym16.h: Likewise.
9790 * config/visium/visium.h: Likewise.
9791 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
9793 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
9795 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
9796 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
9797 FRAME_POINTER_REGNUM point at high end of local var area.
9799 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
9801 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
9802 Move comment around. Do not reset best_edge for a copiable
9803 destination if the copy would cause a partition change.
9804 (better_edge_p): Remove redundant check.
9806 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9808 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
9810 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
9812 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
9813 for math functions that have _Float<N> and _Float<N>X variants.
9814 (mathfn_built_in_2): Add support for math functions that have
9815 _Float<N> and _Float<N>X variants.
9816 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
9817 (expand_builtin_mathfn_ternary): Add support for fma with
9818 _Float<N> and _Float<N>X variants.
9819 (expand_builtin): Likewise.
9820 (fold_builtin_3): Likewise.
9821 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
9822 create math function _Float<N> and _Float<N>X variants as external
9824 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
9825 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
9826 the __builtin_ prefix and if not strict ansi, without the prefix.
9827 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
9828 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
9829 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
9830 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
9831 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
9832 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
9833 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
9834 function signatures for fma _Float<N> and _Float<N>X variants.
9835 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
9836 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
9837 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
9838 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
9839 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
9840 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
9841 * gencfn-macros.c (print_case_cfn): Add support for math functions
9842 that have _Float<N> and _Float<N>X variants.
9843 (print_define_operator_list): Likewise.
9844 (fltfn_suffixes): Likewise.
9846 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
9847 for math functions that have _Float<N> and _Float<N>X variants.
9848 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
9849 and _Float<N>X variants.
9850 (COPYSIGN): Likewise.
9853 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
9854 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
9856 (integer_valued_read_call_p): Likewise.
9857 * fold-const-call.c (fold_const_call_ss): Likewise.
9858 (fold_const_call_sss): Add support for copysign, fmin, and fmax
9859 _Float<N> and _Float<N>X variants.
9860 (fold_const_call_ssss): Add support for fma _Float<N> and
9861 _Float<N>X variants.
9862 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
9863 support for copysign and fma _Float<N> and _Float<N>X variants.
9864 (backprop::process_builtin_call_use): Likewise.
9865 * tree-call-cdce.c (can_test_argument_range); Add support for
9866 sqrt _Float<N> and _Float<N>X variants.
9867 (edom_only_function): Likewise.
9868 (get_no_error_domain): Likewise.
9869 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
9870 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
9871 copysign _Float<N> and _Float<N>X variants.
9872 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
9873 handled by machine independent code.
9874 (FMAF128): Likewise.
9875 * doc/cpp.texi (Common Predefined Macros): Document defining
9876 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
9877 fma _Float<N> and _Float<N>X variants.
9879 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
9882 * config/i386/i386-modes.def (CCFPU): Remove definition.
9883 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
9884 (ix86_cc_modes_compatible): Ditto.
9885 (ix86_expand_carry_flag_compare): Ditto.
9886 (ix86_expand_int_movcc): Ditto.
9887 (ix86_expand_int_addcc): Ditto.
9888 (ix86_reverse_condition): Ditto.
9889 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
9890 Return true/false for unordered/ordered fp comparisons.
9891 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
9892 (ix86_prepare_fp_compare_args): Update for rename.
9893 (ix86_expand_fp_compare): Update for rename. Generate unordered
9894 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
9895 (ix86_expand_sse_compare_and_jump): Ditto.
9896 * config/i386/predicates.md (fcmov_comparison_operator):
9897 Remove CCFPU mode handling.
9898 (ix86_comparison_operator): Ditto.
9899 (ix86_carry_flag_operator): Ditto.
9900 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
9901 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
9902 (*cmpu<mode>_cc_i387): Ditto.
9903 (FPCMP): Remove mode iterator.
9904 (unord): Remove mode attribute.
9905 (unord_subst): New define_subst transformation
9906 (unord): New define_subst attribute.
9908 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
9909 (*cmpi<unord>xf_i387): Ditto.
9910 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
9911 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
9912 using unord_subst transformation.
9913 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
9914 (round_saeonly): Also handle CCFP mode.
9915 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
9916 Remove UNSPEC_SAHF unspec handling.
9918 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9920 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
9922 2017-10-27 Jeff Law <law@redhat.com>
9924 * gimple-ssa-sprintf.c: Include domwalk.h.
9925 (class sprintf_dom_walker): New class, derived from dom_walker.
9926 (sprintf_dom_walker::before_dom_children): New function.
9927 (struct call_info): Moved into sprintf_dom_walker class
9928 (compute_formath_length, handle_gimple_call): Likewise.
9929 (sprintf_length::execute): Call the dominator walker rather
9930 than walking the statements.
9932 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
9933 gimple statement locations.
9934 (check_array_bounds): Corresponding changes. Get the statement's
9935 location directly from wi->stmt.
9937 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
9940 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
9942 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
9944 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
9945 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
9947 2017-10-27 Jakub Jelinek <jakub@redhat.com>
9950 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
9951 * config/i386/i386.c (maybe_get_pool_constant): Removed.
9952 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
9953 maybe_get_pool_constant.
9954 * config/i386/predicates.md (zero_extended_scalar_load_operand):
9957 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9959 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
9960 2.26 caveat. Update gas and gld versions.
9961 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
9964 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9966 * cgraph.h (set_malloc_flag): Declare.
9967 * cgraph.c (set_malloc_flag_1): New function.
9968 (set_malloc_flag): Likewise.
9969 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
9970 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
9972 (read_ipa_call_summary): Add support for reading is_return_callee.
9973 (write_ipa_call_summary): Stream is_return_callee.
9974 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
9975 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
9976 ipa-prop.h, ipa-fnsummary.h.
9977 (pure_const_names): Change to static.
9978 (malloc_state_e): Define.
9979 (malloc_state_names): Define.
9980 (funct_state_d): Add field malloc_state.
9981 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
9982 (check_retval_uses): New function.
9983 (malloc_candidate_p): Likewise.
9984 (analyze_function): Add support for malloc attribute.
9985 (pure_const_write_summary): Stream malloc_state.
9986 (pure_const_read_summary): Add support for reading malloc_state.
9987 (dump_malloc_lattice): New function.
9988 (propagate_malloc): New function.
9989 (warn_function_malloc): New function.
9990 (ipa_pure_const::execute): Call propagate_malloc and
9991 ipa_free_fn_summary.
9992 (pass_local_pure_const::execute): Add support for malloc attribute.
9993 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
9994 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
9996 2017-10-27 Martin Liska <mliska@suse.cz>
9998 PR gcov-profile/82457
9999 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10000 for fork-like functions to be properly instrumented.
10002 2017-10-27 Richard Biener <rguenther@suse.de>
10004 PR middle-end/81659
10005 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10006 info when we redirected EH.
10008 2017-10-26 Michael Collison <michael.collison@arm.com>
10010 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10012 (<optab>_trunchf<GPI:mode>2: New pattern.
10013 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10014 * config/aarch64/iterators.md (wv): New mode attribute.
10015 (vf, VF): New mode attributes.
10016 (vgp, VGP): New mode attributes.
10017 (s): Update attribute with SImode and DImode prefixes.
10019 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10021 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10022 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10023 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10024 (nios2_option_overide): Initialize it. Don't allow R0-relative
10025 addressing with PIC.
10026 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10027 (nios2_symbolic_constant_p): Likewise.
10028 (nios2_legitimate_address_p): Likewise.
10029 (nios2_r0rel_section_name_p): New.
10030 (nios2_symbol_ref_in_r0rel_data_p): New.
10031 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10032 (r0rel_constant_p): New.
10033 (nios2_print_operand_address): Handle r0rel_constant_p.
10034 (nios2_cdx_narrow_form_p): Likewise.
10035 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10036 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10037 (Nios II Options): Document -mr0rel-sec.
10039 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10041 * config/nios2/nios2.c: Include xregex.h.
10042 (nios2_gprel_sec_regex): New.
10043 (nios2_option_overide): Initialize it. Don't allow GP-relative
10044 addressing with PIC.
10045 (nios2_small_section_name_p): Check for regex match.
10046 * config/nios2/nios2.opt (mgprel-sec=): New option.
10047 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10048 (Nios II Options): Document -mgprel-sec.
10050 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10052 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10054 2017-10-26 Tom de Vries <tom@codesourcery.com>
10056 PR tree-optimization/82707
10057 * gimple.c (gimple_copy): Fix unsharing of
10058 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10060 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10062 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10063 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10064 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10065 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10066 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10067 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10068 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10069 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10070 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10071 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10072 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10073 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10074 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10075 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10076 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10077 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10078 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10080 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10082 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10084 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10085 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10086 warning to rs6000.c. Remove the Undocumented flag, since it has
10088 (-mabi=ibmlongdouble): Likewise.
10089 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10090 already set, set the default format for long double.
10091 (rs6000_debug_reg_global): Print whether long double is IBM or
10093 (rs6000_option_override_internal): Rework setting long double
10094 format. Only warn if the user is changing the long double default
10095 and they did not use -Wno-psabi.
10096 * doc/invoke.texi (PowerPC options): Update the documentation for
10097 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10099 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10100 Alan Hayward <alan.hayward@arm.com>
10101 David Sherwood <david.sherwood@arm.com>
10103 * rtl.h (wider_subreg_mode): New function.
10104 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10105 rather than an unsigned int *.
10106 * ira-color.c (regno_max_ref_width): Replace with...
10107 (regno_max_ref_mode): ...this new variable.
10108 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10109 Use wider_subreg_mode.
10110 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10111 rather than an unsigned int *.
10112 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10113 (process_alt_operands): Likewise.
10114 (invariant_p): Likewise.
10115 * lra-spills.c (assign_mem_slot): Likewise.
10116 (add_pseudo_to_slot): Likewise.
10117 * lra.c (collect_non_operand_hard_regs): Likewise.
10118 (add_regs_to_insn_regno_info): Likewise.
10119 * reload1.c (regno_max_ref_width): Replace with...
10120 (regno_max_ref_mode): ...this new variable.
10121 (reload): Update accordingly. Update call to
10122 ira_sort_regnos_for_alter_reg.
10123 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10124 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10125 (scan_paradoxical_subregs): Likewise.
10127 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10129 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10130 (aarch64_frame): Add emit_frame_chain boolean.
10131 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10132 Move eh_return case to aarch64_layout_frame.
10133 (aarch64_layout_frame): Initialize emit_frame_chain.
10134 (aarch64_expand_prologue): Use emit_frame_chain.
10136 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10138 * config/aarch64/aarch64.c (aarch64_layout_frame):
10139 Ensure LR is always stored at the bottom of the callee-saves.
10140 Remove rarely used frame layout which saves callee-saves at top of
10141 frame, so the store of LR can be used as a valid probe in all cases.
10143 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10145 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10146 Improve unaligned TImode/TFmode base/offset split.
10148 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10149 Alan Hayward <alan.hayward@arm.com>
10150 David Sherwood <david.sherwood@arm.com>
10152 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10153 * combine.c (find_single_use_1): Likewise.
10154 (expand_field_assignment): Likewise.
10155 (move_deaths): Likewise.
10156 * lra-constraints.c (simplify_operand_subreg): Likewise.
10157 (curr_insn_transform): Likewise.
10158 * lra.c (collect_non_operand_hard_regs): Likewise.
10159 (add_regs_to_insn_regno_info): Likewise.
10160 * rtlanal.c (reg_referenced_p): Likewise.
10161 (covers_regno_no_parallel_p): Likewise.
10163 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10165 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10166 Don't print any bits outside the precision of the value.
10167 * wide-int.cc (test_printing): Add some new tests.
10169 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10171 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10172 supports -xbrace_comment option.
10173 * configure: Regenerate.
10174 * config.in: Regenerate.
10175 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10176 (ASM_CPU_SPEC): Use it.
10178 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10180 * target.def (static_rtx_alignment): New hook.
10181 * targhooks.h (default_static_rtx_alignment): Declare.
10182 * targhooks.c (default_static_rtx_alignment): New function.
10183 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10184 * doc/tm.texi: Regenerate.
10185 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10186 instead of targetm.constant_alignment. Remove call to
10187 set_mem_attributes.
10188 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10189 (cris_preferred_mininum_alignment): New function, split out from...
10190 (cris_constant_alignment): ...here.
10191 (cris_static_rtx_alignment): New function.
10192 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10194 (ix86_constant_alignment): ...here.
10195 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10196 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10197 (mmix_static_rtx_alignment): New function.
10198 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10199 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10201 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10204 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10205 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10207 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10209 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10210 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10211 mark as a sign-extending load.
10212 (local_pic_load_u): Define.
10214 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10216 PR middle-end/82062
10217 * fold-const.c (operand_equal_for_comparison_p): Also return true
10218 if ARG0 is a simple variant of ARG1 with narrower precision.
10219 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10221 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10223 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10225 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10226 scatter_static, scatter_per_elt.
10227 * x86-tune-costs.h: Add new cost entries.
10229 2017-10-25 Richard Biener <rguenther@suse.de>
10231 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10232 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10233 class pass_fre): Move to ...
10234 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10235 class pass_fre): ... here and adjust for statistics.
10237 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10240 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10241 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10242 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10243 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10246 2017-10-25 Richard Biener <rguenther@suse.de>
10248 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10249 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10250 eliminate_push_avail, eliminate_insert): Move inside...
10251 (class eliminate_dom_walker): ... this class in preparation
10253 (fini_eliminate): Remove by merging with ...
10254 (eliminate): ... this function. Adjust for class changes.
10255 (pass_pre::execute): Remove fini_eliminate call.
10256 (pass_fre::execute): Likewise.
10258 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10261 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10262 (VPERMI2, VPERMI2I): New mode iterators.
10263 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10264 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10266 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10267 mode iterator. Remove 3 old define_insn patterns.
10268 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10269 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10270 VPERMI2 mode iterator, remove the other two expanders.
10271 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10272 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10273 instructions, remove the other two patterns.
10274 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10275 mode iterator, remove the other two patterns.
10276 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10277 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10278 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10279 and adjust argument order accordingly.
10280 (ix86_expand_vec_perm): Adjust caller.
10281 (expand_vec_perm_1): Likewise.
10282 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10283 (expand_vec_perm_vpermt2_vpshub2): ... this.
10284 (ix86_expand_vec_perm_const_1): Adjust caller.
10285 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10288 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10289 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10290 (vec_shl_<mode>): Remove unused expander.
10291 (avx512bw_<shift_insn><mode>3): New define_insn.
10292 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10293 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10295 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10298 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10301 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10303 PR rtl-optimization/82396
10304 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10305 (autopref_multipass_init): Simplify initialization.
10306 (autopref_rank_data): Simplify sort order.
10307 * gcc/sched-int.h (autopref_multipass_data_): Remove
10308 multi_mem_insn_p, min_offset and max_offset.
10310 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10312 PR middle-end/60580
10313 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10314 Check special value of flag_omit_frame_pointer.
10315 (aarch64_can_eliminate): Likewise.
10316 (aarch64_override_options_after_change_1): Simplify handling of
10317 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10319 2017-10-24 Richard Biener <rguenther@suse.de>
10321 PR tree-optimization/82697
10322 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10323 zero for conditional load and unconditional store.
10325 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10327 * doc/install.texi: Document bootstrap-cet.
10329 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10332 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10333 ENDBR instruction at function entrance if function is only
10336 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10339 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10340 patterns to better describe from which operation the CF is computed.
10341 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10342 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10343 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10344 is 0, use _0 suffixed expanders instead of emitting a comparison
10347 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10349 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10350 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10352 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10354 PR middle-end/82569
10355 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10356 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10357 * loop-iv.c (iv_get_reaching_def): Likewise.
10358 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10359 variable is promoted and the partition contains undefined values.
10361 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10363 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10365 (nios2_address_cost): Define.
10366 (nios2_legitimize_address): Recognize (exp + constant) directly.
10367 (TARGET_ADDRESS_COST): Define.
10369 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10371 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10372 (nios2_symbolic_memory_operand_p): Declare.
10373 (nios2_split_large_constant): Declare.
10374 (nios2_split_symbolic_memory_operand): Declare.
10375 * config/nios2/nios2.c: Adjust includes.
10376 (nios2_symbolic_constant_allowed): New.
10377 (nios2_symbolic_constant_p): New.
10378 (nios2_plus_symbolic_constant_p): New.
10379 (nios2_valid_addr_expr_p): Recognize addresses involving
10380 symbolic constants.
10381 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10382 (nios2_symbolic_memory_operand_p): New.
10383 (nios2_large_constant_p): New.
10384 (nios2_split_large_constant): New.
10385 (nios2_split_plus_large_constant): New.
10386 (nios2_split_symbolic_memory_operand): New.
10387 (nios2_legitimize_address): Code refactoring. Handle addresses
10388 involving symbolic constants.
10389 (nios2_emit_move_sequence): Likewise.
10390 (nios2_print_operand): Improve error output.
10391 (nios2_print_operand_address): Handle LO_SUM.
10392 (nios2_cdx_narrow_form_p): Likewise.
10393 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
10394 operands involving symbolic constants.
10395 (movhi_internal, movsi_internal): Likewise.
10396 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
10397 (extendhisi2, extendqi<mode>2): Likewise.
10399 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10401 * tree-pass.h (PROP_rtl_split_insns): Define.
10402 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
10404 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10406 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
10408 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10411 * target.def (const_not_ok_for_debug_p): Default to
10412 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
10413 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
10414 * targhooks.c (default_const_not_ok_for_debug_p): New function.
10415 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
10416 which targetm.const_not_ok_for_debug_p returned true.
10417 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
10419 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
10421 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
10422 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
10423 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
10425 (ix86_const_not_ok_for_debug_p): New function.
10426 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
10427 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
10429 2017-10-23 David Malcolm <dmalcolm@redhat.com>
10432 * system.h: Conditionally include "unique-ptr.h" if
10433 INCLUDE_UNIQUE_PTR is defined.
10434 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
10435 of defining INCLUDE_UNIQUE_PTR before including "system.h".
10437 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
10439 * config/rl78/rl78.md: New define_expand "subdi3".
10441 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
10444 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
10445 DF_REF_INSN if DF_REF_INSN_INFO is false.
10447 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
10449 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
10450 xmm_move instead of sse_move.
10451 (sse_store_index): New function.
10452 (ix86_register_move_cost): Be more sensible about mismatch stall;
10453 model AVX moves correctly; make difference between sse->integer and
10455 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
10456 moves; make difference between SSE and AVX.
10457 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
10458 and zmm_move. Increase size of sse load and store tables;
10459 add unaligned load and store tables; add ssemmx_to_integer.
10460 * x86-tune-costs.h: Update all entries according to real
10461 move latencies from Agner Fog's manual and chip documentation.
10463 2017-10-23 Jakub Jelinek <jakub@redhat.com>
10466 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
10467 * config/i386/constraints.md (Wf): New constraint.
10468 * config/i386/i386.md (UNSPEC_SBB): New unspec.
10469 (cmp<dwi>_doubleword): Removed.
10470 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
10471 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
10472 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
10473 expand with cmp<dwi>_doubleword. For LTU and GEU use
10474 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
10476 * common.opt (gcolumn-info): Enable by default.
10477 * doc/invoke.texi (gcolumn-info): Document new default.
10479 2017-10-23 Richard Biener <rguenther@suse.de>
10481 PR tree-optimization/82672
10482 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
10483 Fold the stmt if we propagated into it.
10485 2017-10-23 Richard Biener <rguenther@suse.de>
10487 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
10488 (bitmap_remove_expr_from_set): ... this. All callers call this
10489 for non-constant values.
10490 (bitmap_set_subtract): Rename to...
10491 (bitmap_set_subtract_expressions): ... this. Adjust and
10493 (bitmap_set_contains_value): Remove superfluous check.
10494 (bitmap_set_replace_value): Inline into single caller ...
10495 (bitmap_value_replace_in_set): ... here and simplify.
10496 (dependent_clean): Merge into ...
10497 (clean): ... this using an overload. Adjust.
10498 (prune_clobbered_mems): Adjust.
10499 (compute_antic_aux): Likewise.
10500 (compute_partial_antic_aux): Likewise.
10502 2017-10-23 Richard Biener <rguenther@suse.de>
10504 PR tree-optimization/82129
10506 2017-08-01 Richard Biener <rguenther@suse.de>
10508 PR tree-optimization/81181
10509 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10510 (compute_antic): ... end of iteration here.
10512 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10514 * target.def (starting_frame_offset): New hook.
10515 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
10516 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
10517 * doc/tm.texi.in: Regenerate.
10518 * hooks.h (hook_hwi_void_0): Declare.
10519 * hooks.c (hook_hwi_void_0): New function.
10520 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
10521 STARTING_FRAME_OFFSET.
10522 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
10523 * reload1.c (reload): Likewise.
10524 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
10525 instead of STARTING_FRAME_OFFSET.
10526 * function.c (try_fit_stack_local): Likewise.
10527 (assign_stack_local_1): Likewise
10528 (instantiate_virtual_regs): Likewise.
10529 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
10530 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
10531 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
10532 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
10533 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
10534 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
10535 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
10536 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
10537 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
10538 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
10539 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
10540 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
10541 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
10542 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
10543 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
10544 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
10545 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
10546 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
10547 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
10548 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
10549 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
10550 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
10551 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
10552 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
10553 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
10554 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
10555 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
10556 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
10557 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
10558 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
10559 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
10560 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
10561 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
10562 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
10563 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
10564 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
10565 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
10566 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
10567 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
10568 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
10569 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
10570 * config/avr/avr.c (avr_starting_frame_offset): Make static and
10571 return a HOST_WIDE_INT.
10572 (avr_builtin_setjmp_frame_value): Use it instead of
10573 STARTING_FRAME_OFFSET.
10574 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10575 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
10576 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
10578 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10579 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
10580 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
10581 (TARGET_CONSTANT_ALIGNMENT): Redefine.
10582 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
10583 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
10584 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10585 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
10586 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
10587 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10588 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
10589 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
10591 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10592 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
10593 * config/mips/mips.c (mips_compute_frame_info): Refer to
10594 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10595 (mips_starting_frame_offset): New function.
10596 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10597 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
10598 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
10599 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
10600 and return a HOST_WIDE_INT.
10601 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10602 (mmix_initial_elimination_offset): Refer to
10603 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
10604 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
10605 * config/pa/pa.c (pa_starting_frame_offset): New function.
10606 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
10607 (pa_expand_prologue): Likewise.
10608 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10609 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
10610 !FRAME_GROWS_DOWNWARD handling to...
10611 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10612 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
10613 !FRAME_GROWS_DOWNWARD handling to...
10614 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10615 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
10616 !FRAME_GROWS_DOWNWARD handling to...
10617 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10618 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
10620 (rs6000_starting_frame_offset): New function.
10621 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
10622 !FRAME_GROWS_DOWNWARD handling to...
10623 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10624 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
10625 !FRAME_GROWS_DOWNWARD handling to...
10626 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10627 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
10628 !FRAME_GROWS_DOWNWARD handling to...
10629 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
10630 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
10631 (rs6000_starting_frame_offset): New function.
10632 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
10633 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
10634 * config/vax/vax.c (vax_starting_frame_offset): New function.
10635 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
10636 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10637 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
10638 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
10639 (TARGET_STARTING_FRAME_OFFSET): Redefine.
10640 * system.h (STARTING_FRAME_OFFSET): Poison.
10642 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10644 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
10645 SCALAR_TYPE_MODE instead of TYPE_MODE.
10647 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10648 Alan Hayward <alan.hayward@arm.com>
10649 David Sherwood <david.sherwood@arm.com>
10651 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
10653 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10654 Alan Hayward <alan.hayward@arm.com>
10655 David Sherwood <david.sherwood@arm.com>
10657 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
10659 2017-10-23 Richard Biener <rguenther@suse.de>
10661 PR tree-optimization/82129
10662 * tree-ssa-pre.c (bitmap_set_and): Remove.
10663 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
10664 canonicalizing expressions in the set to those with lowest
10665 ID rather than taking that from the first edge.
10667 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
10669 * combine.c (rtx_equal_for_field_assignment_p): Use
10670 byte_lowpart_offset.
10672 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10673 Alan Hayward <alan.hayward@arm.com>
10674 David Sherwood <david.sherwood@arm.com>
10676 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
10677 to a SUBREG_PROMOTED_VAR.
10679 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10680 Alan Hayward <alan.hayward@arm.com>
10681 David Sherwood <david.sherwood@arm.com>
10683 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
10684 (expand_debug_source_expr): Likewise.
10685 * combine.c (combine_simplify_rtx): Likewise.
10686 * cse.c (fold_rtx): Likewise.
10687 * optabs.c (expand_float): Likewise.
10688 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10689 (simplify_binary_operation_1): Likewise.
10691 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10692 Alan Hayward <alan.hayward@arm.com>
10693 David Sherwood <david.sherwood@arm.com>
10695 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
10696 (record_promoted_value): Likewise.
10697 * expr.c (expand_expr_real_2): Likewise.
10698 * ree.c (update_reg_equal_equiv_notes): Likewise.
10699 (combine_set_extension): Likewise.
10700 * rtlanal.c (low_bitmask_len): Likewise.
10701 * simplify-rtx.c (neg_const_int): Likewise.
10702 (simplify_binary_operation_1): Likewise.
10704 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10705 Alan Hayward <alan.hayward@arm.com>
10706 David Sherwood <david.sherwood@arm.com>
10708 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
10709 * regcprop.c (maybe_mode_change): Likewise.
10710 * reload1.c (alter_reg): Likewise.
10712 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10714 * inchash.h (inchash::hash::add_wide_int): New function.
10715 * lto-streamer-out.c (hash_tree): Use it.
10717 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
10719 * inchash.h (inchash::hash::add_wide_int): Rename to...
10720 (inchash::hash::add_hwi): ...this.
10721 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
10722 (polymorphic_call_target_hasher::hash): Likewise.
10723 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
10724 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
10725 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
10726 * lto-streamer-out.c (hash_tree): Likewise.
10727 * optc-save-gen.awk: Likewise.
10728 * tree.c (add_expr): Likewise.
10730 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10733 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
10734 for ordered inequality comparisons even with TARGET_IEEE_FP.
10736 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
10739 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
10740 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
10741 Expand with cmp<dwi>_doubleword.
10743 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10745 * extend.texi: Add x86 specific to 'nocf_check' attribute.
10746 List CET intrinsics.
10747 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
10748 specific to -fcf-protection option.
10750 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10752 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
10753 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
10754 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
10755 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
10756 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
10757 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
10758 (extra_objs): Add cet.o for Linux/x86 targets.
10759 (tmake_file): Add i386/t-cet for Linux/x86 targets.
10760 * config/i386/cet.c: New file.
10761 * config/i386/cetintrin.h: Likewise.
10762 * config/i386/t-cet: Likewise.
10763 * config/i386/cpuid.h (bit_SHSTK): New.
10764 (bit_IBT): Likewise.
10765 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
10766 pass IBT and SHSTK bits.
10767 * config/i386/i386-builtin-types.def
10768 (VOID_FTYPE_UNSIGNED_PVOID): New.
10769 (VOID_FTYPE_UINT64_PVOID): Likewise.
10770 * config/i386/i386-builtin.def: Add CET intrinsics.
10771 * config/i386/i386-c.c (ix86_target_macros_internal): Add
10772 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
10773 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
10774 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
10776 * config/i386/i386.c (rest_of_insert_endbranch): New.
10777 (pass_data_insert_endbranch): Likewise.
10778 (pass_insert_endbranch): Likewise.
10779 (make_pass_insert_endbranch): Likewise.
10780 (ix86_notrack_prefixed_insn_p): Likewise.
10781 (ix86_target_string): Add -mibt, -mshstk flags.
10782 (ix86_option_override_internal): Add flag_cf_protection
10784 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
10785 (ix86_print_operand): Add 'notrack' prefix output.
10786 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
10787 (ix86_expand_builtin): Expand CET intrinsics.
10788 (x86_output_mi_thunk): Add 'endbranch' instruction.
10789 * config/i386/i386.h (TARGET_IBT): New.
10790 (TARGET_IBT_P): Likewise.
10791 (TARGET_SHSTK): Likewise.
10792 (TARGET_SHSTK_P): Likewise.
10793 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
10794 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
10795 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
10796 (builtin_setjmp_setup): New pattern.
10797 (builtin_longjmp): Likewise.
10798 (rdssp<mode>): Likewise.
10799 (incssp<mode>): Likewise.
10800 (saveprevssp): Likewise.
10801 (rstorssp): Likewise.
10802 (wrss<mode>): Likewise.
10803 (wruss<mode>): Likewise.
10804 (setssbsy): Likewise.
10805 (clrssbsy): Likewise.
10806 (nop_endbr): Likewise.
10807 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
10809 * config/i386/immintrin.h: Include <cetintrin.h>.
10810 * config/i386/linux-common.h
10811 (file_end_indicate_exec_stack_and_cet): New prototype.
10812 (TARGET_ASM_FILE_END): New.
10814 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10816 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
10817 latencies instead of having separate table; make difference between
10818 integer and float costs.
10819 * i386.h (processor_costs): Remove scalar_stmt_cost,
10820 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
10821 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
10823 * x86-tune-costs.h: Remove entries which has been removed in
10824 procesor_costs from all tables; make cond_taken_branch_cost
10825 and cond_not_taken_branch_cost COST_N_INSNS based.
10827 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
10829 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
10831 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10833 * config/i386/i386.md (isa): Remove fma_avx512f.
10834 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
10835 <avx512>_fmadd_<mode>_mask3<round_name>,
10836 <avx512>_fmsub_<mode>_mask<round_name>,
10837 <avx512>_fmsub_<mode>_mask3<round_name>,
10838 <avx512>_fnmadd_<mode>_mask<round_name>,
10839 <avx512>_fnmadd_<mode>_mask3<round_name>,
10840 <avx512>_fnmsub_<mode>_mask<round_name>,
10841 <avx512>_fnmsub_<mode>_mask3<round_name>,
10842 <avx512>_fmaddsub_<mode>_mask<round_name>,
10843 <avx512>_fmaddsub_<mode>_mask3<round_name>,
10844 <avx512>_fmsubadd_<mode>_mask<round_name>,
10845 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
10846 (*vec_widen_umult_even_v16si<mask_name>,
10847 *vec_widen_smult_even_v16si<mask_name>): Likewise.
10848 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
10850 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10852 * extend.texi: Add 'nocf_check' documentation.
10853 * gimple.texi: Add second parameter to
10854 gimple_build_call_from_tree.
10855 * invoke.texi: Add -fcf-protection documentation.
10856 * rtl.texi: Add REG_CALL_NOTRACK documenation.
10858 2017-10-20 Richard Biener <rguenther@suse.de>
10860 PR tree-optimization/82473
10861 * tree-vect-loop.c (vectorizable_reduction): Properly get at
10862 the largest input type.
10864 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
10866 * c-attribs.c (handle_nocf_check_attribute): New function.
10867 (c_common_attribute_table): Add 'nocf_check' handling.
10868 * gimple-parser.c: Add second argument NULL to
10869 gimple_build_call_from_tree.
10870 * attrib.c (comp_type_attributes): Check nocf_check attribute.
10871 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
10873 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
10874 * common.opt: Add fcf-protection flag.
10875 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
10876 * flag-types.h: Add enum cf_protection_level.
10877 * gimple.c (gimple_build_call_from_tree): Add second parameter.
10878 Add 'nocf_check' attribute propagation to gimple call.
10879 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
10880 (gimple_build_call_from_tree): Update prototype.
10881 (gimple_call_nocf_check_p): New function.
10882 (gimple_call_set_nocf_check): Likewise.
10883 * gimplify.c: Add second argument to gimple_build_call_from_tree.
10884 * ipa-icf.c: Add nocf_check attribute in statement hash.
10885 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
10886 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
10887 * toplev.c (process_options): Add flag_cf_protection handling.
10889 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10891 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
10893 2017-10-20 Richard Biener <rguenther@suse.de>
10895 PR tree-optimization/82603
10896 * tree-if-conv.c (predicate_mem_writes): Make sure to only
10897 remove false predicated stores.
10899 2017-10-20 Richard Biener <rguenther@suse.de>
10901 * graphite-isl-ast-to-gimple.c
10902 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
10903 Remove return value and simplify, dump copied stmt after lhs
10905 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
10906 Reduce dump verbosity.
10907 (gsi_insert_earliest): Likewise.
10908 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
10909 * graphite.c (print_global_statistics): Adjust dumping.
10910 (print_graphite_scop_statistics): Likewise.
10911 (print_graphite_statistics): Do not dump loops here.
10912 (graphite_transform_loops): But here.
10914 2017-10-20 Nicolas Roche <roche@adacore.com>
10916 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
10917 * configure: Regenerate.
10919 2017-10-20 Jakub Jelinek <jakub@redhat.com>
10922 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
10923 functions when optimizing replace GIMPLE_RETURN stmts with
10924 calls to __builtin_unreachable ().
10927 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
10928 for -fsanitize=thread link of executables.
10929 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
10930 link of executables.
10933 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
10934 New mode iterators.
10935 (<shift_insn><mode>3<mask_name>): Change the last of the 3
10936 define_insns for logical vector shifts to use VI248_AVX512BW
10937 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
10938 condition, useless isa and prefix attributes. Change the first
10939 2 of these define_insns to ...
10940 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
10941 define_insn for avx512vl.
10942 (<shift_insn><mode>3): ... and this, new define_insn without
10943 masking for non-avx512vl.
10946 * config/i386/sse.md (*andnot<mode>3,
10947 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
10948 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
10949 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
10950 not applied use empty suffix even for TARGET_AVX512VL.
10951 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
10952 is applied, supply evex,evex or evex,evex,evex instead of just
10955 2017-10-20 Julia Koval <julia.koval@intel.com>
10957 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
10958 (OPTION_MASK_ISA_GFNI_UNSET): New.
10959 (ix86_handle_option): Handle OPT_mgfni.
10960 * config/i386/cpuid.h (bit_GFNI): New.
10961 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
10962 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
10963 * config/i386/i386.c (ix86_target_string): Add -mgfni.
10964 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
10965 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
10966 * config/i386/i386.opt: Add mgfni.
10968 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
10970 * config/msp430/msp430.c (msp430_option_override): Disable
10971 -fdelete-null-pointer-checks.
10972 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
10974 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10976 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
10977 of x87 and SSE instructions.
10979 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
10981 * asan.c (create_cond_insert_point): Do not update edge count.
10982 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
10983 (afdo_propagate_circuit): Likewise.
10984 (afdo_calculate_branch_prob): Likewise.
10985 (afdo_annotate_cfg): Likewise.
10986 * basic-block.h (struct edge_def): Remove count.
10987 (edge_def::count): New accessor.
10988 * bb-reorder.c (rotate_loop): Update.
10989 (find_traces_1_round): Update.
10990 (connect_traces): Update.
10991 (sanitize_hot_paths): Update.
10992 * cfg.c (unchecked_make_edge): Update.
10993 (make_single_succ_edge): Update.
10994 (check_bb_profile): Update.
10995 (dump_edge_info): Update.
10996 (update_bb_profile_for_threading): Update.
10997 (scale_bbs_frequencies_int): Update.
10998 (scale_bbs_frequencies_gcov_type): Update.
10999 (scale_bbs_frequencies_profile_count): Update.
11000 (scale_bbs_frequencies): Update.
11001 * cfganal.c (connect_infinite_loops_to_exit): Update.
11002 * cfgbuild.c (compute_outgoing_frequencies): Update.
11003 (find_many_sub_basic_blocks): Update.
11004 * cfgcleanup.c (try_forward_edges): Update.
11005 (try_crossjump_to_edge): Update
11006 * cfgexpand.c (expand_gimple_cond): Update
11007 (expand_gimple_tailcall): Update
11008 (construct_exit_block): Update
11009 * cfghooks.c (verify_flow_info): Update
11010 (redirect_edge_succ_nodup): Update
11011 (split_edge): Update
11012 (make_forwarder_block): Update
11013 (duplicate_block): Update
11014 (account_profile_record): Update
11015 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11016 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11017 * cfgloopmanip.c (scale_loop_profile): Update.
11019 (lv_adjust_loop_entry_edge): Update.
11020 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11021 (force_nonfallthru_and_redirect): Update.
11022 (purge_dead_edges): Update.
11023 (rtl_flow_call_edges_add): Update.
11024 * cgraphunit.c (init_lowered_empty_function): Update.
11025 (cgraph_node::expand_thunk): Update.
11026 * gimple-pretty-print.c (dump_probability): Update.
11027 (dump_edge_probability): Update.
11028 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11029 * haifa-sched.c (sched_create_recovery_edges): Update.
11030 * hsa-gen.c (convert_switch_statements): Update.
11031 * ifcvt.c (dead_or_predicable): Update.
11032 * ipa-inline-transform.c (inline_transform): Update.
11033 * ipa-split.c (split_function): Update.
11034 * ipa-utils.c (ipa_merge_profiles): Update.
11035 * loop-doloop.c (add_test): Update.
11036 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11037 * lto-streamer-in.c (input_cfg): Update.
11038 (input_function): Update.
11039 * lto-streamer-out.c (output_cfg): Update.
11040 * modulo-sched.c (sms_schedule): Update.
11041 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11042 * predict.c (maybe_hot_edge_p): Update.
11043 (unlikely_executed_edge_p): Update.
11044 (probably_never_executed_edge_p): Update.
11045 (dump_prediction): Update.
11046 (drop_profile): Update.
11047 (propagate_unlikely_bbs_forward): Update.
11048 (determine_unlikely_bbs): Update.
11049 (force_edge_cold): Update.
11050 * profile.c (compute_branch_probabilities): Update.
11051 * reg-stack.c (better_edge): Update.
11052 * shrink-wrap.c (handle_simple_exit): Update.
11053 * tracer.c (better_p): Update.
11054 * trans-mem.c (expand_transaction): Update.
11055 (split_bb_make_tm_edge): Update.
11056 * tree-call-cdce.c: Update.
11057 * tree-cfg.c (gimple_find_sub_bbs): Update.
11058 (gimple_split_edge): Update.
11059 (gimple_duplicate_sese_region): Update.
11060 (gimple_duplicate_sese_tail): Update.
11061 (gimple_flow_call_edges_add): Update.
11062 (insert_cond_bb): Update.
11063 (execute_fixup_cfg): Update.
11064 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11065 * tree-complex.c (expand_complex_div_wide): Update.
11066 * tree-eh.c (lower_resx): Update.
11067 (unsplit_eh): Update.
11068 (cleanup_empty_eh_move_lp): Update.
11069 * tree-inline.c (copy_edges_for_bb): Update.
11070 (freqs_to_counts): Update.
11071 (copy_cfg_body): Update.
11072 * tree-ssa-dce.c (remove_dead_stmt): Update.
11073 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11074 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11075 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11076 (unloop_loops): Update.
11077 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11078 * tree-ssa-loop-split.c (connect_loops): Update.
11079 (split_loop): Update.
11080 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11081 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11082 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11083 * tree-ssa-reassoc.c (branch_fixup): Update.
11084 * tree-ssa-tail-merge.c (replace_block_by): Update.
11085 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11086 (compute_path_counts): Update.
11087 (update_profile): Update.
11088 (recompute_probabilities): Update.
11089 (update_joiner_offpath_counts): Update.
11090 (estimated_freqs_path): Update.
11091 (freqs_to_counts_path): Update.
11092 (clear_counts_path): Update.
11093 (ssa_fix_duplicate_block_edges): Update.
11094 (duplicate_thread_path): Update.
11095 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11096 (case_bit_test_cmp): Update.
11097 (collect_switch_conv_info): Update.
11098 (gen_inbound_check): Update.
11099 (do_jump_if_equal): Update.
11100 (emit_cmp_and_jump_insns): Update.
11101 * tree-tailcall.c (decrease_profile): Update.
11102 (eliminate_tail_call): Update.
11103 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11104 (vect_do_peeling): Update.
11105 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11106 * ubsan.c (ubsan_expand_null_ifn): Update.
11107 (ubsan_expand_ptr_ifn): Update.
11108 * value-prof.c (gimple_divmod_fixed_value): Update.
11109 (gimple_mod_pow2): Update.
11110 (gimple_mod_subtract): Update.
11111 (gimple_ic): Update.
11112 (gimple_stringop_fixed_value): Update.
11114 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11117 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11119 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11121 PR rtl-optimization/82395
11122 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11123 based on non_spilled_static_chain_regno_p.
11125 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11127 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11128 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11129 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11131 2017-10-19 Martin Sebor <msebor@redhat.com>
11133 PR tree-optimization/82596
11134 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11136 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11138 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11139 (get_mem_refs_of_builtin_call): Likewise.
11140 * builtins.c (expand_builtin_apply): Adjust call to
11141 allocate_dynamic_stack_space.
11142 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11143 the third argument to allocate_dynamic_stack_space, otherwise -1.
11144 (expand_builtin): Deal with all alloca variants.
11145 (is_inexpensive_builtin): Likewise.
11146 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11147 * calls.c (special_function_p): Deal with all alloca variants.
11148 (initialize_argument_information): Adjust call to
11149 allocate_dynamic_stack_space.
11150 (expand_call): Likewise.
11151 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11152 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11153 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11154 use it for the stack usage computation.
11155 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11156 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11157 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11158 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11159 (in_loop_p): Remove first argument and useless check.
11160 (pass_walloca::execute): Remove useless test and adjust call to above.
11161 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11162 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11163 (gimplify_call_expr): Deal with all alloca variants.
11164 * hsa-gen.c (gen_hsa_alloca): Likewise.
11165 (gen_hsa_insns_for_call): Likewise.
11166 * ipa-pure-const.c (special_builtin_state): Likewise.
11167 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11168 * tree-object-size.c (alloc_object_size): Likewise.
11169 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11170 (call_may_clobber_ref_p_1): Likewise.
11171 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11172 (ccp_fold_stmt): Likewise.
11173 (optimize_stack_restore): Likewise.
11174 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11175 (mark_all_reaching_defs_necessary_1): Likewise.
11176 (propagate_necessity): Likewise.
11177 (eliminate_unnecessary_stmts): Likewise.
11178 * tree.c (build_common_builtin_nodes): Build
11179 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11180 (build_alloca_call_expr): New function.
11181 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11182 (CASE_BUILT_IN_ALLOCA): Likewise.
11183 (build_alloca_call_expr): Declare.
11184 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11186 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11189 * dwarf2out.c (new_die_raw): New static inline function.
11190 (new_die): Use it to create the DIE.
11191 (add_AT_external_die_ref): Likewise.
11192 (clone_die): Likewise.
11193 (clone_as_declaration): Likewise.
11194 (dwarf2out_vms_debug_main_pointer): Likewise.
11195 (base_type_die): Likewise. Remove early return for corner cases.
11196 Do not call add_pubtype on the DIE here.
11197 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11198 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11199 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11200 native order exists for base types, attach the DIE manually and call
11201 add_pubtype on it. Do not equate a reverse order DIE to the type.
11203 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11205 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11206 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11208 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11210 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11212 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11213 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11214 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11215 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11216 * opts.c (sanitizer_opts): Add builtin.
11217 * ubsan.c (instrument_builtin): New function.
11218 (pass_ubsan::execute): Call it.
11219 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11220 * doc/invoke.texi: Document -fsanitize=builtin.
11222 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11223 builtins, store max (log2 (align), 0) into uchar field instead of
11224 align into uptr field.
11225 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11226 store uchar 0 field instead of uptr 0 field.
11227 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11228 instead of passing one address of struct with 2 locations pass
11229 two addresses of structs with 1 location each.
11230 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11231 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11232 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11233 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11234 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11235 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11236 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11237 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11239 2017-10-19 Martin Liska <mliska@suse.cz>
11242 * file-find.c (remove_prefix): Remove.
11243 * file-find.h (remove_prefix): Likewise.
11244 * gcc-ar.c: Remove smartness of lookup.
11246 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11248 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11249 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11250 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11252 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11255 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11256 (setcc + and to xor + setcc): New peephole2.
11258 2017-10-19 Tom de Vries <tom@codesourcery.com>
11260 * doc/sourcebuild.texi (Test Directives, Variants of
11261 dg-require-support): Add dg-require-stack-size.
11263 2017-10-19 Martin Liska <mliska@suse.cz>
11266 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11267 that have a large alignment.
11268 (gimplify_target_expr): Likewise.
11270 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11272 PR rtl-optimization/82602
11273 * ira.c (rtx_moveable_p): Return false for volatile asm.
11275 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11278 * config/i386/i386-modes.def (CCGZ): New CC mode.
11279 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11280 * config/i386/predicates.md (ix86_comparison_operator):
11282 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11283 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11284 with double-word subtraction.
11285 (put_condition_code): Handle CCGZmode.
11287 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11289 * wide-int.cc (debug (const wide_int &)): New.
11290 (debug (const wide_int *)): New.
11291 (debug (const widest_int &)): New.
11292 (debug (const widest_int *)): New.
11294 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11296 PR middle-end/82556
11297 * lra-constraints.c (curr_insn_transform): Use non-input operand
11298 instead of output one for matched reload.
11300 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11302 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11303 (tree-ssa-loop-ivopts.h): New header file.
11304 (struct builtin_info): New fields.
11305 (classify_builtin_1): Compute and record base and offset parts for
11306 memset builtin partition by calling strip_offset.
11307 (offset_cmp, fuse_memset_builtins): New functions.
11308 (finalize_partitions): Fuse adjacent memset partitions by calling
11310 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11311 Expose the interface.
11312 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11314 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11316 PR tree-optimization/82574
11317 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11318 that data reference must be executed exactly once per iteration
11319 against the outermost loop in nest.
11320 (classify_partition): Update call to above function.
11322 2017-10-18 Richard Biener <rguenther@suse.de>
11324 PR tree-optimization/82591
11325 * graphite.c (graphite_transform_loops): Move code gen message
11327 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11328 Here. Handle scop_to_isl_ast failing.
11329 (scop_to_isl_ast): Limit the number of ISL operations.
11331 2017-10-18 Richard Biener <rguenther@suse.de>
11333 * graphite-isl-ast-to-gimple.c
11334 (translate_isl_ast_to_gimple::set_rename): Simplify.
11335 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11336 (graphite_copy_stmts_from_block): ... here.
11337 (copy_bb_and_scalar_dependences): Simplify.
11338 (add_parameters_to_ivs_params): Canonicalize.
11339 (generate_entry_out_of_ssa_copies): Simplify.
11340 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11341 by passing in ISL dimension.
11342 (parameter_index_in_region_1): Rename to ...
11343 (parameter_index_in_region): ... this.
11344 (extract_affine): Adjust assert, pass down parameter index.
11345 (add_param_constraints): Use range-info when available.
11346 (build_scop_context): Adjust.
11347 * sese.c (new_sese_info): Adjust.
11348 (free_sese_info): Likewise.
11349 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11350 Remove unused typedefs.
11351 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11353 2017-10-18 Martin Liska <mliska@suse.cz>
11355 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11357 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11359 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11360 (s390_sched_init): Do not reset s390_sched_state if we entered the
11361 current basic block via a fallthru edge and all others are unlikely.
11363 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11365 * config/s390/s390.c (NUM_SIDES): New variable.
11366 (LONGRUNNING_THRESHOLD): New variable.
11367 (LATENCY_FACTOR): New variable.
11368 (s390_sched_score): Decrease score for long-running instructions on
11370 (s390_sched_variable_issue): Perform bookkeeping for long-running
11373 2017-10-18 Richard Biener <rguenther@suse.de>
11375 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11376 Simplify with removal of the parameter rename map.
11377 (set_rename): Likewise.
11378 (should_copy_to_new_region): Likewise.
11379 (graphite_copy_stmts_from_block): Likewise.
11380 (copy_bb_and_scalar_dependences): Remove initialization of
11381 unused copied_bb_map.
11382 (copy_def): Remove.
11383 (copy_internal_parameters): Likewise.
11384 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11385 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11386 Use INTEGRAL_TYPE_P.
11387 (parameter_index_in_region_1): Rename to ...
11388 (assign_parameter_index_in_region): ... this. Assert we have
11389 a parameter we handle.
11390 (scan_tree_for_params): Adjust.
11391 * sese.h (parameter_rename_map_t): Remove.
11392 (struct sese_info_t): Remove unused parameter_rename_map and
11393 copied_bb_map members.
11394 * sese.c (new_sese_info): Adjust.
11395 (free_sese_info): Likewise.
11397 2017-10-18 Martin Liska <mliska@suse.cz>
11400 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
11401 on an abnormal edge.
11403 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
11405 * doc/invoke.texi (ffunction-sections and fdata-sections):
11408 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11410 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
11411 the use statement can throw internally.
11413 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
11415 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
11416 any RTX present on the RHS of a SET.
11417 * compare-elim.c (try_eliminate_compare): Restore comment.
11419 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11421 * langhooks.h (struct lang_hooks): Document that tree_size langhook
11422 may be also called on tcc_type nodes.
11423 * langhooks.c (lhd_tree_size): Likewise.
11425 2017-10-17 David Malcolm <dmalcolm@redhat.com>
11427 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
11428 format_warning_at_substring.
11429 (maybe_warn): Convert source_range * param to a location_t. Pass
11430 UNKNOWN_LOCATION rather than NULL to fmtwarn.
11431 (format_directive): Remove code to extract source_ranges and
11432 source_range * in favor of just a location_t.
11433 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
11435 * substring-locations.c (format_warning_va): Convert
11436 source_range * param to a location_t.
11437 (format_warning_at_substring): Likewise.
11438 * substring-locations.h (format_warning_va): Likewise.
11439 (format_warning_at_substring): Likewise.
11441 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
11443 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
11445 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
11446 and scatter/gather ops.
11448 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
11449 vec_gather_load and vec_scatter_store.
11450 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
11451 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
11452 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
11453 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
11454 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
11455 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
11457 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
11459 * reg-stack.c (compare_for_stack_reg): Add bool argument.
11460 Detect FTST instruction and handle its register pops. Only pop
11461 second operand if can_pop_second_op is true.
11462 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
11463 set can_pop_second_op to false in the compare_for_stack_reg call.
11465 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
11466 output_fp_compare for stack register operands.
11467 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
11468 instructions here. Do not emit stack register pops here. Assert
11469 that FCOMPP pops next to top stack register. Rewrite function.
11471 2017-10-17 Nathan Sidwell <nathan@acm.org>
11473 PR middle-end/82577
11474 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
11475 use DECL_ASSEMBLER_NAME_RAW.
11477 PR middle-end/82546
11478 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
11481 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
11482 Wilco Dijkstra <wilco.dijkstra@arm.com>
11484 * builtins.c (expand_builtin_update_setjmp_buf): Add a
11485 converstion to Pmode from the buf_addr.
11487 2017-10-17 Richard Biener <rguenther@suse.de>
11489 * graphite-dependences.c (scop_get_reads_and_writes): Change
11490 output parameters to references.
11492 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
11494 PR 71026/tree-optimization
11495 * fold-const.c (distribute_real_division): Removed.
11496 (fold_binary_loc): Remove calls to distribute_real_divison.
11498 2017-10-17 Richard Biener <rguenther@suse.de>
11500 * graphite-scop-detection.c
11501 (scop_detection::stmt_has_simple_data_refs_p): Always use
11502 the full nest as region.
11503 (try_generate_gimple_bb): Likewise.
11504 * sese.c (scalar_evolution_in_region): Simplify now that
11505 SCEV can handle instantiation in regions.
11506 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
11507 in the non-loop part of a function if requested.
11509 2017-10-17 Richard Biener <rguenther@suse.de>
11511 PR tree-optimization/82563
11512 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
11514 (graphite_regenerate_ast_isl): Call it.
11515 * graphite-scop-detection.c (build_scops): Remove entry edge split.
11517 2017-10-17 Jakub Jelinek <jakub@redhat.com>
11519 PR tree-optimization/82549
11520 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
11521 Formatting fixes. Instead of calling make_bit_field_ref with negative
11524 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
11526 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
11527 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
11528 _mm_maskz_reduce_ss): New.
11529 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
11530 __builtin_ia32_reducess_mask): Ditto..
11531 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
11532 * config/i386/sse.md (reduces<mode>): Renamed to ...
11533 (reduces<mode><mask_scalar_name>): ... this.
11534 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
11536 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
11537 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
11539 2017-10-16 David Malcolm <dmalcolm@redhat.com>
11541 * Makefile.in (OBJS): Add unique-ptr-tests.o.
11542 * selftest-run-tests.c (selftest::run_tests): Call
11543 selftest::unique_ptr_tests_cc_tests.
11544 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
11545 * unique-ptr-tests.cc: New file.
11547 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
11550 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
11552 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
11553 (make_hard_regno_born, make_hard_regno_dead): Update
11554 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
11556 2017-10-16 Jeff Law <law@redhat.com>
11558 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11560 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
11562 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
11564 2017-10-16 Olivier Hainque <hainque@adacore.com>
11566 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
11567 with_cpu if we were configured for an e500v2 target cpu name.
11569 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
11571 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
11572 * doc/invoke.texi: Document +nodsp as a valid extension for
11575 2017-10-16 Martin Liska <mliska@suse.cz>
11577 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
11578 (test_set_range): Likewise.
11579 (test_range_functions): Rename to ...
11580 (test_bit_in_range): ... this.
11581 (sbitmap_c_tests): Add new test.
11583 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11585 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
11587 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
11588 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
11590 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11592 * config/aarch64/aarch64-builtins.c
11593 (aarch64_types_quadopu_lane_qualifiers): New.
11594 (TYPES_QUADOPU_LANE): New.
11595 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
11596 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
11597 (aarch64_<sur>dot_laneq<vsi2qi>): New.
11598 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
11599 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
11600 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
11601 (Vdottype, DOTPROD): New.
11602 (sur): Add SDOT and UDOT.
11604 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11606 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
11607 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
11608 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
11609 Add TARGET_DOTPROD.
11610 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
11611 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
11612 Enable TARGET_DOTPROD.
11613 (cortex-a75.cortex-a55): Likewise.
11614 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11616 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11618 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
11619 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
11621 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
11623 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
11624 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
11625 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
11626 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
11627 * config/arm/types.md (neon_dot, neon_dot_q): New.
11628 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
11630 2017-10-16 Tamar Christina <tamar.christina@arm.com>
11632 * config/arm/arm.h (TARGET_DOTPROD): New.
11633 * config/arm/arm.c (arm_arch_dotprod): New.
11634 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
11635 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
11636 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
11637 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
11638 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
11639 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
11640 * doc/invoke.texi (armv8.2-a): Document dotprod
11642 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11644 * i386.c (ix86_vec_cost): New function.
11645 (ix86_rtx_costs): Handle vector operations better.
11646 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
11647 * x86-tune-costs.h: Add new costs to all tables.
11649 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
11651 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
11653 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
11654 divsd, sqrtss and sqrtsd
11655 * x86-tune-costs.h: Add new entries to all costs.
11656 (znver1_cost): Fix to match real instruction latencies.
11658 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11659 Michael Collison <michael.collison@arm.com>
11661 * compare-elim.c: Include emit-rtl.h.
11662 (can_merge_compare_into_arith): New function.
11663 (try_validate_parallel): Likewise.
11664 (try_merge_compare): Likewise.
11665 (try_eliminate_compare): Call the above when no previous clobber
11667 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
11670 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11672 PR middle-end/62263
11673 PR middle-end/82498
11674 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
11675 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
11677 PR middle-end/62263
11678 PR middle-end/82498
11679 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
11680 to be any operand_equal_p operands. For & (B - 1) require
11681 B to be power of 2. Recognize
11682 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
11684 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11687 * optabs.c (expand_memory_blockage): Fix call of
11688 targetm.have_memory_blockage.
11690 2017-10-14 Jakub Jelinek <jakub@redhat.com>
11693 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
11694 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
11695 objects to extra_objs instead of overwriting it.
11697 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11699 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
11700 Use any_fp_register_operand as operand[3] predicate. Simplify
11701 equality test for operands[2] and operands[4] memory location.
11702 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
11703 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
11704 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
11705 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
11706 any_fp_register_operand as operand[1] predicate. Simplify
11707 equality test for operands[0] and operands[3] memory location.
11708 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
11709 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
11710 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
11712 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
11714 * target-insns.def: Add memory_blockage.
11715 * optabs.c (expand_memory_blockage): New function.
11716 (expand_asm_memory_barrier): Rename ...
11717 (expand_asm_memory_blockage): ... to this.
11718 (expand_mem_thread_fence): Call expand_memory_blockage
11719 instead of expand_asm_memory_barrier.
11720 (expand_mem_singnal_fence): Ditto.
11721 (expand_atomic_load): Ditto.
11722 (expand_atomic_store): Ditto.
11723 * doc/md.texi (Standard Pattern Names For Generation):
11724 Document memory_blockage instruction pattern.
11726 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
11728 * config/rl78/rl78.c (rl78_emit_libcall): New function.
11729 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
11730 * config/rl78/rl78.md: New define_expand "adddi3".
11732 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11734 * cfghooks.c (verify_flow_info): Disable check that all probabilities
11737 2017-10-13 Jeff Law <law@redhat.com>
11739 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
11741 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11744 * internal-fn.c (expand_mul_overflow): If both operands have
11745 the same highpart of -1 or 0 and the topmost bit of lowpart
11746 is different, overflow is if res <= 0 rather than res < 0.
11748 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
11750 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
11751 TARGET_P9_VECTOR code for unaligned_load case.
11753 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
11755 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
11757 2017-10-13 Nathan Sidwell <nathan@acm.org>
11759 * tree-core.h (tree_contains_struct): Make bool.
11760 * tree.c (tree_contains_struct): Likewise.
11761 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
11762 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
11763 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
11764 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
11765 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
11767 2017-10-13 Richard Biener <rguenther@suse.de>
11769 * graphite-isl-ast-to-gimple.c
11770 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
11771 parameters and dominance check.
11772 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
11773 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
11774 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
11775 Do not update SSA form here or do intermediate IL verification.
11776 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
11777 (graphite_initialize): Remove check on the number of loops in
11778 the function and inline into graphite_transform_loops.
11779 (graphite_finalize): Inline into graphite_transform_loops.
11780 (graphite_transform_loops): Perform SSA update and IL verification
11782 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
11784 2017-10-13 Richard Biener <rguenther@suse.de>
11786 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
11787 graphite_expression_type_precision): Avoid global constructor
11789 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
11790 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
11791 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
11792 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
11793 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
11794 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
11796 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
11799 * config/i386/i386.h (ix86_red_zone_size): New.
11800 * config/i386/i386.md (push peephole2s): Replace
11801 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
11803 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11804 Alan Hayward <alan.hayward@arm.com>
11805 David Sherwood <david.sherwood@arm.com>
11807 * combine.c (can_change_dest_mode): Reject changes in
11808 REGMODE_NATURAL_SIZE.
11810 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11811 Alan Hayward <alan.hayward@arm.com>
11812 David Sherwood <david.sherwood@arm.com>
11814 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
11815 (expand_debug_source_expr): Likewise.
11816 * combine.c (combine_simplify_rtx): Likewise.
11817 * cse.c (fold_rtx): Likewise.
11818 * fwprop.c (canonicalize_address): Likewise.
11819 * targhooks.c (default_shift_truncation_mask): Likewise.
11821 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11822 Alan Hayward <alan.hayward@arm.com>
11823 David Sherwood <david.sherwood@arm.com>
11825 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
11826 (widened_mode): Likewise.
11827 (expand_unop): Likewise.
11828 * ree.c (transform_ifelse): Likewise.
11829 (merge_def_and_ext): Likewise.
11830 (combine_reaching_defs): Likewise.
11831 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11833 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11834 Alan Hayward <alan.hayward@arm.com>
11835 David Sherwood <david.sherwood@arm.com>
11837 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
11838 * combine.c (gen_lowpart_for_combine): Likewise.
11839 * dwarf2out.c (rtl_for_decl_location): Likewise.
11840 * final.c (alter_subreg): Likewise.
11841 * rtlhooks.c (gen_lowpart_general): Likewise.
11842 (gen_lowpart_if_possible): Likewise.
11844 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
11845 Alan Hayward <alan.hayward@arm.com>
11846 David Sherwood <david.sherwood@arm.com>
11848 * calls.c (expand_call): Use subreg_lowpart_offset.
11849 * cse.c (cse_insn): Likewise.
11850 * regcprop.c (copy_value): Likewise.
11851 (copyprop_hardreg_forward_1): Likewise.
11853 2017-10-13 Jakub Jelinek <jakub@redhat.com>
11856 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
11857 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
11858 =Q constraints to +Q and into insn condition add check
11859 that operands[0] and operands[1] are equal.
11860 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
11861 =Q constraints to +Q and into insn condition add check
11862 that operands[0] is equal to either operands[1] or operands[2].
11865 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
11866 instead of handling MINUS_EXPR twice (once for each argument),
11867 canonicalize operand order and handle just once, use rtype where
11868 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
11871 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
11872 any values of __C while still being pattern recognizable as a simple
11873 rotate instruction.
11875 2017-10-13 Richard Biener <rguenther@suse.de>
11877 PR tree-optimization/82451
11879 2017-10-02 Richard Biener <rguenther@suse.de>
11881 PR tree-optimization/82355
11882 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11883 a mapping for the enclosing loop but avoid generating one for
11884 the loop tree root.
11885 (copy_bb_and_scalar_dependences): Remove premature codegen
11886 error on PHIs in blocks duplicated into multiple places.
11887 * graphite-scop-detection.c
11888 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11889 in the region use it as loop and nest to analyze the DR in.
11890 (try_generate_gimple_bb): Likewise.
11891 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11892 (add_loop_constraints): For blocks in a loop not in the region
11893 create a dimension with a single iteration.
11894 * sese.h (gbb_loop_at_index): Remove assert.
11896 * cfgloop.c (loop_preheader_edge): For the loop tree root
11897 return the single successor of the entry block.
11898 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11899 Reset the SCEV hashtable and niters.
11900 * graphite-scop-detection.c
11901 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
11902 assert that we only have POLYNOMIAL_CHREC that vary in loops
11903 contained in the region.
11904 (scop_detection::graphite_can_represent_expr): Adjust.
11905 (scop_detection::stmt_has_simple_data_refs_p): For loops
11906 not in the region set loop to NULL. The nest is now the
11907 entry edge to the region.
11908 (try_generate_gimple_bb): Likewise.
11909 * sese.c (scalar_evolution_in_region): Adjust for
11910 instantiate_scev change.
11911 * tree-data-ref.h (graphite_find_data_references_in_stmt):
11912 Make nest parameter the edge into the region.
11913 (create_data_ref): Likewise.
11914 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
11915 entry edge into a region and adjust instantiate_scev calls.
11916 (create_data_ref): Likewise.
11917 (graphite_find_data_references_in_stmt): Likewise.
11918 (find_data_references_in_stmt): Pass the loop preheader edge
11919 from the nest argument.
11920 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
11921 parameter the edge into the region.
11922 (instantiate_parameters): Use the loop preheader edge as entry.
11923 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
11925 (get_instantiated_value_entry): Make instantiate_below parameter
11926 the edge into the region.
11927 (instantiate_scev_name): Likewise. Adjust dominance checks,
11928 when we cannot use loop-based instantiation instantiate by
11929 walking use-def chains.
11930 (instantiate_scev_poly): Adjust.
11931 (instantiate_scev_binary): Likewise.
11932 (instantiate_scev_convert): Likewise.
11933 (instantiate_scev_not): Likewise.
11934 (instantiate_array_ref): Remove.
11935 (instantiate_scev_3): Likewise.
11936 (instantiate_scev_2): Likewise.
11937 (instantiate_scev_1): Likewise.
11938 (instantiate_scev_r): Do not blindly handle N-operand trees.
11939 Do not instantiate array-refs. Handle all constants and invariants.
11940 (instantiate_scev): Make instantiate_below parameter
11941 the edge into the region.
11942 (resolve_mixers): Use the loop preheader edge for the region
11943 parameter to instantiate_scev_r.
11944 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
11946 2017-10-13 Richard Biener <rguenther@suse.de>
11948 PR tree-optimization/82525
11949 * graphite-isl-ast-to-gimple.c
11950 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
11952 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
11953 Fail code generation when we cannot represent the isl integer.
11954 (binary_op_to_tree): Elide modulo operations that are no-ops
11955 in the type we code generate. Remove now superfluous code
11958 2017-10-13 Richard Biener <rguenther@suse.de>
11960 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
11961 (scop_detection::harmful_loop_in_region): Remove premature
11962 IV type restriction.
11963 (scop_detection::graphite_can_represent_scev): We can handle
11964 pointer IVs just fine.
11966 2017-10-13 Alan Modra <amodra@gmail.com>
11968 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
11969 "Clobbers and Scratch Registers". Add paragraph on
11970 alternative to clobbers for scratch registers and OpenBLAS
11973 2017-10-13 Alan Modra <amodra@gmail.com>
11975 * doc/extend.texi (Clobbers): Correct vax example. Delete old
11976 example of a memory input for a string of known length. Move
11977 commentary out of table. Add a number of new examples
11978 covering array memory inputs.
11980 2017-10-12 Martin Liska <mliska@suse.cz>
11982 PR tree-optimization/82493
11983 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
11984 (test_range_functions): New function.
11985 (sbitmap_c_tests): Likewise.
11986 * selftest-run-tests.c (selftest::run_tests): Run new tests.
11987 * selftest.h (sbitmap_c_tests): New function.
11989 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
11991 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11993 * config/rs6000/amo.h: Fix spacing issue.
11995 2017-10-12 Jakub Jelinek <jakub@redhat.com>
11998 * config/i386/i386.md (*ashl<mode>3_mask_1,
11999 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12000 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12003 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12005 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12006 (profile_probability): Set max_probability
12007 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12009 * profile-count.c (profile_probability::differs_from_p): Do not
12010 rely on max_probaiblity == 10000
12012 2017-10-12 Jeff Law <law@redhat.com>
12014 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12017 2017-10-12 Martin Sebor <msebor@redhat.com>
12021 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12022 (handle_alias_pairs): Call it.
12023 * common.opt (-Wattribute-alias): New option.
12024 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12025 * doc/invoke.texi (-Wattribute-alias): Document.
12027 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12030 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12032 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12034 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12035 (make_hard_regno_born, make_hard_regno_dead): Update
12036 bb_killed_pseudos and bb_gen_pseudos.
12038 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12040 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12042 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12044 * config/alpha/alpha.c (alpha_split_conditional_move):
12045 Use std::swap instead of manually swapping.
12046 (alpha_stdarg_optimize_hook): Ditto.
12047 (alpha_canonicalize_comparison): Ditto.
12049 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12051 * tree-loop-distribution.c (struct builtin_info): New struct.
12052 (struct partition): Refactor fields into struct builtin_info.
12053 (partition_free): Free struct builtin_info.
12054 (build_size_arg_loc, build_addr_arg_loc): Delete.
12055 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12056 information from struct builtin_info.
12057 (find_single_drs): New function refactored from classify_partition.
12058 Also moved builtin validity checks to this function.
12059 (compute_access_range, alloc_builtin): New functions.
12060 (classify_builtin_st, classify_builtin_ldst): New functions.
12061 (classify_partition): Refactor code into functions find_single_drs,
12062 classify_builtin_st and classify_builtin_ldst.
12063 (distribute_loop): Don't do runtime alias check when distributing
12065 (find_seed_stmts_for_distribution): New function.
12066 (pass_loop_distribution::execute): Refactor code finding seed
12067 stmts into above function. Support distribution for the innermost
12068 two-level loop nest. Adjust dump information.
12070 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12072 * tree-loop-distribution.c: Adjust the general comment.
12073 (NUM_PARTITION_THRESHOLD): New macro.
12074 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12075 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12076 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12077 in call to build_partition_graph.
12078 (finalize_partitions): New parameter. Make loop distribution more
12079 conservative by fusing more partitions.
12080 (distribute_loop): Don't do runtime alias check in case of loop nest
12082 (find_seed_stmts_for_distribution): New function.
12083 (prepare_perfect_loop_nest): New function.
12084 (pass_loop_distribution::execute): Refactor code finding seed stmts
12085 and loop nest into above functions. Support loop nest distribution.
12086 Adjust dump information accordingly.
12088 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12090 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12091 and set PTYPE_SEQUENTIAL for merged partition.
12093 2017-10-12 Richard Biener <rguenther@suse.de>
12095 PR tree-optimization/69728
12097 2017-09-19 Richard Biener <rguenther@suse.de>
12099 PR tree-optimization/69728
12100 * graphite-sese-to-poly.c (schedule_error): New global.
12101 (add_loop_schedule): Handle empty domain by failing the
12103 (build_original_schedule): Handle schedule_error.
12105 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12106 domain by returning an unchanged schedule.
12108 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12110 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12111 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12113 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12115 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12118 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12121 * expr.c (store_field): Don't optimize away bitsize == 0 store
12122 from CALL_EXPR with addressable return type.
12124 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12126 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12127 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12128 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12129 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12130 TARGET_ISEL instead of TARGET_ISEL<sel>.
12132 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12134 * config/rs6000/rs6000.c
12135 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12137 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12139 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12141 (reg_or_cint_operand): Fix comment.
12142 (reg_or_zero_operand): New predicate.
12143 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12144 * config/rs6000/rs6000.c (output_isel): Delete.
12145 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12146 instead of reg_or_cint_operand. Output instruction directly (not via
12148 (isel_unsigned_<mode>): Ditto.
12149 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12150 gpc_reg_operand. Add an instruction alternative for this. Output
12151 instruction directly.
12152 (*isel_reversed_unsigned_<mode>): Ditto.
12154 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12156 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12157 (TARGET_CANONICALIZE_COMPARISON): Define.
12159 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12162 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12163 Check whether the dest is REG before adding REG_EQUIV note.
12165 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12168 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12170 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12171 (make_hard_regno_born, make_hard_regno_dead): Update
12172 bb_killed_pseudos and bb_gen_pseudos.
12174 2017-10-11 Nathan Sidwell <nathan@acm.org>
12176 * incpath.h (enum incpath_kind): Name enum, prefix values.
12177 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12178 * incpath.c (heads, tails): Use INC_MAX.
12179 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12180 (merge_include_chains, split_quote_chain,
12181 register_include_chains): Update incpath_kind names.
12182 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12183 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12185 (add_framework_path, darwin_register_objc_includes): Likewise.
12186 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12188 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12190 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12191 Do not use float_operator operator predicate.
12192 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12193 * config/i386/predicates.md (float_operator): Remove predicate.
12195 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12197 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12198 (*jccxf_i387): Ditto.
12199 (*jcc<mode>_i387): Ditto.
12200 (*jccu<mode>_i387): Ditto.
12201 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12202 (*jcc_*_i387 splitters): Remove.
12203 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12204 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12205 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12208 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12210 * profile-count.h (slow_safe_scale_64bit): New function.
12211 (safe_scale_64bit): New inline.
12212 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12213 * profile-count.c: Include wide-int.h
12214 (slow_safe_scale_64bit): New.
12216 2017-10-11 Nathan Sidwell <nathan@acm.org>
12218 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12219 HAS_DECL_ASSEMBLER_NAME_P.
12220 * gimple-expr.c (gimple_decl_printable_name: Check
12221 HAS_DECL_ASSEMBLER_NAME_P too.
12222 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12223 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12224 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12225 * passes.c (rest_of_decl_compilation): Check
12226 HAS_DECL_ASSEMBLER_NAME_P too.
12227 * recog.c (verify_changes): Likewise.
12228 * tree-pretty-print.c (dump_decl_name): Likewise.
12229 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12231 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12232 (SET_DECL_ASSEMBLER_NAME): Use it.
12233 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12234 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12235 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12237 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12239 * config.gcc (i386, x86_64): Add extra objects.
12240 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12241 (ix86_min_insn_size): Declare.
12242 (ix86_issue_rate): Declare.
12243 (ix86_adjust_cost): Declare.
12244 (ia32_multipass_dfa_lookahead): Declare.
12245 (ix86_macro_fusion_p): Declare.
12246 (ix86_macro_fusion_pair_p): Declare.
12247 (ix86_bd_has_dispatch): Declare.
12248 (ix86_bd_do_dispatch): Declare.
12249 (ix86_core2i7_init_hooks): Declare.
12250 (ix86_atom_sched_reorder): Declare.
12251 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12252 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12253 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12254 (rip_relative_addr_p): Rename to ...
12255 (ix86_rip_relative_addr_p): ... this one; export.
12256 (memory_address_length): Update.
12257 (ix86_issue_rate): Move to x86-tune-sched.c.
12258 (ix86_flags_dependent): Move to x86-tune-sched.c.
12259 (ix86_agi_dependent): Move to x86-tune-sched.c.
12260 (exact_dependency_1): Move to x86-tune-sched.c.
12261 (exact_store_load_dependency): Move to x86-tune-sched.c.
12262 (ix86_adjust_cost): Move to x86-tune-sched.c.
12263 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12264 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12265 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12266 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12267 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12268 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12269 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12270 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12271 (min_insn_size): Rename to ...
12272 (ix86_min_insn_size): ... this one; export.
12273 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12274 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12275 (core2i7_first_cycle_multipass_backtrack): Move to
12276 x86-tune-sched-core.c.
12277 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12278 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12279 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12280 (ix86_avoid_jump_mispredicts): Update.
12281 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12282 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12283 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12284 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12285 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12286 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12287 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12288 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12289 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12290 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12291 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12292 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12293 (BIG): Move to ix86-tune-sched-bd.c.
12294 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12295 (enum insn_path): Move to ix86-tune-sched-bd.c.
12296 (get_mem_group): Move to ix86-tune-sched-bd.c.
12297 (is_cmp): Move to ix86-tune-sched-bd.c.
12298 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12299 (is_branch): Move to ix86-tune-sched-bd.c.
12300 (is_prefetch): Move to ix86-tune-sched-bd.c.
12301 (init_window): Move to ix86-tune-sched-bd.c.
12302 (allocate_window): Move to ix86-tune-sched-bd.c.
12303 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12304 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12305 (process_end_window): Move to ix86-tune-sched-bd.c.
12306 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12307 (find_constant): Move to ix86-tune-sched-bd.c.
12308 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12309 (has_immediate): Move to ix86-tune-sched-bd.c.
12310 (get_insn_path): Move to ix86-tune-sched-bd.c.
12311 (get_insn_group): Move to ix86-tune-sched-bd.c.
12312 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12313 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12314 (add_insn_window): Move to ix86-tune-sched-bd.c.
12315 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12316 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12317 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12318 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12319 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12320 (do_dispatch): Move to ix86-tune-sched-bd.c.
12321 (has_dispatch): Move to ix86-tune-sched-bd.c.
12322 * config/i386/t-i386: Add new object files.
12323 * config/i386/x86-tune-costs.h: New file.
12324 * config/i386/x86-tune-sched-atom.c: New file.
12325 * config/i386/x86-tune-sched-bd.c: New file.
12326 * config/i386/x86-tune-sched-core.c: New file.
12327 * config/i386/x86-tune-sched.c: New file.
12329 2017-10-11 Liu Hao <lh_mouse@126.com>
12331 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12332 the generic version below instead.
12333 (should_colorize): Recognize Windows consoles as terminals
12335 * pretty-print.c [__MINGW32__] (write_all): New function.
12336 [__MINGW32__] (find_esc_head): Likewise.
12337 [__MINGW32__] (find_esc_terminator): Likewise.
12338 [__MINGW32__] (eat_esc_sequence): Likewise.
12339 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12341 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12344 2017-10-11 Richard Biener <rguenther@suse.de>
12346 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12347 Properly call analyze_scalar_evolution with the loop of the stmt.
12349 2017-10-11 Richard Biener <rguenther@suse.de>
12351 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12352 * tree-core.h (tree_base): Add chrec_var union member.
12353 * tree.h (CHREC_VAR): Remove.
12354 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12355 * tree-chrec.h (build_polynomial_chrec): Adjust.
12356 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12357 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12359 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12361 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12362 * match.pd: ... here.
12363 ((T) X == (T) Y): Relax condition.
12365 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12367 PR tree-optimization/82472
12368 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12370 (break_alias_scc_partitions): Update postorder number.
12372 2017-10-11 Martin Liska <mliska@suse.cz>
12375 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12377 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12379 2017-10-11 Martin Liska <mliska@suse.cz>
12381 * print-rtl.c (print_insn): Move declaration of idbuf
12382 to same scope as name.
12384 2017-10-11 Martin Liska <mliska@suse.cz>
12389 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12390 volatile arguments.
12392 2017-10-11 Martin Liska <mliska@suse.cz>
12395 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12396 volatile arguments.
12398 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
12400 * config.gcc (default_gnu_indirect_function): Default to yes for
12401 arm*-*-linux* with glibc.
12403 2017-10-11 Richard Biener <rguenther@suse.de>
12405 * tree-scalar-evolution.c (get_scalar_evolution): Handle
12406 default-defs and types we do not want to analyze.
12407 (interpret_loop_phi): Replace unreachable code with an assert.
12408 (compute_scalar_evolution_in_loop): Remove and inline ...
12409 (analyze_scalar_evolution_1): ... here, replacing condition with
12410 what makes the intent clearer. Remove handling of cases
12411 get_scalar_evolution now handles.
12413 2017-10-10 Jim Wilson <wilson@tuliptree.org>
12415 PR rtl-optimization/81434
12416 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
12417 comment for main loop. In sched_group_found if, also add checks for
12418 pass and min_cost_group.
12420 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
12422 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
12423 (rs6000_insn_cost): New function.
12424 * config/rs6000/rs6000.md (cost): New attribute.
12426 2017-10-10 Jakub Jelinek <jakub@redhat.com>
12427 H.J. Lu <hongjiu.lu@intel.com>
12431 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
12432 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
12433 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
12434 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
12435 to OPTION_MASK_ISA_AVX512VL - builtins that have both
12436 OPTION_MASK_ISA_MMX and some other bit set require both
12437 mmx and the ISAs without the mmx bit.
12438 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
12439 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
12440 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
12441 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
12442 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
12443 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
12444 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
12445 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
12446 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
12447 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
12448 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
12449 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
12450 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
12451 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
12452 Add OPTION_MASK_ISA_MMX.
12454 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
12456 * config.gcc (armv7*-*-freebsd*): New target.
12457 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
12459 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
12461 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
12462 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
12465 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
12467 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
12468 a template parameter.
12469 (WIDE_INT_REF_FOR): Update accordingly.
12470 * tree.h (wi::int_traits <const_tree>): Delete.
12471 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
12472 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
12473 (wi::tree_to_wide_ref): New typedef.
12474 (wi::to_wide): New function.
12475 * calls.c (get_size_range): Use wi::to_wide when operating on
12476 trees as wide_ints.
12477 * cgraph.c (cgraph_node::create_thunk): Likewise.
12478 * config/i386/i386.c (ix86_data_alignment): Likewise.
12479 (ix86_local_alignment): Likewise.
12480 * dbxout.c (stabstr_O): Likewise.
12481 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
12482 * expr.c (const_vector_from_tree): Likewise.
12483 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
12484 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
12485 (fold_negate_expr_1, int_const_binop_1, const_binop)
12486 (fold_convert_const_int_from_real, optimize_bit_field_compare)
12487 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
12488 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
12489 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
12490 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
12491 (fold_not_const, round_up_loc): Likewise.
12492 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
12493 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
12494 (alloca_call_type): Likewise.
12495 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
12496 * godump.c (go_output_typedef): Likewise.
12497 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
12498 * internal-fn.c (get_min_precision): Likewise.
12499 * ipa-cp.c (ipcp_store_vr_results): Likewise.
12500 * ipa-polymorphic-call.c
12501 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
12502 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
12503 (ipa_modify_call_arguments): Likewise.
12504 * match.pd: Likewise.
12505 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
12506 * print-tree.c (print_node_brief, print_node): Likewise.
12507 * stmt.c (expand_case): Likewise.
12508 * stor-layout.c (layout_type): Likewise.
12509 * tree-affine.c (tree_to_aff_combination): Likewise.
12510 * tree-cfg.c (group_case_labels_stmt): Likewise.
12511 * tree-data-ref.c (dr_analyze_indices): Likewise.
12512 (prune_runtime_alias_test_list): Likewise.
12513 * tree-dump.c (dequeue_and_dump): Likewise.
12514 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
12515 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
12516 * tree-pretty-print.c (dump_generic_node): Likewise.
12517 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
12518 (simple_iv_with_niters): Likewise.
12519 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
12520 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
12521 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
12522 * tree-ssa-loop-niter.c (split_to_var_and_offset)
12523 (refine_value_range_using_guard, number_of_iterations_ne_max)
12524 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
12525 (get_cst_init_from_scev, record_nonwrapping_iv)
12526 (scev_var_range_cant_overflow): Likewise.
12527 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
12528 * tree-ssa-pre.c (compute_avail): Likewise.
12529 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
12530 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
12531 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
12532 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
12533 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
12534 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
12535 (dump_case_nodes, try_switch_expansion): Likewise.
12536 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
12537 (vect_do_peeling): Likewise.
12538 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
12539 * tree-vect-stmts.c (vectorizable_load): Likewise.
12540 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
12541 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
12542 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
12543 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
12544 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
12545 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
12546 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
12547 (evrp_dom_walker::before_dom_children): Likewise.
12548 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
12549 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
12550 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
12551 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
12552 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
12553 (get_range_pos_neg): Likewise.
12554 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
12555 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
12556 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
12557 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
12558 * config/avr/avr.c (avr_fold_builtin): Likewise.
12559 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
12560 * config/msp430/msp430.c (msp430_attr): Likewise.
12561 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
12562 * config/powerpcspe/powerpcspe-c.c
12563 (altivec_resolve_overloaded_builtin): Likewise.
12564 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
12565 (rs6000_expand_ternop_builtin): Likewise.
12566 * config/rs6000/rs6000-c.c
12567 (altivec_resolve_overloaded_builtin): Likewise.
12568 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
12569 (rs6000_expand_ternop_builtin): Likewise.
12570 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
12572 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12574 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
12575 when copying loop nest with only one inner loop.
12577 2017-10-10 Richard Biener <rguenther@suse.de>
12579 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
12580 blocks if SCEV is active.
12581 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
12583 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
12584 (scev_initialize): Assert we are not yet initialized.
12586 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12588 * tree-loop-distribution.c (generate_loops_for_partition): Remove
12589 inner loop's exit stmt by making it always exit the loop, otherwise
12590 we would generate an infinite empty loop.
12592 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12594 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
12595 renaming variables in new preheader if it's deleted.
12597 2017-10-10 Bin Cheng <bin.cheng@arm.com>
12599 * tree-loop-distribution.c (struct partition): Remove unused field
12600 loops of the structure.
12601 (partition_alloc, partition_free): Ditto.
12602 (build_rdg_partition_for_vertex): Ditto.
12604 2017-10-09 Jeff Law <law@redhat.com>
12606 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
12607 return type to match prototype and documentation.
12609 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12611 * config/rs6000/rs6000.c (processor_costs): Move to ...
12612 * config/rs6000/rs6000.h: ... here.
12613 (rs6000_cost): Declare.
12615 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
12617 * except.c (setjmp_fn): New global variable.
12618 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
12619 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
12620 if DONT_USE_BUILTIN_SETJMP is defined.
12622 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12624 * target.def (insn_cost): New hook.
12625 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
12626 * doc/tm.texi: Regenerate.
12627 * rtlanal.c (insn_cost): Use the new hook.
12629 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12631 * combine.c (combine_validate_cost): Compute the new insn_cost,
12632 not just pattern_cost.
12633 (try_combine): Adjust comment.
12635 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
12637 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
12639 * combine.c (uid_insn_cost): Adjust comment.
12640 (combine_validate_cost): Adjust comment. Use pattern_cost instead
12642 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
12643 * dse.c (find_shift_sequence): Ditto.
12644 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
12645 (bb_valid_for_noce_process_p): Use pattern_cost.
12646 * rtl.h (insn_rtx_cost): Delete.
12647 (pattern_cost): New prototype.
12648 (insn_cost): New prototype.
12649 * rtlanal.c (insn_rtx_cost): Rename to...
12650 (pattern_cost): ... this.
12653 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
12655 * config/i386/i386.md (*jcc_2): Remove insn pattern.
12656 (*jcc<mode>_0_r_i387): Ditto.
12657 (*jccxf_r_i387): Ditto.
12658 (*jcc<mode>_r_i387): Ditto.
12659 (*jccu<mode>_r_i387): Ditto.
12660 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
12661 (*jcc): Rename from *jcc_1.
12663 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12665 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
12666 deferred rescans after the lvx/stvx recombination pre-pass.
12668 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
12670 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
12671 memory operation instruction support.
12672 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
12673 (rs6000-ibm-aix[789]*): Likewise.
12674 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
12675 Document new functions.
12677 2017-10-09 Richard Biener <rguenther@suse.de>
12679 PR tree-optimization/82397
12680 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
12681 equality only for semantically equal trees.
12683 2017-10-09 Richard Biener <rguenther@suse.de>
12685 PR tree-optimization/82449
12686 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
12687 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
12688 allow constant addresses.
12689 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
12692 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12694 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
12697 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12700 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
12703 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12706 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
12708 2017-10-09 Jakub Jelinek <jakub@redhat.com>
12711 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
12712 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
12714 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
12716 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
12717 (WI_BINARY_PREDICATE_RESULT): Likewise.
12718 (wi::binary_traits::operator_result): New type.
12719 (wi::binary_traits::predicate_result): Likewise.
12720 (generic_wide_int::operator~, unary generic_wide_int::operator-)
12721 (generic_wide_int::operator==, generic_wide_int::operator!=)
12722 (generic_wide_int::operator&, generic_wide_int::and_not)
12723 (generic_wide_int::operator|, generic_wide_int::or_not)
12724 (generic_wide_int::operator^, generic_wide_int::operator+
12725 (binary generic_wide_int::operator-, generic_wide_int::operator*):
12727 (operator~, unary operator-, operator==, operator!=, operator&)
12728 (operator|, operator^, operator+, binary operator-, operator*): New
12730 * expr.c (get_inner_reference): Use wi::bit_and_not.
12731 * fold-const.c (fold_binary_loc): Likewise.
12732 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
12733 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
12734 (bit_value_binop): Likewise.
12735 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
12736 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
12737 (extract_range_from_binary_expr_1): Likewise.
12738 (masked_increment): Likewise.
12739 (simplify_bit_ops_using_ranges): Likewise.
12741 2017-10-09 Martin Jambor <mjambor@suse.cz>
12744 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
12745 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
12746 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
12748 (hsa_fixup_mov_insn_type): New function.
12749 (hsa_op_with_type::get_in_type): Use it.
12750 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
12751 immediates in an assert.
12752 (hsa_op_with_type::extend_int_to_32bit): New method.
12753 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
12754 types. Convert to dest type if necessary.
12755 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
12756 (reg_for_gimple_ssa): Pass false as min32int to
12757 hsa_type_for_scalar_tree_type.
12758 (gen_hsa_addr): Fixup type when creating addresable temporary.
12759 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
12760 (gen_hsa_unary_operation): Extend operands and convert to dest type if
12761 necessary. Call hsa_fixup_mov_insn_type.
12762 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
12763 extend operands and convert to dest type if necessary.
12764 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
12765 to dest type if necessary.
12766 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
12767 if conversion nt necessary and size matches.
12768 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
12769 to dest type if necessary.
12770 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
12771 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
12773 (gen_hsa_clrsb): Likewise.
12774 (gen_hsa_ffs): Likewise.
12775 (gen_hsa_divmod): Extend operands and convert to dest type if
12777 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
12779 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12781 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
12782 Remove empty default arguments. Use a brace block as output
12784 (conditional return): Ditto.
12786 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
12787 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
12789 (group_ending_nop): Ditto.
12790 (doloop_end): Ditto.
12791 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
12792 (splitters for those): Ditto.
12794 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
12796 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
12797 a conditional jump (and the compare for it) so that pc_rtx is the
12799 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
12800 for the deleted and renamed ctr<mode>_internal[234] patterns.
12801 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
12802 Delete second conditional return pattern.
12803 (ctr<mode>_internal2): Delete this second bdnz pattern.
12804 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
12805 (ctr<mode>_internal4): Delete this second bdz pattern.
12807 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12809 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
12810 (always_initialized_rtx_for_ssa_name_p): New predicate.
12811 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
12812 (finish_out_of_ssa): Free new field of SA.
12813 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
12814 * tree-ssa-coalesce.c: Include tree-ssa.h.
12815 (get_parm_default_def_partitions): Remove extern keyword.
12816 (get_undefined_value_partitions): New function.
12817 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
12818 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
12819 uninitialized bits.
12820 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
12822 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
12824 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
12826 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12828 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
12829 for targets that preffer 128bit.
12831 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
12833 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
12835 2017-10-08 Olivier Hainque <hainque@adacore.com>
12837 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
12838 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
12840 (thumb_set_return_address): Likewise.
12842 2017-10-08 Olivier Hainque <hainque@adacore.com>
12844 * common/config/arm/arm-common.c (arm_except_unwind_info):
12845 Handle DWARF2_UNWIND_INFO.
12847 2017-10-07 Michael Collison <michael.collison@arm.com>
12849 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
12852 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
12854 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
12855 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
12856 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
12857 defined, force the creation of a new block for a dispatch label.
12859 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
12861 * invoke.texi (Wsuggest-attribute=cold): Document.
12862 * common.opt (Wsuggest-attribute=cold): New
12863 * ipa-pure-const.c (warn_function_cold): New function.
12864 * predict.c (compute_function_frequency): Use it.
12865 * predict.h (warn_function_cold): Declare.
12867 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
12869 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
12872 2017-10-06 Martin Liska <mliska@suse.cz>
12874 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
12875 keyword for member functions.
12876 (struct sanopt_tree_couple): New struct.
12877 (struct sanopt_tree_couple_hash): New function.
12878 (struct sanopt_ctx): Add new hash_map.
12879 (has_dominating_ubsan_ptr_check): New function.
12880 (record_ubsan_ptr_check_stmt): Likewise.
12881 (maybe_optimize_ubsan_ptr_ifn): Likewise.
12882 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
12883 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
12885 2017-10-06 Sudakshina Das <sudi.das@arm.com>
12888 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
12889 aarch64_simd_valid_immediate on CONST_VECTORs.
12890 (aarch64_reg_or_bic_imm): Likewise.
12892 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
12894 PR rtl-optimization/82396
12895 * haifa-sched.c (ready_sort_real): Disable qsort checking.
12897 2017-10-06 Sebastian Pop <sebpop@gmail.com>
12899 * graphite-dependences.c (scop_get_reads): Move code to...
12900 (scop_get_must_writes): Move code to...
12901 (scop_get_may_writes): Move code to...
12902 (scop_get_reads_and_writes): ... here.
12903 (scop_get_dependences): Call scop_get_reads_and_writes.
12905 2017-10-06 Jakub Jelinek <jakub@redhat.com>
12907 PR tree-optimization/82434
12908 * fold-const.h (can_native_encode_type_p,
12909 can_native_encode_string_p): Remove.
12910 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
12911 don't encode anything, just return what would be otherwise returned.
12912 (native_encode_fixed, native_encode_complex, native_encode_vector):
12914 (native_encode_string): Likewise. Inline by hand
12915 can_native_encode_string_p.
12916 (can_native_encode_type_p): Remove.
12917 (can_native_encode_string_p): Remove.
12918 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
12919 STRING_CSTs using can_native_encode_string_p, test all
12920 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
12921 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
12922 argument from native_encode_expr.
12923 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
12924 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
12927 2017-10-06 Richard Biener <rguenther@suse.de>
12929 PR tree-optimization/82397
12930 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
12931 operand_equal_p but rely on data_ref_compare_tree for detecting
12933 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
12934 to match up with dr_group_sort_cmp.
12936 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12939 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
12941 * config/s390/s390-builtin-types.def: Regenerate.
12943 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12946 * config/s390/s390-builtin-types.def: Regenerate.
12947 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
12948 Change flag from B_VXE to B_VX.
12949 (s390_vec_min_dbl): Remove B_VXE flag.
12951 2017-10-06 Richard Biener <rguenther@suse.de>
12953 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
12954 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
12955 translate_isl_ast_to_gimple::is_valid_rename,
12956 translate_isl_ast_to_gimple::get_rename,
12957 translate_isl_ast_to_gimple::get_def_bb_for_const,
12958 translate_isl_ast_to_gimple::get_new_name,
12959 translate_isl_ast_to_gimple::collect_all_ssa_names,
12960 translate_isl_ast_to_gimple::copy_loop_phi_args,
12961 translate_isl_ast_to_gimple::collect_all_ssa_names,
12962 translate_isl_ast_to_gimple::copy_loop_phi_args,
12963 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
12964 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
12965 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
12966 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
12967 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
12968 translate_isl_ast_to_gimple::copy_cond_phi_args,
12969 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
12970 translate_isl_ast_to_gimple::edge_for_new_close_phis,
12971 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
12972 translate_isl_ast_to_gimple::rename_uses,
12973 translate_isl_ast_to_gimple::rename_all_uses): Remove.
12974 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
12975 (set_rename_for_each_def): Likewise.
12976 (graphite_copy_stmts_from_block): Handle debug stmt resetting
12977 here. Handle rewriting SCEV analyzable uses here.
12978 (copy_bb_and_scalar_dependences): Generate code for PHI
12980 (graphite_regenerate_ast_isl): Adjust.
12981 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
12982 (add_write, add_read): New functions.
12983 (build_cross_bb_scalars_def): Use it and simplify.
12984 (build_cross_bb_scalars_use): Likewise.
12985 (graphite_find_cross_bb_scalar_vars): Inline into...
12986 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
12987 simulating out-of-SSA. Compute liveout and add dependencies.
12988 (build_scops): Force an empty entry block.
12989 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
12991 (sese_build_liveouts): Declare.
12992 (sese_trivially_empty_bb_p): Likewise.
12993 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
12994 compute liveout and debug_liveout.
12995 (sese_bad_liveouts_use): Remove.
12996 (sese_reset_debug_liveouts_bb): Likewise.
12997 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
12998 (sese_build_liveouts): Build liveout and debug_liveout and store
13000 (new_sese_info): Adjust.
13001 (free_sese_info): Likewise.
13002 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13003 do not build liveout here.
13004 (move_sese_in_condition): Adjust region entry.
13005 (scev_analyzable_p): Match up with chrec_apply requirements.
13006 (sese_trivially_empty_bb_p): New.
13007 * tree-into-ssa.c (get_reaching_def): Properly support generating
13008 default-defs for incremental rewrite of anonymous names.
13010 2017-10-06 Richard Biener <rguenther@suse.de>
13012 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13013 precision do not perform modulo reduction.
13015 2017-10-06 Richard Biener <rguenther@suse.de>
13017 PR tree-optimization/82436
13018 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13019 conservatively choose the vectorization factor when checking
13020 whether we can perform the required load permutation.
13021 (vect_transform_slp_perm_load): Assert when we may not fail.
13023 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13025 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13026 message for incompatible -msdata=* and -mcall-* options.
13028 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13030 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13031 rate for post-reload scheduling.
13033 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13035 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13037 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13039 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13040 to improve monte carlo in scimark.
13042 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13044 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13045 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13046 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13047 pentium4_cost, nocona_cost): Set reassociation width to 1.
13048 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13049 width to 2 for fp operations and 1 otherwise.
13050 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13052 (atom_cost): Set reassociation width to 2.
13053 (slm_cost, generic_cost): Set fp reassociation width
13054 to 2 and 1 otherwise.
13055 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13056 (core_cost): Set fp reassociation width to 4 and vector to 2.
13057 (ix86_reassociation_width): Rewrite using cost table; special case
13058 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13059 and TARGET_AVX128_OPTIMAL.
13060 * config/i386/i386.h (processor_costs): Add
13061 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13062 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13063 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13064 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13065 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13066 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13068 2017-10-05 Nathan Sidwell <nathan@acm.org>
13070 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13072 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13074 * config/arm/arm.c (arm_test_fpu_data): New.
13075 (arm_run_selftests): Call arm_test_fpu_data.
13077 2017-10-04 Nathan Sidwell <nathan@acm.org>
13079 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13081 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13083 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13085 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13086 check type for aarch64_simd_valid_immediate.
13087 (aarch64_output_simd_mov_immediate): Update prototype.
13088 (aarch64_simd_valid_immediate): Update prototype.
13089 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13090 support for ORR-immediate.
13091 (and<mode>3): modified pattern to add support for BIC-immediate.
13092 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13093 now checks for valid immediate for BIC and ORR based on new enum
13095 (aarch64_output_simd_mov_immediate): Function now used to output
13096 BIC/ORR imm as well based on new enum argument.
13097 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13099 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13100 (aarch64_reg_or_bic_imm): Likewise.
13102 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13104 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13105 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13107 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13111 PR rtl-optimization/82396
13112 * haifa-sched.c (autopref_multipass_init): Simplify
13114 (autopref_rank_data): Simplify sort order.
13115 * sched-int.h (autopref_multipass_data_): Remove
13116 multi_mem_insn_p, min_offset and max_offset.
13118 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13120 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13122 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13124 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13125 vect_doubleint_cvt.
13127 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13129 * doc/sourcebuild.texi: Document vect_long_mult.
13131 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13133 PR tree-optimization/82413
13134 * fold-const.c (build_range_check): Use widest_int when comparing
13135 the maximum ETYPE value with HIGH.
13137 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13139 PR rtl-optimization/82396
13140 * haifa-sched.c (autopref_multipass_init): Simplify
13142 (autopref_rank_data): Simplify sort order.
13143 * sched-int.h (autopref_multipass_data_): Remove
13144 multi_mem_insn_p, min_offset and max_offset.
13146 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13148 PR tree-optimization/82381
13149 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13150 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13153 PR tree-optimization/82374
13154 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13155 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13156 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13157 current_function_decl to the new decl.
13159 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13161 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13162 helper macro for IEEE float128 hardware built-in functions.
13163 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13165 (TRUNCF128_ODD): Likewise.
13166 (ADDF128_ODD): Likewise.
13167 (SUBF128_ODD): Likewise.
13168 (MULF128_ODD): Likewise.
13169 (DIVF128_ODD): Likewise.
13170 (FMAF128_ODD): Likewise.
13171 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13172 UNSPEC_TRUNC_ROUND_TO_ODD.
13173 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13174 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13175 floating point round to odd instructions.
13176 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13177 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13178 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13179 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13180 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13181 (trunc<mode>sf2_hw): Change the truncate with round to odd
13182 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13183 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13184 to odd hardware instructions.
13185 (sub<mode>3_odd): Likewise.
13186 (mul<mode>3_odd): Likewise.
13187 (div<mode>3_odd): Likewise.
13188 (sqrt<mode>2_odd): Likewise.
13189 (fma<mode>4_odd): Likewise.
13190 (fms<mode>4_odd): Likewise.
13191 (nfma<mode>4_odd): Likewise.
13192 (nfms<mode>4_odd): Likewise.
13193 (trunc<mode>df2_odd): Change the truncate with round to odd
13194 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13196 * doc/extend.texi (PowerPC built-in functions): Update documentation
13197 for existing IEEE float128-bit built-in functions. Add built-in
13198 functions that generate the IEEE 128-bit floating point round to
13201 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13203 PR rtl-optimization/77729
13204 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13205 to (X&(C1&~C2))|C2 transformations.
13207 2017-10-03 Martin Jambor <mjambor@suse.cz>
13209 PR tree-optimization/82363
13210 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13211 mismatch, mark lacc written regardless of racc.
13213 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13215 PR tree-optimization/82381
13216 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13217 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13218 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13219 fallthrough into reassoc_stmt_dominates_stmt_p.
13222 * combine.c (combine_instructions): Don't combine in unreachable
13225 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13228 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13229 function to not use the have_cpu variable. Do not set cpu_index,
13230 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13231 or the default cpu.
13232 (rs6000_valid_attribute_p): Remove duplicate initializations of
13233 old_optimize and func_optimize.
13234 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13235 (rs6000_activate_target_options): Make global.
13236 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13239 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13241 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13242 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13243 Set *poffset to 0 and *psize and *pmax_size to -1 if
13244 *poffset + *psize overflows in HOST_WIDE_INT.
13246 PR tree-optimization/82387
13247 PR tree-optimization/82388
13248 PR tree-optimization/82389
13249 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13250 instead of live_bytes non-NULL.
13252 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13255 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13258 2017-10-02 Richard Biener <rguenther@suse.de>
13260 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13261 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13262 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13264 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13266 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13267 requested precision matches the type's.
13268 * calls.c (alloc_max_size): Calculate the new candidate size as
13269 a widest_int and use wi::to_widest when comparing it with the
13270 current candidate size.
13271 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13272 zero rather than integer_zero_node.
13273 * match.pd: Check for a no-op conversion before using wi::add
13274 rather than after. Use tree_to_uhwi when summing small shift
13275 counts into an unsigned int.
13277 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13278 Alan Hayward <alan.hayward@arm.com>
13279 David Sherwood <david.sherwood@arm.com>
13282 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13283 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13284 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13285 POINTER_AND_FP_REGS.
13287 2017-10-02 Richard Biener <rguenther@suse.de>
13289 PR tree-optimization/82355
13290 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13291 a mapping for the enclosing loop but avoid generating one for
13292 the loop tree root.
13293 (copy_bb_and_scalar_dependences): Remove premature codegen
13294 error on PHIs in blocks duplicated into multiple places.
13295 * graphite-scop-detection.c
13296 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13297 in the region use it as loop and nest to analyze the DR in.
13298 (try_generate_gimple_bb): Likewise.
13299 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13300 (add_loop_constraints): For blocks in a loop not in the region
13301 create a dimension with a single iteration.
13302 * sese.h (gbb_loop_at_index): Remove assert.
13304 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13306 * omp-expand.c (adjust_context_scope): New function.
13307 (expand_parallel_call): Call adjust_context_scope.
13309 2017-10-01 Jeff Law <law@redhat.com>
13311 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13312 dom_opt_dom_walker class with direct access to private members.
13313 Add comments. Call test_for_singularity.
13314 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13315 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13316 m_dummy_cond anymore.
13317 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13319 (pass_dominator:execute): Build the dummy_cond here and pass it
13320 to the dom_opt_dom_walker ctor.
13321 (test_for_singularity): New function.
13323 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13324 Maya Rashish <coypu@sdf.org>
13326 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13327 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13328 (alpha*-*-netbsd*) Use nbsd_tm_file.
13329 (arm*-*-netbsdelf*) Likewise.
13330 (i[34567]86-*-netbsdelf*) Likewise.
13331 (x86_64-*-netbsd*) Likewise.
13332 (mips*-*-netbsd*) Likewise.
13333 (powerpc-*-netbsd*) Likewise.
13334 (sh*-*-netbsd*) Likewise.
13335 (sparc-*-netbsdelf*) Likewise.
13336 (sparc64-*-netbsd*) Likewise.
13337 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13339 (vax-*-netbsdelf*) Likewise.
13340 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13341 (UINT_FAST8_TYPE) Likewise.
13342 (INT_FAST16_TYPE) Check CHAR_FAST16.
13343 (UINT_FAST16_TYPE) Likewise.
13345 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13348 * config/i386/i386.md
13349 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13350 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13351 *divmodsi4_zext_2): New define_insn_and_split.
13352 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13353 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13354 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13355 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13356 New define_insn_and_split.
13357 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13358 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13359 operands[1] having DImode when mode is SImode.
13361 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13362 always SImode for DIV and MOD in REG_EQUAL notes.
13364 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13366 PR middle-end/82319
13367 * match.pd: Fix handling of NaNs in pattern.
13369 2017-09-29 Jeff Law <law@redhat.com>
13371 * sbitmap.c (bitmap_bit_in_range_p): New function.
13372 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13373 * tree-ssa-dse.c (live_bytes_read): New function.
13374 (dse_classify_store): Ignore reads of dead bytes.
13376 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13377 typos and whitespace errors.
13378 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13379 * config/s390/s390.c (s390_emit_prologue): Likewise.
13381 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13384 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13385 with a symbol for LRA.
13387 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13389 PR rtl-optimization/82338
13390 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13392 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
13394 * genmodes.c (calc_wider_mode): Suppress qsort macro.
13395 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
13396 (qsort_chk): Declare.
13397 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
13398 (qsort_chk): New function.
13400 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13402 PR tree-optimization/82337
13403 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
13404 phi definition if the PHI result appears in an abnormal PHI.
13405 (find_basis_for_base_expr): Don't record a basis if the LHS of the
13406 basis appears in an abnormal PHI.
13408 2017-09-29 Richard Biener <rguenther@suse.de>
13410 * graphite-isl-ast-to-gimple.c
13411 (translate_isl_ast_to_gimple::set_codegen_error): New function.
13412 (binary_op_to_tree): Use it.
13413 (get_rename_from_scev): Likewise.
13414 (copy_loop_phi_nodes): Likewise.
13415 (copy_bb_and_scalar_dependences): Likewise.
13416 (translate_pending_phi_nodes): Likewise.
13418 2017-09-29 Jakub Jelinek <jakub@redhat.com>
13421 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
13422 for movabsq $(i32 << shift), r64.
13424 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
13426 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
13427 index when encoding %esp as %rsp to avoid 0x67 prefix.
13429 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
13431 * config/i386/i386.md (*movsf_internal, *movdf_internal):
13432 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
13434 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
13436 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
13437 Extensions with more than 16 double VFP registers.
13438 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
13439 to_clear_mask and all code related to it. Replace the remaining
13440 entry by a sbitmap and adapt code accordingly.
13442 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
13444 * brig-builtins.def: Change pure attributes to const.
13446 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13448 * config.gcc (default_gnu_indirect_function): Default to yes for
13449 sparc*-*-linux* with glibc.
13451 2017-09-28 Joseph Myers <joseph@codesourcery.com>
13453 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
13454 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
13455 when creating .init_array and .fini_array sections with priority
13458 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
13461 * config/aarch64/aarch64.c
13462 (aarch64_builtin_support_vector_misalignment): Always return false
13463 when misalignment is unknown.
13465 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13467 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
13468 this function to return false if the definition used by the swap
13469 instruction is artificial, or if the memory address from which the
13470 constant value is loaded is not represented by a base address held
13471 in a register or if the base address register is a frame or stack
13472 pointer. Additionally, return false if the base address of the
13473 loaded constant is a SYMBOL_REF but is not considered to be a
13475 (replace_swapped_load_constant): New function.
13476 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
13477 loaded constant vector with a load of a swapped constant vector.
13479 2017-09-27 Carl Love <cel@us.ibm.com>
13481 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
13482 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
13483 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
13486 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
13488 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
13489 first, always call autopref_rank_data otherwise.
13491 2017-09-27 Richard Biener <rguenther@suse.de>
13493 * graphite-scop-detection.c (find_scop_parameters): Move
13494 loop bound handling ...
13495 (gather_bbs::before_dom_children): ... here, avoiding the need
13496 to build scop_info->loop_nest.
13497 (record_loop_in_sese): Remove.
13498 * sese.h (sese_info_t::loop_nest): Remove.
13499 * sese.c (new_sese_info): Do not allocate loop_nest.
13500 (free_sese_info): Do not free loop_nest.
13502 2017-09-27 Jakub Jelinek <jakub@redhat.com>
13505 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
13506 lhs from calls if the lhs has addressable type.
13508 2017-09-27 Richard Biener <rguenther@suse.de>
13510 * graphite.h (scop::max_alias_set): New member.
13511 * graphite-scop-detection.c: Remove references to non-existing
13512 --param in comments.
13513 (build_alias_sets): Record the maximum alias set used for drs.
13514 (build_scops): Support zero as unlimited for
13515 --param graphite-max-arrays-per-scop.
13516 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
13517 and inline into ...
13518 (build_poly_sr_1): ... here. Compute alias set based on the
13519 maximum alias set used for drs rather than
13520 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
13522 2017-09-27 Richard Biener <rguenther@suse.de>
13524 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
13525 --param loop-block-tile-size=0 to disable tiling.
13527 2017-09-27 Richard Biener <rguenther@suse.de>
13529 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
13530 (graphite-max-nb-scop-params): Document special value zero.
13531 * domwalk.h (dom_walker::STOP): New symbolical constant.
13532 (dom_walker::dom_walker): Add optional parameter for bb to
13534 (dom_walker::~dom_walker): Declare.
13535 (dom_walker::before_dom_children): Document STOP return value.
13536 (dom_walker::m_user_bb_to_rpo): New member.
13537 (dom_walker::m_bb_to_rpo): Likewise.
13538 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
13539 mapping here if not provided by the user.
13540 (dom_walker::~dom_walker): Free bb to RPO mapping if not
13541 provided by the user.
13542 (dom_walker::STOP): Define.
13543 (dom_walker::walk): Do not compute bb to RPO mapping here.
13544 Support STOP return value from before_dom_children to stop
13546 * graphite-optimize-isl.c (optimize_isl): If the schedule
13547 is the same still generate code if -fgraphite-identity
13548 or -floop-parallelize-all are given.
13549 * graphite-scop-detection.c: Include cfganal.h.
13550 (gather_bbs::gather_bbs): Get and pass through bb to RPO
13552 (gather_bbs::before_dom_children): Return STOP for BBs
13554 (build_scops): Compute bb to RPO mapping and pass it to
13555 the domwalk. Treat --param graphite-max-nb-scop-params=0
13556 as not limiting the number of params.
13557 * graphite.c (graphite_initialize): Remove limit on the
13558 number of basic-blocks in a function.
13559 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
13560 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
13561 default value of 10.
13563 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13565 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
13566 Adjust code to eliminate needing to do the shift right 32-bits
13567 operation after XSCVDPSPN.
13569 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13571 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
13572 ((X / Y) != 0 -> X >= Y): Likewise.
13574 2017-09-26 Carl Love <cel@us.ibm.com>
13576 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
13577 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
13578 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
13579 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
13580 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
13581 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
13582 definitions and overloading.
13583 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
13584 statement for P9V_BUILTIN_XST_LEN_R.
13585 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
13586 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
13587 define_expand and define_insn for the instructions and builtins.
13588 * doc/extend.texi: Update the built-in documentation file for the new
13589 built-in functions.
13590 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
13591 define_insn for the instructions
13593 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
13596 * gcc/config/netbsd-protos.h: New file.
13597 * gcc/config/netbsd.c: New file.
13598 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
13599 * gcc/config/t-netbsd: New file.
13600 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
13601 (tmake_file) Add t-netbsd.
13602 (extra_objs) Add netbsd.o.
13604 2017-09-26 Janus Weil <janus@gcc.gnu.org>
13608 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
13610 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13612 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
13613 sign extension from a vector register to a GPR by doing a 32-bit
13614 direct move and then an EXTSW.
13615 (extendsi<mode>2 splitter): Likewise.
13616 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
13617 right or vector extract after doing XSCVDPSPN. Use
13618 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
13620 (movdi_from_sf_zero_ext): Likewise.
13621 (reload_gpr_from_vsxsf): Likewise.
13622 (p8_mfvsrd_4_disf): Delete, no longer used.
13623 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
13624 then needing to move the SFmode to a GPR to use the XSCVDPSP
13625 instruction instead of FRSP and XSCVDPSPN.
13626 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
13627 it is adjacent to the other XSCVSPDP insns.
13628 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
13629 SFmode to be in traditional Altivec registers.
13630 (vsx_xscvdpspn): Eliminate useless alternative constraint.
13631 (vsx_xscvspdpn): Likewise.
13632 (vsx_xscvspdpn_scalar): Likewise.
13634 2017-09-26 Martin Jambor <mjambor@suse.cz>
13636 * tree-sra.c (compare_access_positions): Put integral types first,
13637 stabilize sorting of integral types, remove conditions putting
13638 non-full-precision integers last.
13639 (sort_and_splice_var_accesses): Disable scalarization if a
13640 non-integert would be represented by a non-full-precision integer.
13642 2017-09-26 Joseph Myers <joseph@codesourcery.com>
13644 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
13645 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
13646 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
13647 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
13648 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
13649 conditionals inside the function instead of around it. Call
13650 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
13651 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
13653 2017-09-26 Richard Biener <rguenther@suse.de>
13655 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
13657 (scop_detection::build_scop_breadth): ... this. Removed.
13658 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
13659 (scop_detection::harmful_stmt_in_bb): Likewise.
13660 (scop_detection::graphite_can_represent_stmt): Likewise.
13661 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
13662 (scop_detection::can_represent_loop): Remove recursion, fold in ...
13663 (scop_detection::can_represent_loop_1): ... this. Removed.
13664 (scop_detection::harmful_loop_in_region): Simplify after inlining
13665 the above and remove more quadraticness.
13666 (build_scops): Adjust.
13667 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
13670 2017-09-26 Jakub Jelinek <jakub@redhat.com>
13673 * config/i386/i386.c (ix86_print_operand_address_as): Only test
13674 REGNO (base) == SP_REG if base is a REG.
13676 PR middle-end/35691
13677 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
13678 if it is different SSA_NAME.
13679 (optimize_range_tests_cmp_bitwise): New function.
13680 (optimize_range_tests): Call it.
13682 2017-09-26 Richard Biener <rguenther@suse.de>
13684 PR tree-optimization/82321
13685 * graphite.c (canonicalize_loop_closed_ssa): Properly check
13686 for the def being inside the loop.
13688 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13690 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
13692 * config/s390/s390-builtins.def: Fix constraint on op4.
13694 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13696 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
13697 independent expanders.
13698 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
13699 ("vec_ordered", "vec_unordered"): New expanders.
13701 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13703 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
13706 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13708 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
13709 vec_unpacks_lo_v16qi.
13710 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
13712 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13714 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
13715 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
13716 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
13718 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13720 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
13722 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
13724 ("*vec_slb<mode>"): New insn pattern.
13725 ("vec_shr_<mode>"): New expander.
13726 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
13727 and force the shift count operand to V16QImode.
13728 ("vec_srb<mode>"): Set shift count mode to V16QI.
13730 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13732 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
13733 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
13734 ("vec_widen_smult_hi_<mode>"): New expander definitions.
13736 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
13739 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
13741 2017-09-26 Richard Biener <rguenther@suse.de>
13743 PR tree-optimization/82320
13744 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
13747 2017-09-25 Jeff Law <law@redhat.com>
13749 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
13750 prototype for new argument.
13751 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
13752 mostly extracted from rs6000_emit_allocate_stack.
13753 (rs6000_emit_probe_stack_range_stack_clash): New function.
13754 (rs6000_emit_allocate_stack): Call
13755 rs6000_emit_probe_stack_range_stack_clash as needed.
13756 (rs6000_emit_probe_stack_range): Add additional argument
13757 to call to gen_probe_stack_range{si,di}.
13758 (output_probe_stack_range): New.
13759 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
13760 (output_probe_stack_range_stack_clash): New.
13761 (rs6000_emit_prologue): Emit notes into dump file as requested.
13762 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
13763 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
13764 Add additional operand and pass it to output_probe_stack_range.
13766 2017-09-25 Bin Cheng <bin.cheng@arm.com>
13768 PR tree-optimization/82163
13769 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
13770 (checking_verify_loop_closed_ssa): New parameter.
13771 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
13772 (check_loop_closed_ssa_stmt): Delete.
13773 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
13774 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
13775 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
13778 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
13780 * brig-builtins.def: Treat HSAIL barrier builtins as
13781 setjmp/longjump style functions.
13783 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13785 * target.def (constant_alignment): New hook.
13786 * defaults.h (CONSTANT_ALIGNMENT): Delete.
13787 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
13788 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
13789 * doc/tm.texi: Regenerate.
13790 * targhooks.h (default_constant_alignment): Declare.
13791 (constant_alignment_word_strings): Likewise.
13792 * targhooks.c (default_constant_alignment): New function.
13793 (constant_alignment_word_strings): Likewise.
13794 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
13795 instead of CONSTANT_ALIGNMENT.
13796 * varasm.c (align_variable, get_variable_align, build_constant_desc)
13797 (force_const_mem): Likewise.
13798 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
13799 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
13800 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
13801 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13802 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
13804 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
13805 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13806 constant_alignment_word_strings.
13807 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
13808 (CONSTANT_ALIGNMENT): Likewise.
13809 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13810 (arm_constant_alignment): New function.
13811 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
13812 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13813 constant_alignment_word_strings.
13814 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
13815 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13816 constant_alignment_word_strings.
13817 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
13818 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13819 (cris_constant_alignment): New function.
13820 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
13821 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13822 (epiphany_constant_alignment): New function.
13823 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
13824 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13825 constant_alignment_word_strings.
13826 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
13827 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13828 constant_alignment_word_strings.
13829 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
13830 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13831 constant_alignment_word_strings.
13832 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
13833 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
13834 * config/i386/i386.c (ix86_constant_alignment): Make static.
13835 Use the same interface as the target hook.
13836 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13837 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
13838 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13839 constant_alignment_word_strings.
13840 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
13841 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
13842 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13843 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
13844 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13845 constant_alignment_word_strings.
13846 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
13847 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13848 constant_alignment_word_strings.
13849 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
13850 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13851 constant_alignment_word_strings.
13852 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
13853 * config/microblaze/microblaze.c (microblaze_constant_alignment):
13855 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13856 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
13857 * config/mips/mips.c (mips_constant_alignment): New function.
13858 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13859 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
13860 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
13861 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13862 (mmix_constant_alignment): Make static. Use the same interface
13863 as the target hook.
13864 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
13865 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13866 constant_alignment_word_strings.
13867 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
13868 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13869 constant_alignment_word_strings.
13870 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
13871 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13872 constant_alignment_word_strings.
13873 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
13874 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13875 (rs6000_constant_alignment): New function.
13876 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
13877 * config/riscv/riscv.c (riscv_constant_alignment): New function.
13878 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13879 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
13880 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13881 (rs6000_constant_alignment): New function.
13882 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
13883 * config/s390/s390.c (s390_constant_alignment): New function.
13884 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13885 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
13886 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13887 constant_alignment_word_strings.
13888 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
13889 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13890 (sparc_constant_alignment): New function.
13891 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
13892 * config/spu/spu.c (spu_constant_alignment): New function.
13893 (TARGET_CONSTANT_ALIGNMENT): Redefine.
13894 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
13895 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13896 constant_alignment_word_strings.
13897 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
13898 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13899 constant_alignment_word_strings.
13900 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
13901 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
13902 constant_alignment_word_strings.
13903 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
13904 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13905 (visium_constant_alignment): New function.
13906 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
13907 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
13908 (xtensa_constant_alignment): New function.
13909 * system.h (CONSTANT_ALIGNMENT): Poison.
13911 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
13913 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13914 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
13915 (rs6000_builtin_valid_without_lhs): New helper function.
13916 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13917 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
13919 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
13921 * target.h (vec_perm_indices): Use unsigned short rather than
13923 (auto_vec_perm_indices): Likewise.
13924 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
13925 Use unsigned int rather than unsigned char.
13926 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
13928 2017-09-25 Richard Biener <rguenther@suse.de>
13930 * cfgloop.h (sort_sibling_loops): Declare.
13931 * cfgloop.c (sort_sibling_loops_cmp): New helper.
13932 (sort_sibling_loops): New function sorting the sibling loop list
13934 * graphite.c (graphite_transform_loops): Sort sibling loops.
13936 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
13938 * target.def (vec_perm_const_ok): Change sel parameter to
13940 * optabs-query.c (can_vec_perm_p): Update accordingly.
13941 * doc/tm.texi: Regenerate.
13942 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
13943 auto_vec_perm_indices and remove separate nelt field.
13944 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
13945 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
13946 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
13947 (aarch64_expand_vec_perm_const): Update accordingly.
13948 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
13949 to vec_perm_indices.
13950 * config/arm/arm.c (expand_vec_perm_d): Change perm to
13951 auto_vec_perm_indices and remove separate nelt field.
13952 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
13953 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
13954 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
13956 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
13957 to vec_perm_indices.
13958 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
13959 sel to vec_perm_indices.
13960 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
13961 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
13962 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
13964 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
13967 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
13970 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
13971 on the FUNCTION_DECL function context if it has a DIE that is a
13974 2017-09-25 Richard Biener <rguenther@suse.de>
13976 PR tree-optimization/82285
13977 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
13980 2017-09-25 Tom de Vries <tom@codesourcery.com>
13984 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
13987 2017-09-25 Richard Biener <rguenther@suse.de>
13989 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
13990 ISL errors other than isl_error_quota happen. Dump if the
13991 schedule is the same.
13992 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
13993 errors instead of aborting inside ISL.
13995 2017-09-25 Iain Sandoe <iain@codesourcery.com>
13998 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
13999 of libgcc_eh for m64.
14000 * config/i386/darwin64.h: Likewise.
14002 2017-09-25 Richard Biener <rguenther@suse.de>
14004 PR middle-end/82144
14005 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14006 attribute for incomplete types nor twice for complete ones.
14008 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14011 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14012 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14015 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14018 * config/i386/i386.opt (mprefer-avx256): Use
14019 ix86_target_flags variable.
14020 * config/i386/i386.c (ix86_target_string): Move
14021 -mprefer-avx256 to flag2_opts.
14023 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14025 PR middle-end/35691
14026 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14027 and x != -1 | y != -1 into (x & y) != -1.
14029 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14031 * config.gcc: Add new case statement to set
14032 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14033 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14034 s390x-*-linux* case statements. Added aarch64 to the list of
14035 supported architectures.
14037 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14039 PR tree-optimization/82289
14040 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14041 STMT_VINFO_RELEVANT_P.
14043 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14044 Alan Hayward <alan.hayward@arm.com>
14045 David Sherwood <david.sherwood@arm.com>
14047 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14048 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14049 (extract_range_from_binary_expr_1): Don't call
14050 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14052 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14053 Alan Hayward <alan.hayward@arm.com>
14054 David Sherwood <david.sherwood@arm.com>
14056 * target.def (preferred_vector_alignment): New hook.
14057 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14059 * doc/tm.texi: Regenerate.
14060 * targhooks.h (default_preferred_vector_alignment): Declare.
14061 * targhooks.c (default_preferred_vector_alignment): New function.
14062 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14064 (DR_TARGET_ALIGNMENT): New macro.
14065 (aligned_access_p): Update commentary.
14066 (vect_known_alignment_in_bytes): New function.
14067 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14069 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14070 Calculate the misalignment based on the target alignment rather than
14072 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14073 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14074 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14075 the target alignment, rather than masking the element misalignment
14076 with the number of elements in a vector. Also use the target
14077 alignment when calculating the maximum number of peels.
14078 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14079 instead of TYPE_ALIGN_UNIT.
14080 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14081 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14082 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14083 (vect_create_data_ref_ptr): Likewise.
14084 (vect_setup_realignment): Realign by ANDing with
14085 -DR_TARGET_MISALIGNMENT.
14086 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14087 the number of peels based on DR_TARGET_ALIGNMENT.
14088 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14089 with the guaranteed alignment boundary when deciding whether
14091 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14092 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14093 (ensure_base_align): Remove stmt_info parameter. Get the
14094 target base alignment from DR_TARGET_ALIGNMENT.
14095 (vectorizable_store): Update call accordingly. Interpret
14096 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14098 (vectorizable_load): Likewise.
14100 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14101 Alan Hayward <alan.hayward@arm.com>
14102 David Sherwood <david.sherwood@arm.com>
14104 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14105 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14106 (vect_enhance_data_refs_alignment): Likewise.
14108 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14110 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14111 error. Only quit immediately if parsing is complete.
14112 (BEGIN): Initialize fatal_err and parse_done.
14113 (begin fpu, end fpu): Check number of arguments.
14114 (begin arch, end arch): Likewise.
14115 (begin cpu, end cpu): Likewise.
14116 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14117 (optalias): Likewise.
14119 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14121 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14122 * config/arm/arm-isa.h: Delete. Move definitions to ...
14123 * arm-cpus.in: ... here. Use new feature and fgroup values.
14124 * config/arm/arm.c (arm_option_override): Use lower case for feature
14126 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14127 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14128 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14130 (print_isa_bits_for): New function.
14131 (gen_isa): New function.
14132 (gen_comm_data): Use print_isa_bits_for.
14133 (define feature): New keyword.
14134 (define fgroup): New keyword.
14135 * config/arm/t-arm (TM_H): Remove.
14136 (GTM_H): Add arm-isa.h.
14137 (arm-isa.h): Add rule to generate file.
14138 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14139 case for feature bit names.
14141 2017-09-22 Richard Biener <rguenther@suse.de>
14143 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14145 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14146 print of no dependency loops ...
14147 * graphite.c (graphite_transform_loops): ... here.
14148 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14150 (same_close_phi_node, remove_duplicate_close_phi,
14151 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14152 (canonicalize_loop_closed_ssa): ... here and simplify.
14153 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14154 (optimize_isl): Use dump_printf_loc to tell when we stopped
14155 optimizing because of an ISL timeout.
14157 2017-09-22 Richard Biener <rguenther@suse.de>
14159 PR tree-optimization/82291
14160 * tree-if-conv.c (predicate_mem_writes): Make sure to
14161 remove writes in blocks predicated with false.
14163 2017-09-22 Richard Biener <rguenther@suse.de>
14165 * sese.c: Include cfganal.h.
14166 (if_region_set_false_region): Remove.
14167 (create_if_region_on_edge): Likewise.
14168 (move_sese_in_condition): Re-implement without destroying
14171 2017-09-22 Richard Biener <rguenther@suse.de>
14173 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14174 Verify both BBs contain loop PHI nodes before dispatching to
14175 copy_loop_phi_args.
14176 (graphite_regenerate_ast_isl): Do not recompute dominators,
14177 do not verify three times. Restructure for clarity.
14178 * graphite-scop-detection.c (same_close_phi_node,
14179 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14180 defined_in_loop_p, canonicalize_loop_closed_ssa,
14181 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14182 checking and SSA rewrite, move to ...
14183 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14184 (graphite_initialize): Do not pass in ctx, do not reset the
14185 SCEV cache, compute only dominators.
14186 (graphite_transform_loops): Allocate ISL ctx after
14187 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14188 Maintain post-dominators only around build_scops.
14189 * sese.c (if_region_set_false_region): Make static. Free
14190 and recompute dominators.
14191 (move_sese_in_condition): Assert we don't get called with
14192 post-dominators computed.
14193 * sese.h (if_region_set_false_region): Remove.
14195 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14197 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14198 mode attribute for TARGET_AVX512VL.
14200 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14202 * config/i386/i386.opt (mprefer-avx256): New option.
14203 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14205 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14206 for TARGET_PREFER_AVX256.
14207 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14209 2017-09-21 Jeff Law <law@redhat.com>
14211 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14212 Fix dump output if the only stack space is for pushed registers.
14214 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14216 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14219 2017-09-21 Martin Sebor <msebor@redhat.com>
14222 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14223 code (in C++) or code that triggers warnings.
14225 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14227 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14229 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14231 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14232 * sched-rgn.c: Ditto.
14233 * sel-sched-ir.c: Ditto.
14235 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14237 * toplev.h (set_random_seed): Adjust return type.
14238 * toplev.c (init_local_tick): Move eager initialization of random_seed
14239 to get_random_seed. Adjust comment.
14240 (init_random_seed): Inline to get_random_seed, delete.
14241 (get_random_seed): Initialize random_seed lazily.
14242 (set_random_seed): Do not return previous value.
14243 (print_switch_value): Do not call get_random_seed.
14245 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14247 * cgraph.c (delete_function_version): New, broken out from...
14248 (cgraph_node::delete_function_version): ...here. Rename to
14249 cgraph_node::delete_function_version_by_decl. Update all uses.
14250 (cgraph_node::remove): Call delete_function_version.
14252 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14255 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14256 VAR_DECLs to which addressable non-volatile parameters are mapped
14257 and for id->retvar after the return value assignment. Clear
14258 id->retval and id->retbnd after inlining.
14260 2017-09-21 Richard Biener <rguenther@suse.de>
14262 PR tree-optimization/82276
14263 PR tree-optimization/82244
14264 * tree-vrp.c (build_assert_expr_for): Set
14265 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14267 (remove_range_assertions): Revert earlier change.
14269 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14272 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14274 2017-09-21 Richard Biener <rguenther@suse.de>
14276 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14277 Restore valid IL after code generation errors.
14278 * graphite.c (graphite_transform_loops): Diagnose code
14279 generation issues as MSG_MISSED_OPTIMIZATION and continue
14280 with processing SCOPs.
14282 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14283 Alan Hayward <alan.hayward@arm.com>
14284 David Sherwood <david.sherwood@arm.com>
14286 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14287 rather than choosing between plus_constant and gen_rtx_<CODE>.
14288 * expr.c (emit_push_insn): Likewise.
14289 (expand_expr_real_2): Likewise.
14291 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14292 Alan Hayward <alan.hayward@arm.com>
14293 David Sherwood <david.sherwood@arm.com>
14295 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14297 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14298 Alan Hayward <alan.hayward@arm.com>
14299 David Sherwood <david.sherwood@arm.com>
14301 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14302 calling tree_to_uhwi.
14304 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14305 Alan Hayward <alan.hayward@arm.com>
14306 David Sherwood <david.sherwood@arm.com>
14308 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14309 INTEGER_CST rather than a negative test for ADDR_EXPR.
14311 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14312 Alan Hayward <alan.hayward@arm.com>
14313 David Sherwood <david.sherwood@arm.com>
14315 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14316 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14318 2017-09-21 Richard Biener <rguenther@suse.de>
14320 PR tree-optimization/71351
14321 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14322 graphite_create_new_loop_guard): Remove, fold remaining parts
14324 (translate_isl_ast_node_for): ... here and simplify.
14326 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14329 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14330 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14331 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14332 alternative always use QI mode, for -Os imov (=R,R) alternative
14333 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14336 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14337 Jeff Law <law@redhat.com>
14339 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14340 (allocate_stack_space): New function, partially extracted from
14341 s390_emit_prologue.
14342 (s390_emit_prologue): Track offset to most recent stack probe.
14343 Code to allocate space moved into allocate_stack_space.
14344 Dump actions when no stack is allocated.
14345 (s390_prologue_plus_offset): New function.
14346 (s390_emit_stack_probe): Likewise.
14348 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14350 * common.opt (Wa, Wl, Wp, g, gz=): Add
14352 (gno-column-info): Remove.
14353 (gcolumn-info): Drop RejectNegative.
14354 (gno-): New prefix.
14355 (gno-record-gcc-switches): Remove.
14356 (grecord-gcc-switches): Drop RejectNegative.
14357 (gno-split-dwarf): Remove.
14358 (gsplit-dwarf): Drop RejectNegative.
14359 (gno-strict-dwarf): Remove.
14360 (gstrict-dwarf): Drop RejectNegative.
14361 * config/darwin.opt (gfull, gused): Add RejectNegative.
14362 * dwarf2out.c (gen_producer_string): Drop
14363 gno-record-gcc-switches handler.
14364 * optc-gen.awk: Add g to prefixes with negative forms.
14365 * opts-common.c (remapping_prefix_p): New.
14366 (find_opt): Check it.
14367 (generate_canonical_option): Test g prefix.
14368 (option_map): Add -gno- mapping.
14369 (add_misspelling_candidates): Check remapping_prefix_p.
14371 2017-09-20 Jeff Law <law@redhat.com>
14373 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14374 thinko in stack clash protection support.
14376 * explow.c (compute_stack_clash_protection_loop_data): Use
14377 CONST_INT_P instead of explicit test. Verify object is a
14378 CONST_INT_P before looking at INTVAL.
14379 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14380 instead of explicit test.
14382 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14385 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14386 address instead of to r1 and r11.
14388 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14390 * config.gcc: Support "knm".
14391 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14392 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14394 * config/i386/i386.c (m_KNM): Define.
14395 (processor_target_table): Add "knm".
14397 (ix86_option_override_internal): Add "knm".
14398 (ix86_issue_rate): Add PROCESSOR_KNM.
14399 (ix86_adjust_cost): Ditto.
14400 (ia32_multipass_dfa_lookahead): Ditto.
14401 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
14402 (fold_builtin_cpu): Add M_INTEL_KNM.
14403 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
14404 (processor_type): Add PROCESSOR_KNM.
14405 * config/i386/x86-tune.def: Add m_KNM.
14406 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
14408 2017-09-20 Richard Biener <rguenther@suse.de>
14410 PR tree-optimization/80213
14411 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
14412 are allowed in empty BBs as well.
14413 (canonicalize_loop_closed_ssa): Also look for other complex
14415 (scop_detection::get_sese): Include the loop-closed PHI block
14417 (scop_detection::merge_sese): Remove code adding extra blocks.
14418 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
14419 (build_scops): Assert the final returned scop is invalid.
14421 2017-09-20 Richard Biener <rguenther@suse.de>
14423 PR tree-optimization/82264
14424 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
14426 (vn_phi_lookup): Likewise.
14427 (vn_phi_insert): Likewise.
14429 2017-09-20 Jakub Jelinek <jakub@redhat.com>
14431 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
14432 that fits into uhwi or shwi, add DW_AT_const_value regardless
14433 of early_dwarf without going through RTL, using add_AT_unsigned
14436 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
14437 (DEBUG_LTO_ABBREV_SECTION): Likewise.
14438 (DEBUG_LTO_MACINFO_SECTION): Likewise.
14439 (DEBUG_MACRO_SECTION): Likewise.
14440 (DEBUG_LTO_MACRO_SECTION): Likewise.
14441 (DEBUG_STR_DWO_SECTION): Likewise.
14442 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
14443 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
14444 (DEBUG_LTO_DWO_LINE_SECTION): Define.
14445 (DEBUG_LTO_LINE_STR_SECTION): Define.
14446 (init_sections_and_labels): Initialize debug_line_str_section
14447 variable. Initialize debug_loc_section for -gdwarf-5 to
14448 DEBUG_LOCLISTS_SECTION. Formatting fixes.
14450 2017-09-20 Richard Biener <rguenther@suse.de>
14452 * graphite-sese-to-poly.c (extract_affine): Properly handle
14453 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
14455 2017-09-20 Richard Biener <rguenther@suse.de>
14457 PR tree-optimization/81373
14458 * graphite-scop-detection.c (build_cross_bb_scalars_def):
14459 Force SESE live-out defs to be handled even if they are
14462 2017-09-19 Jeff Law <law@redhat.com>
14464 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
14465 nothing for stack adjustments with REG_STACK_CHECK.
14466 * sched-deps.c (parse_add_or_inc): Reject insns with
14467 REG_STACK_CHECK from dependency breaking.
14468 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
14469 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
14470 * reg-notes.def (STACK_CHECK): New note.
14472 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
14473 (ix86_expand_prologue): Dump stack clash info as needed.
14474 Call ix86_adjust_stack_and_probe_stack_clash as needed.
14476 * function.c (dump_stack_clash_frame_info): New function.
14477 * function.h (dump_stack_clash_frame_info): Prototype.
14478 (enum stack_clash_probes): New enum.
14480 * config/alpha/alpha.c (alpha_expand_prologue): Also check
14481 flag_stack_clash_protection.
14482 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
14483 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
14484 (arm_frame_pointer_required): Likewise.
14485 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
14486 (ia64_expand_prologue): Likewise.
14487 * config/mips/mips.c (mips_expand_prologue): Likewise.
14488 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
14489 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14490 (sparc_flat_expand_prologue): Likewise.
14491 * config/spu/spu.c (spu_expand_prologue): Likewise.
14493 * explow.c: Include "params.h".
14494 (anti_adjust_stack_and_probe_stack_clash): New function.
14495 (get_stack_check_protect): Likewise.
14496 (compute_stack_clash_protection_loop_data): Likewise.
14497 (emit_stack_clash_protection_loop_start): Likewise.
14498 (emit_stack_clash_protection_loop_end): Likewise.
14499 (allocate_dynamic_stack_space): Use get_stack_check_protect.
14500 Use anti_adjust_stack_and_probe_stack_clash.
14501 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
14502 (emit_stack_clash_protection_loop_start): Likewise.
14503 (emit_stack_clash_protection_loop_end): Likewise.
14504 * rtl.h (get_stack_check_protect): Prototype.
14505 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
14506 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
14507 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
14509 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
14511 * doc/tm.texi: Rebuilt.
14512 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
14513 get_stack_check_protect.
14514 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
14515 * config/arm/arm.c (arm_expand_prologue): Likewise.
14516 (arm_frame_pointer_required): Likewise.
14517 * config/i386/i386.c (ix86_expand_prologue): Likewise.
14518 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
14519 * config/mips/mips.c (mips_expand_prologue): Likewise.
14520 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
14521 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
14522 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
14523 (sparc_flat_expand_prologue): Likewise.
14525 * common.opt (-fstack-clash-protection): New option.
14526 * flag-types.h (enum stack_check_type): Note difference between
14527 -fstack-check= and -fstack-clash-protection.
14528 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
14529 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
14530 * toplev.c (process_options): Issue warnings/errors for cases
14531 not handled with -fstack-clash-protection.
14532 * doc/invoke.texi (-fstack-clash-protection): Document new option.
14533 (-fstack-check): Note additional problem with -fstack-check=generic.
14534 Note that -fstack-check is primarily for Ada and refer users
14535 to -fstack-clash-protection for stack-clash-protection.
14536 Document new params for stack clash protection.
14538 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
14540 * config/i386/i386.c (ix86_split_long_move): Do not handle
14541 address used for LEA in a special way.
14543 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
14545 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
14547 2017-09-19 Martin Sebor <msebor@redhat.com>
14550 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
14551 of incompatible types.
14553 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
14555 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14556 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
14557 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14558 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
14560 2017-09-19 Richard Biener <rguenther@suse.de>
14562 PR tree-optimization/82244
14563 * tree-vrp.c (remove_range_assertions): Do not propagate
14564 a constant to abnormals but replace the assert with a copy.
14566 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
14568 PR rtl-optimization/57878
14569 PR rtl-optimization/68988
14570 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
14571 avoidance test involving non_reload_pseudos. Move frequency test
14572 below the general fragmentation avoidance test.
14574 2017-09-19 Richard Biener <rguenther@suse.de>
14576 PR tree-optimization/69728
14577 * graphite-sese-to-poly.c (schedule_error): New global.
14578 (add_loop_schedule): Handle empty domain by failing the
14580 (build_original_schedule): Handle schedule_error.
14582 2017-09-19 Richard Biener <rguenther@suse.de>
14584 * graphite-scop-detection.c (scop_detection::can_represent_loop):
14585 Do not iterate to sibling loops but only to siblings of inner
14588 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
14591 * config/m68k/m68k.md (moveq feeding equality comparison): Check
14592 that the registers are different.
14594 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
14596 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
14597 to processor_model and "amdfam17h" to arch_names_table.
14598 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
14600 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14603 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
14605 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14606 Alan Hayward <alan.hayward@arm.com>
14607 David Sherwood <david.sherwood@arm.com>
14609 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
14611 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
14613 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
14614 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
14615 vect_schedule_slp_instance.
14616 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
14617 Update call to vect_slp_analyze_node_operations. Simplify return
14619 (vect_slp_analyze_bb_1): Update call accordingly.
14620 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
14621 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
14622 (vect_schedule_slp): Update call accordingly.
14624 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14625 Alan Hayward <alan.hayward@arm.com>
14626 David Sherwood <david.sherwood@arm.com>
14628 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
14629 with types that aren't in fact scalar.
14631 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14633 * tree-vect-slp.c (vect_record_max_nunits): New function,
14635 (vect_build_slp_tree_1): ...here.
14636 (vect_build_slp_tree_2): Call it for phis too.
14638 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14640 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
14641 to vect_get_vec_def_for_operand when getting the mask operand.
14643 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14644 Alan Hayward <alan.hayward@arm.com>
14645 David Sherwood <david.sherwood@arm.com>
14647 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
14650 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14651 Alan Hayward <alan.hayward@arm.com>
14652 David Sherwood <david.sherwood@arm.com>
14654 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
14655 calculation for vector booleans.
14657 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
14658 Alan Hayward <alan.hayward@arm.com>
14659 David Sherwood <david.sherwood@arm.com>
14661 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
14663 (vect_transform_stmt): ...here.
14664 (vect_analyze_stmt): Use it instead of calling
14665 vectorizable_live_operation directly.
14667 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
14669 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
14670 non-SIMT targets in acc vector loops.
14672 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
14674 * configure.ac: Add arc and check if assembler supports gdwarf2.
14675 * configure: Regenerate.
14677 2017-09-18 Richard Biener <rguenther@suse.de>
14679 PR tree-optimization/82220
14680 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
14681 epilogue niters from the min_profitable_iters compute.
14683 2017-09-18 Jakub Jelinek <jakub@redhat.com>
14686 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
14687 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
14688 (ix86_init_pic_reg): Revert 2017-09-01 changes.
14690 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
14693 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
14694 switching to a new text section.
14696 2017-09-18 Richard Biener <rguenther@suse.de>
14698 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
14700 (build_alias_set): Reject aliases with no access function.
14702 2017-09-18 Richard Biener <rguenther@suse.de>
14704 PR tree-optimization/79622
14705 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
14707 (build_cross_bb_scalars_use): Likewise.
14709 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
14711 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
14713 2017-09-18 Alan Modra <amodra@gmail.com>
14716 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
14717 stack_pointer_rtx for count 0. Update comments. Break up
14718 large rtl expression.
14720 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
14722 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
14723 Increase to 20 bytes.
14724 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
14725 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
14726 or avx version of the stub.
14728 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
14731 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
14732 compute the minimum stack alignment. Also update preferred stack
14733 boundary for leaf functions.
14735 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
14737 PR tree-optimization/82228
14738 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
14741 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14743 * common/config/nds32/nds32-common.c
14744 (nds32_option_optimization_table): Refine formatting.
14745 (nds32_option_optimization_table): Use -fsched-pressure and
14746 -fomit-frame-pointer for specific optimization level.
14748 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
14750 * config/nds32/nds32.c: Refine formatting and comments.
14751 * config/nds32/nds32.h: Likewise.
14752 * config/nds32/nds32.md: Likewise.
14753 * config/nds32/nds32-cost.c: Likewise.
14754 * config/nds32/nds32-isr.c: Likewise.
14755 * config/nds32/nds32-md-auxiliary.c: Likewise.
14756 * config/nds32/nds32-multiple.md: Likewise.
14757 * config/nds32/nds32-predicates.c: Likewise.
14759 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
14760 Jakub Jelinek <jakub@redhat.com>
14762 Add support for -std=c++2a.
14763 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
14765 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
14767 2017-09-15 Steve Ellcey <sellcey@cavium.com>
14770 * doc/extend.texi (Common Function Attributes): Add
14771 references to ARM, AArch64, and S/390 specific attributes.
14772 (Function Specific Option Pragmas): Add AArch64 and S/390
14773 to list of back ends that support the target pragma.
14775 2017-09-15 Nathan Sidwell <nathan@acm.org>
14777 * doc/standards.texi: Fix C++17 description. Update URLs for
14780 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14782 * common.opt (Wcast-align=strict): New warning option.
14783 * doc/invoke.texi: Document -Wcast-align=strict.
14785 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
14787 * cgraph.h (cgraph_thunk_info): Add comments.
14788 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
14789 assert for VIRTUAL_* arguments stricter.
14791 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
14793 PR tree-optimization/71026
14794 * match.pd: Move RDIV patterns from fold-const.c
14795 * fold-const.c (distribute_real_division): Removed.
14796 (fold_binary_loc): Remove calls to distribute_real_divison.
14798 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14800 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
14801 c++1z and gnu++1z as deprecated. Change other references to
14802 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
14803 Change -Wc++1z-compat to -Wc++17-compat.
14804 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
14805 * dwarf2out.c (highest_c_language): Handle C++17.
14806 (gen_compile_unit_die): Likewise.
14808 2017-09-15 Jakub Jelinek <jakub@redhat.com>
14810 PR rtl-optimization/82192
14811 * combine.c (make_extraction): Don't look through non-paradoxical
14812 SUBREGs or TRUNCATE if pos + len is or might be bigger than
14815 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14816 Alan Hayward <alan.hayward@arm.com>
14817 David Sherwood <david.sherwood@arm.com>
14819 * target.def (function_arg_offset): New hook.
14820 * targhooks.h (default_function_arg_offset): Declare.
14821 * targhooks.c (default_function_arg_offset): New function.
14822 * function.c (locate_and_pad_parm): Use
14823 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
14824 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
14825 (TARGET_FUNCTION_ARG_OFFSET): ...this.
14826 * doc/tm.texi: Regenerate.
14827 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
14828 * config/spu/spu.c (spu_function_arg_offset): New function.
14829 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
14830 * system.h (FUNCTION_ARG_OFFSET): Poison.
14832 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14833 Alan Hayard <alan.hayward@arm.com>
14834 David Sherwood <david.sherwood@arm.com>
14836 * target.def (truly_noop_truncation): New hook.
14837 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
14838 than TRULY_NOOP_TRUNCATION.
14839 * hooks.h (hook_bool_uint_uint_true): Declare.
14840 * hooks.c (hook_bool_uint_uint_true): New function.
14841 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
14842 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
14843 * doc/tm.texi: Regenerate.
14844 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
14845 rather than TRULY_NOOP_TRUNCATION in comments.
14846 (simplify_comparison): Likewise.
14847 (record_truncated_value): Likewise.
14848 * expmed.c (extract_bit_field_1): Likewise.
14849 (extract_split_bit_field): Likewise.
14850 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
14851 instead of TRULY_NOOP_TRUNCATION.
14852 * function.c (assign_parm_setup_block): Likewise.
14853 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
14854 * rtlhooks.c: Include target.h.
14855 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
14856 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
14857 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
14858 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
14859 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
14860 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
14861 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
14862 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
14863 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
14864 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
14865 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
14866 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
14867 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
14868 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
14869 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
14870 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
14871 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
14872 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
14873 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
14874 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
14875 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
14876 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
14877 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
14878 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
14879 * config/mips/mips.c (mips_truly_noop_truncation): New function.
14880 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14881 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14882 rather than TRULY_NOOP_TRUNCATION in comments.
14883 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
14884 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
14885 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
14886 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
14887 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
14888 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
14889 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
14890 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
14891 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
14892 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
14893 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
14894 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14895 rather than TRULY_NOOP_TRUNCATION in comments.
14896 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
14897 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
14898 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
14899 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
14900 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
14901 TRULY_NOOP_TRUNCATION condition.
14902 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
14903 (TRULY_NOOP_TRUNCATION): Delete.
14904 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
14905 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
14906 * config/spu/spu.c (spu_truly_noop_truncation): New function.
14907 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14908 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
14909 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
14910 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
14911 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
14912 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
14913 rather than TRULY_NOOP_TRUNCATION in comments.
14914 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
14915 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
14916 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
14917 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
14918 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
14919 * system.h (TRULY_NOOP_TRUNCATION): Poison.
14921 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
14924 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
14925 (*cmp_ior): Likewise.
14926 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
14927 (*ior_scc_scc_cmp): Likewise.
14928 (*and_scc_scc): Likewise.
14929 (*and_scc_scc_cmp): Likewise.
14931 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
14932 Alan Hayard <alan.hayward@arm.com>
14933 David Sherwood <david.sherwood@arm.com>
14935 * target.def (can_change_mode_class): New hook.
14936 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
14937 (hard_regno_nregs): Likewise.
14938 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
14939 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
14940 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
14941 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
14942 (LOAD_EXTEND_OP): Update accordingly.
14943 * doc/tm.texi: Regenerate.
14944 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
14945 CANNOT_CHANGE_MODE_CLASS.
14946 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
14947 (REG_CAN_CHANGE_MODE_P): ...this new macro.
14948 * combine.c (simplify_set): Update accordingly.
14949 * emit-rtl.c (validate_subreg): Likewise.
14950 * recog.c (general_operand): Likewise.
14951 * regcprop.c (mode_change_ok): Likewise.
14952 * reload1.c (choose_reload_regs): Likewise.
14953 (inherit_piecemeal_p): Likewise.
14954 * rtlanal.c (simplify_subreg_regno): Likewise.
14955 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
14956 instead of CANNOT_CHANGE_MODE_CLASS.
14957 (reload_cse_simplify_operands): Likewise.
14958 * reload.c (push_reload): Use targetm.can_change_mode_class
14959 instead of CANNOT_CHANGE_MODE_CLASS.
14960 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
14961 REG_CANNOT_CHANGE_MODE_P.
14962 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14963 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
14964 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14965 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14966 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14967 (arm_can_change_mode_class): New function.
14968 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
14969 than CANNOT_CHANGE_MODE_CLASS in comments.
14970 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14971 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
14972 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
14973 (ix86_can_change_mode_class): ...this new function, inverting the
14974 sense of the return value.
14975 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14976 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14977 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14978 (ia64_can_change_mode_class): New function.
14979 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14980 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
14981 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
14982 (m32c_can_change_mode_class): ...this new function, inverting the
14983 sense of the return value.
14984 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14985 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14986 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
14987 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
14988 (mips_can_change_mode_class): ...this new function, inverting the
14989 sense of the return value.
14990 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14991 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14992 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14993 (msp430_can_change_mode_class): New function.
14994 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14995 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
14996 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
14997 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14998 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
14999 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15000 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15001 (pa_cannot_change_mode_class): Replace with...
15002 (pa_can_change_mode_class): ...this new function, inverting the
15003 sense of the return value.
15004 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15005 than CANNOT_CHANGE_MODE_CLASS in comments.
15006 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15007 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15008 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15009 (pdp11_cannot_change_mode_class): Replace with...
15010 (pdp11_can_change_mode_class): ...this new function, inverting the
15011 sense of the return value.
15012 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15013 * config/powerpcspe/powerpcspe-protos.h
15014 (rs6000_cannot_change_mode_class_ptr): Delete.
15015 * config/powerpcspe/powerpcspe.c
15016 (rs6000_cannot_change_mode_class_ptr): Delete.
15017 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15018 (rs6000_option_override_internal): Assign to
15019 targetm.can_change_mode_class instead of
15020 rs6000_cannot_change_mode_class_ptr.
15021 (rs6000_cannot_change_mode_class): Replace with...
15022 (rs6000_can_change_mode_class): ...this new function, inverting the
15023 sense of the return value.
15024 (rs6000_debug_cannot_change_mode_class): Replace with...
15025 (rs6000_debug_can_change_mode_class): ...this new function.
15026 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15027 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15028 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15029 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15030 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15032 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15033 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15034 (rs6000_option_override_internal): Assign to
15035 targetm.can_change_mode_class instead of
15036 rs6000_cannot_change_mode_class_ptr.
15037 (rs6000_cannot_change_mode_class): Replace with...
15038 (rs6000_can_change_mode_class): ...this new function, inverting the
15039 sense of the return value.
15040 (rs6000_debug_cannot_change_mode_class): Replace with...
15041 (rs6000_debug_can_change_mode_class): ...this new function.
15042 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15043 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15044 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15045 (s390_can_change_mode_class): ...this new function, inverting the
15046 sense of the return value.
15047 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15048 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15049 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15050 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15051 (sh_cannot_change_mode_class): Replace with...
15052 (sh_can_change_mode_class): ...this new function, inverting the
15053 sense of the return value.
15054 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15055 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15056 (sparc_can_change_mode_class): New function.
15057 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15058 * config/spu/spu.c (spu_can_change_mode_class): New function.
15059 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15060 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15061 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15062 (visium_can_change_mode_class): New function.
15063 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15065 2017-09-15 Richard Biener <rguenther@suse.de>
15067 PR tree-optimization/82217
15068 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15069 but not undefined case.
15071 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15074 * postreload.c (reload_cse_simplify_operands): Skip
15075 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15077 2017-09-15 Richard Biener <rguenther@suse.de>
15079 PR tree-optimization/68823
15080 * graphite-scop-detection.c (build_alias_set): If we have a
15081 possible dependence check whether we can handle them by just
15082 looking at the DRs DR_ACCESS_FNs.
15083 (build_scops): If build_alias_set fails, fail the SCOP.
15085 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15087 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15088 to support float128 built-in functions that require the ISA 3.0
15090 (BU_FLOAT128_3_HW): Likewise.
15091 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15092 built-in functions.
15093 (FMAF128): Likewise.
15095 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15096 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15097 floating point instructions.
15098 (rs6000_invalid_builtin): Likewise.
15099 (rs6000_builtin_mask_names): Likewise.
15100 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15101 (RS6000_BTM_FLOAT128_HW): Likewise.
15102 (RS6000_BTM_COMMON): Likewise.
15103 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15105 * doc/extend.texi (RS/6000 built-in functions): Document the
15106 IEEE 128-bit floating point square root and fused multiply-add
15107 built-in functions.
15109 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15111 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15112 reg (r2) isn't in the set of registers defined in the prologue.
15114 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15115 Alan Hayward <alan.hayward@arm.com>
15116 David Sherwood <david.sherwood@arm.com>
15118 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15119 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15120 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15121 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15122 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15124 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15125 max_vectorization_factor.
15126 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15128 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15129 Alan Hayward <alan.hayward@arm.com>
15130 David Sherwood <david.sherwood@arm.com>
15132 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15133 (vect_worthwhile_without_simd_p): Declare.
15134 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15135 (vectorizable_reduction): Use it.
15136 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15137 (vectorizable_operation): Likewise.
15139 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15140 Alan Hayward <alan.hayward@arm.com>
15141 David Sherwood <david.sherwood@arm.com>
15143 * tree-vectorizer.h (vect_get_num_copies): New function.
15144 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15145 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15146 (vectorizable_induction): Likewise.
15147 (vectorizable_live_operation): Likewise.
15148 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15149 (vectorizable_bswap): Likewise.
15150 (vectorizable_call): Likewise.
15151 (vectorizable_conversion): Likewise.
15152 (vectorizable_assignment): Likewise.
15153 (vectorizable_shift): Likewise.
15154 (vectorizable_operation): Likewise.
15155 (vectorizable_store): Likewise.
15156 (vectorizable_load): Likewise.
15157 (vectorizable_condition): Likewise.
15158 (vectorizable_comparison): Likewise.
15159 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15161 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15162 Alan Hayward <alan.hayward@arm.com>
15163 David Sherwood <david.sherwood@arm.com>
15165 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15166 of vect_init_vector.
15168 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15169 Alan Hayward <alan.hayward@arm.com>
15170 David Sherwood <david.sherwood@arm.com>
15172 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15173 an inline wrapper that provides a location.
15174 (gimple_build_vector): Likewise.
15175 * gimple-fold.c (gimple_build_vector_from_val): New function.
15176 (gimple_build_vector): Likewise.
15177 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15178 functions to build the initial value. Always return a gimple value.
15179 (get_initial_defs_for_reduction): Likewise. Only compute
15181 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15182 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15183 (vectorizable_induction): Use gimple_build_vector rather than
15186 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15187 Alan Hayward <alan.hayward@arm.com>
15188 David Sherwood <david.sherwood@arm.com>
15190 * target.h (vec_perm_indices): New typedef.
15191 (auto_vec_perm_indices): Likewise.
15192 * optabs-query.h: Include target.h
15193 (can_vec_perm_p): Take a vec_perm_indices *.
15194 * optabs-query.c (can_vec_perm_p): Likewise.
15195 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15196 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15197 * tree-vect-generic.c (lower_vec_perm): Likewise.
15198 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15199 (vect_grouped_load_supported): Likewise.
15200 (vect_shift_permute_load_chain): Likewise.
15201 (vect_permute_store_chain): Use auto_vec_perm_indices.
15202 (vect_permute_load_chain): Likewise.
15203 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15204 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15205 Update uses of can_vec_perm_p.
15206 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15207 mode with a number of elements. Take a vec_perm_indices *.
15208 (vect_create_epilog_for_reduction): Update accordingly.
15209 Use auto_vec_perm_indices.
15210 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15211 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15212 (vect_transform_slp_perm_load): Likewise.
15213 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15214 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15215 (vect_gen_perm_mask_checked): Likewise.
15216 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15217 (vect_gen_perm_mask_checked): Likewise.
15218 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15219 (vectorizable_store): Likewise.
15220 (vectorizable_load): Likewise.
15221 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15222 (vectorizable_bswap): Likewise.
15224 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15225 Alan Hayward <alan.hayward@arm.com>
15226 David Sherwood <david.sherwood@arm.com>
15228 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15229 * tree.c (build_vector): Likewise.
15230 (build_vector_from_ctor): Update accordingly.
15231 (build_vector_from_val): Likewise.
15232 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15233 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15234 * tree-vect-generic.c (add_rshift): Likewise.
15235 (expand_vector_divmod): Likewise.
15236 (optimize_vector_constructor): Likewise.
15237 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15238 (vect_transform_slp_perm_load): Likewise.
15239 (vect_schedule_slp_instance): Likewise.
15240 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15241 (vectorizable_call): Likewise.
15242 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15243 * expmed.c (make_tree): Likewise.
15244 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15245 a vector passed to build_vector.
15246 (fold_convert_const): Likewise.
15247 (exact_inverse): Likewise.
15248 (fold_ternary_loc): Likewise.
15249 (fold_relational_const): Likewise.
15250 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15251 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15252 (const_unop): Likewise. Store the reduction accumulator in a
15253 variable rather than an array.
15254 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15255 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15256 the new vector, rather than constructing it after the input arrays.
15257 (native_interpret_vector): Use auto_vec<tree> when building
15258 a vector passed to build_vector. Add elements in order.
15259 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15260 auto_vec<tree> when building a vector passed to build_vector.
15261 (vect_create_epilog_for_reduction): Likewise.
15262 (vectorizable_induction): Likewise.
15263 (get_initial_def_for_reduction): Likewise. Fix indentation of
15265 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15267 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15268 passed to build_vector.
15270 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15271 Alan Hayward <alan.hayward@arm.com>
15272 David Sherwood <david.sherwood@arm.com>
15274 * tree-core.h (tree_base::u): Add an "nelts" field.
15275 (tree_vector): Use VECTOR_CST_NELTS as the length.
15276 * tree.c (tree_size): Likewise.
15277 (make_vector): Initialize VECTOR_CST_NELTS.
15278 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15279 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15280 TYPE_VECTOR_SUBPARTS.
15281 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15282 as the number of units, setting it from VECTOR_CST_NELTS.
15283 (const_vector_from_tree): Likewise.
15284 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15285 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15286 (fold_negate_expr_1): Likewise.
15287 (fold_convert_const): Likewise.
15288 (const_binop): Likewise. Differentiate the number of output and
15290 (const_unop): Likewise.
15291 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15292 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15293 in cases that did the opposite.
15295 2017-09-14 Richard Biener <rguenther@suse.de>
15297 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15300 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15302 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15304 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15307 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15308 if and where to split a bb, except for splitting before debug insn
15309 sequences followed by non-label real insn. Delete debug insns
15310 in between basic blocks.
15312 * combine.c (make_compound_operation_int): Formatting fixes.
15314 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15315 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15316 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15317 * config/sol2.h (LINK_EH_SPEC): Likewise.
15318 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15319 * config/s390/linux.h (LINK_SPEC): Likewise.
15320 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15321 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15322 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15323 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15324 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15325 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15326 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15327 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15329 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15331 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15333 (ENDFILE_LINUX_SPEC): Likewise.
15334 (LINK_EH_SPEC): Likewise.
15335 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15336 (LINK_OS_LINUX_SPEC32): Likewise.
15337 (LINK_OS_LINUX_SPEC64): Likewise.
15338 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15339 (LINK_OS_LINUX_SPEC): Likewise.
15341 2017-09-13 Martin Liska <mliska@suse.cz>
15343 PR middle-end/82154
15344 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15345 CASE_HIGH is NULL_TREE.
15347 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15348 Alan Hayward <alan.hayward@arm.com>
15349 David Sherwood <david.sherwood@arm.com>
15351 * target.def (secondary_memory_needed): New hook.
15352 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15353 instead of SECONDARY_MEMORY_NEEDED.
15354 (secondary_memory_needed_mode): Likewise.
15355 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15356 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15357 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15358 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15359 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15360 * doc/tm.texi: Regenerate.
15361 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15362 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15363 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15364 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15365 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15366 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15367 mode argument first and change the reg_class arguments to reg_class_t.
15368 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15369 Make static. Update the call to inline_secondary_memory_needed.
15370 (ix86_register_move_cost): Update the call to
15371 inline_secondary_memory_needed.
15372 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15373 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15375 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15376 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15378 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15379 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15380 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15381 and match hook interface. Add comment from mips.h.
15382 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15383 * config/mmix/mmix.md (truncdfsf2): Refer to
15384 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15386 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15387 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15388 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15389 (pa_secondary_memory_needed): New function.
15390 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15391 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15392 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15393 (pdp11_secondary_memory_needed): Make static and match hook interface.
15394 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
15395 * config/powerpcspe/powerpcspe-protos.h
15396 (rs6000_secondary_memory_needed_ptr): Delete.
15397 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
15399 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15400 (rs6000_option_override_internal): Assign to
15401 targetm.secondary_memory_needed rather than
15402 rs6000_secondary_memory_needed_ptr.
15403 (rs6000_secondary_memory_needed): Match hook interface.
15404 (rs6000_debug_secondary_memory_needed): Likewise.
15405 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
15406 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
15407 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
15408 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15409 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
15410 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
15412 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
15413 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15414 (rs6000_option_override_internal): Assign to
15415 targetm.secondary_memory_needed rather than
15416 rs6000_secondary_memory_needed_ptr.
15417 (rs6000_secondary_memory_needed): Match hook interface.
15418 (rs6000_debug_secondary_memory_needed): Likewise.
15419 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
15420 * config/s390/s390.c (s390_secondary_memory_needed): New function.
15421 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15422 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
15423 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15424 (sparc_secondary_memory_needed): New function.
15425 * lra-constraints.c (check_and_process_move): Refer to
15426 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15428 (curr_insn_transform): Likewise.
15429 (process_alt_operands): Use targetm.secondary_memory_needed
15430 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15431 (check_secondary_memory_needed_p): Likewise.
15432 (choose_split_class): Likewise.
15433 * reload.c: Unconditionally include code that was previously
15434 conditional on SECONDARY_MEMORY_NEEDED.
15435 (push_secondary_reload): Use targetm.secondary_memory_needed
15436 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15437 (push_reload): Likewise.
15438 * reload1.c: Unconditionally include code that was previously
15439 conditional on SECONDARY_MEMORY_NEEDED.
15440 (choose_reload_regs): Use targetm.secondary_memory_needed
15441 instead of TARGET_SECONDARY_MEMORY_NEEDED.
15442 (gen_reload): Likewise.
15443 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
15445 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15446 Alan Hayward <alan.hayward@arm.com>
15447 David Sherwood <david.sherwood@arm.com>
15449 * target.def (secondary_memory_needed_mode): New hook:
15450 * targhooks.c (default_secondary_memory_needed_mode): Declare.
15451 * targhooks.h (default_secondary_memory_needed_mode): New function.
15452 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
15453 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
15454 * doc/tm.texi: Regenerate.
15455 * lra-constraints.c (check_and_process_move): Use
15456 targetm.secondary_memory_needed_mode instead of
15457 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
15458 (curr_insn_transform): Likewise.
15459 * reload.c (get_secondary_mem): Likewise.
15460 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15461 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
15463 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15464 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15465 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
15466 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15467 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
15469 * config/powerpcspe/powerpcspe-protos.h
15470 (rs6000_secondary_memory_needed_mode): Delete.
15471 * config/powerpcspe/powerpcspe.c
15472 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15473 (rs6000_secondary_memory_needed_mode): Make static.
15474 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15475 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
15477 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15479 (rs6000_secondary_memory_needed_mode): Make static.
15480 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15481 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
15482 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
15483 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
15484 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
15486 (sparc_secondary_memory_needed_mode): New function.
15487 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
15489 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
15491 * config/aarch64/constraints.md (Umq): New constraint.
15492 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
15494 (mov<mode>): Update condition.
15496 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15498 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
15499 when bitposition is the same.
15501 2017-09-13 Richard Biener <rguenther@suse.de>
15503 * dwarf2out.c (output_die_symbol): Remove.
15504 (output_die): Do not output a DIEs symbol.
15506 2017-09-13 Richard Biener <rguenther@suse.de>
15508 PR middle-end/82128
15509 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
15510 default-def to avoid breaking iterator update with the weird
15511 interaction with cgraph_update_edges_for_call_stmt_node.
15513 2017-09-13 Richard Biener <rguenther@suse.de>
15515 * tree-cfg.c (verify_gimple_assign_binary): Add verification
15516 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
15517 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
15518 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
15520 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15522 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
15523 Disable pc relative literal load irrespective of
15524 TARGET_FIX_ERR_A53_84341 for default.
15526 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
15528 * config/sparc/sparc.c (output_return): Output the source location of
15529 the insn in the delay slot, if any.
15530 (output_sibcall): Likewise.
15532 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
15535 * common.opt (-static-pie): New alias.
15536 (shared): Negate static-pie.
15537 (-no-pie): Update help text.
15539 (static-pie): New option.
15540 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
15541 -static-pie support.
15542 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
15543 (LINK_EH_SPEC): Likewise.
15544 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15545 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15546 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
15547 * gcc.c (LINK_COMMAND_SPEC): Likewise.
15548 (init_gcc_specs): Likewise.
15549 (init_spec): Likewise.
15550 (display_help): Update help message for -pie.
15551 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
15554 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
15556 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
15557 (movdi_aarch64): Likewise.
15558 (movti_aarch64): Likewise.
15560 2017-09-12 Simon Wright <simon@pushface.org>
15563 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
15564 calculation of the minor version, always output as 0.
15566 2017-09-12 Jakub Jelinek <jakub@redhat.com>
15569 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15570 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
15571 on it early, rather than manual conversion late. For
15572 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
15573 instead of performing manual conversion.
15575 2017-09-12 Carl Love <cel@us.ibm.com>
15577 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
15578 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
15580 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15581 VMULOSW): Add definitions.
15582 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15583 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15584 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15585 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15586 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15588 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15590 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
15592 (movti_aarch64): Likewise.
15593 (movdf_aarch64): Likewise.
15594 (movtf_aarch64): Likewise.
15595 (load_pairdi): Likewise.
15596 (store_pairdi): Likewise.
15597 (load_pairdf): Likewise.
15598 (store_pairdf): Likewise.
15599 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
15600 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
15601 (ldr_got_small_<mode>): Likewise.
15602 (ldr_got_small_28k_<mode>): Likewise.
15603 (ldr_got_tiny): Likewise.
15604 * config/aarch64/iterators.md (ldst_sz): New.
15605 (ldpstp_sz): Likewise.
15606 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
15608 (thunderx_load): Split load_8 to load_16.
15609 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
15611 (thunderx2t99_storepair_basic): Split store_8 to store_16.
15612 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
15613 (xgene1_store_pair): Split store_8 to store_16.
15614 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
15615 (falkor_st_0_st_sd): Split store_8 to store_16.
15617 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
15619 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
15620 and store1/2/3/4 to store_4/8/12/16.
15621 * config/aarch64/aarch64.md: Update for rename.
15622 * config/arm/arm.md: Likewise.: Likewise.
15623 * config/arm/arm.c: Likewise.
15624 * config/arm/thumb1.md: Likewise.
15625 * config/arm/thumb2.md: Likewise.
15626 * config/arm/vfp.md: Likewise.
15627 * config/arm/arm-generic.md: Likewise.
15628 * config/arm/arm1020e.md: Likewise.
15629 * config/arm/arm1026ejs.md: Likewise.
15630 * config/arm/arm1136jfs.md: Likewise.
15631 * config/arm/arm926ejs.md: Likewise.
15632 * config/arm/cortex-a15.md: Likewise.
15633 * config/arm/cortex-a17.md: Likewise.
15634 * config/arm/cortex-a5.md: Likewise.
15635 * config/arm/cortex-a53.md: Likewise.
15636 * config/arm/cortex-a57.md: Likewise.
15637 * config/arm/cortex-a7.md: Likewise.
15638 * config/arm/cortex-a8.md: Likewise.
15639 * config/arm/cortex-a9.md: Likewise.
15640 * config/arm/cortex-m4.md: Likewise.
15641 * config/arm/cortex-m7.md: Likewise.
15642 * config/arm/cortex-r4.md: Likewise.
15643 * config/arm/exynos-m1.md: Likewise.
15644 * config/arm/fa526.md: Likewise.
15645 * config/arm/fa606te.md: Likewise.
15646 * config/arm/fa626te.md: Likewise.
15647 * config/arm/fa726te.md: Likewise.
15648 * config/arm/fmp626.md: Likewise.
15649 * config/arm/iwmmxt.md: Likewise.
15650 * config/arm/ldmstm.md: Likewise.
15651 * config/arm/marvell-pj4.md: Likewise.
15652 * config/arm/xgene1.md: Likewise.
15653 * config/aarch64/thunderx.md: Likewise.
15654 * config/aarch64/thunderx2t99.md: Likewise.
15655 * config/aarch64/falkor.md: Likewise.
15657 2017-09-12 Martin Liska <mliska@suse.cz>
15659 * attribs.c (private_lookup_attribute): New function.
15660 * attribs.h (private_lookup_attribute): Declared here.
15661 (lookup_attribute): Called from this place.
15663 2017-09-12 Richard Biener <rguenther@suse.de>
15665 PR tree-optimization/82157
15666 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
15667 stmts with side-effects.
15669 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15670 Alan Hayward <alan.hayward@arm.com>
15671 David Sherwood <david.sherwood@arm.com>
15673 * target.def (hard_regno_nregs): New hook.
15674 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
15675 * targhooks.h (default_hard_regno_nregs): Declare.
15676 * targhooks.c (default_hard_regno_nregs): New function.
15677 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
15678 (TARGET_HARD_REGNO_NREGS): ...this hook.
15679 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
15680 (CLASS_MAX_NREGS): Likewise.
15681 * doc/tm.texi: Regenerate.
15682 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
15683 instead of HARD_REGNO_NREGS.
15684 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
15685 HARD_REGNO_NREGS in the comment.
15686 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
15687 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
15688 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
15689 Return an unsigned int.
15690 (TARGET_HARD_REGNO_NREGS): Redefine.
15691 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
15692 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
15693 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15694 (arc_hard_regno_nregs): New function.
15695 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
15696 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
15697 (arm_hard_regno_nregs): New function.
15698 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
15699 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
15700 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
15701 (TARGET_HARD_REGNO_NREGS): Redefine.
15702 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
15703 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
15704 (HARD_REGNO_NREGS): Delete.
15705 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
15706 (cr16_hard_regno_nregs): New function.
15707 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
15708 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
15709 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
15710 (cris_hard_regno_nregs): New function.
15711 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
15712 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
15713 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15714 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
15715 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
15716 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
15717 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
15718 (frv_hard_regno_nregs): Make static. Take and return an
15720 (frv_class_max_nregs): Remove outdated copy of documentation.
15721 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
15722 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
15723 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
15724 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
15725 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
15726 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
15727 (TARGET_HARD_REGNO_NREGS): Redefine.
15728 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
15729 (CLASS_MAX_NREGS): Update comment.
15730 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
15731 (ia64_hard_regno_nregs): New function.
15732 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
15733 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
15734 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
15735 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
15736 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
15738 (m32c_hard_regno_nregs): Likewise. Make static.
15739 (TARGET_HARD_REGNO_NREGS): Redefine.
15740 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
15741 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
15742 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
15743 (m68k_hard_regno_nregs): New function.
15744 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
15745 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
15746 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
15747 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
15748 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
15749 Take and return an unsigned int.
15750 (TARGET_HARD_REGNO_NREGS): Redefine.
15751 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
15752 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
15753 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
15754 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
15755 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
15756 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
15757 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
15758 (msp430_hard_regno_nregs): Make static. Take and return an
15760 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
15761 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
15762 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
15763 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
15764 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
15765 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
15766 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
15767 (TARGET_HARD_REGNO_NREGS): Redefine.
15768 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
15769 (PA_HARD_REGNO_NREGS): ...this.
15770 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
15771 (PA_HARD_REGNO_NREGS): ...this.
15772 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15773 (pa_hard_regno_nregs): New function.
15774 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
15775 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
15776 (pdp11_hard_regno_nregs): New function.
15777 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
15778 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
15779 (rs6000_hard_regno_nregs_hook): New function.
15780 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
15781 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
15782 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
15783 Take and return an unsigned int. Move earlier in file.
15784 (TARGET_HARD_REGNO_NREGS): Redefine.
15785 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
15786 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
15787 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
15788 (rl78_hard_regno_nregs): Make static. Take and return an
15790 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
15791 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
15792 (rs6000_hard_regno_nregs_hook): New function.
15793 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
15794 * config/rx/rx.c (rx_hard_regno_nregs): New function.
15795 (TARGET_HARD_REGNO_NREGS): Redefine.
15796 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
15797 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
15798 instead of HARD_REGNO_NREGS.
15799 (s390_hard_regno_nregs): New function.
15800 (s390_hard_regno_mode_ok): Add comment from s390.h.
15801 (TARGET_HARD_REGNO_NREGS): Redefine.
15802 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
15803 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
15804 (sh_hard_regno_nregs): New function.
15805 (sh_pass_in_reg_p): Use it.
15806 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
15807 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
15808 (sparc_hard_regno_nregs): New function.
15809 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
15810 * config/spu/spu.c (spu_hard_regno_nregs): New function.
15811 (spu_function_arg_advance): Use it, supplying a valid register number.
15812 (TARGET_HARD_REGNO_NREGS): Redefine.
15813 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
15814 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
15815 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
15816 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
15817 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
15818 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
15819 (CLASS_MAX_NREGS): Remove copy of old documentation.
15820 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
15821 (visium_hard_regno_nregs): New function.
15822 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
15823 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
15824 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
15825 xtensa_hard_regno_nregs): New function.
15826 * system.h (HARD_REGNO_NREGS): Poison.
15828 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15830 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
15831 hard_regno_nregs instead of HARD_REGNO_NREGS.
15832 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
15833 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
15834 (c6x_expand_epilogue): Likewise.
15835 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
15836 (frv_read_iacc_argument): Likewise.
15837 * config/sh/sh.c: Include regs.h.
15838 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15839 (regs_used): Likewise.
15840 (output_stack_adjust): Likewise.
15841 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
15842 * expmed.c: Include regs.h.
15843 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
15844 * ree.c: Include regs.h.
15845 (combine_reaching_defs): Use hard_regno_nregs instead of
15847 (add_removable_extension): Likewise.
15849 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15851 * regs.h (hard_regno_nregs): Turn into a function.
15852 (end_hard_regno): Update accordingly.
15853 * caller-save.c (setup_save_areas): Likewise.
15854 (save_call_clobbered_regs): Likewise.
15855 (replace_reg_with_saved_mem): Likewise.
15856 (insert_restore): Likewise.
15857 (insert_save): Likewise.
15858 * combine.c (can_change_dest_mode): Likewise.
15859 (move_deaths): Likewise.
15860 (distribute_notes): Likewise.
15861 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
15862 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
15863 (rs6000_split_multireg_move): Likewise.
15864 (rs6000_register_move_cost): Likewise.
15865 (rs6000_memory_move_cost): Likewise.
15866 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
15867 (rs6000_split_multireg_move): Likewise.
15868 (rs6000_register_move_cost): Likewise.
15869 (rs6000_memory_move_cost): Likewise.
15870 * cselib.c (cselib_reset_table): Likewise.
15871 (cselib_lookup_1): Likewise.
15872 * emit-rtl.c (set_mode_and_regno): Likewise.
15873 * function.c (aggregate_value_p): Likewise.
15874 * ira-color.c (setup_profitable_hard_regs): Likewise.
15875 (check_hard_reg_p): Likewise.
15876 (calculate_saved_nregs): Likewise.
15877 (assign_hard_reg): Likewise.
15878 (improve_allocation): Likewise.
15879 (calculate_spill_cost): Likewise.
15880 * ira-emit.c (modify_move_list): Likewise.
15881 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
15882 (ira_hard_reg_in_set_p): Likewise.
15883 * ira.c (setup_reg_mode_hard_regset): Likewise.
15884 (clarify_prohibited_class_mode_regs): Likewise.
15885 (check_allocation): Likewise.
15886 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15887 (lra_setup_reg_renumber): Likewise.
15888 (setup_try_hard_regno_pseudos): Likewise.
15889 (spill_for): Likewise.
15890 (assign_hard_regno): Likewise.
15891 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
15892 * lra-constraints.c (in_class_p): Likewise.
15893 (lra_constraint_offset): Likewise.
15894 (simplify_operand_subreg): Likewise.
15895 (lra_constraints): Likewise.
15896 (split_reg): Likewise.
15897 (split_if_necessary): Likewise.
15898 (invariant_p): Likewise.
15899 (inherit_in_ebb): Likewise.
15900 * lra-lives.c (process_bb_lives): Likewise.
15901 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
15902 (get_hard_regs): Likewise.
15903 (do_remat): Likewise.
15904 * lra-spills.c (assign_spill_hard_regs): Likewise.
15905 * mode-switching.c (create_pre_exit): Likewise.
15906 * postreload.c (reload_combine_recognize_pattern): Likewise.
15907 * recog.c (peep2_find_free_register): Likewise.
15908 * regcprop.c (kill_value_regno): Likewise.
15909 (set_value_regno): Likewise.
15910 (copy_value): Likewise.
15911 (maybe_mode_change): Likewise.
15912 (find_oldest_value_reg): Likewise.
15913 (copyprop_hardreg_forward_1): Likewise.
15914 * regrename.c (check_new_reg_p): Likewise.
15915 (regrename_do_replace): Likewise.
15916 * reload.c (push_reload): Likewise.
15917 (combine_reloads): Likewise.
15918 (find_dummy_reload): Likewise.
15919 (operands_match_p): Likewise.
15920 (find_reloads): Likewise.
15921 (find_equiv_reg): Likewise.
15922 (reload_adjust_reg_for_mode): Likewise.
15923 * reload1.c (count_pseudo): Likewise.
15924 (count_spilled_pseudo): Likewise.
15925 (find_reg): Likewise.
15926 (clear_reload_reg_in_use): Likewise.
15927 (free_for_value_p): Likewise.
15928 (allocate_reload_reg): Likewise.
15929 (choose_reload_regs): Likewise.
15930 (reload_adjust_reg_for_temp): Likewise.
15931 (emit_reload_insns): Likewise.
15932 (delete_output_reload): Likewise.
15933 * rtlanal.c (subreg_get_info): Likewise.
15934 * sched-deps.c (sched_analyze_reg): Likewise.
15935 * sel-sched.c (init_regs_for_mode): Likewise.
15936 (mark_unavailable_hard_regs): Likewise.
15937 (choose_best_reg_1): Likewise.
15938 (verify_target_availability): Likewise.
15939 * valtrack.c (dead_debug_insert_temp): Likewise.
15940 * var-tracking.c (track_loc_p): Likewise.
15941 (emit_note_insn_var_location): Likewise.
15942 * varasm.c (make_decl_rtl): Likewise.
15943 * reginfo.c (choose_hard_reg_mode): Likewise.
15944 (init_reg_modes_target): Refer directly to
15945 this_target_regs->x_hard_regno_nregs.
15947 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15949 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
15950 instead of hard_regno_nregs.
15952 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15954 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
15955 end_hard_regno instead of hard_regno_nregs.
15956 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
15957 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
15958 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
15959 * ira-color.c (improve_allocation): Likewise.
15960 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15961 * lra-lives.c (mark_regno_live): Likewise.
15962 (mark_regno_dead): Likewise.
15963 * lra-remat.c (operand_to_remat): Likewise.
15964 * lra.c (collect_non_operand_hard_regs): Likewise.
15965 * postreload.c (reload_combine_note_store): Likewise.
15966 (move2add_valid_value_p): Likewise.
15967 * reload.c (regno_clobbered_p): Likewise.
15969 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15971 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
15973 * config/v850/v850.c (v850_reorg): Likewise.
15974 * reload.c (refers_to_regno_for_reload_p): Likewise.
15975 (find_equiv_reg): Likewise.
15976 * reload1.c (reload_reg_reaches_end_p): Likewise.
15978 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
15980 * caller-save.c (add_used_regs): Use REG_NREGS instead of
15982 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
15983 * config/arm/arm.c (output_move_neon): Likewise.
15984 (arm_attr_length_move_neon): Likewise.
15985 (neon_split_vcombine): Likewise.
15986 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
15987 (c6x_mark_reg_written): Likewise.
15988 (c6x_dwarf_register_span): Likewise.
15989 * config/i386/i386.c (ix86_save_reg): Likewise.
15990 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
15991 (rws_access_reg): Likewise.
15992 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
15993 * mode-switching.c (create_pre_exit): Likewise.
15994 * ree.c (combine_reaching_defs): Likewise.
15995 (add_removable_extension): Likewise.
15996 * regcprop.c (find_oldest_value_reg): Likewise.
15997 (copyprop_hardreg_forward_1): Likewise.
15998 * reload.c (reload_inner_reg_of_subreg): Likewise.
15999 (push_reload): Likewise.
16000 (combine_reloads): Likewise.
16001 (find_dummy_reload): Likewise.
16002 (reload_adjust_reg_for_mode): Likewise.
16003 * reload1.c (find_reload_regs): Likewise.
16004 (forget_old_reloads_1): Likewise.
16005 (reload_reg_free_for_value_p): Likewise.
16006 (reload_adjust_reg_for_temp): Likewise.
16007 (emit_reload_insns): Likewise.
16008 (delete_output_reload): Likewise.
16009 * sel-sched.c (choose_best_reg_1): Likewise.
16010 (choose_best_pseudo_reg): Likewise.
16012 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16013 Alan Hayward <alan.hayward@arm.com>
16014 David Sherwood <david.sherwood@arm.com>
16016 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16017 * target.def (slow_unaligned_access): New hook.
16018 * targhooks.h (default_slow_unaligned_access): Declare.
16019 * targhooks.c (default_slow_unaligned_access): New function.
16020 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16021 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16022 * doc/tm.texi: Regenerate.
16023 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16024 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16025 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16027 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16028 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16030 (rs6000_slow_unaligned_access): New function.
16031 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16032 (expand_block_compare): Likewise.
16033 (expand_strn_compare): Likewise.
16034 (rs6000_rtx_costs): Likewise.
16035 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16036 (riscv_slow_unaligned_access): Likewise.
16037 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16038 (riscv_slow_unaligned_access_p): ...this and make static.
16039 (riscv_option_override): Update accordingly.
16040 (riscv_slow_unaligned_access): New function.
16041 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16042 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16043 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16044 (rs6000_slow_unaligned_access): New function.
16045 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16046 (rs6000_rtx_costs): Likewise.
16047 * config/rs6000/rs6000-string.c (expand_block_compare)
16048 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16049 of SLOW_UNALIGNED_ACCESS.
16050 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16051 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16052 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16053 of SLOW_UNALIGNED_ACCESS.
16054 * expmed.c (simple_mem_bitfield_p): Likewise.
16055 * expr.c (alignment_for_piecewise_move): Likewise.
16056 (emit_group_load_1): Likewise.
16057 (emit_group_store): Likewise.
16058 (copy_blkmode_from_reg): Likewise.
16059 (emit_push_insn): Likewise.
16060 (expand_assignment): Likewise.
16061 (store_field): Likewise.
16062 (expand_expr_real_1): Likewise.
16063 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16064 * lra-constraints.c (simplify_operand_subreg): Likewise.
16065 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16066 * gimple-ssa-store-merging.c: Likewise in block comment at start
16068 * tree-ssa-strlen.c: Include target.h.
16069 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16070 of SLOW_UNALIGNED_ACCESS.
16071 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16073 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16075 PR rtl-optimization/82185
16076 * expmed.c (emit_store_flag_int): Only test tem if it has been
16079 2017-09-12 Richard Biener <rguenther@suse.de>
16081 PR middle-end/82149
16082 * match.pd ((FTYPE) N CMP CST): Fix typo.
16084 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16086 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16088 (mips_near_type_p): Add 'short_call' attribute as a synonym
16090 * doc/extend.texi (short_call): Document new function attribute.
16092 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16095 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16096 assertion check that in the condition.
16097 (get_atomic_generic_size): Likewise. Before testing if parameter
16098 has pointer type, if it has array type, call for C++
16099 default_conversion to perform array-to-pointer conversion.
16101 2017-09-12 Richard Biener <rguenther@suse.de>
16103 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16104 for operations we cannot scalarize.
16106 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16108 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16109 vectors heap vectors. Clean up comments.
16110 Make visited_bbs a reference.
16111 (profitable_jump_thread_path): Make GC
16112 vectors heap vectors. Clean up comments.
16114 (convert_and_register_jump_thread_path): Make GC vectors heap
16116 (check_subpath_and_update_thread_path): Same. Clean up comments.
16117 Make visited_bbs a reference.
16118 (handle_phi): Abstract common code to to
16119 register_jump_thread_path_if_profitable.
16120 Rename VAR_BB to DEF_BB.
16122 Make GC vectors heap vectors.
16123 Make visited_bbs a reference.
16124 (handle_assignment): Same.
16125 (register_jump_thread_path_if_profitable): New.
16126 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16128 Make GC vectors heap vectors. Clean up comments.
16129 Make visited_bbs a reference.
16130 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16131 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16134 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16137 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16138 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16140 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16142 Revert r251800 and r251799.
16144 2017-09-11 Martin Jambor <mjambor@suse.cz>
16147 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16148 arguments in advance.
16149 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16150 use it to find predecessor edges.
16151 (naive_outof_ssa): Collect vector of predecessors.
16153 2017-09-08 Jason Merrill <jason@redhat.com>
16155 PR c++/70029 - ICE with ref-qualifier and -flto
16156 * langhooks.h (struct lang_hooks_for_types): Add
16157 copy_lang_qualifiers.
16158 * attribs.c (build_type_attribute_qual_variant): Use it.
16159 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16161 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16162 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16164 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16167 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16168 (*mulsi3_sp64): New instruction.
16169 (mulsi3): New expander.
16171 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16173 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16175 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16177 * sancov.c: Include memmodel.h.
16179 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16182 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16185 2017-09-07 Carl Love <cel@us.ibm.com>
16187 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16188 the sldi instruction.
16190 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16192 * sancov.c: Include tm_p.h.
16194 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16197 * output.h (switch_to_other_text_partition): New declaration.
16198 * varasm.c (switch_to_other_text_partition): New function.
16199 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16200 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16201 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16202 to the other text partition before emitting LCL label and switch back
16203 after emitting the word after it.
16205 2017-09-07 Richard Biener <rguenther@suse.de>
16207 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16208 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16209 critical edges here, after loop init.
16210 (pass_data_pre): Remove PROP_no_crit_edges flags.
16211 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16212 for valueization of call args to avoid leaking VN_TOP.
16213 (visit_use): Assert we do not visit default defs.
16214 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16215 Use error_mark_node to more easily detect leaking VN_TOP.
16216 All default-defs are varying, not VN_TOP. Mark them visited.
16217 (run_scc_vn): Make code match comment.
16219 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16221 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16222 OPTION_MASK_FLOAT128_KEYWORD.
16223 (POWERPC_MASKS): Likewise.
16224 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16225 support for the -mfloat128-type option, and make -mfloat128
16226 default on PowerPC Linux systems. Define or undefine
16227 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16228 Define __float128 to be __ieee128 if IEEE 128-bit support is
16229 enabled, or undefine it.
16230 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16231 Delete defining __FLOAT128_TYPE__.
16232 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16233 -mfloat128-type option and make -mfloat128 default on PowerPC
16235 (TARGET_FLOAT128_TYPE): Likewise.
16236 (-mfloat128-type): Likewise.
16237 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16238 Delete the -mfloat128-type option and make -mfloat128 default on
16239 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16240 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16241 128-bit floating point is enabled. Change tests from using
16242 -mfloat128-type to -mfloat128.
16243 (rs6000_mangle_type): Use the correct mangling for the __float128
16244 type even if normal long double is restricted to 64-bits.
16245 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16247 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16248 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16249 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16251 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16252 documentation for -mfloat128.
16254 2017-09-06 Olivier Hainque <hainque@adacore.com>
16256 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16258 2017-09-06 Wish Wu <wishwu007@gmail.com>
16259 Jakub Jelinek <jakub@redhat.com>
16261 * asan.c (initialize_sanitizer_builtins): Add
16262 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16263 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16264 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16265 BT_FN_VOID_UINT64_PTR variables.
16266 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16267 (BT_FN_VOID_UINT16_UINT16): Likewise.
16268 (BT_FN_VOID_UINT32_UINT32): Likewise.
16269 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16270 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16271 (BT_FN_VOID_UINT64_PTR): Likewise.
16272 * common.opt (flag_sanitize_coverage): New variable.
16273 (fsanitize-coverage=trace-pc): Remove.
16274 (fsanitize-coverage=): Add.
16275 * flag-types.h (enum sanitize_coverage_code): New enum.
16276 * fold-const.c (fold_range_test): Disable non-short-circuit
16277 optimization if flag_sanitize_coverage.
16278 (fold_truth_andor): Likewise.
16279 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16280 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16281 (coverage_sanitizer_opts): New array.
16282 (get_closest_sanitizer_option): Add OPTS argument, handle also
16283 OPT_fsanitize_coverage_.
16284 (parse_sanitizer_options): Adjusted to also handle
16285 OPT_fsanitize_coverage_.
16286 (common_handle_option): Add OPT_fsanitize_coverage_.
16287 * sancov.c (instrument_comparison, instrument_switch): New function.
16288 (sancov_pass): Add trace-cmp support.
16289 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16290 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16291 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16292 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16293 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16294 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16295 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16296 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16297 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16298 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16300 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16302 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16303 error. Only quit immediately if parsing is complete.
16304 (BEGIN): Initialize fatal_err and parse_done.
16305 (begin fpu, end fpu): Check number of arguments.
16306 (begin arch, end arch): Likewise.
16307 (begin cpu, end cpu): Likewise.
16308 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16309 (optalias): Likewise.
16311 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16313 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16314 * config/arm/arm-isa.h: Delete. Move definitions to ...
16315 * arm-cpus.in: ... here. Use new feature and fgroup values.
16316 * config/arm/arm.c (arm_option_override): Use lower case for feature
16318 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16319 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16320 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16322 (print_isa_bits_for): New function.
16323 (gen_isa): New function.
16324 (gen_comm_data): Use print_isa_bits_for.
16325 (define feature): New keyword.
16326 (define fgroup): New keyword.
16327 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16328 (arm-isa.h): Add rule to generate file.
16329 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16330 case for feature bit names.
16332 2017-09-06 Richard Biener <rguenther@suse.de>
16334 * tree-ssa-pre.c (NECESSARY): Remove.
16335 (create_expression_by_pieces): Do not touch pass-local flags.
16336 (insert_into_preds_of_block): Likewise.
16337 (do_pre_regular_insertion): Likewise.
16338 (eliminate_insert): Likewise.
16339 (eliminate_dom_walker::before_dom_children): Likewise.
16340 (fini_eliminate): Do not look at inserted_exprs.
16341 (mark_operand_necessary): Remove.
16342 (remove_dead_inserted_code): Replace with simple work-list
16343 algorithm based on inserted_exprs and SSA uses.
16344 (pass_pre::execute): Re-order fini_eliminate and
16345 remove_dead_inserted_code.
16347 2017-09-06 Olivier Hainque <hainque@adacore.com>
16349 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16350 for VxWorks 7. Adjust surrounding comments.
16352 2017-09-06 Richard Biener <rguenther@suse.de>
16354 * gimple-ssa-strength-reduction.c
16355 (find_candidates_dom_walker::before_dom_children): Also allow
16358 2017-09-06 Richard Biener <rguenther@suse.de>
16360 PR tree-optimization/82108
16361 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16362 for gap in the non-permutation SLP case.
16364 2017-09-06 Martin Jambor <mjambor@suse.cz>
16366 PR tree-optimization/82078
16367 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16368 add_access_to_work_queue...
16369 (build_accesses_from_assign): ...here.
16370 (propagate_all_subaccesses): Make sure racc is the group
16371 representative, if there is one.
16373 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16375 PR middle-end/82095
16376 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16379 2017-09-06 Richard Biener <rguenther@suse.de>
16381 * gimple-ssa-strength-reduction.c
16382 (find_candidates_dom_walker::before_doom_children): Use a
16383 type and not a mode check.
16385 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16388 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16390 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16392 2017-09-05 Jeff Law <law@redhat.com>
16394 PR tree-optimization/64910
16395 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
16396 cases where we have 3 or more operands.
16398 2017-09-05 Jakub Jelinek <jakub@redhat.com>
16400 PR middle-end/81768
16401 * omp-low.c (lower_omp_for): Recompute tree invariant if
16402 gimple_omp_for_initial/final is ADDR_EXPR.
16404 PR middle-end/81768
16405 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
16406 into gimple val before gimplification fo the COND_EXPR.
16408 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
16410 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
16411 REGION_COPY argument.
16412 (thread_through_all_blocks): Remove unused argument to
16413 duplicate_thread_path.
16415 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16416 Alan Hayward <alan.hayward@arm.com>
16417 David Sherwood <david.sherwood@arm.com>
16419 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
16420 Take a scalar_mode rather than a machine_mode.
16421 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16422 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
16423 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16424 (aarch64_gen_adjusted_ldpstp): Likewise.
16425 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
16427 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16428 Alan Hayward <alan.hayward@arm.com>
16429 David Sherwood <david.sherwood@arm.com>
16431 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
16432 Take a scalar_int_mode instead of a machine_mode.
16433 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16434 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16435 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16436 (aarch64_simd_attr_length_rglist): Delete.
16437 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
16438 a scalar_int_mode instead of a machine_mode.
16439 (aarch64_add_offset): Likewise.
16440 (aarch64_internal_mov_immediate): Likewise
16441 (aarch64_add_constant_internal): Likewise.
16442 (aarch64_add_constant): Likewise.
16443 (aarch64_movw_imm): Likewise.
16444 (aarch64_rtx_arith_op_extract_p): Likewise.
16445 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
16446 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
16447 Remove assert that the mode isn't a vector.
16448 (aarch64_output_scalar_simd_mov_immediate): Likewise.
16449 (aarch64_expand_mov_immediate): Update calls after above changes.
16450 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
16451 (aarch64_and_bitmask_imm): Check for scalar integer modes.
16452 (aarch64_move_imm): Likewise.
16453 (aarch64_can_const_movi_rtx_p): Likewise.
16454 (aarch64_strip_extend): Likewise.
16455 (aarch64_extr_rtx_p): Likewise.
16456 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
16457 a CONST_INT when the mode parameter is VOIDmode.
16458 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
16460 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16462 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
16463 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16464 (bitwise_type_for_mode): Update accordingly.
16466 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16468 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
16469 * stor-layout.c (mode_for_size_tree): Likewise.
16470 (mode_for_array): Update accordingly.
16471 (layout_decl): Likewise.
16472 (compute_record_mode): Likewise. Only set the mode once.
16474 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16476 * target.def (get_mask_mode): Change return type to opt_mode.
16478 * doc/tm.texi: Regenerate.
16479 * targhooks.h (default_get_mask_mode): Return an opt_mode.
16480 * targhooks.c (default_get_mask_mode): Likewise.
16481 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16482 * optabs-query.c (can_vec_mask_load_store_p): Update use of
16483 targetm.get_mask_mode.
16484 * tree.c (build_truth_vector_type): Likewise.
16486 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16488 * machmode.h (mode_for_vector): Return an opt_mode.
16489 * stor-layout.c (mode_for_vector): Likewise.
16490 (mode_for_int_vector): Update accordingly.
16491 (layout_type): Likewise.
16492 * config/i386/i386.c (emit_memmov): Likewise.
16493 (ix86_expand_set_or_movmem): Likewise.
16494 (ix86_expand_vector_init): Likewise.
16495 (ix86_get_mask_mode): Likewise.
16496 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
16498 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
16499 * expmed.c (extract_bit_field_1): Likewise.
16500 * expr.c (expand_expr_real_2): Likewise.
16501 * optabs-query.c (can_vec_perm_p): Likewise.
16502 (can_vec_mask_load_store_p): Likewise.
16503 * optabs.c (expand_vec_perm): Likewise.
16504 * targhooks.c (default_get_mask_mode): Likewise.
16505 * tree-vect-stmts.c (vectorizable_store): Likewise.
16506 (vectorizable_load): Likewise.
16507 (get_vectype_for_scalar_type_and_size): Likewise.
16509 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16511 * machmode.h (mode_for_int_vector): New function.
16512 * stor-layout.c (mode_for_int_vector): Likewise.
16513 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
16514 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
16515 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16516 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16517 (s390_expand_vcond): Likewise.
16519 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16521 * machmode.h (opt_machine_mode): New type.
16522 (opt_mode<T>): Allow construction from anything that can be
16524 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
16525 (mode_for_size): Return an opt_machine_mode.
16526 * stor-layout.c (mode_for_size): Likewise.
16527 (mode_for_size_tree): Update call accordingly.
16528 (bitwise_mode_for_mode): Likewise.
16529 (make_fract_type): Likewise.
16530 (make_accum_type): Likewise.
16531 * caller-save.c (replace_reg_with_saved_mem): Update call
16533 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16534 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16535 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16536 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
16537 * expmed.c (extract_bit_field_1): Likewise.
16538 * reload.c (get_secondary_mem): Likewise.
16539 * varasm.c (assemble_integer): Likewise.
16540 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
16543 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16545 * machmode.h (decimal_float_mode_for_size): New function.
16546 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
16547 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
16548 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
16549 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
16550 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
16551 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
16553 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16555 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
16556 (get_builtin_sync_mode): Likewise.
16557 (expand_ifn_atomic_compare_exchange): Likewise.
16558 (expand_builtin_atomic_clear): Likewise.
16559 (expand_builtin_atomic_test_and_set): Likewise.
16560 (fold_builtin_atomic_always_lock_free): Likewise.
16561 * calls.c (compute_argument_addresses): Likewise.
16562 (emit_library_call_value_1): Likewise.
16563 (store_one_arg): Likewise.
16564 * combine.c (combine_instructions): Likewise.
16565 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
16566 * config/arm/arm.c (arm_function_value): Likewise.
16567 (aapcs_allocate_return_reg): Likewise.
16568 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
16569 * config/i386/i386.c (construct_container): Likewise.
16570 (ix86_gimplify_va_arg): Likewise.
16571 (ix86_expand_sse_cmp): Likewise.
16572 (emit_memmov): Likewise.
16573 (emit_memset): Likewise.
16574 (expand_small_movmem_or_setmem): Likewise.
16575 (ix86_expand_pextr): Likewise.
16576 (ix86_expand_pinsr): Likewise.
16577 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
16578 * config/microblaze/microblaze.c (microblaze_block_move_straight):
16580 * config/mips/mips.c (mips_function_value_1) Likewise.
16581 (mips_block_move_straight): Likewise.
16582 (mips_expand_ins_as_unaligned_store): Likewise.
16583 * config/powerpcspe/powerpcspe.c
16584 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16585 (rs6000_darwin64_record_arg_flush): Likewise.
16586 * config/rs6000/rs6000.c
16587 (rs6000_darwin64_record_arg_advance_flush): Likewise.
16588 (rs6000_darwin64_record_arg_flush): Likewise.
16589 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
16590 (sparc_function_value_1): Likewise.
16591 * config/spu/spu.c (adjust_operand): Likewise.
16592 (spu_emit_branch_or_set): Likewise.
16593 (arith_immediate_p): Likewise.
16594 * emit-rtl.c (gen_lowpart_common): Likewise.
16595 * expr.c (expand_expr_real_1): Likewise.
16596 * function.c (assign_parm_setup_block): Likewise.
16597 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
16598 * reload1.c (alter_reg): Likewise.
16599 * stor-layout.c (mode_for_vector): Likewise.
16600 (layout_type): Likewise.
16602 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
16604 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
16605 (spu_convert_move): Likewise.
16606 * lower-subreg.c (resolve_simple_move): Likewise.
16608 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16611 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
16612 define_insn to a define_expand.
16613 (altivec_vsum2sws_direct): New define_insn.
16614 (altivec_vsumsws): Convert from a define_insn to a define_expand.
16616 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16618 * config/arm/arm.c (arm_option_params_internal): Improve setting of
16621 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
16626 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
16627 realign stack if stack alignment needed is less than incoming
16630 2017-09-05 Marek Polacek <polacek@redhat.com>
16633 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
16636 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
16638 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
16640 2017-09-05 Richard Biener <rguenther@suse.de>
16642 PR tree-optimization/82084
16643 * fold-const.c (can_native_encode_string_p): Handle wide characters.
16645 2017-09-05 Richard Biener <rguenther@suse.de>
16647 PR tree-optimization/82102
16648 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
16650 2017-09-05 Martin Liska <mliska@suse.cz>
16652 PR tree-optimization/82032
16653 * tree-cfg.c (generate_range_test): New function.
16654 * tree-cfg.h (generate_range_test): Declared here.
16655 * tree-cfgcleanup.c (convert_single_case_switch): New function.
16656 (cleanup_control_expr_graph): Use it.
16657 * tree-switch-conversion.c (try_switch_expansion): Remove
16659 (emit_case_nodes): Use generate_range_test.
16661 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16664 * config/i386/i386.md (*<btsc><mode>_mask): Add
16665 TARGET_USE_BT to insn constraint.
16666 (*btr<mode>_mask): Ditto.
16668 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
16670 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
16671 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
16673 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
16676 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
16677 TARGET_NEON and TARGET_IWMMXT.
16678 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
16679 TARGET_NEON and TARGET_IWMMXT.
16680 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
16682 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
16684 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
16685 (ix86_rewrite_tls_address): Ditto.
16686 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
16687 (ix86_rewrite_tls_address_1): Ditto.
16688 (ix86_rewrite_tls_address): Ditto.
16689 * config/i386/predicates.md (tls_address_pattern): New predicate.
16690 * config/i386/i386.md (TLS address splitter): New splitter.
16692 2017-09-04 Richard Biener <rguenther@suse.de>
16694 PR tree-optimization/82084
16695 * fold-const.h (can_native_encode_string_p): Declare.
16696 * fold-const.c (can_native_encode_string_p): Factor out from ...
16697 (native_encode_string): ... here.
16698 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
16699 vectorizing stores from constants we later cannot handle.
16701 2017-09-04 Marek Polacek <polacek@redhat.com>
16704 * doc/invoke.texi: Update -Wtautological-compare documentation.
16706 2017-09-04 Jeff Law <law@redhat.com>
16708 PR tree-optimization/64910
16709 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
16710 swap the first and last operand if the last is a constant.
16712 2017-09-04 Marek Polacek <polacek@redhat.com>
16715 * convert.c (do_narrow): When sanitizing signed integer overflows,
16716 bail out for signed types.
16717 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
16719 2017-09-04 Richard Biener <rguenther@suse.de>
16721 PR tree-optimization/82060
16722 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
16723 Move devirtualization after stmt folding and before EH/AB/noreturn
16724 cleanup to get the stmt refs canonicalized. Use a bool instead
16725 of gimple_modified_p since that doesn't work for NOPs. Schedule
16726 NOPs generated by folding for removal.
16728 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16729 Alan Hayward <alan.hayward@arm.com>
16730 David Sherwood <david.sherwood@arm.com>
16732 * coretypes.h (pad_direction): New enum.
16733 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
16734 (FUNCTION_ARG_PADDING): Likewise.
16735 * target.def (function_arg_padding): New hook.
16736 * targhooks.h (default_function_arg_padding): Declare.
16737 * targhooks.c (default_function_arg_padding): New function.
16738 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
16739 (TARGET_FUNCTION_ARG_PADDING): ...this.
16740 * doc/tm.texi: Regenerate.
16741 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
16742 instead of direction.
16743 (compute_argument_addresses): Likewise.
16744 (load_register_parameters): Likewise.
16745 (emit_library_call_value_1): Likewise.
16746 (store_one_arg): Use targetm.calls.function_arg_padding instead
16747 of FUNCTION_ARG_PADDING.
16748 (must_pass_in_stack_var_size_or_pad): Likewise.
16749 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
16750 (emit_group_store): Likewise.
16751 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
16752 instead of FUNCTION_ARG_PADDING.
16753 (emit_push_insn): Likewise, and propagate enum change throughout
16755 * function.h (direction): Delete.
16756 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
16758 * function.c (assign_parm_find_stack_rtl): Likewise.
16759 (assign_parm_setup_block_p): Likewise.
16760 (assign_parm_setup_block): Likewise.
16761 (gimplify_parameters): Likewise.
16762 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
16763 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
16765 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
16766 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16767 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
16768 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
16769 (aarch64_function_arg_padding): ...this new function.
16770 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
16771 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16772 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
16773 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16774 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
16775 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16776 (arm_pad_arg_upward): Replace with...
16777 (arm_function_arg_padding): ...this new function.
16778 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
16780 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
16781 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
16782 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16783 (ia64_hpux_function_arg_padding): Replace with...
16784 (ia64_function_arg_padding): ...this new function. Use pad_direction
16785 instead of direction. Check for TARGET_HPUX.
16786 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
16787 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16788 (iq2000_function_arg_padding): New function.
16789 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
16790 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
16791 (mips_function_arg_padding): ...this new function.
16792 (mips_pad_reg_upward): Update accordingly.
16793 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16794 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
16795 targetm.calls.function_arg_padding.
16796 (FUNCTION_ARG_PADDING): Delete.
16797 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
16798 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
16799 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16800 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
16801 (nios2_block_reg_padding): Return pad_direction instead of direction.
16802 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
16803 instead of direction.
16804 (nios2_function_arg_padding): Likewise. Make static.
16805 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16806 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
16807 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
16808 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
16809 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16810 (pa_function_arg_padding): Make static. Return pad_direction instead
16812 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
16813 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16814 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
16815 instead of direction. Use targetm.calls.function_arg_padding.
16816 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
16817 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16818 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
16819 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
16820 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
16822 (function_arg_padding): Rename to...
16823 (rs6000_function_arg_padding): ...this. Make static. Return
16824 pad_direction instead of direction.
16825 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16826 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
16827 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
16828 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
16829 instead of direction. Use targetm.calls.function_arg_padding.
16830 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
16831 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
16832 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
16833 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
16834 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16835 (function_arg_padding): Rename to...
16836 (rs6000_function_arg_padding): ...this. Make static. Return
16837 pad_direction instead of direction.
16838 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
16839 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
16840 * config/s390/s390.c (s390_function_arg_padding): New function.
16841 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16842 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
16843 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
16844 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
16845 (function_arg_padding): Rename to...
16846 (sparc_function_arg_padding): ...this. Make static. Return
16847 pad_direction instead of direction.
16848 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
16849 * config/spu/spu.c (spu_function_arg_padding): New function.
16850 (TARGET_FUNCTION_ARG_PADDING): Redefine.
16851 * system.h (FUNCTION_ARG_PADDING): Poison.
16853 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16854 Alan Hayward <alan.hayward@arm.com>
16855 David Sherwood <david.sherwood@arm.com>
16857 * target.def (modes_tieable_p): New hook.
16858 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
16859 (TARGET_MODES_TIEABLE_P): ...this.
16860 * doc/tm.texi.in: Regenerate.
16861 * hooks.h (hook_bool_mode_mode_true): Declare.
16862 * hooks.c (hook_bool_mode_mode_true): New function.
16863 * combine.c (subst): Use targetm.modes_tieable_p instead of
16865 * dse.c (find_shift_sequence): Likewise.
16866 * expmed.c (extract_low_bits): Likewise.
16867 * lower-subreg.c: Include target.h.
16868 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
16870 * rtlanal.c (rtx_cost): Likewise.
16871 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
16872 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
16873 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
16874 (TARGET_MODES_TIEABLE_P): Redefine.
16875 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
16876 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
16877 (TARGET_MODES_TIEABLE_P): Redefine.
16878 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
16879 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
16880 (arc_modes_tieable_p): New function.
16881 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
16882 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
16883 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
16884 (arm_modes_tieable_p): Make static.
16885 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
16886 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
16887 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
16888 (TARGET_MODES_TIEABLE_P): Redefine.
16889 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
16890 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
16891 (TARGET_MODES_TIEABLE_P): Redefine.
16892 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
16893 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
16894 (cr16_modes_tieable_p): New function.
16895 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
16896 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
16897 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
16898 (TRULY_NOOP_TRUNCATION): Update comment.
16899 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
16900 (TRULY_NOOP_TRUNCATION): Update comment.
16901 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
16902 (frv_modes_tieable_p): New function.
16903 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
16904 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
16905 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
16906 (TARGET_MODES_TIEABLE_P): Redefine.
16907 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
16908 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
16909 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
16910 (TARGET_MODES_TIEABLE_P): Redefine.
16911 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
16912 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
16913 (ia64_modes_tieable_p): New function.
16914 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
16915 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
16916 (iq2000_modes_tieable_p): New function.
16917 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
16918 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
16919 (lm32_modes_tieable_p): New function.
16920 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
16921 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
16922 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
16923 (TARGET_MODES_TIEABLE_P): Redefine.
16924 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
16925 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
16926 (m32r_modes_tieable_p): New function.
16927 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
16928 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
16929 (m68k_modes_tieable_p): New function.
16930 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
16931 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
16932 (mcore_modes_tieable_p): New function.
16933 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
16934 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
16936 (TARGET_MODES_TIEABLE_P): Redefine.
16937 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
16938 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
16939 * config/mips/mips.c (mips_modes_tieable_p): Make static.
16940 (TARGET_MODES_TIEABLE_P): Redefine.
16941 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
16942 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
16943 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
16944 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
16945 (mn10300_modes_tieable_p): ...this and make static.
16946 (TARGET_MODES_TIEABLE_P): Redefine.
16947 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
16948 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
16949 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
16950 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
16951 (msp430_modes_tieable_p): Make static.
16952 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
16953 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
16954 (TARGET_MODES_TIEABLE_P): Redefine.
16955 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
16956 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
16957 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
16958 (TARGET_MODES_TIEABLE_P): Redefine.
16959 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
16960 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
16961 * config/pa/pa.c (pa_modes_tieable_p): Make static.
16962 (TARGET_MODES_TIEABLE_P): Redefine.
16963 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
16964 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
16965 (pdp11_modes_tieable_p): New function.
16966 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
16967 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
16968 (rs6000_modes_tieable_p): New function.
16969 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16970 * config/powerpcspe/powerpcspe.md: Update comment.
16971 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
16972 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
16973 (TARGET_MODES_TIEABLE_P): Redefine.
16974 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
16975 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
16976 (rl78_modes_tieable_p): New function.
16977 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
16978 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
16979 (rs6000_modes_tieable_p): New function.
16980 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
16981 * config/rs6000/rs6000.md: Update comment.
16982 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
16983 * config/rx/rx.c (rx_modes_tieable_p): New function.
16984 (TARGET_MODES_TIEABLE_P): Redefine.
16985 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
16986 * config/s390/s390.c (s390_modes_tieable_p): New function.
16987 (TARGET_MODES_TIEABLE_P): Redefine.
16988 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
16989 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
16990 (sh_modes_tieable_p): New function.
16991 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
16992 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
16993 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
16994 (sparc_modes_tieable_p): Make static.
16995 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
16996 * config/spu/spu.c (spu_modes_tieable_p): New function.
16997 (TARGET_MODES_TIEABLE_P): Redefine.
16998 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
16999 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17000 (TARGET_MODES_TIEABLE_P): Redefine.
17001 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17002 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17003 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17004 * config/v850/v850.c (v850_modes_tieable_p): New function.
17005 (TARGET_MODES_TIEABLE_P): Redefine.
17006 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17007 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17008 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17009 (visium_modes_tieable_p): New function.
17010 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17011 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17012 (xtensa_modes_tieable_p): New function.
17013 * system.h (MODES_TIEABLE_P): Poison.
17015 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17016 Alan Hayward <alan.hayward@arm.com>
17017 David Sherwood <david.sherwood@arm.com>
17019 * target.def (hard_regno_mode_ok): New hook.
17020 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17021 (TARGET_HARD_REGNO_MODE_OK): ...this.
17022 * doc/tm.texi.in: Regenerate.
17023 * hooks.h (hook_bool_uint_mode_true): Declare.
17024 * hooks.c (hook_bool_uint_mode_true): New function.
17025 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17026 HARD_REGNO_MODE_OK.
17027 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17028 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17029 instead of HARD_REGNO_MODE_OK.
17030 * caller-save.c: Include target.h.
17031 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17032 HARD_REGNO_MODE_OK.
17033 * combine.c (can_combine_p): Likewise.
17034 (combinable_i3pat): Likewise.
17035 (can_change_dest_mode): Likewise.
17036 * expr.c (init_expr_target): Likewise.
17037 (convert_move): Likewise.
17038 (convert_modes): Likewise.
17039 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17040 (setup_prohibited_mode_move_regs): Likewise.
17041 * ira.h (target_ira): Likewise.
17042 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17043 * lra-constraints.c (process_alt_operands): Likewise.
17044 (split_reg): Likewise.
17045 * recog.c (peep2_find_free_register): Likewise.
17046 * ree.c (combine_reaching_defs): Likewise.
17047 * regcprop.c (maybe_mode_change): Likewise.
17048 * reginfo.c (init_reg_sets_1): Likewise.
17049 (choose_hard_reg_mode): Likewise.
17050 (simplifiable_subregs): Likewise.
17051 * regrename.c (check_new_reg_p): Likewise.
17052 * reload.c (find_valid_class): Likewise.
17053 (find_valid_class_1): Likewise.
17054 (reload_inner_reg_of_subreg): Likewise.
17055 (push_reload): Likewise.
17056 (combine_reloads): Likewise.
17057 (find_dummy_reload): Likewise.
17058 (find_reloads): Likewise.
17059 * reload1.c (find_reg): Likewise.
17060 (set_reload_reg): Likewise.
17061 (allocate_reload_reg): Likewise.
17062 (choose_reload_regs): Likewise.
17063 (reload_adjust_reg_for_temp): Likewise.
17064 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17065 (simplify_subreg_regno): Likewise.
17066 * sel-sched.c (init_regs_for_mode): Likewise.
17067 * varasm.c (make_decl_rtl): Likewise.
17068 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17069 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17070 HARD_REGNO_MODE_OK.
17071 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17072 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17073 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17074 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17075 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17076 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17077 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17078 (arc_mode_class): Delete.
17079 (HARD_REGNO_MODE_OK): Delete.
17080 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17081 (arc_hard_regno_mode_ok): Rename old array to...
17082 (arc_hard_regno_mode_ok_modes): ...this.
17083 (arc_conditional_register_usage): Update accordingly.
17084 (arc_mode_class): Make static.
17085 (arc_hard_regno_mode_ok): New function.
17086 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17087 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17088 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17089 (arm_hard_regno_mode_ok): Make static.
17090 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17091 HARD_REGNO_MODE_OK.
17092 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17093 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17094 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17096 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17097 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17098 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17099 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17100 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17101 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17102 * config/bfin/predicates.md (valid_reg_operand): Use
17103 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17104 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17105 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17106 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17107 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17108 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17109 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17110 (cr16_hard_regno_mode_ok): Make static and return a bool.
17111 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17112 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17113 (cris_hard_regno_mode_ok): New function.
17114 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17115 (epiphany_mode_class): Delete.
17116 (HARD_REGNO_MODE_OK): Delete.
17117 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17118 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17119 (hard_regno_mode_ok): Rename to...
17120 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17121 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17122 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17123 HARD_REGNO_MODE_OK.
17124 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17125 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17126 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17127 (frv_hard_regno_mode_ok): Make static and return a bool.
17128 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17129 HARD_REGNO_MODE_OK.
17130 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17131 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17132 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17133 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17135 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17136 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17137 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17138 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17140 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17141 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17142 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17143 (ia64_hard_regno_mode_ok): New function.
17144 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17145 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17146 (iq2000_hard_regno_mode_ok): New function.
17147 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17148 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17149 (lm32_hard_regno_mode_ok): New function.
17150 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17151 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17152 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17153 instead of HARD_REGNO_MODE_OK.
17154 (m32c_hard_regno_ok): Rename to...
17155 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17156 (m32c_cannot_change_mode_class): Update accordingly.
17157 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17158 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17159 (m32r_mode_class): Delete.
17160 (HARD_REGNO_MODE_OK): Delete.
17161 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17162 (m32r_hard_regno_mode_ok): Rename to...
17163 (m32r_hard_regno_modes): ...this.
17164 (m32r_mode_class): Make static.
17165 (m32r_hard_regno_mode_ok): New function.
17166 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17167 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17168 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17169 (m68k_hard_regno_mode_ok): Make static.
17170 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17171 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17172 (mcore_hard_regno_mode_ok): New function.
17173 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17174 (HARD_REGNO_MODE_OK): Delete.
17175 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17177 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17178 (microblaze_hard_regno_mode_ok): New function.
17179 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17180 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17181 (mips_hard_regno_mode_ok): Delete.
17182 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17183 (mips_hard_regno_mode_ok_p): ...this and make static.
17184 (mips_hard_regno_mode_ok_p): Rename to...
17185 (mips_hard_regno_mode_ok_uncached): ...this.
17186 (mips_hard_regno_mode_ok): New function.
17187 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17188 of HARD_REGNO_MODE_OK.
17189 (mips_option_override): Update after above name changes.
17190 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17191 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17192 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17193 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17194 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17195 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17196 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17197 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17198 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17199 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17200 (msp430_hard_regno_mode_ok): Make static and return a bool.
17201 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17202 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17203 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17205 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17206 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17207 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17208 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17209 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17210 (PA_HARD_REGNO_MODE_OK): ...this
17211 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17212 (PA_HARD_REGNO_MODE_OK): ...this.
17213 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17214 (pa_hard_regno_mode_ok): New function.
17215 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17216 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17217 (pdp11_hard_regno_mode_ok): New function.
17218 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17219 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17221 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17223 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17224 (rs6000_hard_regno_mode_ok): Rename to...
17225 (rs6000_hard_regno_mode_ok_uncached): ...this.
17226 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17227 (rs6000_hard_regno_mode_ok): New function.
17228 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17229 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17230 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17231 (riscv_hard_regno_mode_ok): ...this and make static.
17232 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17233 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17234 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17235 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17236 (rl78_hard_regno_mode_ok): Make static and return bool.
17237 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17238 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17240 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17241 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17242 (rs6000_hard_regno_mode_ok): Rename to...
17243 (rs6000_hard_regno_mode_ok_uncached): ...this.
17244 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17245 (rs6000_hard_regno_mode_ok): New function.
17246 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17247 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17248 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17249 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17250 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17251 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17252 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17253 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17254 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17255 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17256 (sh_hard_regno_mode_ok): Make static.
17257 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17258 instead of HARD_REGNO_MODE_OK.
17259 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17260 (sparc_mode_class): Delete.
17261 (HARD_REGNO_MODE_OK): Delete.
17262 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17263 (hard_regno_mode_classes): Make static.
17264 (sparc_mode_class): Likewise.
17265 (sparc_hard_regno_mode_ok): New function.
17266 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17267 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17268 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17270 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17271 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17272 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17273 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17274 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17275 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17276 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17277 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17278 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17279 (visium_hard_regno_mode_ok): New function.
17280 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17281 instead of HARD_REGNO_MODE_OK.
17282 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17283 (HARD_REGNO_MODE_OK): Delete.
17284 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17285 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17286 (xtensa_option_override): Update accordingly.
17287 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17288 (xtensa_hard_regno_mode_ok): New function.
17289 * system.h (HARD_REGNO_MODE_OK): Poison.
17291 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17292 Alan Hayward <alan.hayward@arm.com>
17293 David Sherwood <david.sherwood@arm.com>
17295 * target.def (hard_regno_call_part_clobbered): New hook.
17296 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17297 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17298 * doc/tm.texi: Regenerate.
17299 * hooks.h (hook_bool_uint_mode_false): Declare.
17300 * hooks.c (hook_bool_uint_mode_false): New function.
17301 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17302 * cselib.c (cselib_process_insn): Use
17303 targetm.hard_regno_call_part_clobbered instead of
17304 HARD_REGNO_CALL_PART_CLOBBERED.
17305 * ira-conflicts.c (ira_build_conflicts): Likewise.
17306 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17307 * lra-constraints.c (need_for_call_save_p): Likewise.
17308 * lra-lives.c: Include target.h.
17309 (check_pseudos_live_through_calls): Use
17310 targetm.hard_regno_call_part_clobbered instead of
17311 HARD_REGNO_CALL_PART_CLOBBERED.
17312 * regcprop.c: Include target.h.
17313 (copyprop_hardreg_forward_1): Use
17314 targetm.hard_regno_call_part_clobbered instead of
17315 HARD_REGNO_CALL_PART_CLOBBERED.
17316 * reginfo.c (choose_hard_reg_mode): Likewise.
17317 * regrename.c (check_new_reg_p): Likewise.
17318 * reload.c (find_equiv_reg): Likewise.
17319 * reload1.c (emit_reload_insns): Likewise.
17320 * sched-deps.c (deps_analyze_insn): Likewise.
17321 * sel-sched.c (init_regs_for_mode): Likewise.
17322 (mark_unavailable_hard_regs): Likewise.
17323 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17324 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17325 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17327 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17328 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17329 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17331 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17333 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17334 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17335 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17337 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17338 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17339 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17341 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17342 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17344 * config/powerpcspe/powerpcspe.c
17345 (rs6000_hard_regno_call_part_clobbered): New function.
17346 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17347 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17348 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17350 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17351 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17352 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17354 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17355 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17356 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17358 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17359 Alan Hayward <alan.hayward@arm.com>
17360 David Sherwood <david.sherwood@arm.com>
17362 * rtl.h (subreg_memory_offset): Declare.
17363 * emit-rtl.c (subreg_memory_offset): New function.
17364 * expmed.c (store_bit_field_1): Use it.
17365 * expr.c (undefined_operand_subword_p): Likewise.
17366 * simplify-rtx.c (simplify_subreg): Likewise.
17368 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17370 PR rtl-optimization/57448
17373 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17374 using atomic_load pattern.
17375 (expand_atomic_store): Likewise.
17377 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17380 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17381 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17382 handling. Use replace_call_with_value with NULL instead of
17383 gsi_replace, unlink_stmt_vdef and release_defs.
17385 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17388 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17390 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17393 * rtl.h (emit_library_call_value_1): Declare.
17394 (emit_library_call): Replace declaration with a series of overloads.
17395 Remove the parameter count argument.
17396 (emit_library_call_value): Likewise.
17397 * calls.c (emit_library_call_value_1): Make global. Replace varargs
17398 with an "rtx_mode_t *".
17399 (emit_library_call_value): Delete.
17400 (emit_library_call): Likewise.
17401 * asan.c (asan_emit_stack_protection): Update calls accordingly.
17402 (asan_emit_allocas_unpoison): Likewise.
17403 * builtins.c (expand_builtin_powi): Likewise.
17404 (expand_asan_emit_allocas_unpoison): Likewise.
17405 * cfgexpand.c (expand_main_function): Likewise.
17406 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
17407 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
17408 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
17409 * config/arm/arm.c (arm_trampoline_init): Likewise.
17410 (arm_call_tls_get_addr): Likewise.
17411 (arm_expand_divmod_libfunc): Likewise.
17412 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
17413 (smulsi3_highpart): Likewise.
17414 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
17415 (c6x_expand_compare): Likewise.
17416 (c6x_expand_movmem): Likewise.
17417 * config/frv/frv.c (frv_trampoline_init): Likewise.
17418 * config/i386/i386.c (ix86_trampoline_init): Likewise.
17419 (ix86_expand_divmod_libfunc): Likewise.
17420 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
17421 (ia64_expand_compare): Likewise.
17422 (ia64_profile_hook): Likewise.
17423 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
17424 (nonlocal_goto): Likewise.
17425 (restore_stack_nonlocal): Likewise.
17426 * config/m32r/m32r.c (block_move_call): Likewise.
17427 (m32r_trampoline_init): Likewise.
17428 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
17429 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
17430 (m68k_call_m68k_read_tp): Likewise.
17431 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
17432 (microblaze_expand_divide): Likewise.
17433 * config/mips/mips.h (mips_args): Likewise.
17434 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
17435 (MIPS_ICACHE_SYNC): Likewise.
17436 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
17437 (nios2_trampoline_init): Likewise.
17438 * config/pa/pa.c (hppa_tls_call): Likewise.
17439 (pa_trampoline_init): Likewise.
17440 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
17441 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
17442 (expand_strn_compare): Likewise.
17443 (rs6000_generate_compare): Likewise.
17444 (rs6000_expand_float128_convert): Likewise.
17445 (output_profile_hook): Likewise.
17446 (rs6000_trampoline_init): Likewise.
17447 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
17448 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
17449 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
17450 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
17451 (rs6000_generate_compare): Likewise.
17452 (rs6000_expand_float128_convert): Likewise.
17453 (output_profile_hook): Likewise.
17454 (rs6000_trampoline_init): Likewise.
17455 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
17456 * config/sh/sh.c (sh_trampoline_init): Likewise.
17457 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
17458 (sparc_emit_float_lib_cmp): Likewise.
17459 (sparc32_initialize_trampoline): Likewise.
17460 (sparc64_initialize_trampoline): Likewise.
17461 (sparc_profile_hook): Likewise.
17462 * config/spu/spu.c (ea_load_store): Likewise.
17463 * config/spu/spu.md (floatunssidf2): Likewise.
17464 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
17465 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
17466 * config/visium/visium.c (expand_block_move_4): Likewise.
17467 (expand_block_move_2): Likewise.
17468 (expand_block_move_1): Likewise.
17469 (expand_block_set_4): Likewise.
17470 (expand_block_set_2): Likewise.
17471 (expand_block_set_1): Likewise.
17472 (visium_trampoline_init): Likewise.
17473 (visium_profile_hook): Likewise.
17474 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
17475 (xtensa_setup_frame_addresses): Likewise.
17476 (xtensa_trampoline_init): Likewise.
17477 * except.c (sjlj_emit_function_enter): Likewise.
17478 (sjlj_emit_function_exit): Likewise.
17479 * explow.c (allocate_dynamic_stack_space): Likewise.
17480 (probe_stack_range): Likewise.
17481 * expr.c (convert_mode_scalar): Likewise.
17482 * optabs.c (expand_binop): Likewise.
17483 (expand_twoval_binop_libfunc): Likewise.
17484 (expand_unop): Likewise.
17485 (prepare_cmp_insn): Likewise.
17486 (prepare_float_lib_cmp): Likewise.
17487 (expand_float): Likewise.
17488 (expand_fix): Likewise.
17489 (expand_fixed_convert): Likewise.
17490 (maybe_emit_sync_lock_test_and_set): Likewise.
17491 (expand_atomic_compare_and_swap): Likewise.
17492 (expand_mem_thread_fence): Likewise.
17493 (expand_atomic_fetch_op): Likewise.
17495 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
17497 * doc/generic.texi (OpenACC): Adjust URL.
17498 * doc/invoke.texi (C Dialect Options): Ditto.
17500 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
17502 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
17503 predicate for operand 1. Add (m,<S>) constraint.
17504 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
17505 Prevent memory operand 1 with register operand 2.
17507 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
17509 PR rtl-optimization/82024
17510 * combine.c (try_combine): If the combination result is a PARALLEL,
17511 and we only need to retain the SET in there that would be placed
17512 at I2, check that we can place that at I3 instead, before doing so.
17514 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17517 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
17519 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
17520 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
17523 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
17524 Jeff Law <law@redhat.com>
17526 * varasm.c (bss_initializer_p): Do not put constants into .bss
17527 (categorize_decl_for_section): Handle bss_initializer_p returning
17528 false when DECL_INITIAL is NULL.
17530 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17533 * config/s390/s390.c (s390_can_inline_p): New function.
17535 2017-09-01 Jeff Law <law@redhat.com>
17537 PR tree-optimization/82052
17538 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
17539 Always initialize the returned slot after a hash table miss
17540 when INSERT is true.
17542 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
17544 * config/s390/s390.md (mem_signal_fence): Remove.
17545 * doc/md.texi (mem_signal_fence): Remove.
17546 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
17548 * target-insns.def (mem_signal_fence): Remove.
17550 2017-09-01 Jakub Jelinek <jakub@redhat.com>
17553 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
17556 * asan.c (create_odr_indicator): Strip name encoding from assembler
17557 name before appending it after __odr_asan_.
17559 2017-09-01 Martin Liska <mliska@suse.cz>
17561 PR tree-optimization/82059
17562 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
17563 frequency only when an edge is redirected.
17565 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17567 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
17568 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
17569 (arc_conditional_register_usage): Remove ARC600 lp_count
17571 (arc_file_start): Emit Tag_ARC_CPU_variation.
17572 (arc_can_use_doloop_p): New conditions to use ZOLs.
17573 (hwloop_fail): New function.
17574 (hwloop_optimize): Likewise.
17575 (hwloop_pattern_reg): Likewise.
17576 (arc_doloop_hooks): New struct, to be used with reorg_loops.
17577 (arc_reorg_loops): New function, calls reorg_loops.
17578 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
17579 (arc600_corereg_hazard): Remove ZOL checking, case handled by
17581 (arc_loop_hazard): Remove function, functionality moved into
17583 (arc_hazard): Remove arc_loop_hazard call.
17584 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
17585 into hwloop_optimize.
17586 (arc_label_align): Remove ZOL handling.
17587 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
17588 * config/arc/arc.md (doloop_begin): Remove pattern.
17589 (doloop_begin_i): Likewise.
17590 (doloop_end_i): Likewise.
17591 (doloop_fallback): Likewise.
17592 (doloop_fallback_m): Likewise.
17593 (doloop_end): Reimplement expand.
17594 (arc_lp): New pattern for LP instruction.
17595 (loop_end): New pattern.
17596 (loop_fail): Likewise.
17597 (decrement_and_branch_until_zero): Likewise.
17598 * config/arc/arc.opt (mlpc-width): New option.
17599 * doc/invoke.texi (mlpc-width): Document option.
17601 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17603 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
17604 (arc_ccfsm_advance): Fix checking for delay slots.
17605 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
17607 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17609 * config/arc/arc.md (movqi_insn): Add stores to save constant long
17611 (movhi_insn): Update store instruction constraint which are saving
17612 6-bit short immediates.
17613 (movsi_insn): Consider also short scaled load operations.
17614 (zero_extendhisi2_i): Use Usd constraint instead of T.
17615 (extendhisi2_i): Add q constraint.
17616 (arc_clzsi2): Add type and length attributes.
17617 (arc_ctzsi2): Likewise.
17618 * config/arc/constraints.md (Usc): Update constraint, the
17619 assembler can parse two relocations for a single instruction.
17621 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
17623 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
17624 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
17626 2017-08-31 Olivier Hainque <hainque@adacore.com>
17628 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
17629 match as powerpc-wrs-vxworks*.
17631 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
17633 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
17634 register constraint for by-element operand.
17635 (aarch64_mls_elt_merge<mode>): Likewise.
17637 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17639 * config/arc/arc.c (arc_can_follow_jump): Check for short
17642 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17644 * config.gcc: Use g.opt for arc.
17645 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
17646 functionality moved to ...
17647 (legitimate_scaled_address_p): New function, ...here.
17648 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
17649 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
17650 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
17652 (arc_override_options): Handle G option.
17653 (arc_output_pic_addr_const): Correct function definition.
17654 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
17655 (arc_decl_anon_ns_mem_p): Delete.
17656 (arc_in_small_data_p): Overhaul this function to take into
17657 consideration the value given via G option.
17658 (arc_rewrite_small_data_1): Renamed and corrected old
17659 arc_rewrite_small_data function.
17660 (arc_rewrite_small_data): New function.
17661 (small_data_pattern): Don't use pic_offset_table_rtx.
17662 * config/arc/arc.h (CC1_SPEC): Recognize G option.
17663 * config/arc/simdext.md (movmisalignv2hi): Use
17664 prepare_move_operands function.
17666 (movmisalign*): Likewise.
17667 * doc/invoke.texi (ARC options): Document -G option.
17669 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
17671 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
17673 * config/arc/arc.c (arc_print_operand): Output scalled address for
17674 sdata whenever is possible.
17675 (arc_in_small_data_p): Allow sdata for 64bit datum when double
17676 load/stores are available.
17677 (compact_sda_memory_operand): Check for the alignment required by
17678 code density instructions.
17679 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
17681 * config/arc/constraints.md (Usd): Update constraint.
17682 (Us0): New constraint.
17683 (Usc): Update constraint.
17685 2017-08-31 Richard Biener <rguenther@suse.de>
17687 PR middle-end/82054
17688 * dwarf2out.c (dwarf2out_early_global_decl): Process each
17689 function only once.
17691 2017-08-31 Tamar Christina <tamar.christina@arm.com>
17693 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
17694 Resize type_signature.
17696 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17697 Alan Hayward <alan.hayward@arm.com>
17698 David Sherwood <david.sherwood@arm.com>
17700 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
17701 subregs whose inner modes can be stored in GPRs.
17702 (aarch64_classify_index): Likewise.
17704 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17705 Alan Hayward <alan.hayward@arm.com>
17706 David Sherwood <david.sherwood@arm.com>
17708 * config/aarch64/iterators.md (V_cmp_result): Rename to...
17709 (V_INT_EQUIV): ...this.
17710 (v_cmp_result): Rename to...
17711 (v_int_equiv): ...this.
17712 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
17713 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
17714 (copysign<mode>3): Likewise.
17715 (aarch64_simd_bsl<mode>_internal): Likewise.
17716 (aarch64_simd_bsl<mode>): Likewise.
17717 (vec_cmp<mode><mode>): Likewise.
17718 (vcond<mode><mode>): Likewise.
17719 (vcond<v_cmp_mixed><mode>): Likewise.
17720 (vcondu<mode><v_cmp_mixed>): Likewise.
17721 (aarch64_cm<optab><mode>): Likewise.
17722 (aarch64_cmtst<mode>): Likewise.
17723 (aarch64_fac<optab><mode>): Likewise.
17724 (vec_perm_const<mode>): Likewise.
17725 (vcond_mask_<mode><v_cmp_result>): Rename to...
17726 (vcond_mask_<mode><v_int_equiv>): ...this.
17727 (vec_cmp<mode><v_cmp_result>): Rename to...
17728 (vec_cmp<mode><v_int_equiv>): ...this.
17730 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
17731 Alan Hayward <alan.hayward@arm.com>
17732 David Sherwood <david.sherwood@arm.com>
17734 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
17736 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
17737 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
17738 (UNSPEC_LD4_DREG): New unspecs.
17739 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
17740 (aarch64_ld2<mode>_dreg_be): Replace with...
17741 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
17743 (aarch64_ld3<mode>_dreg_le)
17744 (aarch64_ld3<mode>_dreg_be): Replace with...
17745 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
17747 (aarch64_ld4<mode>_dreg_le)
17748 (aarch64_ld4<mode>_dreg_be): Replace with...
17749 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
17752 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17754 PR tree-optimization/81987
17755 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
17756 insert an initializer in a location not dominated by the stride
17759 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17761 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
17762 on the entire header of the finally block in the fallthru case.
17764 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
17766 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
17768 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
17770 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
17771 rs6000_emit_move_from_cr and call renamed function.
17772 (rs6000_emit_prologue): Call renamed functions.
17773 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
17774 movesi_from_cr, remove volatile CRs.
17776 2017-08-30 Jon Beniston <jon@beniston.com>
17777 Richard Biener <rguenther@suse.de>
17779 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
17780 of VECTOR_MODE_P check.
17781 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
17782 element vector types.
17784 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17786 * df.h (df_read_modify_subreg_p): Remove in favor of...
17787 * rtl.h (read_modify_subreg_p): ...this new function. Take a
17788 const_rtx instead of an rtx.
17789 * cprop.c (local_cprop_find_used_regs): Update accordingly.
17790 * df-problems.c (df_word_lr_mark_ref): Likewise.
17791 * ira-lives.c (mark_pseudo_reg_live): Likewise.
17792 (mark_pseudo_reg_dead): Likewise.
17793 (mark_ref_dead): Likewise.
17794 * reginfo.c (init_subregs_of_mode): Likewise.
17795 * sched-deps.c (sched_analyze_1): Likewise.
17796 * df-scan.c (df_def_record_1): Likewise.
17797 (df_uses_record): Likewise.
17798 (df_read_modify_subreg_p): Remove in favor of...
17799 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
17800 const_rtx instead of an rtx.
17802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17803 Alan Hayward <alan.hayward@arm.com>
17804 David Sherwood <david.sherwood@arm.com>
17806 * rtl.h (partial_subreg_p): New function.
17807 * caller-save.c (save_call_clobbered_regs): Use it.
17808 * calls.c (expand_call): Likewise.
17809 * combine.c (combinable_i3pat): Likewise.
17810 (simplify_set): Likewise.
17811 (make_extraction): Likewise.
17812 (make_compound_operation_int): Likewise.
17813 (gen_lowpart_or_truncate): Likewise.
17814 (force_to_mode): Likewise.
17815 (make_field_assignment): Likewise.
17816 (reg_truncated_to_mode): Likewise.
17817 (record_truncated_value): Likewise.
17818 (move_deaths): Likewise.
17819 * cse.c (record_jump_cond): Likewise.
17820 (cse_insn): Likewise.
17821 * cselib.c (cselib_lookup_1): Likewise.
17822 * expmed.c (extract_bit_field_using_extv): Likewise.
17823 * function.c (assign_parm_setup_reg): Likewise.
17824 * ifcvt.c (noce_convert_multiple_sets): Likewise.
17825 * ira-build.c (create_insn_allocnos): Likewise.
17826 * lra-coalesce.c (merge_pseudos): Likewise.
17827 * lra-constraints.c (match_reload): Likewise.
17828 (simplify_operand_subreg): Likewise.
17829 (curr_insn_transform): Likewise.
17830 * lra-lives.c (process_bb_lives): Likewise.
17831 * lra.c (new_insn_reg): Likewise.
17832 (lra_substitute_pseudo): Likewise.
17833 * regcprop.c (mode_change_ok): Likewise.
17834 (maybe_mode_change): Likewise.
17835 (copyprop_hardreg_forward_1): Likewise.
17836 * reload.c (push_reload): Likewise.
17837 (find_reloads): Likewise.
17838 (find_reloads_subreg_address): Likewise.
17839 * reload1.c (alter_reg): Likewise.
17840 (eliminate_regs_1): Likewise.
17841 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17843 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
17845 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
17846 back to if statements, including unpack.
17848 2017-08-30 Martin Liska <mliska@suse.cz>
17850 PR inline-asm/82001
17851 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
17853 (func_checker::compare_asm_inputs_outputs): ... this function.
17854 (func_checker::compare_gimple_asm): Use the function to compare
17855 also ASM constrains.
17856 * ipa-icf-gimple.h: Rename the function.
17858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17859 Alan Hayward <alan.hayward@arm.com>
17860 David Sherwood <david.sherwood@arm.com>
17862 * coretypes.h (complex_mode): New type.
17863 * gdbhooks.py (build_pretty_printer): Handle it.
17864 * machmode.h (complex_mode): New class.
17865 (complex_mode::includes_p): New function.
17866 (is_complex_int_mode): Likewise.
17867 (is_complex_float_mode): Likewise.
17868 * genmodes.c (get_mode_class): Handle complex mode classes.
17869 * function.c (expand_function_end): Use is_complex_int_mode.
17871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17872 Alan Hayward <alan.hayward@arm.com>
17873 David Sherwood <david.sherwood@arm.com>
17875 * coretypes.h (scalar_mode_pod): New typedef.
17876 * gdbhooks.py (build_pretty_printer): Handle it.
17877 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
17878 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
17879 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
17880 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
17881 as_a <scalar_mode>.
17883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17884 Alan Hayward <alan.hayward@arm.com>
17885 David Sherwood <david.sherwood@arm.com>
17887 * machmode.h (mode_for_vector): Take a scalar_mode instead
17889 * stor-layout.c (mode_for_vector): Likewise.
17890 * explow.c (promote_mode): Use as_a <scalar_mode>.
17891 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
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 * target.def (preferred_simd_mode): Take a scalar_mode
17898 instead of a machine_mode.
17899 * targhooks.h (default_preferred_simd_mode): Likewise.
17900 * targhooks.c (default_preferred_simd_mode): Likewise.
17901 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
17902 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
17903 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
17904 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
17905 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
17906 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
17907 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
17908 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
17910 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
17911 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
17912 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
17913 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
17914 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
17915 * doc/tm.texi: Regenerate.
17916 * optabs-query.c (can_vec_mask_load_store_p): Return false for
17919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17920 Alan Hayward <alan.hayward@arm.com>
17921 David Sherwood <david.sherwood@arm.com>
17923 * target.def (scalar_mode_supported_p): Take a scalar_mode
17924 instead of a machine_mode.
17925 * targhooks.h (default_scalar_mode_supported_p): Likewise.
17926 * targhooks.c (default_scalar_mode_supported_p): Likewise.
17927 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
17928 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17929 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
17930 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
17931 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
17932 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
17933 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
17934 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
17935 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
17936 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
17937 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
17938 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
17940 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
17941 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
17942 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17943 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17944 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
17946 * doc/tm.texi: Regenerate.
17948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17949 Alan Hayward <alan.hayward@arm.com>
17950 David Sherwood <david.sherwood@arm.com>
17952 * coretypes.h (opt_scalar_mode): New typedef.
17953 * gdbhooks.py (build_pretty_printers): Handle it.
17954 * machmode.h (mode_iterator::get_2xwider): Add overload for
17956 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
17958 * expr.c (convert_mode_scalar): Likewise.
17959 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17960 * optabs.c (expand_float): Likewise.
17961 (expand_fix): Likewise.
17962 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17965 Alan Hayward <alan.hayward@arm.com>
17966 David Sherwood <david.sherwood@arm.com>
17968 * optabs.c (expand_float): Explicitly check for scalars before
17969 using a branching expansion.
17970 (expand_fix): Likewise.
17972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17973 Alan Hayward <alan.hayward@arm.com>
17974 David Sherwood <david.sherwood@arm.com>
17976 * expr.c (convert_mode): Split scalar handling out into...
17977 (convert_mode_scalar): ...this new function. Treat the modes
17980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17981 Alan Hayward <alan.hayward@arm.com>
17982 David Sherwood <david.sherwood@arm.com>
17984 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
17986 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
17988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17989 Alan Hayward <alan.hayward@arm.com>
17990 David Sherwood <david.sherwood@arm.com>
17992 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
17993 rather than a machine_mode.
17994 (fixed_from_string): Likewise.
17995 (fixed_convert): Likewise.
17996 (fixed_convert_from_int): Likewise.
17997 (fixed_convert_from_real): Likewise.
17998 (real_convert_from_fixed): Likewise.
17999 * fixed-value.c (fixed_from_double_int): Likewise.
18000 (fixed_from_string): Likewise.
18001 (fixed_convert): Likewise.
18002 (fixed_convert_from_int): Likewise.
18003 (fixed_convert_from_real): Likewise.
18004 (real_convert_from_fixed): Likewise.
18005 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18008 Alan Hayward <alan.hayward@arm.com>
18009 David Sherwood <david.sherwood@arm.com>
18011 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18012 of separate mode class checks. Do not allow vector modes here.
18013 (immed_wide_int_const): Use as_a <scalar_mode>.
18014 * explow.c (trunc_int_for_mode): Likewise.
18015 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18016 (wi::shwi): Likewise.
18017 (wi::min_value): Likewise.
18018 (wi::max_value): Likewise.
18019 * dwarf2out.c (loc_descriptor): Likewise.
18020 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18021 for CONST_WIDE_INT.
18023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18024 Alan Hayward <alan.hayward@arm.com>
18025 David Sherwood <david.sherwood@arm.com>
18027 * tree.h (SCALAR_TYPE_MODE): New macro.
18028 * expr.c (expand_expr_addr_expr_1): Use it.
18029 (expand_expr_real_2): Likewise.
18030 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18031 (fold_convert_const_fixed_from_int): Likewise.
18032 (fold_convert_const_fixed_from_real): Likewise.
18033 (native_encode_fixed): Likewise
18034 (native_encode_complex): Likewise
18035 (native_encode_vector): Likewise.
18036 (native_interpret_fixed): Likewise.
18037 (native_interpret_real): Likewise.
18038 (native_interpret_complex): Likewise.
18039 (native_interpret_vector): Likewise.
18040 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18041 (simd_clone_adjust_argument_types): Likewise.
18042 (simd_clone_init_simd_arrays): Likewise.
18043 (simd_clone_adjust): Likewise.
18044 * stor-layout.c (layout_type): Likewise.
18045 * tree.c (build_minus_one_cst): Likewise.
18046 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18047 * tree-inline.c (estimate_move_cost): Likewise.
18048 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18049 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18050 (vectorizable_reduction): Likewise.
18051 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18052 (vect_recog_mixed_size_cond_pattern): Likewise.
18053 (check_bool_pattern): Likewise.
18054 (adjust_bool_pattern): Likewise.
18055 (search_type_for_mask_1): Likewise.
18056 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18057 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18058 (vectorizable_load): Likewise.
18059 (vectorizable_store): Likewise.
18060 * ubsan.c (ubsan_encode_value): Likewise.
18061 * varasm.c (output_constant): Likewise.
18063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18064 Alan Hayward <alan.hayward@arm.com>
18065 David Sherwood <david.sherwood@arm.com>
18067 * coretypes.h (scalar_mode): New class.
18068 * machmode.h (scalar_mode): Likewise.
18069 (scalar_mode::includes_p): New function.
18070 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18071 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18072 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18073 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18074 * expmed.c (store_bit_field_1): Likewise.
18075 (extract_bit_field_1): Likewise.
18076 * expr.c (write_complex_part): Likewise.
18077 (read_complex_part): Likewise.
18078 (emit_move_complex_push): Likewise.
18079 (expand_expr_real_2): Likewise.
18080 * function.c (assign_parm_setup_reg): Likewise.
18081 (assign_parms_unsplit_complex): Likewise.
18082 * optabs.c (expand_binop): Likewise.
18083 * rtlanal.c (subreg_get_info): Likewise.
18084 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18085 * varasm.c (output_constant_pool_2): Likewise.
18087 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18088 Alan Hayward <alan.hayward@arm.com>
18089 David Sherwood <david.sherwood@arm.com>
18091 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18093 (expmed_mult_highpart_optab): Likewise.
18094 (expmed_mult_highpart): Likewise.
18096 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18097 Alan Hayward <alan.hayward@arm.com>
18098 David Sherwood <david.sherwood@arm.com>
18100 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18101 instead of a machine_mode.
18102 (builtin_memset_read_str): Likewise.
18103 * builtins.c (c_readstr): Likewise.
18104 (builtin_memcpy_read_str): Likewise.
18105 (builtin_strncpy_read_str): Likewise.
18106 (builtin_memset_read_str): Likewise.
18107 (builtin_memset_gen_str): Likewise.
18108 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18109 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18110 instead of a machine_mode.
18111 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18113 (make_extraction): Likewise.
18114 (try_widen_shift_mode): Take and return scalar_int_modes instead
18116 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18117 a scalar_int_mode instead of a machine_mode.
18118 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18119 (avr_addr_space_pointer_mode): Likewise.
18120 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18121 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18122 (msp430_unwind_word_mode): Likewise.
18123 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18124 (spu_addr_space_pointer_mode): Likewise.
18125 (spu_addr_space_address_mode): Likewise.
18126 (spu_libgcc_cmp_return_mode): Likewise.
18127 (spu_libgcc_shift_count_mode): Likewise.
18128 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18129 (rl78_addr_space_pointer_mode): Likewise.
18130 (fl78_unwind_word_mode): Likewise.
18131 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18133 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18134 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18135 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18136 (mips_valid_pointer_mode): Likewise.
18137 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18138 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18139 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18141 (ft32_addr_space_address_mode): Likewise.
18142 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18143 scalar_int_mode instead of a machine_mode.
18144 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18146 (m32c_addr_space_address_mode): Likewise.
18147 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18148 (rs6000_eh_return_filter_mode): Likewise.
18149 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18150 (rs6000_eh_return_filter_mode): Likewise.
18151 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18152 (s390_libgcc_shift_count_mode): Likewise.
18153 (s390_unwind_word_mode): Likewise.
18154 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18156 * target.def (mode_rep_extended): Likewise.
18157 (valid_pointer_mode): Likewise.
18158 (addr_space.valid_pointer_mode): Likewise.
18159 (eh_return_filter_mode): Return a scalar_int_mode rather than
18161 (libgcc_cmp_return_mode): Likewise.
18162 (libgcc_shift_count_mode): Likewise.
18163 (unwind_word_mode): Likewise.
18164 (addr_space.pointer_mode): Likewise.
18165 (addr_space.address_mode): Likewise.
18166 * doc/tm.texi: Regenerate.
18167 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18169 (do_jump): Use scalar_int_mode for local variables.
18170 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18171 rather than a machine_mode.
18172 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18173 (scompare_loc_descriptor_wide): Likewise.
18174 (scompare_loc_descriptor_narrow): Likewise.
18175 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18177 * except.c (sjlj_emit_dispatch_table): Likewise.
18178 (expand_builtin_eh_copy_values): Likewise.
18179 * explow.c (convert_memory_address_addr_space_1): Likewise.
18180 Take a scalar_int_mode rather than a machine_mode.
18181 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18182 than a machine_mode.
18183 (memory_address_addr_space): Use scalar_int_mode for local variables.
18184 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18185 rather than a machine_mode.
18186 * expmed.c (mask_rtx): Likewise.
18187 (init_expmed_one_conv): Likewise.
18188 (expand_mult_highpart_adjust): Likewise.
18189 (extract_high_half): Likewise.
18190 (expmed_mult_highpart_optab): Likewise.
18191 (expmed_mult_highpart): Likewise.
18192 (expand_smod_pow2): Likewise.
18193 (expand_sdiv_pow2): Likewise.
18194 (emit_store_flag_int): Likewise.
18195 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18197 (extract_low_bits): Likewise.
18198 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18200 * expr.c (pieces_addr::adjust): Likewise.
18201 (can_store_by_pieces): Likewise.
18202 (store_by_pieces): Likewise.
18203 (clear_by_pieces_1): Likewise.
18204 (expand_expr_addr_expr_1): Likewise.
18205 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18206 (expand_expr_real_1): Likewise.
18207 (try_casesi): Likewise.
18208 * final.c (shorten_branches): Likewise.
18209 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18210 type of "mode" to machine_mode.
18211 * internal-fn.c (expand_arith_overflow_result_store): Take a
18212 scalar_int_mode rather than a machine_mode.
18213 (expand_mul_overflow): Use scalar_int_mode for local variables.
18214 * loop-doloop.c (doloop_modify): Likewise.
18215 (doloop_optimize): Likewise.
18216 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18217 than a machine_mode.
18218 (expand_doubleword_shift_condmove): Likewise.
18219 (expand_doubleword_shift): Likewise.
18220 (expand_doubleword_clz): Likewise.
18221 (expand_doubleword_popcount): Likewise.
18222 (expand_doubleword_parity): Likewise.
18223 (expand_absneg_bit): Use scalar_int_mode for local variables.
18224 (prepare_float_lib_cmp): Likewise.
18225 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18226 rather than a machine_mode.
18227 (convert_memory_address_addr_space): Likewise.
18228 (get_mode_bounds): Likewise.
18229 (get_address_mode): Return a scalar_int_mode rather than a
18231 * rtlanal.c (get_address_mode): Likewise.
18232 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18233 than a machine_mode.
18234 * targhooks.c (default_mode_rep_extended): Likewise.
18235 (default_valid_pointer_mode): Likewise.
18236 (default_addr_space_valid_pointer_mode): Likewise.
18237 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18238 than a machine_mode.
18239 (default_libgcc_cmp_return_mode): Likewise.
18240 (default_libgcc_shift_count_mode): Likewise.
18241 (default_unwind_word_mode): Likewise.
18242 (default_addr_space_pointer_mode): Likewise.
18243 (default_addr_space_address_mode): Likewise.
18244 * targhooks.h (default_eh_return_filter_mode): Likewise.
18245 (default_libgcc_cmp_return_mode): Likewise.
18246 (default_libgcc_shift_count_mode): Likewise.
18247 (default_unwind_word_mode): Likewise.
18248 (default_addr_space_pointer_mode): Likewise.
18249 (default_addr_space_address_mode): Likewise.
18250 (default_mode_rep_extended): Take a scalar_int_mode rather than
18252 (default_valid_pointer_mode): Likewise.
18253 (default_addr_space_valid_pointer_mode): Likewise.
18254 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18256 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18257 rather than a machine_mode.
18258 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18259 for local variables.
18260 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18261 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18262 than a machine_mode.
18264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18265 Alan Hayward <alan.hayward@arm.com>
18266 David Sherwood <david.sherwood@arm.com>
18268 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18269 the mode argument to scalar_int_mode.
18270 (do_jump_by_parts_zero_rtx): Likewise.
18271 (do_jump_by_parts_equality_rtx): Likewise.
18272 (do_jump_by_parts_greater): Take a mode argument.
18273 (do_jump_by_parts_equality): Likewise.
18274 (do_jump_1): Update calls accordingly.
18276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18277 Alan Hayward <alan.hayward@arm.com>
18278 David Sherwood <david.sherwood@arm.com>
18280 * is-a.h (safe_dyn_cast): New function.
18281 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18282 (jump_table_for_label): Likewise.
18283 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18284 instead of an rtx_insn *.
18285 (shorten_branches): Use dyn_cast instead of LABEL_P and
18286 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18287 rtx_jump_table_data::get_data_mode.
18288 (final_scan_insn): Likewise.
18290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18291 Alan Hayward <alan.hayward@arm.com>
18292 David Sherwood <david.sherwood@arm.com>
18294 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18295 trying to combine a full-register integer set with a subreg
18298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18299 Alan Hayward <alan.hayward@arm.com>
18300 David Sherwood <david.sherwood@arm.com>
18302 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18303 that is always either address_mode or pointer_mode.
18305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18306 Alan Hayward <alan.hayward@arm.com>
18307 David Sherwood <david.sherwood@arm.com>
18309 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18310 when the two are known to be equal.
18312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18313 Alan Hayward <alan.hayward@arm.com>
18314 David Sherwood <david.sherwood@arm.com>
18316 * simplify-rtx.c (simplify_const_unary_operation): Use
18317 is_a <scalar_int_mode> instead of checking for a nonzero
18318 precision. Forcibly convert op_mode to a scalar_int_mode
18319 in that case. More clearly differentiate the operand and
18320 result modes and use the former when deciding what the value
18321 of a count-bits operation should be. Use is_int_mode instead
18322 of checking for a MODE_INT. Remove redundant check for whether
18323 this mode has a zero precision.
18325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18326 Alan Hayward <alan.hayward@arm.com>
18327 David Sherwood <david.sherwood@arm.com>
18329 * optabs.c (widen_leading): Change the type of the mode argument
18330 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18331 (widen_bswap): Likewise.
18332 (expand_parity): Likewise.
18333 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18334 (expand_ffs): Likewise.
18335 (epand_unop): Check for scalar integer modes before calling the
18338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18339 Alan Hayward <alan.hayward@arm.com>
18340 David Sherwood <david.sherwood@arm.com>
18342 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18344 (expand_expr_real_1): Update call accordingly.
18346 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18347 Alan Hayward <alan.hayward@arm.com>
18348 David Sherwood <david.sherwood@arm.com>
18350 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18351 value_mode arguments. Use scalar_int_mode internally.
18352 (store_bit_field_1): Rename the new integer mode from imode
18353 to op0_mode and use it instead of GET_MODE (op0). Update calls
18354 to store_split_bit_field, store_bit_field_using_insv and
18355 store_fixed_bit_field.
18356 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18357 Use scalar_int_mode internally. Use a bit count rather than a mode
18358 when calculating the largest bit size for get_best_mode.
18359 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18360 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18361 Remove assertion that OP0 has a scalar integer mode.
18362 (store_split_bit_field): Add op0_mode and value_mode arguments.
18363 Update calls to extract_fixed_bit_field.
18364 (extract_bit_field_using_extv): Add an op0_mode argument.
18365 Use scalar_int_mode internally.
18366 (extract_bit_field_1): Rename the new integer mode from imode to
18367 op0_mode and use it instead of GET_MODE (op0). Update calls to
18368 extract_split_bit_field, extract_bit_field_using_extv and
18369 extract_fixed_bit_field.
18370 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18371 to extract_split_bit_field and extract_fixed_bit_field_1.
18372 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18373 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18374 on the target mode.
18375 (extract_split_bit_field): Add an op0_mode argument. Update call
18376 to extract_fixed_bit_field.
18378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18379 Alan Hayward <alan.hayward@arm.com>
18380 David Sherwood <david.sherwood@arm.com>
18382 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18383 * explow.c (hard_function_value): Likewise.
18384 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18385 convert_to_mode call outside the loop.
18386 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18387 for the mode iterator. Require the mode specified by max_pieces
18389 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18391 (copy_blkmode_to_reg): Likewise.
18392 (set_storage_via_setmem): Likewise.
18393 * optabs.c (prepare_cmp_insn): Likewise.
18394 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
18395 * stor-layout.c (finish_bitfield_representative): Likewise.
18397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18398 Alan Hayward <alan.hayward@arm.com>
18399 David Sherwood <david.sherwood@arm.com>
18401 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
18402 * expr.c (convert_move): Use them.
18403 (convert_modes): Likewise.
18404 (store_expr_with_bounds): Likewise.
18406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18407 Alan Hayward <alan.hayward@arm.com>
18408 David Sherwood <david.sherwood@arm.com>
18410 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
18411 parameter for the mode of "x". Remove the "known_x", "known_mode"
18412 and "known_ret" arguments. Change the type of the mode argument
18413 to scalar_int_mode.
18414 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
18415 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
18416 (reg_num_sign_bit_copies_for_combine): Likewise.
18417 * rtlanal.c (nonzero_bits1): Likewise.
18418 (num_sign_bit_copies1): Likewise.
18419 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
18420 (reg_num_sign_bit_copies_general): Likewise.
18421 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
18422 (reg_nonzero_bits_general): Likewise.
18424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18425 Alan Hayward <alan.hayward@arm.com>
18426 David Sherwood <david.sherwood@arm.com>
18428 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
18429 than in subroutines. Return 1 for non-integer modes.
18430 (cached_num_sign_bit_copies): Change the type of the mode parameter
18431 to scalar_int_mode.
18432 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
18433 classes. Handle CONST_INT_P first and then check whether X also
18434 has a scalar integer mode. Check the same thing for inner registers
18435 of a SUBREG and for values that are being extended or truncated.
18437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18438 Alan Hayward <alan.hayward@arm.com>
18439 David Sherwood <david.sherwood@arm.com>
18441 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
18442 in subroutines. Return the mode mask for non-integer modes.
18443 (cached_nonzero_bits): Change the type of the mode parameter
18444 to scalar_int_mode.
18445 (nonzero_bits1): Likewise. Remove early exit for other mode
18446 classes. Handle CONST_INT_P first and then check whether X
18447 also has a scalar integer mode.
18449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18450 Alan Hayward <alan.hayward@arm.com>
18451 David Sherwood <david.sherwood@arm.com>
18453 * expr.c (widest_int_mode_for_size): Make the comment match the code.
18454 Return a scalar_int_mode and assert that the size is greater than
18456 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
18457 (op_by_pieces_d::op_by_pieces_d): Likewise.
18458 (op_by_pieces_d::run): Likewise.
18459 (can_store_by_pieces): Likewise.
18461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18462 Alan Hayward <alan.hayward@arm.com>
18463 David Sherwood <david.sherwood@arm.com>
18465 * combine.c (extract_left_shift): Add a mode argument and update
18467 (make_compound_operation_int): Change the type of the mode parameter
18468 to scalar_int_mode and update the call to extract_left_shift.
18470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18471 Alan Hayward <alan.hayward@arm.com>
18472 David Sherwood <david.sherwood@arm.com>
18474 * combine.c (simplify_and_const_int): Change the type of the mode
18475 parameter to scalar_int_mode.
18476 (simplify_and_const_int_1): Likewise. Update recursive call.
18478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18479 Alan Hayward <alan.hayward@arm.com>
18480 David Sherwood <david.sherwood@arm.com>
18482 * combine.c (simplify_compare_const): Check that the mode is a
18483 scalar_int_mode (rather than VOIDmode) before testing its
18485 (simplify_comparison): Move COMPARISON_P handling out of the
18486 loop and restrict the latter part of the loop to scalar_int_modes.
18487 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
18488 and when considering SUBREG_REGs. Use is_int_mode instead of
18489 checking GET_MODE_CLASS against MODE_INT.
18491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18492 Alan Hayward <alan.hayward@arm.com>
18493 David Sherwood <david.sherwood@arm.com>
18495 * combine.c (try_widen_shift_mode): Move check for equal modes to...
18496 (simplify_shift_const_1): ...here. Use scalar_int_mode for
18497 shift_unit_mode and for modes involved in scalar shifts.
18499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18500 Alan Hayward <alan.hayward@arm.com>
18501 David Sherwood <david.sherwood@arm.com>
18503 * combine.c (force_int_to_mode): New function, split out from...
18504 (force_to_mode): ...here. Keep xmode up-to-date and use it
18505 instead of GET_MODE (x).
18507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18508 Alan Hayward <alan.hayward@arm.com>
18509 David Sherwood <david.sherwood@arm.com>
18511 * optabs-query.h (extraction_insn::struct_mode): Change type to
18512 opt_scalar_int_mode and update comment.
18513 (extraction_insn::field_mode): Change type to scalar_int_mode.
18514 (extraction_insn::pos_mode): Likewise.
18515 * combine.c (make_extraction): Update accordingly.
18516 * optabs-query.c (get_traditional_extraction_insn): Likewise.
18517 (get_optab_extraction_insn): Likewise.
18518 * recog.c (simplify_while_replacing): Likewise.
18519 * expmed.c (narrow_bit_field_mem): Change the type of the mode
18520 parameter to opt_scalar_int_mode.
18522 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18523 Alan Hayward <alan.hayward@arm.com>
18524 David Sherwood <david.sherwood@arm.com>
18526 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
18527 to a scalar_int_mode instead of a machine_mode.
18528 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
18529 (get_best_mode): Return a boolean and use a pointer argument to store
18530 the selected mode. Replace the limit mode parameter with a bit limit.
18531 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
18532 for the values returned by bit_field_mode_iterator::next_mode.
18533 (store_bit_field): Update call to get_best_mode.
18534 (store_fixed_bit_field): Likewise.
18535 (extract_fixed_bit_field): Likewise.
18536 * expr.c (optimize_bitfield_assignment_op): Likewise.
18537 * fold-const.c (optimize_bit_field_compare): Likewise.
18538 (fold_truth_andor_1): Likewise.
18539 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
18540 Update for new type of m_mode.
18541 (get_best_mode): As above.
18543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18544 Alan Hayward <alan.hayward@arm.com>
18545 David Sherwood <david.sherwood@arm.com>
18547 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
18548 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
18549 (store_bit_field): Check is_a <scalar_int_mode> before calling
18550 strict_volatile_bitfield_p.
18551 (extract_bit_field): Likewise.
18553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18554 Alan Hayward <alan.hayward@arm.com>
18555 David Sherwood <david.sherwood@arm.com>
18557 * target.def (cstore_mode): Return a scalar_int_mode.
18558 * doc/tm.texi: Regenerate.
18559 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
18560 * targhooks.h (default_cstore_mode): Likewise.
18561 * targhooks.c (default_cstore_mode): Likewise, using a forced
18563 * expmed.c (emit_cstore): Expect the target of the cstore to be
18566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18567 Alan Hayward <alan.hayward@arm.com>
18568 David Sherwood <david.sherwood@arm.com>
18570 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
18572 (niter_desc): Likewise mode.
18573 (iv_analyze): Add a mode parameter.
18575 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
18576 and change its type to scalar_int_mode.
18577 * loop-iv.c: Update commentary at head of file.
18578 (iv_constant): Pass the mode paraeter before the rtx it describes
18579 and change its type to scalar_int_mode. Remove VOIDmode handling.
18580 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
18581 (iv_extend): Likewise.
18582 (shorten_into_mode): Likewise.
18583 (iv_add): Use scalar_int_mode.
18584 (iv_mult): Likewise.
18585 (iv_shift): Likewise.
18586 (canonicalize_iv_subregs): Likewise.
18587 (get_biv_step_1): Pass the outer_mode parameter before the rtx
18588 it describes and change its mode to scalar_int_mode. Also change
18589 the type of the returned inner_mode to scalar_int_mode.
18590 (get_biv_step): Likewise, turning outer_mode from a pointer
18591 into a direct parameter. Update call to get_biv_step_1.
18592 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
18593 iv_constant and get_biv_step.
18594 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
18595 and change its type to scalar_int_mode. Don't initialise iv->mode
18596 to VOIDmode and remove later checks for its still being VOIDmode.
18597 Update calls to iv_analyze_op and iv_analyze_expr. Check
18598 is_a <scalar_int_mode> when changing the mode under consideration.
18599 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
18600 Update call to iv_analyze_expr.
18601 (iv_analyze_op): Add a mode parameter. Reject subregs whose
18602 inner register is not also a scalar_int_mode. Update call to
18604 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
18605 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
18606 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
18607 separate mode class checks. Update calls to iv_analyze. Remove
18608 fix-up of VOIDmodes after iv_analyze_biv.
18609 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
18610 don't have a scalar_int_mode. Update call to biv_p.
18612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18613 Alan Hayward <alan.hayward@arm.com>
18614 David Sherwood <david.sherwood@arm.com>
18616 * cfgexpand.c (convert_debug_memory_address): Use
18617 as_a <scalar_int_mode>.
18618 * combine.c (expand_compound_operation): Likewise.
18619 (make_extraction): Likewise.
18620 (change_zero_ext): Likewise.
18621 (simplify_comparison): Likewise.
18622 * cse.c (cse_insn): Likewise.
18623 * dwarf2out.c (minmax_loc_descriptor): Likewise.
18624 (mem_loc_descriptor): Likewise.
18625 (loc_descriptor): Likewise.
18626 * expmed.c (init_expmed_one_mode): Likewise.
18627 (synth_mult): Likewise.
18628 (emit_store_flag_1): Likewise.
18629 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
18630 of a comparison with size.
18631 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
18632 (reduce_to_bit_field_precision): Likewise.
18633 * function.c (expand_function_end): Likewise.
18634 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
18635 * loop-doloop.c (doloop_modify): Likewise.
18636 * optabs.c (expand_binop): Likewise.
18637 (expand_unop): Likewise.
18638 (expand_copysign_absneg): Likewise.
18639 (prepare_cmp_insn): Likewise.
18640 (maybe_legitimize_operand): Likewise.
18641 * recog.c (const_scalar_int_operand): Likewise.
18642 * rtlanal.c (get_address_mode): Likewise.
18643 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18644 (simplify_cond_clz_ctz): Likewise.
18645 * tree-nested.c (get_nl_goto_field): Likewise.
18646 * tree.c (build_vector_type_for_mode): Likewise.
18647 * var-tracking.c (use_narrower_mode): Likewise.
18649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18650 Alan Hayward <alan.hayward@arm.com>
18651 David Sherwood <david.sherwood@arm.com>
18653 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
18654 * builtins.c (expand_builtin_signbit): Use it.
18655 * cfgexpand.c (expand_debug_expr): Likewise.
18656 * dojump.c (do_jump): Likewise.
18657 (do_compare_and_jump): Likewise.
18658 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
18659 * expmed.c (make_tree): Likewise.
18660 * expr.c (expand_expr_real_2): Likewise.
18661 (expand_expr_real_1): Likewise.
18662 (try_casesi): Likewise.
18663 * fold-const-call.c (fold_const_call_ss): Likewise.
18664 * fold-const.c (unextend): Likewise.
18665 (extract_muldiv_1): Likewise.
18666 (fold_single_bit_test): Likewise.
18667 (native_encode_int): Likewise.
18668 (native_encode_string): Likewise.
18669 (native_interpret_int): Likewise.
18670 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
18671 * internal-fn.c (expand_addsub_overflow): Likewise.
18672 (expand_neg_overflow): Likewise.
18673 (expand_mul_overflow): Likewise.
18674 (expand_arith_overflow): Likewise.
18675 * match.pd: Likewise.
18676 * stor-layout.c (layout_type): Likewise.
18677 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18678 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
18679 * tree-ssanames.c (get_range_info): Likewise.
18680 * tree-switch-conversion.c (array_value_type) Likewise.
18681 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
18682 (vect_recog_divmod_pattern): Likewise.
18683 (vect_recog_mixed_size_cond_pattern): Likewise.
18684 * tree-vrp.c (extract_range_basic): Likewise.
18685 (simplify_float_conversion_using_ranges): Likewise.
18686 * tree.c (int_fits_type_p): Likewise.
18687 * ubsan.c (instrument_bool_enum_load): Likewise.
18688 * varasm.c (mergeable_string_section): Likewise.
18689 (narrowing_initializer_constant_valid_p): Likewise.
18690 (output_constant): Likewise.
18692 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18693 Alan Hayward <alan.hayward@arm.com>
18694 David Sherwood <david.sherwood@arm.com>
18696 * machmode.h (NARROWEST_INT_MODE): New macro.
18697 * expr.c (alignment_for_piecewise_move): Use it instead of
18698 GET_CLASS_NARROWEST_MODE (MODE_INT).
18699 (push_block): Likewise.
18700 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
18702 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * postreload.c (move2add_valid_value_p): Change the type of the
18709 mode parameter to scalar_int_mode.
18710 (move2add_use_add2_insn): Add a mode parameter and use it instead
18712 (move2add_use_add3_insn): Likewise.
18713 (reload_cse_move2add): Update accordingly.
18715 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18716 Alan Hayward <alan.hayward@arm.com>
18717 David Sherwood <david.sherwood@arm.com>
18719 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
18721 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
18722 * optabs.c (expand_unop): Likewise.
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 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
18729 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
18730 (popcount_loc_descriptor): Likewise.
18731 (bswap_loc_descriptor): Likewise.
18732 (rotate_loc_descriptor): Likewise.
18733 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
18734 calling the functions above.
18736 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18737 Alan Hayward <alan.hayward@arm.com>
18738 David Sherwood <david.sherwood@arm.com>
18740 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
18742 (try_combine): Likewise.
18743 (simplify_if_then_else): Likewise.
18744 * cse.c (cse_insn): Likewise.
18745 * dwarf2out.c (mem_loc_descriptor): Likewise.
18746 * emit-rtl.c (gen_lowpart_common): Likewise.
18747 * simplify-rtx.c (simplify_truncation): Likewise.
18748 (simplify_binary_operation_1): Likewise.
18749 (simplify_const_relational_operation): Likewise.
18750 (simplify_ternary_operation): Likewise.
18751 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
18753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18754 Alan Hayward <alan.hayward@arm.com>
18755 David Sherwood <david.sherwood@arm.com>
18757 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
18758 * reload.c (push_reload): Likewise.
18759 (find_reloads): Likewise.
18761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18762 Alan Hayward <alan.hayward@arm.com>
18763 David Sherwood <david.sherwood@arm.com>
18765 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
18766 (make_compound_operation_int): Likewise.
18767 (change_zero_ext): Likewise.
18768 * expr.c (convert_move): Likewise.
18769 (convert_modes): Likewise.
18770 * fwprop.c (forward_propagate_subreg): Likewise.
18771 * loop-iv.c (get_biv_step_1): Likewise.
18772 * optabs.c (widen_operand): Likewise.
18773 * postreload.c (move2add_valid_value_p): Likewise.
18774 * recog.c (simplify_while_replacing): Likewise.
18775 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18776 (simplify_binary_operation_1): Likewise. Remove redundant
18777 mode equality check.
18779 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18780 Alan Hayward <alan.hayward@arm.com>
18781 David Sherwood <david.sherwood@arm.com>
18783 * combine.c (combine_simplify_rtx): Add checks for
18784 is_a <scalar_int_mode>.
18785 (simplify_if_then_else): Likewise.
18786 (make_field_assignment): Likewise.
18787 (simplify_comparison): Likewise.
18788 * ifcvt.c (noce_try_bitop): Likewise.
18789 * loop-invariant.c (canonicalize_address_mult): Likewise.
18790 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18792 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18793 Alan Hayward <alan.hayward@arm.com>
18794 David Sherwood <david.sherwood@arm.com>
18796 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
18797 is_a <scalar_int_mode> instead of != BLKmode.
18799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18800 Alan Hayward <alan.hayward@arm.com>
18801 David Sherwood <david.sherwood@arm.com>
18803 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
18804 instead of != VOIDmode.
18805 * combine.c (if_then_else_cond): Likewise.
18806 (change_zero_ext): Likewise.
18807 * dwarf2out.c (mem_loc_descriptor): Likewise.
18808 (loc_descriptor): Likewise.
18809 * rtlanal.c (canonicalize_condition): Likewise.
18810 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
18812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18813 Alan Hayward <alan.hayward@arm.com>
18814 David Sherwood <david.sherwood@arm.com>
18816 * simplify-rtx.c (simplify_binary_operation_1): Use
18817 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
18819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18820 Alan Hayward <alan.hayward@arm.com>
18821 David Sherwood <david.sherwood@arm.com>
18823 * wide-int.h (int_traits<unsigned char>) New class.
18824 (int_traits<unsigned short>) Likewise.
18825 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
18826 Use GET_MODE_UNIT_PRECISION and remove redundant test for
18828 * combine.c (set_nonzero_bits_and_sign_copies): Use
18829 is_a <scalar_int_mode>.
18830 (find_split_point): Likewise.
18831 (combine_simplify_rtx): Likewise.
18832 (simplify_logical): Likewise.
18833 (expand_compound_operation): Likewise.
18834 (expand_field_assignment): Likewise.
18835 (make_compound_operation): Likewise.
18836 (extended_count): Likewise.
18837 (change_zero_ext): Likewise.
18838 (simplify_comparison): Likewise.
18839 * dwarf2out.c (scompare_loc_descriptor): Likewise.
18840 (ucompare_loc_descriptor): Likewise.
18841 (minmax_loc_descriptor): Likewise.
18842 (mem_loc_descriptor): Likewise.
18843 (loc_descriptor): Likewise.
18844 * expmed.c (init_expmed_one_mode): Likewise.
18845 * lra-constraints.c (lra_constraint_offset): Likewise.
18846 * optabs.c (prepare_libcall_arg): Likewise.
18847 * postreload.c (move2add_note_store): Likewise.
18848 * reload.c (operands_match_p): Likewise.
18849 * rtl.h (load_extend_op): Likewise.
18850 * rtlhooks.c (gen_lowpart_general): Likewise.
18851 * simplify-rtx.c (simplify_truncation): Likewise.
18852 (simplify_unary_operation_1): Likewise.
18853 (simplify_binary_operation_1): Likewise.
18854 (simplify_const_binary_operation): Likewise.
18855 (simplify_const_relational_operation): Likewise.
18856 (simplify_subreg): Likewise.
18857 * stor-layout.c (bitwise_mode_for_mode): Likewise.
18858 * var-tracking.c (adjust_mems): Likewise.
18859 (prepare_call_arguments): Likewise.
18861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18862 Alan Hayward <alan.hayward@arm.com>
18863 David Sherwood <david.sherwood@arm.com>
18865 * machmode.h (is_int_mode): New fuction.
18866 * combine.c (find_split_point): Use it.
18867 (combine_simplify_rtx): Likewise.
18868 (simplify_if_then_else): Likewise.
18869 (simplify_set): Likewise.
18870 (simplify_shift_const_1): Likewise.
18871 (simplify_comparison): Likewise.
18872 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
18873 * cse.c (notreg_cost): Likewise.
18874 (cse_insn): Likewise.
18875 * cselib.c (cselib_lookup_1): Likewise.
18876 * dojump.c (do_jump_1): Likewise.
18877 (do_compare_rtx_and_jump): Likewise.
18878 * dse.c (get_call_args): Likewise.
18879 * dwarf2out.c (rtl_for_decl_init): Likewise.
18880 (native_encode_initializer): Likewise.
18881 * expmed.c (emit_store_flag_1): Likewise.
18882 (emit_store_flag): Likewise.
18883 * expr.c (convert_modes): Likewise.
18884 (store_field): Likewise.
18885 (expand_expr_real_1): Likewise.
18886 * fold-const.c (fold_read_from_constant_string): Likewise.
18887 * gimple-ssa-sprintf.c (get_format_string): Likewise.
18888 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
18889 * optabs.c (expand_binop): Likewise.
18890 (expand_unop): Likewise.
18891 (expand_abs_nojump): Likewise.
18892 (expand_one_cmpl_abs_nojump): Likewise.
18893 * simplify-rtx.c (mode_signbit_p): Likewise.
18894 (val_signbit_p): Likewise.
18895 (val_signbit_known_set_p): Likewise.
18896 (val_signbit_known_clear_p): Likewise.
18897 (simplify_relational_operation_1): Likewise.
18898 * tree.c (vector_type_mode): Likewise.
18900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18901 Alan Hayward <alan.hayward@arm.com>
18902 David Sherwood <david.sherwood@arm.com>
18904 * machmode.h (smallest_mode_for_size): Fix formatting.
18905 (smallest_int_mode_for_size): New function.
18906 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
18907 instead of smallest_mode_for_size.
18908 * combine.c (make_extraction): Likewise.
18909 * config/arc/arc.c (arc_expand_movmem): Likewise.
18910 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
18911 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
18912 * config/s390/s390.c (s390_expand_insv): Likewise.
18913 * config/sparc/sparc.c (assign_int_registers): Likewise.
18914 * config/spu/spu.c (spu_function_value): Likewise.
18915 (spu_function_arg): Likewise.
18916 * coverage.c (get_gcov_type): Likewise.
18917 (get_gcov_unsigned_t): Likewise.
18918 * dse.c (find_shift_sequence): Likewise.
18919 * expmed.c (store_bit_field_1): Likewise.
18920 * expr.c (convert_move): Likewise.
18921 (store_field): Likewise.
18922 * internal-fn.c (expand_arith_overflow): Likewise.
18923 * optabs-query.c (get_best_extraction_insn): Likewise.
18924 * optabs.c (expand_twoval_binop_libfunc): Likewise.
18925 * stor-layout.c (layout_type): Likewise.
18926 (initialize_sizetypes): Likewise.
18927 * targhooks.c (default_get_mask_mode): Likewise.
18928 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
18930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18931 Alan Hayward <alan.hayward@arm.com>
18932 David Sherwood <david.sherwood@arm.com>
18934 * machmode.h (opt_mode::else_blk): New function.
18935 (int_mode_for_mode): Declare.
18936 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
18937 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
18939 * cfgexpand.c (expand_debug_expr): Likewise.
18940 * combine.c (gen_lowpart_or_truncate): Likewise.
18941 (gen_lowpart_for_combine): Likewise.
18942 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
18943 * config/avr/avr.c (avr_to_int_mode): Likewise.
18944 (avr_out_plus_1): Likewise.
18945 (avr_out_plus): Likewise.
18946 (avr_out_round): Likewise.
18947 * config/i386/i386.c (ix86_split_to_parts): Likewise.
18948 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
18949 (s390_expand_vcond): Likewise.
18950 * config/spu/spu.c (spu_split_immediate): Likewise.
18951 (spu_expand_mov): Likewise.
18952 * dse.c (get_stored_val): Likewise.
18953 * expmed.c (store_bit_field_1): Likewise.
18954 (convert_extracted_bit_field): Use int_mode_for_mode instead of
18956 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
18957 (extract_low_bits): Likewise.
18958 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
18959 handling rather than repeating the check.
18960 (emit_group_store): Likewise.
18961 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
18962 * optabs.c (expand_absneg_bit): Likewise.
18963 (expand_copysign_absneg): Likewise.
18964 (expand_copysign_bit): Likewise.
18965 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
18966 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
18967 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
18968 * var-tracking.c (prepare_call_arguments): Likewise.
18969 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
18970 int_mode_for_mode instead of mode_for_size.
18971 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
18973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18974 Alan Hayward <alan.hayward@arm.com>
18975 David Sherwood <david.sherwood@arm.com>
18977 * machmode.h (int_mode_for_size): New function.
18978 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
18979 instead of mode_for_size.
18980 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
18982 * combine.c (expand_field_assignment): Use int_mode_for_size
18983 instead of mode_for_size.
18984 (make_extraction): Likewise.
18985 (simplify_shift_const_1): Likewise.
18986 (simplify_comparison): Likewise.
18987 * dojump.c (do_jump): Likewise.
18988 * dwarf2out.c (mem_loc_descriptor): Likewise.
18989 * emit-rtl.c (init_derived_machine_modes): Likewise.
18990 * expmed.c (flip_storage_order): Likewise.
18991 (convert_extracted_bit_field): Likewise.
18992 * expr.c (copy_blkmode_from_reg): Likewise.
18993 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
18994 * internal-fn.c (expand_mul_overflow): Likewise.
18995 * lower-subreg.c (simple_move): Likewise.
18996 * optabs-libfuncs.c (init_optabs): Likewise.
18997 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18998 * tree.c (vector_type_mode): Likewise.
18999 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19000 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19001 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19002 * tree-vect-stmts.c (vectorizable_load): Likewise.
19003 (vectorizable_store): Likewise.
19005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19006 Alan Hayward <alan.hayward@arm.com>
19007 David Sherwood <david.sherwood@arm.com>
19009 * coretypes.h (pod_mode): New type.
19010 (scalar_int_mode_pod): New typedef.
19011 * machmode.h (pod_mode): New class.
19012 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19013 * genmodes.c (emit_mode_int_n): Update accordingly.
19014 * lower-subreg.h (target_lower_subreg): Change type to
19015 scalar_int_mode_pod.
19016 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19017 scalar_int_mode_pod.
19019 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19020 Alan Hayward <alan.hayward@arm.com>
19021 David Sherwood <david.sherwood@arm.com>
19023 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19024 machine_mode to scalar_int_mode.
19025 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19026 (rs6000_option_override_internal): Remove cast to int.
19027 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19028 machine_mode to scalar_int_mode.
19029 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19030 (rs6000_option_override_internal): Remove cast to int.
19031 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19032 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19034 * config/s390/s390.c (s390_expand_builtin): Likewise.
19035 * coretypes.h (scalar_int_mode): New type.
19036 (opt_scalar_int_mode): New typedef.
19037 * machmode.h (scalar_int_mode): New class.
19038 (scalar_int_mode::includes_p): New function.
19039 (byte_mode): Change type to scalar_int_mode.
19040 (word_mode): Likewise.
19041 (ptr_mode): Likewise.
19042 * emit-rtl.c (byte_mode): Likewise.
19043 (word_mode): Likewise.
19044 (ptr_mode): Likewise.
19045 (init_derived_machine_modes): Update accordingly.
19046 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19047 and MODE_PARTIAL_INT.
19048 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19049 opt_scalar_int_mode.
19051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19052 Alan Hayward <alan.hayward@arm.com>
19053 David Sherwood <david.sherwood@arm.com>
19055 * target.def (libgcc_floating_mode_supported_p): Take a
19057 * doc/tm.texi: Regenerate.
19058 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19060 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19061 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19064 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19065 Alan Hayward <alan.hayward@arm.com>
19066 David Sherwood <david.sherwood@arm.com>
19068 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19069 * doc/tm.texi: Regenerate.
19070 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19071 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19072 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19073 * targhooks.h (default_floatn_mode): Likewise.
19074 * targhooks.c (default_floatn_mode): Likewise.
19075 * tree.c (build_common_tree_nodes): Update accordingly.
19077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19078 Alan Hayward <alan.hayward@arm.com>
19079 David Sherwood <david.sherwood@arm.com>
19081 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19082 (mode_iterator::iterate_p): Likewise.
19083 (mode_iterator::get_wider): Likewise.
19084 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19087 Alan Hayward <alan.hayward@arm.com>
19088 David Sherwood <david.sherwood@arm.com>
19090 * coretypes.h (opt_scalar_float_mode): New typedef.
19091 * machmode.h (float_mode_for_size): New function.
19092 * emit-rtl.c (double_mode): Delete.
19093 (init_emit_once): Use float_mode_for_size.
19094 * stor-layout.c (layout_type): Likewise.
19095 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19098 Alan Hayward <alan.hayward@arm.com>
19099 David Sherwood <david.sherwood@arm.com>
19101 * output.h (assemble_real): Take a scalar_float_mode.
19102 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19103 * config/arm/arm.md (consttable_4): Likewise.
19104 (consttable_8): Likewise.
19105 (consttable_16): Likewise.
19106 * config/mips/mips.md (consttable_float): Likewise.
19107 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19108 * varasm.c (assemble_real): Take a scalar_float_mode.
19109 (output_constant_pool_2): Update accordingly.
19110 (output_constant): Likewise.
19112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19113 Alan Hayward <alan.hayward@arm.com>
19114 David Sherwood <david.sherwood@arm.com>
19116 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19117 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19118 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19119 (native_encode_real): Likewise.
19120 (native_interpret_real): Likewise.
19121 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19122 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19124 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19125 Alan Hayward <alan.hayward@arm.com>
19126 David Sherwood <david.sherwood@arm.com>
19128 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19129 <scalar_float_mode>. Simplify.
19130 (gen_extend_conv_libfunc): Likewise.
19132 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19133 Alan Hayward <alan.hayward@arm.com>
19134 David Sherwood <david.sherwood@arm.com>
19136 * coretypes.h (scalar_float_mode): New type.
19137 * machmode.h (mode_traits::from_int): Use machine_mode if
19138 USE_ENUM_MODES is defined.
19139 (is_a): New function.
19141 (dyn_cast): Likewise.
19142 (scalar_float_mode): New class.
19143 (scalar_float_mode::includes_p): New function.
19144 (is_float_mode): Likewise.
19145 * gdbhooks.py (MachineModePrinter): New class.
19146 (build_pretty_printer): Use it for scalar_float_mode.
19147 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19148 (format_helper::format_helper): Turn into a template.
19149 * genmodes.c (get_mode_class): New function.
19150 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19151 or machine_mode if none.
19152 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19153 as_a <scalar_float_mode>.
19154 * dwarf2out.c (mem_loc_descriptor): Likewise.
19155 (insert_float): Likewise.
19156 (add_const_value_attribute): Likewise.
19157 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19158 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19159 (expand_unop): Update accordingly.
19160 (expand_abs_nojump): Likewise.
19161 (expand_copysign_absneg): Take a scalar_float_mode.
19162 (expand_copysign_bit): Likewise.
19163 (expand_copysign): Update accordingly.
19165 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19166 Alan Hayward <alan.hayward@arm.com>
19167 David Sherwood <david.sherwood@arm.com>
19169 * coretypes.h (opt_mode): New class.
19170 * machmode.h (opt_mode): Likewise.
19171 (opt_mode::else_void): New function.
19172 (opt_mode::require): Likewise.
19173 (opt_mode::exists): Likewise.
19174 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19175 (GET_MODE_2XWIDER_MODE): Likewise.
19176 (mode_iterator::get_wider): Update accordingly.
19177 (mode_iterator::get_2xwider): Likewise.
19178 (mode_iterator::get_known_wider): Likewise, turning into a template.
19179 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19180 forcing a wider mode to exist.
19181 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19182 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19183 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19184 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19185 * lower-subreg.c (init_lower_subreg): Likewise.
19186 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19187 on the final iteration.
19188 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19189 a wider mode exists before asking for a move pattern.
19190 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19191 forcing a wider mode to exist.
19192 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19193 returning false if no such mode exists.
19194 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19195 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19196 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19197 Avoid checking for a MODE_INT if we already know the mode is not a
19199 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19200 forcing a wider mode to exist.
19201 (expmed_mult_highpart_optab): Likewise.
19202 (expmed_mult_highpart): Likewise.
19203 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19205 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19206 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19207 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19208 * internal-fn.c (expand_mul_overflow): Update use of
19209 GET_MODE_2XWIDER_MODE.
19210 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19211 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19212 GET_MODE_WIDER_MODE.
19213 (convert_plusminus_to_widen): Likewise.
19214 * tree-switch-conversion.c (array_value_type): Likewise.
19215 * var-tracking.c (emit_note_insn_var_location): Likewise.
19216 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19217 Return false inside rather than outside the loop if no wider mode
19219 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19220 and GET_MODE_2XWIDER_MODE
19221 (can_compare_p): Use else_void.
19222 * gdbhooks.py (OptMachineModePrinter): New class.
19223 (build_pretty_printer): Use it for opt_mode.
19225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19226 Alan Hayward <alan.hayward@arm.com>
19227 David Sherwood <david.sherwood@arm.com>
19229 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19230 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19232 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19233 Alan Hayward <alan.hayward@arm.com>
19234 David Sherwood <david.sherwood@arm.com>
19236 * machmode.h (mode_traits): New structure.
19237 (get_narrowest_mode): New function.
19238 (mode_iterator::start): Likewise.
19239 (mode_iterator::iterate_p): Likewise.
19240 (mode_iterator::get_wider): Likewise.
19241 (mode_iterator::get_known_wider): Likewise.
19242 (mode_iterator::get_2xwider): Likewise.
19243 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19244 (FOR_EACH_MODE): Likewise.
19245 (FOR_EACH_MODE_FROM): Likewise.
19246 (FOR_EACH_MODE_UNTIL): Likewise.
19247 (FOR_EACH_WIDER_MODE): Likewise.
19248 (FOR_EACH_2XWIDER_MODE): Likewise.
19249 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19250 * combine.c (simplify_comparison): Likewise
19251 * config/i386/i386.c (type_natural_mode): Likewise.
19252 * cse.c (cse_insn): Likewise.
19253 * dse.c (find_shift_sequence): Likewise.
19254 * emit-rtl.c (init_derived_machine_modes): Likewise.
19255 (init_emit_once): Likewise.
19256 * explow.c (hard_function_value): Likewise.
19257 * expmed.c (extract_fixed_bit_field_1): Likewise.
19258 (extract_bit_field_1): Likewise.
19259 (expand_divmod): Likewise.
19260 (emit_store_flag_1): Likewise.
19261 * expr.c (init_expr_target): Likewise.
19262 (convert_move): Likewise.
19263 (alignment_for_piecewise_move): Likewise.
19264 (widest_int_mode_for_size): Likewise.
19265 (emit_block_move_via_movmem): Likewise.
19266 (copy_blkmode_to_reg): Likewise.
19267 (set_storage_via_setmem): Likewise.
19268 (compress_float_constant): Likewise.
19269 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19270 * optabs-query.c (get_best_extraction_insn): Likewise.
19271 * optabs.c (expand_binop): Likewise.
19272 (expand_twoval_unop): Likewise.
19273 (expand_twoval_binop): Likewise.
19274 (widen_leading): Likewise.
19275 (widen_bswap): Likewise.
19276 (expand_parity): Likewise.
19277 (expand_unop): Likewise.
19278 (prepare_cmp_insn): Likewise.
19279 (prepare_float_lib_cmp): Likewise.
19280 (expand_float): Likewise.
19281 (expand_fix): Likewise.
19282 (expand_sfix_optab): Likewise.
19283 * postreload.c (move2add_use_add2_insn): Likewise.
19284 * reg-stack.c (reg_to_stack): Likewise.
19285 * reginfo.c (choose_hard_reg_mode): Likewise.
19286 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19287 * stor-layout.c (mode_for_size): Likewise.
19288 (smallest_mode_for_size): Likewise.
19289 (mode_for_vector): Likewise.
19290 (finish_bitfield_representative): Likewise.
19291 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19292 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19293 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19294 * var-tracking.c (prepare_call_arguments): Likewise.
19296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19297 Alan Hayward <alan.hayward@arm.com>
19298 David Sherwood <david.sherwood@arm.com>
19300 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19301 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19302 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19303 * machmode.h (mode_size): Move earlier in file.
19304 (mode_precision): Likewise.
19305 (mode_inner): Likewise.
19306 (mode_nunits): Likewise.
19307 (mode_unit_size): Likewise.
19308 (unit_unit_precision): Likewise.
19309 (mode_wider): Likewise.
19310 (mode_2xwider): Likewise.
19311 (machine_mode): New class.
19312 (mode_to_bytes): New function.
19313 (mode_to_bits): Likewise.
19314 (mode_to_precision): Likewise.
19315 (mode_to_inner): Likewise.
19316 (mode_to_unit_size): Likewise.
19317 (mode_to_unit_precision): Likewise.
19318 (mode_to_nunits): Likewise.
19319 (GET_MODE_SIZE): Use mode_to_bytes.
19320 (GET_MODE_BITSIZE): Use mode_to_bits.
19321 (GET_MODE_PRECISION): Use mode_to_precision.
19322 (GET_MODE_INNER): Use mode_to_inner.
19323 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19324 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19325 (GET_MODE_NUNITS): Use mode_to_nunits.
19326 * system.h (ALWAYS_INLINE): New macro.
19327 * config/powerpcspe/powerpcspe-c.c
19328 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19329 int for arg1_mode and arg2_mode.
19331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19332 Alan Hayward <alan.hayward@arm.com>
19333 David Sherwood <david.sherwood@arm.com>
19335 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19336 Prefix mode names with E_ in case statements.
19337 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19338 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19339 (aarch64_split_simd_move): Likewise.
19340 (aarch64_gen_storewb_pair): Likewise.
19341 (aarch64_gen_loadwb_pair): Likewise.
19342 (aarch64_gen_store_pair): Likewise.
19343 (aarch64_gen_load_pair): Likewise.
19344 (aarch64_get_condition_code_1): Likewise.
19345 (aarch64_constant_pool_reload_icode): Likewise.
19346 (get_rsqrte_type): Likewise.
19347 (get_rsqrts_type): Likewise.
19348 (get_recpe_type): Likewise.
19349 (get_recps_type): Likewise.
19350 (aarch64_gimplify_va_arg_expr): Likewise.
19351 (aarch64_simd_container_mode): Likewise.
19352 (aarch64_emit_load_exclusive): Likewise.
19353 (aarch64_emit_store_exclusive): Likewise.
19354 (aarch64_expand_compare_and_swap): Likewise.
19355 (aarch64_gen_atomic_cas): Likewise.
19356 (aarch64_emit_bic): Likewise.
19357 (aarch64_emit_atomic_swap): Likewise.
19358 (aarch64_emit_atomic_load_op): Likewise.
19359 (aarch64_evpc_trn): Likewise.
19360 (aarch64_evpc_uzp): Likewise.
19361 (aarch64_evpc_zip): Likewise.
19362 (aarch64_evpc_ext): Likewise.
19363 (aarch64_evpc_rev): Likewise.
19364 (aarch64_evpc_dup): Likewise.
19365 (aarch64_gen_ccmp_first): Likewise.
19366 (aarch64_gen_ccmp_next): Likewise.
19367 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19368 (alpha_emit_xfloating_libcall): Likewise.
19369 (emit_insxl): Likewise.
19370 (alpha_arg_type): Likewise.
19371 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19372 (arc_preferred_simd_mode): Likewise.
19373 (arc_secondary_reload): Likewise.
19374 (get_arc_condition_code): Likewise.
19375 (arc_print_operand): Likewise.
19376 (arc_legitimate_constant_p): Likewise.
19377 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19378 * config/arc/arc.md (casesi_load): Likewise.
19379 (casesi_compact_jump): Likewise.
19380 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19381 (cc_use_register): Likewise.
19382 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19383 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19384 (arm_init_iwmmxt_builtins): Likewise.
19385 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19386 (neon_expand_vector_init): Likewise.
19387 (arm_attr_length_move_neon): Likewise.
19388 (maybe_get_arm_condition_code): Likewise.
19389 (arm_emit_vector_const): Likewise.
19390 (arm_preferred_simd_mode): Likewise.
19391 (arm_output_iwmmxt_tinsr): Likewise.
19392 (thumb1_output_casesi): Likewise.
19393 (thumb2_output_casesi): Likewise.
19394 (arm_emit_load_exclusive): Likewise.
19395 (arm_emit_store_exclusive): Likewise.
19396 (arm_expand_compare_and_swap): Likewise.
19397 (arm_evpc_neon_vuzp): Likewise.
19398 (arm_evpc_neon_vzip): Likewise.
19399 (arm_evpc_neon_vrev): Likewise.
19400 (arm_evpc_neon_vtrn): Likewise.
19401 (arm_evpc_neon_vext): Likewise.
19402 (arm_validize_comparison): Likewise.
19403 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
19404 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
19405 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
19406 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
19407 (c6x_preferred_simd_mode): Likewise.
19408 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
19409 (epiphany_rtx_costs): Likewise.
19410 * config/epiphany/predicates.md (proper_comparison_operator):
19412 * config/frv/frv.c (condexec_memory_operand): Likewise.
19413 (frv_emit_move): Likewise.
19414 (output_move_single): Likewise.
19415 (output_condmove_single): Likewise.
19416 (frv_hard_regno_mode_ok): Likewise.
19417 (frv_matching_accg_mode): Likewise.
19418 * config/h8300/h8300.c (split_adds_subs): Likewise.
19419 (h8300_rtx_costs): Likewise.
19420 (h8300_print_operand): Likewise.
19421 (compute_mov_length): Likewise.
19422 (output_logical_op): Likewise.
19423 (compute_logical_op_length): Likewise.
19424 (compute_logical_op_cc): Likewise.
19425 (h8300_shift_needs_scratch_p): Likewise.
19426 (output_a_shift): Likewise.
19427 (compute_a_shift_length): Likewise.
19428 (compute_a_shift_cc): Likewise.
19429 (expand_a_rotate): Likewise.
19430 (output_a_rotate): Likewise.
19431 * config/i386/i386.c (classify_argument): Likewise.
19432 (function_arg_advance_32): Likewise.
19433 (function_arg_32): Likewise.
19434 (function_arg_64): Likewise.
19435 (function_value_64): Likewise.
19436 (ix86_gimplify_va_arg): Likewise.
19437 (ix86_legitimate_constant_p): Likewise.
19438 (put_condition_code): Likewise.
19439 (split_double_mode): Likewise.
19440 (ix86_avx256_split_vector_move_misalign): Likewise.
19441 (ix86_expand_vector_logical_operator): Likewise.
19442 (ix86_split_idivmod): Likewise.
19443 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
19444 (ix86_build_const_vector): Likewise.
19445 (ix86_build_signbit_mask): Likewise.
19446 (ix86_match_ccmode): Likewise.
19447 (ix86_cc_modes_compatible): Likewise.
19448 (ix86_expand_branch): Likewise.
19449 (ix86_expand_sse_cmp): Likewise.
19450 (ix86_expand_sse_movcc): Likewise.
19451 (ix86_expand_int_sse_cmp): Likewise.
19452 (ix86_expand_vec_perm_vpermi2): Likewise.
19453 (ix86_expand_vec_perm): Likewise.
19454 (ix86_expand_sse_unpack): Likewise.
19455 (ix86_expand_int_addcc): Likewise.
19456 (ix86_split_to_parts): Likewise.
19457 (ix86_vectorize_builtin_gather): Likewise.
19458 (ix86_vectorize_builtin_scatter): Likewise.
19459 (avx_vpermilp_parallel): Likewise.
19460 (inline_memory_move_cost): Likewise.
19461 (ix86_tieable_integer_mode_p): Likewise.
19462 (x86_maybe_negate_const_int): Likewise.
19463 (ix86_expand_vector_init_duplicate): Likewise.
19464 (ix86_expand_vector_init_one_nonzero): Likewise.
19465 (ix86_expand_vector_init_one_var): Likewise.
19466 (ix86_expand_vector_init_concat): Likewise.
19467 (ix86_expand_vector_init_interleave): Likewise.
19468 (ix86_expand_vector_init_general): Likewise.
19469 (ix86_expand_vector_set): Likewise.
19470 (ix86_expand_vector_extract): Likewise.
19471 (emit_reduc_half): Likewise.
19472 (ix86_emit_i387_round): Likewise.
19473 (ix86_mangle_type): Likewise.
19474 (ix86_expand_round_sse4): Likewise.
19475 (expand_vec_perm_blend): Likewise.
19476 (canonicalize_vector_int_perm): Likewise.
19477 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
19478 (expand_vec_perm_1): Likewise.
19479 (expand_vec_perm_interleave3): Likewise.
19480 (expand_vec_perm_even_odd_pack): Likewise.
19481 (expand_vec_perm_even_odd_1): Likewise.
19482 (expand_vec_perm_broadcast_1): Likewise.
19483 (ix86_vectorize_vec_perm_const_ok): Likewise.
19484 (ix86_expand_vecop_qihi): Likewise.
19485 (ix86_expand_mul_widen_hilo): Likewise.
19486 (ix86_expand_sse2_abs): Likewise.
19487 (ix86_expand_pextr): Likewise.
19488 (ix86_expand_pinsr): Likewise.
19489 (ix86_preferred_simd_mode): Likewise.
19490 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
19491 * config/i386/sse.md (*andnot<mode>3): Likewise.
19492 (<mask_codefor><code><mode>3<mask_name>): Likewise.
19493 (*<code><mode>3): Likewise.
19494 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
19495 (ia64_expand_atomic_op): Likewise.
19496 (ia64_arg_type): Likewise.
19497 (ia64_mode_to_int): Likewise.
19498 (ia64_scalar_mode_supported_p): Likewise.
19499 (ia64_vector_mode_supported_p): Likewise.
19500 (expand_vec_perm_broadcast): Likewise.
19501 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
19502 (iq2000_function_arg_advance): Likewise.
19503 (iq2000_function_arg): Likewise.
19504 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
19505 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
19506 (m68k_libcall_value): Likewise.
19507 (m68k_function_value): Likewise.
19508 (sched_attr_op_type): Likewise.
19509 * config/mcore/mcore.c (mcore_output_move): Likewise.
19510 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
19512 (microblaze_function_arg): Likewise.
19513 * config/mips/mips.c (mips16_build_call_stub): Likewise.
19514 (mips_print_operand): Likewise.
19515 (mips_mode_ok_for_mov_fmt_p): Likewise.
19516 (mips_vector_mode_supported_p): Likewise.
19517 (mips_preferred_simd_mode): Likewise.
19518 (mips_expand_vpc_loongson_even_odd): Likewise.
19519 (mips_expand_vec_unpack): Likewise.
19520 (mips_expand_vi_broadcast): Likewise.
19521 (mips_expand_vector_init): Likewise.
19522 (mips_expand_vec_reduc): Likewise.
19523 (mips_expand_msa_cmp): Likewise.
19524 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
19525 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
19526 (cc_flags_for_mode): Likewise.
19527 * config/msp430/msp430.c (msp430_print_operand): Likewise.
19528 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
19529 (nds32_output_casesi_pc_relative): Likewise.
19530 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19531 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
19532 (nvptx_gen_unpack): Likewise.
19533 (nvptx_gen_pack): Likewise.
19534 (nvptx_gen_shuffle): Likewise.
19535 (nvptx_gen_wcast): Likewise.
19536 (nvptx_preferred_simd_mode): Likewise.
19537 * config/pa/pa.c (pa_secondary_reload): Likewise.
19538 * config/pa/predicates.md (base14_operand): Likewise.
19539 * config/powerpcspe/powerpcspe-c.c
19540 (altivec_resolve_overloaded_builtin): Likewise.
19541 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
19543 (rs6000_preferred_simd_mode): Likewise.
19544 (output_vec_const_move): Likewise.
19545 (rs6000_expand_vector_extract): Likewise.
19546 (rs6000_split_vec_extract_var): Likewise.
19547 (reg_offset_addressing_ok_p): Likewise.
19548 (rs6000_legitimate_offset_address_p): Likewise.
19549 (rs6000_legitimize_address): Likewise.
19550 (rs6000_emit_set_const): Likewise.
19551 (rs6000_const_vec): Likewise.
19552 (rs6000_emit_move): Likewise.
19553 (spe_build_register_parallel): Likewise.
19554 (rs6000_darwin64_record_arg_recurse): Likewise.
19555 (swap_selector_for_mode): Likewise.
19556 (spe_init_builtins): Likewise.
19557 (paired_init_builtins): Likewise.
19558 (altivec_init_builtins): Likewise.
19559 (do_load_for_compare): Likewise.
19560 (rs6000_generate_compare): Likewise.
19561 (rs6000_expand_float128_convert): Likewise.
19562 (emit_load_locked): Likewise.
19563 (emit_store_conditional): Likewise.
19564 (rs6000_output_function_epilogue): Likewise.
19565 (rs6000_handle_altivec_attribute): Likewise.
19566 (rs6000_function_value): Likewise.
19567 (emit_fusion_gpr_load): Likewise.
19568 (emit_fusion_p9_load): Likewise.
19569 (emit_fusion_p9_store): Likewise.
19570 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
19571 (fusion_gpr_mem_load): Likewise.
19572 (fusion_addis_mem_combo_load): Likewise.
19573 (fusion_addis_mem_combo_store): Likewise.
19574 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
19575 (fusion_gpr_mem_load): Likewise.
19576 (fusion_addis_mem_combo_load): Likewise.
19577 (fusion_addis_mem_combo_store): Likewise.
19578 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19580 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
19581 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
19582 (rs6000_preferred_simd_mode): Likewise.
19583 (output_vec_const_move): Likewise.
19584 (rs6000_expand_vector_extract): Likewise.
19585 (rs6000_split_vec_extract_var): Likewise.
19586 (reg_offset_addressing_ok_p): Likewise.
19587 (rs6000_legitimate_offset_address_p): Likewise.
19588 (rs6000_legitimize_address): Likewise.
19589 (rs6000_emit_set_const): Likewise.
19590 (rs6000_const_vec): Likewise.
19591 (rs6000_emit_move): Likewise.
19592 (rs6000_darwin64_record_arg_recurse): Likewise.
19593 (swap_selector_for_mode): Likewise.
19594 (paired_init_builtins): Likewise.
19595 (altivec_init_builtins): Likewise.
19596 (rs6000_expand_float128_convert): Likewise.
19597 (emit_load_locked): Likewise.
19598 (emit_store_conditional): Likewise.
19599 (rs6000_output_function_epilogue): Likewise.
19600 (rs6000_handle_altivec_attribute): Likewise.
19601 (rs6000_function_value): Likewise.
19602 (emit_fusion_gpr_load): Likewise.
19603 (emit_fusion_p9_load): Likewise.
19604 (emit_fusion_p9_store): Likewise.
19605 * config/rx/rx.c (rx_gen_move_template): Likewise.
19606 (flags_from_mode): Likewise.
19607 * config/s390/predicates.md (s390_alc_comparison): Likewise.
19608 (s390_slb_comparison): Likewise.
19609 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
19610 (s390_vector_mode_supported_p): Likewise.
19611 (s390_cc_modes_compatible): Likewise.
19612 (s390_match_ccmode_set): Likewise.
19613 (s390_canonicalize_comparison): Likewise.
19614 (s390_emit_compare_and_swap): Likewise.
19615 (s390_branch_condition_mask): Likewise.
19616 (s390_rtx_costs): Likewise.
19617 (s390_secondary_reload): Likewise.
19618 (__SECONDARY_RELOAD_CASE): Likewise.
19619 (s390_expand_cs): Likewise.
19620 (s390_preferred_simd_mode): Likewise.
19621 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
19622 * config/sh/sh.c (sh_print_operand): Likewise.
19623 (dump_table): Likewise.
19624 (sh_secondary_reload): Likewise.
19625 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19626 * config/sh/sh.md (casesi_worker_1): Likewise.
19627 (casesi_worker_2): Likewise.
19628 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
19629 (fcc_comparison_operator): Likewise.
19630 * config/sparc/sparc.c (sparc_expand_move): Likewise.
19631 (emit_soft_tfmode_cvt): Likewise.
19632 (sparc_preferred_simd_mode): Likewise.
19633 (output_cbranch): Likewise.
19634 (sparc_print_operand): Likewise.
19635 (sparc_expand_vec_perm_bmask): Likewise.
19636 (vector_init_bshuffle): Likewise.
19637 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
19638 (spu_vector_mode_supported_p): Likewise.
19639 (spu_expand_insv): Likewise.
19640 (spu_emit_branch_or_set): Likewise.
19641 (spu_handle_vector_attribute): Likewise.
19642 (spu_builtin_splats): Likewise.
19643 (spu_builtin_extract): Likewise.
19644 (spu_builtin_promote): Likewise.
19645 (spu_expand_sign_extend): Likewise.
19646 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
19647 (tilegx_simd_int): Likewise.
19648 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
19649 (tilepro_simd_int): Likewise.
19650 * config/v850/v850.c (const_double_split): Likewise.
19651 (v850_print_operand): Likewise.
19652 (ep_memory_offset): Likewise.
19653 * config/vax/vax.c (vax_rtx_costs): Likewise.
19654 (vax_output_int_move): Likewise.
19655 (vax_output_int_add): Likewise.
19656 (vax_output_int_subtract): Likewise.
19657 * config/visium/predicates.md (visium_branch_operator): Likewise.
19658 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
19659 (visium_print_operand_address): Likewise.
19660 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19661 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
19662 (xtensa_expand_conditional_branch): Likewise.
19663 (xtensa_copy_incoming_a7): Likewise.
19664 (xtensa_output_literal): Likewise.
19665 * dfp.c (decimal_real_maxval): Likewise.
19666 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19668 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19669 Alan Hayward <alan.hayward@arm.com>
19670 David Sherwood <david.sherwood@arm.com>
19672 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
19673 (mode_nunits_inline): Likewise.
19674 (mode_inner_inline): Likewise.
19675 (mode_unit_size_inline): Likewise.
19676 (mode_unit_precision_inline): Likewise.
19677 (emit_insn_modes_h): Likewise. Also emit a #define of the
19679 (emit_mode_wider): Add an E_ prefix to mode names.
19680 (emit_mode_complex): Likewise.
19681 (emit_mode_inner): Likewise.
19682 (emit_mode_adjustments): Likewise.
19683 (emit_mode_int_n): Likewise.
19684 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
19685 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
19686 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
19687 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
19688 (CRC32_BUILTIN, ENTRY): Likewise.
19689 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
19690 (aarch64_pop_regs): Likewise.
19691 (aarch64_process_components): Likewise.
19692 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
19693 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
19694 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
19695 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
19696 * config/arm/arm.c (arm_init_libfuncs): Likewise.
19697 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
19699 * config/i386/i386-builtin.def (pcmpestr): Likewise.
19700 (pcmpistr): Likewise.
19701 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
19702 * config/mmix/mmix.c (mmix_output_condition): Likewise.
19703 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
19705 * config/rl78/rl78.c (mduc_regs): Likewise.
19706 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
19707 (htm_expand_builtin): Likewise.
19708 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
19709 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
19710 * config/xtensa/xtensa.c (print_operand): Likewise.
19711 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
19712 (NUM_MODE_VECTOR_INT): Likewise.
19713 * genoutput.c (null_operand): Likewise.
19714 (output_operand_data): Likewise.
19715 * genrecog.c (print_parameter_value): Likewise.
19716 * lra.c (debug_operand_data): Likewise.
19718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19720 * dwarf2out.c (scompare_loc_descriptor_wide)
19721 (scompare_loc_descriptor_narrow): New functions, split out from...
19722 (scompare_loc_descriptor): ...here.
19723 * expmed.c (emit_store_flag_int): New function, split out from...
19724 (emit_store_flag): ...here.
19726 2017-08-30 Richard Biener <rguenther@suse.de>
19728 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
19729 (dwarf2out_early_finish): Move setting of AT_pubnames from
19730 early debug output to early finish.
19732 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19734 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
19735 and -mdata-region to the assembler.
19737 2017-08-30 Richard Biener <rguenther@suse.de>
19739 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
19741 (gen_subprogram_die): Add DW_AT_object_pointer only early.
19742 (dwarf2out_early_global_decl): Only generate a DIE for the
19743 abstract origin if it doesn't already exist or is a declaration DIE.
19744 (resolve_addr): Do not add the linkage name twice when
19745 generating a stub DIE for the DW_TAG_GNU_call_site target.
19747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19749 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19750 Use machine_mode rather than int for arg1_mode.
19752 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19755 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
19756 that the second argument of the built-in functions to unpack
19757 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
19758 switch statement instead a lot of if statements.
19759 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
19760 Allow 64-bit values to be in Altivec registers as well as
19761 traditional floating point registers.
19762 (pack<mode>, FMOVE128_VSX iterator): Likewise.
19764 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
19766 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
19767 MAX_REGS_PER_ADDRESS == 1.
19769 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
19771 * config/i386/i386.opt (flag_fentry): Do not init to -1.
19772 * config/i386/i386.c (ix86_option_override_internal): Simplify
19773 setting of opts->x_flag_entry.
19775 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19776 Jakub Jelinek <jakub@redhat.com>
19777 Richard Biener <rguenther@suse.de>
19779 PR tree-optimization/81503
19780 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
19781 folded constant fits in the target type; reorder tests for clarity.
19783 2017-08-29 Martin Liska <mliska@suse.cz>
19785 * passes.def: Include pass_lower_switch.
19786 * stmt.c (dump_case_nodes): Remove and move to
19787 tree-switch-conversion.
19788 (case_values_threshold): Likewise.
19789 (expand_switch_as_decision_tree_p): Likewise.
19790 (emit_case_decision_tree): Likewise.
19791 (expand_case): Likewise.
19792 (balance_case_nodes): Likewise.
19793 (node_has_low_bound): Likewise.
19794 (node_has_high_bound): Likewise.
19795 (node_is_bounded): Likewise.
19796 (emit_case_nodes): Likewise.
19797 (struct simple_case_node): New struct.
19798 (add_case_node): Remove.
19799 (emit_case_dispatch_table): Use vector instead of case_list.
19800 (reset_out_edges_aux): Remove.
19801 (compute_cases_per_edge): Likewise.
19802 (expand_case): Build list of simple_case_node.
19803 (expand_sjlj_dispatch_table): Use it.
19804 * tree-switch-conversion.c (struct case_node): Moved from
19805 stmt.c and adjusted.
19806 (emit_case_nodes): Likewise.
19807 (node_has_low_bound): Likewise.
19808 (node_has_high_bound): Likewise.
19809 (node_is_bounded): Likewise.
19810 (case_values_threshold): Likewise.
19811 (reset_out_edges_aux): Likewise.
19812 (compute_cases_per_edge): Likewise.
19813 (add_case_node): Likewise.
19814 (dump_case_nodes): Likewise.
19815 (balance_case_nodes): Likewise.
19816 (expand_switch_as_decision_tree_p): Likewise.
19817 (emit_jump): Likewise.
19818 (emit_case_decision_tree): Likewise.
19819 (try_switch_expansion): Likewise.
19820 (do_jump_if_equal): Likewise.
19821 (emit_cmp_and_jump_insns): Likewise.
19822 (fix_phi_operands_for_edge): New function.
19823 (record_phi_operand_mapping): Likewise.
19824 (class pass_lower_switch): New pass.
19825 (pass_lower_switch::execute): New function.
19826 (make_pass_lower_switch): Likewise.
19827 (conditional_probability):
19828 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
19829 * tree-pass.h: Add make_pass_lower_switch.
19831 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19834 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
19837 2017-08-29 Richard Biener <rguenther@suse.de>
19839 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
19840 we do not add a DW_AT_inline attribute twice.
19841 (gen_subprogram_die): Remove code setting DW_AT_inline on
19842 DECL_ABSTRACT_P nodes.
19844 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
19846 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
19847 calls to internal functions.
19848 (gimplify_modify_expr): Likewise.
19849 * tree-call-cdce.c (use_internal_fn): Likewise.
19850 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
19851 (convert_to_divmod): Set the nothrow flag.
19852 * tree-if-conv.c (predicate_mem_writes): Likewise.
19853 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
19854 (vectorizable_call): Likewise.
19855 (vectorizable_store): Likewise.
19856 (vectorizable_load): Likewise.
19857 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
19858 (vect_recog_mask_conversion_pattern): Likewise.
19860 2017-08-29 Martin Liska <mliska@suse.cz>
19863 * gcc.c (driver_handle_option): Add new argument.
19864 * opts-common.c (handle_option): Pass
19865 target_option_override_hook.
19866 * opts-global.c (lang_handle_option): Add new option.
19867 (set_default_handlers): Add new argument.
19868 (decode_options): Likewise.
19869 * opts.c (target_handle_option): Likewise.
19870 (common_handle_option): Call target_option_override_hook.
19871 * opts.h (struct cl_option_handler_func): Add hook for
19872 target option override.
19873 (struct cl_option_handlers): Likewise.
19874 (set_default_handlers): Add new argument.
19875 (decode_options): Likewise.
19876 (common_handle_option): Likewise.
19877 (target_handle_option): Likewise.
19878 * toplev.c (toplev::main): Pass targetm.target_option.override
19881 2017-08-29 Richard Biener <rguenther@suse.de>
19882 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
19884 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
19885 life to the active subtree.
19887 2017-08-28 Jeff Law <law@redhat.com>
19889 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
19890 derive_equivalences.
19891 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
19892 Code moved into....
19893 (edge_info::derive_equivalences): New private member function
19895 * tree-ssa-dom.c (class edge_info): Changed from a struct
19896 to a class. Add ctor/dtor, methods and data members.
19897 (edge_info::edge_info): Renamed from allocate_edge_info.
19898 Initialize additional members.
19899 (edge_info::~edge_info): New.
19900 (free_dom_edge_info): Delete the edge info.
19901 (record_edge_info): Use new class & associated member functions.
19902 Tighten forms for testing for edge equivalences.
19903 (record_temporary_equivalences): Iterate over the simple
19904 equivalences rather than assuming there's only one per edge.
19905 (cprop_into_successor_phis): Iterate over the simple
19906 equivalences rather than assuming there's only one per edge.
19907 (optimize_stmt): Use operand_equal_p rather than pointer
19908 equality for mini-DSE code.
19910 2017-08-28 Nathan Sidwell <nathan@acm.org>
19912 * gcc.c (execute): Fold SIGPIPE handling into switch
19913 statement. Adjust internal error message.
19915 2017-08-28 Richard Biener <rguenther@suse.de>
19918 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
19919 Do nothing for removed DIEs.
19921 2017-08-28 Richard Biener <rguenther@suse.de>
19923 PR tree-optimization/81977
19924 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
19927 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
19930 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
19931 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
19932 using targetm.gen_mem_thread_fence.
19934 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
19937 * config/i386/i386.md (*<btsc><mode>): Change operand 2
19938 predicate to register_operand. Reorder operands.
19939 (*btr<mode>): Ditto.
19940 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
19941 (*btr<mode>_mask): Ditto.
19943 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
19945 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
19946 * config/rs6000/xmmintrin.h: New file.
19947 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
19949 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19952 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
19953 parameter and_insn and return it.
19954 (recombine_lvx_pattern): Insert a copy to ensure availability of
19955 the base register of the copied masking operation at the point of
19956 the instruction replacement.
19957 (recombine_stvx_pattern): Likewise.
19959 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
19961 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
19962 undocumented switches.
19963 (-mpower9-dform-vector): Likewise.
19964 (-mpower9-dform): Likewise.
19965 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19966 comments to delete references to -mpower9-dform* switches.
19967 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
19968 Delete reference to -mpower9-dform* switches, test for
19969 -mpower9-vector instead.
19970 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
19971 (OTHER_P9_VECTOR_MASKS): Likewise.
19972 (POWERPC_MASKS): Likewise.
19973 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
19974 tests against -mpower9-dform* to -mpower9-vector. Delete code
19975 that checked for -mpower9-dform* consistancy with other options.
19976 Add test for -mpower9-misc to enable other power9 switches.
19977 (rs6000_init_hard_regno_mode_ok): Likewise.
19978 (rs6000_option_override_internal): Likewise.
19979 (rs6000_emit_prologue): Likewise.
19980 (rs6000_emit_epilogue): Likewise.
19981 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
19982 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
19983 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
19985 (emit_fusion_p9_store): Likewise.
19986 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
19987 resetting these macros if the assembler does not support ISA 3.0
19989 (TARGET_P9_DFORM_VECTOR): Likewise.
19990 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
19991 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
19993 2017-08-25 Alan Modra <amodra@gmail.com>
19995 PR rtl-optimization/81747
19996 * cse.c (cse_extended_basic_block): Don't attempt to record
19997 equivalences for degenerate conditional branches that branch
19998 to their fall-through.
20000 2017-08-24 Martin Sebor <msebor@redhat.com>
20002 PR middle-end/81908
20003 * gimple-fold.c (size_must_be_zero_p): New function.
20004 (gimple_fold_builtin_memory_op): Call it.
20006 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20008 * config/rs6000/mm_malloc.h: New file.
20010 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20012 PR tree-optimization/81913
20013 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20014 analysis when either IVs in condition can wrap.
20016 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20018 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20019 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20021 2017-08-24 Richard Biener <rguenther@suse.de>
20024 * targhooks.c (default_target_can_inline_p): Properly
20025 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20026 is present and always compare.
20027 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20028 infer -mfpmath=sse from TARGET_SSE_P.
20029 (ix86_can_inline_p): Properly use target_option_default_node when
20030 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20032 2017-08-24 Richard Biener <rguenther@suse.de>
20035 * dwarf2out.c (output_die): Handle flag_generate_offload like
20037 (output_comp_unit): Likewise.
20038 (gen_array_type_die): Likewise.
20039 (dwarf2out_early_finish): Likewise.
20040 (note_variable_value_in_expr): Likewise.
20041 (dwarf2out_finish): Likewise. Adjust assert.
20042 * cgraphunit.c (symbol_table::compile): Move setting of
20043 flag_generate_offload earlier ...
20044 (symbol_table::finalize_compilation_unit): ... here, before
20045 early debug finalization.
20047 2017-08-24 Richard Biener <rguenther@suse.de>
20049 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20050 and ipa-fnsummary.h.
20051 (ix86_can_inline_p): When ix86_fpmath flags do not match
20052 check whether the callee uses FP math at all.
20054 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20056 PR middle-end/81931
20057 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20058 instead of TYPE_PRECISION.
20060 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20061 Alan Hayward <alan.hayward@arm.com>
20062 David Sherwood <david.sherwood@arm.com>
20064 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20066 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20068 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20070 (ix86_frame::stack_realign_allocate): New field.
20071 (struct machine_frame_state): Modify comments.
20072 (machine_frame_state::sp_realigned_fp_end): New field.
20073 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20074 layout calculation.
20075 (sp_valid_at): Add assertion to assure no attempt to access invalid
20076 offset of a realigned stack.
20077 (fp_valid_at): Likewise.
20078 (choose_baseaddr): Modify comments.
20079 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20080 ix86_expand_prologue.
20081 (ix86_expand_prologue): Modify stack realignment and allocation.
20082 (ix86_expand_epilogue): Modify comments.
20083 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20084 avx2_runtime, avx512f, and avx512f_runtime.
20086 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20088 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20089 (mstackrealign): Do not init to -1.
20090 * config/i386/i386.c (ix86_option_override_internal):
20091 Check opts_set, not opts when setting default value of
20092 opts->x_ix86_force_align_arg_pointer.
20094 2017-08-23 Richard Biener <rguenther@suse.de>
20096 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20097 lang_hooks.decl_printable_name.
20098 * print-rtl-function.c (print_rtx_function): Likewise.
20099 * tree-pretty-print.c (dump_function_header): Likewise.
20101 2017-08-23 Richard Biener <rguenther@suse.de>
20104 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20105 -g0 at compile-time.
20107 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20109 PR middle-end/19706
20110 * doc/sourcebuild.texi (Other hardware attributes):
20113 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20115 PR middle-end/19706
20116 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20117 Fix single-use check.
20119 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20121 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20123 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20125 * doc/install.texi: Modify to add more details on running selected
20128 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20130 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20131 is combined with -mabi=ms.
20132 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20135 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20137 PR tree-optimization/81488
20138 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20139 and cached_basis fields.
20140 (MAX_SPREAD): New constant.
20141 (alloc_cand_and_find_basis): Initialize new fields.
20142 (clear_visited): New function.
20143 (create_phi_basis_1): Rename from create_phi_basis, set visited
20144 and cached_basis fields.
20145 (create_phi_basis): New wrapper function.
20146 (phi_add_costs_1): Rename from phi_add_costs, add spread
20147 parameter, set visited field, short-circuit when limits reached.
20148 (phi_add_costs): New wrapper function.
20149 (record_phi_increments_1): Rename from record_phi_increments, set
20151 (record_phi_increments): New wrapper function.
20152 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20153 (phi_incr_cost): New wrapper function.
20154 (all_phi_incrs_profitable_1): Rename from
20155 all_phi_incrs_profitable, set visited field.
20156 (all_phi_incrs_profitable): New wrapper function.
20158 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20159 Alan Hayward <alan.hayward@arm.com>
20160 David Sherwood <david.sherwood@arm.com>
20162 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20163 that takes the outer and inner modes.
20164 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20165 comparison as the canonical test for a paradoxical subreg.
20166 * combine.c (simplify_set): Use paradoxical_subreg_p.
20167 (make_extraction): Likewise.
20168 (force_to_mode): Likewise.
20169 (rtx_equal_for_field_assignment_p): Likewise.
20170 (gen_lowpart_for_combine): Likewise.
20171 (simplify_comparison): Likewise.
20172 * cse.c (equiv_constant): Likewise.
20173 * expmed.c (store_bit_field_1): Likewise.
20174 * final.c (alter_subreg): Likewise.
20175 * fwprop.c (propagate_rtx): Likewise.
20176 (forward_propagate_subreg): Likewise.
20177 * ira-conflicts.c (ira_build_conflicts): Likewise.
20178 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20179 * lra-constraints.c (curr_insn_transform): Likewise.
20180 (split_reg): Likewise.
20181 * lra-eliminations.c (move_plus_up): Likewise.
20182 (lra_eliminate_regs_1): Likewise.
20183 * recog.c (general_operand): Likewise.
20184 * ree.c (combine_reaching_defs): Likewise.
20185 * reload.c (push_reload): Likewise.
20186 (find_reloads): Likewise.
20187 * reload1.c (elimination_effects): Likewise.
20188 (compute_reload_subreg_offset): Likewise.
20189 (choose_reload_regs): Likewise.
20190 * rtlanal.c (subreg_lsb_1): Likewise.
20191 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20192 (simplify_subreg): Likewise.
20193 * var-tracking.c (track_loc_p): Likewise.
20194 * emit-rtl.c (byte_lowpart_offset): Likewise.
20195 (paradoxical_subreg_p): Delete out-of-line definition.
20197 2017-08-22 Jeff Law <law@redhat.com>
20199 PR tree-optimization/81741
20200 PR tree-optimization/71947
20201 * tree-ssa-dom.c: Include tree-inline.h.
20202 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20203 equivalences if one is more expensive to compute than the other.
20204 * tree-ssa-scopedtables.h (class const_or_copies): Make
20205 record_const_or_copy_raw method private.
20206 (class avail_exprs_stack): New method simplify_binary_operation.
20207 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20208 avail_exprs_stack::simplify_binary_operation as needed.
20209 (avail_exprs_stack::simplify_binary_operation): New function.
20211 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20213 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20214 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20215 (DOT_SYMBOLS): Likewise.
20216 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20217 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20218 (RS6000_ABI_NAME): Likewise.
20219 (TARGET_CMODEL): Likewise.
20220 (TOC_SECTION_ASM_OP): Likewise.
20221 (SET_CMODEL): New macro.
20222 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20224 2017-08-22 Richard Biener <rguenther@suse.de>
20226 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20227 to allow for free-lang-data replacements similar to verify_type_variant.
20229 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20230 Alan Hayward <alan.hayward@arm.com>
20231 David Sherwood <david.sherwood@arm.com>
20233 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20234 VOIDmode for the LABEL_REF.
20236 2017-08-22 Richard Biener <rguenther@suse.de>
20238 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20240 2017-08-22 Richard Biener <rguenther@suse.de>
20242 * common.opt (feliminate-dwarf2-dups): Ignore.
20243 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20244 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20245 same_die_p_wrap, compute_section_prefix,
20246 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20247 (comdat_symbol_id, comdat_symbol_number): Likewise.
20248 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20250 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20251 (output_die): Mark unreachable path unreachable.
20252 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20253 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20254 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20255 (dwarf2out_early_finish): Likewise.
20257 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20259 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20261 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20264 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20265 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20266 (avr_attribute_table) <io, io_low, address>: Initialize
20267 .decl_required with true.
20269 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20271 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20272 undocumented debugging options.
20273 (-mvsx-scalar-double): Likewise.
20274 (-mallow-df-permute): Likewise.
20275 (-mvectorize-builtins): Likewise.
20276 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20277 (rs6000_builtin_vectorized_function): Likewise.
20278 (rs6000_builtin_md_vectorized_function): Likewise.
20279 (rs6000_opt_vars): Likewise.
20281 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20284 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20285 (*btrq_imm): Rename from *btrq.
20286 (*btcq_imm): Rename from *btcq.
20287 (btsc): New code attribute.
20288 (*<btsc><mode>): New insn pattern.
20289 (*btr<mode>): Ditto.
20290 (*<btsc><mode>_mask): New insn_and_split pattern.
20291 (*btr<mode>_mask): Ditto.
20293 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20294 Alan Hayward <alan.hayward@arm.com>
20295 David Sherwood <david.sherwood@arm.com>
20297 * function.c (pad_below): Simplify padding calculation.
20299 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20300 Alan Hayward <alan.hayward@arm.com>
20301 David Sherwood <david.sherwood@arm.com>
20303 * target.def (function_prologue): Remove frame size argument.
20304 (function_epilogue): Likewise.
20305 * doc/tm.texi: Regenerate.
20306 * final.c (final_start_function): Update call to function_prologue.
20307 (final_end_function): Update call to function_epilogue.
20308 (default_function_pro_epilogue): Remove frame size argument.
20309 * output.h (default_function_pro_epilogue): Likewise.
20310 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20311 (arm_output_function_prologue): Likewise.
20312 * config/frv/frv.c (frv_function_prologue): Likewise.
20313 (frv_function_epilogue): Likewise.
20314 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20315 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20316 (ia64_output_function_epilogue): Likewise.
20317 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20318 (m32r_output_function_epilogue): Likewise.
20319 * config/microblaze/microblaze.c (microblaze_function_prologue)
20320 (microblaze_function_epilogue): Likewise.
20321 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20322 (mips_output_function_epilogue): Likewise.
20323 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20324 (mmix_target_asm_function_epilogue): Likewise.
20325 * config/msp430/msp430.c (msp430_start_function): Likewise.
20326 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20327 (nds32_asm_function_epilogue): Likewise.
20328 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20329 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20330 (pa_output_function_epilogue): Likewise.
20331 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20332 (rs6000_output_function_epilogue): Likewise.
20333 * config/rl78/rl78.c (rl78_start_function): Likewise.
20334 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20335 (rs6000_output_function_epilogue): Likewise.
20336 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20337 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20338 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20339 (sparc_asm_function_epilogue): Likewise.
20341 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20343 * tree.h (type_has_mode_precision_p): New function.
20344 * convert.c (convert_to_integer_1): Use it.
20345 * expr.c (expand_expr_real_2): Likewise.
20346 (expand_expr_real_1): Likewise.
20347 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20348 * match.pd: Likewise.
20349 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20350 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20351 * tree-tailcall.c (process_assignment): Likewise.
20352 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20353 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20354 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20355 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20356 (vectorizable_assignment): Likewise.
20357 (vectorizable_shift): Likewise.
20358 (vectorizable_operation): Likewise.
20359 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20361 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20363 * match.pd: Add pow (C, x) simplification.
20365 2017-08-21 Richard Biener <rguenther@suse.de>
20367 PR tree-optimization/81900
20368 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20369 for blocks with abnormal predecessors.
20370 (compute_antic): Do not set visited flag prematurely.
20372 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20375 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20377 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20379 * stor-layout.h (vector_type_mode): Move to...
20380 * tree.h (vector_type_mode): ...here.
20381 * stor-layout.c (vector_type_mode): Move to...
20382 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20384 2017-08-21 Richard Biener <rguenther@suse.de>
20386 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20387 register_external_die hooks.
20388 (debug_false_tree_charstarstar_uhwistar): Declare.
20389 (debug_nothing_tree_charstar_uhwi): Likewise.
20390 * debug.c (do_nothing_debug_hooks): Adjust.
20391 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20392 (debug_nothing_tree_charstar_uhwi): Likewise.
20393 * dbxout.c (dbx_debug_hooks): Adjust.
20394 (xcoff_debug_hooks): Likewise.
20395 * sdbout.c (sdb_debug_hooks): Likewise.
20396 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20397 * dwarf2out.c (macinfo_label_base): New global.
20398 (dwarf2out_register_external_die): New function for the
20399 register_external_die hook.
20400 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
20401 (dwarf2_debug_hooks): Use them.
20402 (dwarf2_lineno_debug_hooks): Adjust.
20403 (struct die_struct): Add with_offset flag.
20404 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
20405 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
20406 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
20407 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
20408 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
20409 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
20410 defining section names for the early LTO debug variants.
20411 (reset_indirect_string): New helper.
20412 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
20413 (print_dw_val): Add support for offsetted symbol references.
20414 (get_ultimate_context): Split out from is_cxx.
20415 (is_cxx): Use get_ultimate_context.
20416 (is_fortran): Add decl overload.
20417 (compute_comp_unit_symbol): Split out worker from
20418 compute_section_prefix.
20419 (compute_section_prefix): Call compute_comp_unit_symbol and
20420 set comdat_type_p here.
20421 (output_die): Skip DIE symbol output for the LTO added one.
20422 Handle DIE symbol references with offset.
20423 (output_comp_unit): Guard section name mangling properly.
20424 For LTO debug sections emit a symbol at the section beginning
20425 which we use to refer to its DIEs.
20426 (add_abstract_origin_attribute): For DIEs registered via
20427 dwarf2out_register_external_die directly refer to the early
20428 DIE rather than indirectly through the shadow one we created.
20429 Remove obsolete call to dwarf2out_abstract_function for
20430 non-function/block origins.
20431 (gen_array_type_die): When generating early LTO debug do
20432 not emit DW_AT_string_length.
20433 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
20434 late when in LTO. As suggested place a gcc_unreachable for
20435 the DECL_ABSTRACT_P case.
20436 (gen_subprogram_die): Avoid another specification DIE
20437 for early built declarations/definitions for the late LTO case.
20438 (gen_variable_die): Add type references for late duplicated VLA dies
20440 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
20441 we have the abstract instance already.
20442 (process_scope_var): Adjust decl DIE contexts in LTO which
20443 first puts them in limbo.
20444 (gen_decl_die): Do not generate type DIEs late apart from
20445 types for VLAs or for decls we do not yet have a DIE. Do not
20446 call dwarf2out_abstract_function late.
20447 (dwarf2out_early_global_decl): Make sure to create DIEs
20448 for abstract instances of a decl first.
20449 (dwarf2out_late_global_decl): Adjust comment.
20450 (output_macinfo_op): With multiple macro sections use
20451 macinfo_label_base to distinguish labels.
20452 (output_macinfo): Likewise. Update macinfo_label_base.
20453 Pass in the line info label.
20454 (note_variable_value_in_expr): When generating LTO resolve
20455 all variable values here by generating DIEs as needed.
20456 (init_sections_and_labels): Add early LTO debug flag parameter
20457 and generate different sections and names if set. Add generation
20458 counter for the labels so we can have multiple of them.
20459 (reset_dies): Helper to allow DIEs to be output multiple times.
20460 (dwarf2out_finish): When outputting DIEs to the fat part of an
20461 LTO object first reset DIEs.
20462 (dwarf2out_early_finish): Output early DIEs when generating LTO.
20463 (modified_type_die): Check for decl_ultimate_origin being self
20465 (gen_type_die_with_usage): Likewise.
20466 (gen_typedef_die): Allow decl_ultimate_origin being self.
20467 (set_decl_abstract_flags): Remove.
20468 (set_block_abstract_flags): Likewise.
20469 (dwarf2out_abstract_function): Treat the early generated DIEs
20470 as the abstract copy and only add DW_AT_inline and
20471 DW_AT_artificial here and call set_decl_origin_self.
20472 If the DIE has an abstract origin don't do anything.
20473 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
20474 if we have none yet (Go fails to build one, PR78628).
20475 (variably_modified_type_p): Prevent endless recursion for Ada
20476 cyclic pointer types.
20477 * lto-streamer-in.c: Include debug.h.
20478 (dref_queue): New global.
20479 (lto_read_tree_1): Stream in DIE references.
20480 (lto_input_tree): Register DIE references.
20481 (input_function): Stream DECL_DEBUG_ARGS.
20482 * lto-streamer-out.c: Include debug.h.
20483 (lto_write_tree_1): Output DIE references.
20484 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
20485 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
20486 (output_function): Stream DECL_DEBUG_ARGS.
20487 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
20488 Stream DECL_ABSTRACT_ORIGIN.
20489 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
20490 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
20491 DECL_CONTEXT for file-scope decls.
20492 * lto-streamer.h (struct dref_entry): Declare.
20493 (dref_queue): Likewise.
20494 * cfgexpand.c (pass_expand::execute): Do not call the
20495 outlining_inline_function hook here.
20496 * lto-wrapper.c (debug_obj): New global.
20497 (tool_cleanup): Unlink it if required.
20498 (debug_objcopy): New function.
20499 (run_gcc): Handle early debug sections in the IL files by
20500 extracting them to separate files, partially linkin them and
20501 feeding the result back as result to the linker.
20502 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
20503 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
20504 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
20505 sections into a separate segment.
20506 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
20508 (darwin_asm_dwarf_section): Likewise.
20509 (darwin_asm_output_dwarf_offset): Likewise.
20510 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
20512 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20513 Alan Hayward <alan.hayward@arm.com>
20514 David Sherwood <david.sherwood@arm.com>
20516 * read-md.h (md_reader::record_potential_iterator_use): Replace
20517 pointer argument with an rtx and an index.
20518 * read-rtl.c (iterator_group::apply_iterator): Likewise.
20519 (apply_mode_iterator): Likewise.
20520 (apply_code_iterator): Likewise.
20521 (apply_int_iterator): Likewise.
20522 (apply_subst_iterator): Likewise.
20523 (record_iterator_use): Likewise.
20524 (record_attribute_use): Likewise.
20525 (md_reader::record_potential_iterator_use): Likewise. Update calls
20526 to record_iterator_use and apply_iterator.
20527 (iterator_use): Replace ptr with x and index.
20528 (attribute_use): Likewise.
20529 (apply_attribute_uses): Update calls to apply_iterator.
20530 (apply_iterators): Likewise. Update initialization of iterator_use.
20531 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
20532 and record_potential_iterator_use.
20533 (rtx_reader::read_rtx_operand): Likewise.
20535 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20536 Alan Hayward <alan.hayward@arm.com>
20537 David Sherwood <david.sherwood@arm.com>
20539 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
20542 2017-08-21 Richard Biener <rguenther@suse.de>
20544 PR middle-end/81884
20545 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
20546 at struct end conservatively when comparing common bases.
20548 2017-08-21 Richard Biener <rguenther@suse.de>
20550 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
20551 (mem_ref_in_stmt): Remove.
20552 (determine_max_movement): Use ref index to get at the reference.
20553 (invariantness_dom_walker::before_dom_children): Deal with
20554 lim data already initialized.
20555 (gather_mem_refs_stmt): Initialize lim data and record ref index.
20557 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20559 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
20560 (TARGET_ISA_ROUND): Ditto.
20561 (TARGET_ROUND): Ditto.
20562 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
20563 * config/i386/i386.md: Ditto.
20564 * config/i386/sse.md: Ditto.
20565 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
20566 with OPTION_MASK_ISA_SSE4_1.
20568 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
20571 * doc/extend.texi (x86 Built-in Functions): Correct the name of
20572 __builtin_ia32_lzcnt_u16.
20574 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
20577 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
20578 (rs6000_set_current_function): Rewrite function to use it.
20580 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
20583 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
20584 and TYPE_WARN_IF_NOT_ALIGN.
20585 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
20586 (handle_warn_if_not_align): New.
20587 (place_union_field): Call handle_warn_if_not_align.
20588 (place_field): Call handle_warn_if_not_align.
20589 Copy TYPE_WARN_IF_NOT_ALIGN.
20590 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
20591 (layout_type): Likewise.
20592 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
20594 (tree_decl_common): Add warn_if_not_align.
20595 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
20596 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
20597 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
20598 (DECL_WARN_IF_NOT_ALIGN): Likewise.
20599 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
20600 * doc/extend.texi: Document warn_if_not_aligned attribute.
20601 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
20603 2017-08-17 Martin Liska <mliska@suse.cz>
20606 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
20607 (get_data_dependence): Use it as pointer type.
20608 (distribute_loop): Likewise.
20610 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20612 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
20613 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
20614 (p8_vmrgow_<mode>_direct): New define_insn.
20615 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
20616 handle endianness for vmrgew and vmrgow permute patterns.
20618 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20620 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
20621 * config/rs6000/rs6000-cpus.def: Remove comment.
20622 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
20623 (POWERPC_MASKS): Likewise.
20624 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
20625 use of TARGET_VSX_TIMODE.
20626 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
20627 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
20628 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
20629 (rs6000_option_override_internal): Remove dead code.
20630 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
20631 (rs6000_legitimize_reload_address): Likewise.
20632 (rs6000_legitimate_address_p): Likewise.
20633 (rs6000_opt_masks): Delete "vsx-timode".
20634 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
20635 from function comment.
20636 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
20637 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
20638 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
20640 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
20641 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
20642 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
20643 (VSX_TI): Likewise.
20645 (define_peephole2): Likewise.
20647 2017-08-17 Martin Sebor <msebor@redhat.com>
20650 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
20651 past the end of an array.
20652 (test_pp_format): Add test cases.
20654 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
20656 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
20657 missing ECF_NOTHROW flags.
20659 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
20662 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
20663 operands residing in integer registers.
20664 (*vsx_le_perm_load_<mode>): Likewise.
20665 (*vsx_le_perm_store_<mode>): Likewise.
20666 (define_peephole2): Add peepholes to optimize the above.
20668 2017-08-17 Marek Polacek <polacek@redhat.com>
20670 PR middle-end/81814
20671 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
20672 to mimic what shorten_compare did. Change the return type to bool.
20673 (fold_cond_expr_with_comparison): Update call to
20674 operand_equal_for_comparison_p.
20675 (fold_ternary_loc): Likewise.
20677 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
20679 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
20681 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
20683 2017-08-17 Richard Biener <rguenther@suse.de>
20685 * tree-ssa-structalias.c (solve_graph): When propagating
20686 to successors update the graphs succ edges and avoid duplicate work.
20688 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
20691 * config/i386/i386.c (ix86_option_override_internal): Save target
20692 specific options after ix86_stack_protector_guard_reg was changed.
20694 2017-08-17 Richard Biener <rguenther@suse.de>
20696 PR tree-optimization/81827
20697 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
20698 (new_var_info): Initialize it conservatively.
20699 (get_call_vi): Mark register vars.
20700 (new_scalar_tmp_constraint_exp): Likewise.
20701 (handle_rhs_call): Likewise.
20702 (handle_const_call): Likewise.
20703 (create_function_info_for): Likewise.
20704 (solve_constraints): Sort varinfos to separate register from
20705 non-register vars to pack points-to solution bitmaps during
20708 2017-08-17 Marek Polacek <polacek@redhat.com>
20710 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
20712 2017-08-17 Richard Biener <rguenther@suse.de>
20714 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
20715 to true when overflow is undefined and we saturated the result.
20717 2017-08-17 Alan Modra <amodra@gmail.com>
20720 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
20721 Don't use store multiple if only one reg needs saving.
20722 (interesting_frame_related_regno): New function.
20723 (rs6000_frame_related): Don't emit frame info for regs that
20725 (rs6000_emit_epilogue): Likewise.
20727 2017-08-16 Nathan Sidwell <nathan@acm.org>
20729 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
20730 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
20731 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
20732 * tree.c (free_lang_data_in_type): Use else-if chain. Always
20733 clear TYPE_LANG_1. Remove obsolete member-function stripping.
20734 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
20735 (verify_type): Adjust for TYPE_BINFO move.
20736 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
20737 process TYPE_BINFO directly.
20738 (hash_tree): Likewise.
20739 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
20741 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
20744 2017-08-16 David Malcolm <dmalcolm@redhat.com>
20746 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
20748 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
20751 * config/i386/i386.md (*anddi_1_btr): Change predicates of
20752 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
20753 Add ix86_binary_operator_ok to insn constraint.
20754 (*iordi_1_bts): Ditto.
20755 (*xordi_1_btc): Ditto.
20756 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
20757 Update corresponding peephole2 pattern.
20761 2017-08-16 Bin Cheng <bin.cheng@arm.com>
20763 PR tree-optimization/81832
20764 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
20765 copy loop header which has IFN_LOOP_DIST_ALIAS call.
20767 2017-08-16 Marek Polacek <polacek@redhat.com>
20770 * fold-const.c (fold_indirect_ref_1): Restore original behavior
20771 regarding size_zero_node.
20773 2017-08-16 Martin Liska <mliska@suse.cz>
20776 * config.gcc: Respect previously set extra_objs in case
20779 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
20781 PR tree-optimization/81835
20782 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
20783 the phi SSA_NAME. Check that the condition in a COND_EXPR does
20784 not depend on the phi.
20786 2017-08-16 Alan Modra <amodra@gmail.com>
20788 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
20791 2017-08-16 Alan Modra <amodra@gmail.com>
20793 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
20794 (save_reg_p): ..into this. Update all callers.
20795 (first_reg_to_save): Simplify.
20797 2017-08-16 Alan Modra <amodra@gmail.com>
20799 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
20802 2017-08-15 Joseph Myers <joseph@codesourcery.com>
20806 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
20807 constant count if that count is less than 32.
20809 2017-08-15 Nathan Sidwell <nathan@acm.org>
20811 * gcc.c (execute): Emit friendlier message if inferior is killed
20812 by an external cause.
20814 2017-08-15 Richard Biener <rguenther@suse.de>
20816 PR tree-optimization/81790
20817 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
20818 CONSTRUCTORs from simplifying and VN.
20820 2017-08-14 Martin Sebor <msebor@redhat.com>
20822 * builtin-attrs.def: Add comments.
20824 2017-08-14 Martin Sebor <msebor@redhat.com>
20827 * doc/extend.texi (attribute nonstring): Document new attribute.
20829 2017-08-14 Martin Sebor <msebor@redhat.com>
20832 * tree-diagnostic.c (default_tree_printer): Handle %G.
20833 * gimple-pretty-print.h (percent_G_format): Declare new function.
20834 * gimple-pretty-print.c (percent_G_format): Define.
20835 * tree-pretty-print.c (percent_K_format): Add argument.
20837 2017-08-14 Martin Sebor <msebor@redhat.com>
20839 PR translation/79998
20840 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
20841 Remove a stray space.
20843 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
20846 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
20847 (*iordi_1_bts): Ditto.
20848 (*xordi_1_btc): Ditto.
20850 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20853 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
20854 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20856 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
20857 quoted strings, and make more translator-friendly.
20858 (darwin_rs6000_override_options): Likewise.
20859 (rs6000_option_override_internal): Likewise.
20860 (rs6000_return_in_memory): Fix overlong line.
20861 (init_cmulative_args): Use quoted strings, and make more
20862 translator-friendly.
20863 (rs6000_pass_by_reference): Fix overlong line.
20864 (def_builtin): Use quoted strings.
20865 (altivec_expand_predicate_builtin): Use quoted strings, and make
20866 more translator-friendly.
20867 (htm_expand_builtin): Use quoted strings.
20868 (cpu_expand_builtin): Use quoted strings, and make more
20869 translator-friendly.
20870 (altivec_expand_builtin): Likewise.
20871 (paired_expand_predicate_builtin): Likewise.
20872 (rs6000_invalid_builtin): Likewise.
20873 (builtin_function_type): Use quoted strings.
20874 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
20875 more translator-friendly.
20876 (rs6000_trampoline_init): Likewise.
20877 (rs6000_handle_altivec_attribute): Likewise.
20878 (rs6000_inner_target_options): Use quoted strings.
20879 (rs6000_disable_incompatible_switches): Likewise.
20880 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
20881 strings, and make more translator-friendly.
20882 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
20884 2017-08-14 Bin Cheng <bin.cheng@arm.com>
20886 PR tree-optimization/81799
20887 * tree-loop-distribution.c (version_loop_by_alias_check): Force
20888 cond_expr to simple gimple operand.
20890 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
20892 PR middle-end/46932
20893 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
20895 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
20899 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
20900 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
20901 TARGET_GASISR_PROLOGUES.
20902 * config/avr/avr.c (avr_option_override): Same.
20903 (avr_pass_pre_proep::execute): Same.
20905 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
20908 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
20909 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
20911 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
20913 * config/i386/i386.md (*load_tp_<mode>): Redefine as
20914 define_insn_and_split. Split to a memory load from 0 in
20915 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
20916 using PTR mode iterator.
20917 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
20918 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
20919 (*add_tp_<mode>): Redefine as define_insn_and_split.
20920 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
20921 address space. Merge with *add_tp_x32 using PTR mode iterator.
20922 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
20923 Split to an add with a memory load from 0 in
20924 DEFAULT_TLS_SEG_REG address space.
20926 2017-08-12 Andrew Pinski <apinski@cavium.com>
20928 * config/aarch64/aarch64-option-extensions.def (rdma):
20929 Fix feature string to what Linux prints out in /proc/cpuinfo.
20931 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
20934 * dwarf2out.c (modified_type_die): For C typedef types that have
20935 an ultimate origin, process the ultimate origin instead of the
20937 (gen_typedef_die): Assert that input DECLs have no ultimate
20939 (gen_type_die_with_usage): For typedef variants that have an
20940 ultimate origin, just call gen_decl_die on the original DECL.
20941 (process_scope_var): Avoid creating DIEs for local typedefs and
20942 concrete static variables.
20944 2017-08-12 Alan Modra <amodra@gmail.com>
20948 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
20949 match gnu-user.h startfile.
20950 (ENDFILE_LINUX_SPEC): Similarly.
20952 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
20955 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
20957 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
20959 2017-08-11 Tamar Christina <tamar.christina@arm.com>
20960 * config/aarch64/aarch64.md (mov<mode>): Change.
20961 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20962 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
20963 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
20965 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
20967 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
20968 for storage order barriers.
20970 2017-08-11 Martin Liska <mliska@suse.cz>
20973 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
20974 variables of void type.
20976 2017-08-11 Martin Liska <mliska@suse.cz>
20978 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
20979 TARGET_SUPPORTS_ALIASES.
20980 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
20981 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
20982 (optimize_weakref): Likewise.
20983 * symtab.c (symtab_node::noninterposable_alias): Likewise.
20984 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
20985 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
20987 2017-08-11 Martin Liska <mliska@suse.cz>
20990 * config/i386/i386.c (make_resolver_func): Do complete
20991 refactoring of the function.
20993 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
20996 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
20997 * config/i386/i386.c (ix86_stack_protect_guard): Use
20998 ix86_stack_protect_guard_symbol_str to generate varible declaration.
20999 * doc/invoke.texi (x86 Options): Document
21000 -mstack-protector-guard-symbol= option.
21002 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21004 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21005 * config/i386/i386.c (ix86_split_stack_guard): New function.
21006 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21007 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21008 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21009 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21010 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21011 (split_stack_space_check): Call ix86_split_stack_guard.
21013 2017-08-10 Martin Sebor <msebor@redhat.com>
21015 * print-tree.c (print_node): Print location using the established
21017 Replace spaces with colons.
21018 (debug_raw, debug): Ditto.
21020 2017-08-10 Martin Sebor <msebor@redhat.com>
21023 * pretty-print.c (pp_format): Correct the handling of %s precision.
21025 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21028 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21030 (ix86_finalize_stack_frame_flags): This. Also clear
21031 frame_pointer_needed if -fno-omit-frame-pointer is used without
21033 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21034 with ix86_finalize_stack_frame_flags.
21035 (ix86_expand_epilogue): Likewise.
21036 (ix86_expand_split_stack_prologue): Likewise.
21037 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21039 2017-08-10 Martin Liska <mliska@suse.cz>
21042 * c-attribs.c (handle_target_attribute):
21043 Report warning for an empty string argument of target attribute.
21045 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21048 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21050 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21051 or DECL_NONLOCAL labels.
21052 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21053 or DECL_NONLOCAL labels here.
21055 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21057 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21058 to indicate when early gimple folding has been disabled.
21059 (rs6000_gimple_fold_builtin): Add debug content.
21060 (rs6000_invalid_builtin): Fix whitespace.
21061 (rs6000_expand_builtin): Fix whitespace.
21062 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21064 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21067 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21068 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21070 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21072 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21073 (qdf24xx): Likewise.
21074 * config/aarch64/aarch64.md: Include falkor.md.
21075 * config/aarch64/falkor.md: New.
21077 2017-08-09 Marek Polacek <polacek@redhat.com>
21080 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21081 * diagnostic.c (emit_diagnostic): Add a comment.
21082 (emit_diagnostic_valist): New function.
21084 2017-08-09 Marek Polacek <polacek@redhat.com>
21087 * input.c (make_location): New overload.
21088 * input.h (make_location): Declare.
21090 2017-08-08 Alan Modra <amodra@gmail.com>
21091 H.J. Lu <hongjiu.lu@intel.com>
21094 * gcc.c (NO_PIE_SPEC): Delete.
21095 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21097 (LINK_PIE_SPEC): ..to here.
21098 (LINK_COMMAND_SPEC): Support -no-pie.
21099 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21100 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21102 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21103 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21104 (ENDFILE_CRTEND_SPEC): Similarly.
21106 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21109 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21110 (mstack-protector-guard-offset=): Ditto.
21111 * config/i386/i386.c (ix86_option_override): Handle
21112 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21114 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21115 ix86_stack_protect_guard_offset variables.
21116 (TARGET_STACK_PROTECT_GUARD): Always define.
21117 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21118 and -mstack-protector-guard-offset= options.
21120 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21122 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21123 boundary case for the last candidate.
21125 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21127 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21128 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21130 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21132 PR middle-end/19706
21133 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21134 * config/aarch64/aarch64-builtins.c
21135 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21136 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21137 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21139 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21140 Andrew Pinski <pinskia@gmail.com>
21142 PR middle-end/19706
21143 * internal-fn.def (XORSIGN): New.
21144 * optabs.def (xorsign_optab): New.
21145 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21146 (convert_expand_mult_copysign): New.
21147 (pass_optimize_widening_mul::execute): Call
21148 convert_expand_mult_copysign.
21150 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21152 PR tree-optimization/81354
21153 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21154 Insert on edges rather than explicitly creating landing pads.
21155 (analyze_candidates_and_replace): Commit edge inserts.
21157 2017-08-08 Richard Biener <rguenther@suse.de>
21159 PR middle-end/81719
21160 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21161 (expand_simple_operations): Also look through ADDR_EXPRs with
21162 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21164 2017-08-08 Richard Biener <rguenther@suse.de>
21166 PR tree-optimization/81723
21167 * tree-vect-slp.c (struct bst_traits): New hash traits.
21168 (bst_fail): New global.
21169 (vect_build_slp_tree_2): New worker, split out from ...
21170 (vect_build_slp_tree): ... this now wrapping it with using
21171 bst_fail set to cache SLP tree build fails. Properly handle
21173 (vect_analyze_slp_instance): Allocate and free bst_fail.
21175 2017-08-08 Martin Liska <mliska@suse.cz>
21178 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21179 LABEL_DECLs that can be from a different function.
21181 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21183 PR tree-optimization/81744
21184 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21185 loop's number of iterations.
21187 2017-08-08 Martin Liska <mliska@suse.cz>
21189 * asan.c: Include header files.
21190 * attribs.c (build_decl_attribute_variant): New function moved
21192 (build_type_attribute_qual_variant): Likewise.
21193 (cmp_attrib_identifiers): Likewise.
21194 (simple_cst_list_equal): Likewise.
21195 (omp_declare_simd_clauses_equal): Likewise.
21196 (attribute_value_equal): Likewise.
21197 (comp_type_attributes): Likewise.
21198 (build_type_attribute_variant): Likewise.
21199 (lookup_ident_attribute): Likewise.
21200 (remove_attribute): Likewise.
21201 (merge_attributes): Likewise.
21202 (merge_type_attributes): Likewise.
21203 (merge_decl_attributes): Likewise.
21204 (merge_dllimport_decl_attributes): Likewise.
21205 (handle_dll_attribute): Likewise.
21206 (attribute_list_equal): Likewise.
21207 (attribute_list_contained): Likewise.
21208 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21209 (lookup_attribute_by_prefix): Likewise.
21210 * bb-reorder.c: Include header files.
21211 * builtins.c: Likewise.
21212 * calls.c: Likewise.
21213 * cfgexpand.c: Likewise.
21214 * cgraph.c: Likewise.
21215 * cgraphunit.c: Likewise.
21216 * convert.c: Likewise.
21217 * dwarf2out.c: Likewise.
21218 * final.c: Likewise.
21219 * fold-const.c: Likewise.
21220 * function.c: Likewise.
21221 * gimple-expr.c: Likewise.
21222 * gimple-fold.c: Likewise.
21223 * gimple-pretty-print.c: Likewise.
21224 * gimple.c: Likewise.
21225 * gimplify.c: Likewise.
21226 * hsa-common.c: Likewise.
21227 * hsa-gen.c: Likewise.
21228 * internal-fn.c: Likewise.
21229 * ipa-chkp.c: Likewise.
21230 * ipa-cp.c: Likewise.
21231 * ipa-devirt.c: Likewise.
21232 * ipa-fnsummary.c: Likewise.
21233 * ipa-inline.c: Likewise.
21234 * ipa-visibility.c: Likewise.
21236 * lto-cgraph.c: Likewise.
21237 * omp-expand.c: Likewise.
21238 * omp-general.c: Likewise.
21239 * omp-low.c: Likewise.
21240 * omp-offload.c: Likewise.
21241 * omp-simd-clone.c: Likewise.
21242 * opts-global.c: Likewise.
21243 * passes.c: Likewise.
21244 * predict.c: Likewise.
21245 * sancov.c: Likewise.
21246 * sanopt.c: Likewise.
21247 * symtab.c: Likewise.
21248 * toplev.c: Likewise.
21249 * trans-mem.c: Likewise.
21250 * tree-chkp.c: Likewise.
21251 * tree-eh.c: Likewise.
21252 * tree-into-ssa.c: Likewise.
21253 * tree-object-size.c: Likewise.
21254 * tree-parloops.c: Likewise.
21255 * tree-profile.c: Likewise.
21256 * tree-ssa-ccp.c: Likewise.
21257 * tree-ssa-live.c: Likewise.
21258 * tree-ssa-loop.c: Likewise.
21259 * tree-ssa-sccvn.c: Likewise.
21260 * tree-ssa-structalias.c: Likewise.
21261 * tree-ssa.c: Likewise.
21262 * tree-streamer-in.c: Likewise.
21263 * tree-vectorizer.c: Likewise.
21264 * tree-vrp.c: Likewise.
21265 * tsan.c: Likewise.
21266 * ubsan.c: Likewise.
21267 * varasm.c: Likewise.
21268 * varpool.c: Likewise.
21269 * tree.c: Remove functions moved to attribs.[ch].
21270 * tree.h: Likewise.
21271 * config/aarch64/aarch64.c: Add attrs.h header file.
21272 * config/alpha/alpha.c: Likewise.
21273 * config/arc/arc.c: Likewise.
21274 * config/arm/arm.c: Likewise.
21275 * config/avr/avr.c: Likewise.
21276 * config/bfin/bfin.c: Likewise.
21277 * config/c6x/c6x.c: Likewise.
21278 * config/cr16/cr16.c: Likewise.
21279 * config/cris/cris.c: Likewise.
21280 * config/darwin.c: Likewise.
21281 * config/epiphany/epiphany.c: Likewise.
21282 * config/fr30/fr30.c: Likewise.
21283 * config/frv/frv.c: Likewise.
21284 * config/ft32/ft32.c: Likewise.
21285 * config/h8300/h8300.c: Likewise.
21286 * config/i386/winnt.c: Likewise.
21287 * config/ia64/ia64.c: Likewise.
21288 * config/iq2000/iq2000.c: Likewise.
21289 * config/lm32/lm32.c: Likewise.
21290 * config/m32c/m32c.c: Likewise.
21291 * config/m32r/m32r.c: Likewise.
21292 * config/m68k/m68k.c: Likewise.
21293 * config/mcore/mcore.c: Likewise.
21294 * config/microblaze/microblaze.c: Likewise.
21295 * config/mips/mips.c: Likewise.
21296 * config/mmix/mmix.c: Likewise.
21297 * config/mn10300/mn10300.c: Likewise.
21298 * config/moxie/moxie.c: Likewise.
21299 * config/msp430/msp430.c: Likewise.
21300 * config/nds32/nds32-isr.c: Likewise.
21301 * config/nds32/nds32.c: Likewise.
21302 * config/nios2/nios2.c: Likewise.
21303 * config/nvptx/nvptx.c: Likewise.
21304 * config/pa/pa.c: Likewise.
21305 * config/pdp11/pdp11.c: Likewise.
21306 * config/powerpcspe/powerpcspe.c: Likewise.
21307 * config/riscv/riscv.c: Likewise.
21308 * config/rl78/rl78.c: Likewise.
21309 * config/rx/rx.c: Likewise.
21310 * config/s390/s390.c: Likewise.
21311 * config/sh/sh.c: Likewise.
21312 * config/sol2.c: Likewise.
21313 * config/sparc/sparc.c: Likewise.
21314 * config/spu/spu.c: Likewise.
21315 * config/stormy16/stormy16.c: Likewise.
21316 * config/tilegx/tilegx.c: Likewise.
21317 * config/tilepro/tilepro.c: Likewise.
21318 * config/v850/v850.c: Likewise.
21319 * config/vax/vax.c: Likewise.
21320 * config/visium/visium.c: Likewise.
21321 * config/xtensa/xtensa.c: Likewise.
21323 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21326 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21327 constraints since the -mupper-regs-* switches have been
21329 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21330 into a vector from a double word element that was extracted from
21331 another vector, and eliminate extra XXPERMDI instructions.
21332 (vsx_concat_<mode>_2): Likewise.
21333 (vsx_concat_<mode>_3): Likewise.
21334 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21335 concat to allow optimizing inserts from previous extracts.
21337 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21339 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21340 memory reference to a SSP offset in TLS address space.
21341 (ix86_print_operand) <case '@'>: Remove.
21342 (ix86_print_operand_punct_valid_p): Remove '@' code.
21343 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21344 UNSPEC_SP_TLS_TEST.
21345 (stack_tls_protect_set_<mode>): Remove.
21346 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21347 (stack_tls_protect_test_<mode>): Remove.
21348 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21350 2017-08-07 Olivier Hainque <hainque@adacore.com>
21353 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21355 2017-08-07 Douglas Rupp <rupp@adacore.com>
21357 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21358 variable was referenced as multidir in command.
21360 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21363 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21366 2017-08-07 Martin Liska <mliska@suse.cz>
21368 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21369 * config/rl78/rl78.c: Add include of attribs.h.
21370 * config/sh/sh.c: Likewise.
21371 * config/v850/v850.c: Likewise.
21373 2017-08-07 Tom de Vries <tom@codesourcery.com>
21375 PR middle-end/78266
21376 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21378 2017-08-07 Martin Liska <mliska@suse.cz>
21380 * config/mips/mips.c: Include attribs.h.
21382 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21385 * doc/invoke.texi: Document change in behvaior for -Ofast for
21388 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21390 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21392 (aarch64_pop_regs): Likewise.
21393 (aarch64_gen_load_pair): Likewise.
21394 (aarch64_save_callee_saves): Likewise.
21395 (aarch64_restore_callee_saves): Likewise.
21397 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21399 * config/i386/i386.c: Revert the last change.
21401 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21404 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21406 (ix86_finalize_stack_frame_flags): This. Also clear
21407 frame_pointer_needed if -fno-omit-frame-pointer is used without
21409 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21410 with ix86_finalize_stack_frame_flags.
21411 (ix86_expand_epilogue): Likewise.
21412 (ix86_expand_split_stack_prologue): Likewise.
21414 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
21417 * config/i386/i386.c (get_builtin_code_for_version): Set priority
21418 to P_AES for Westmere.
21420 2017-08-07 Jonathan Yong <10walls@gmail.com>
21422 * config/i386/mingw.opt (fset-stack-executable): Removed.
21423 * config/i386/cygming.opt (fset-stack-executable): Moved
21425 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
21427 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
21429 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
21431 2017-08-07 Marek Polacek <polacek@redhat.com>
21433 PR middle-end/81737
21434 * fold-const.c (fold_indirect_ref_1): Check type_domain.
21436 2017-08-07 Martin Liska <mliska@suse.cz>
21438 * attribs.h (canonicalize_attr_name): New function.
21439 (cmp_attribs): Move from c-format.c and adjusted.
21440 (is_attribute_p): Moved from tree.h.
21441 * tree-inline.c: Add new includes.
21442 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
21443 (private_is_attribute_p): Remove.
21444 (private_lookup_attribute): Likewise.
21445 (private_lookup_attribute_by_prefix): Simplify.
21446 (remove_attribute): Use is_attribute_p.
21447 * tree.h: Remove removed declarations.
21449 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21451 PR middle-end/81698
21452 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
21453 instead of computing it in the function. Formatting fix.
21454 (expand_case): Don't rely on default_edge being the first edge,
21455 clear it if removing it, pass default_edge to
21456 emit_case_dispatch_table.
21457 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
21460 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
21462 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
21463 insn in the function, emit NOP after the insn.
21465 2017-08-06 Tom de Vries <tom@codesourcery.com>
21467 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
21470 2017-08-06 Tom de Vries <tom@codesourcery.com>
21472 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
21475 2017-08-04 Yury Gribov <tetra2005@gmail.com>
21477 PR tree-optimization/57371
21478 * match.pd: New pattern.
21480 2017-08-04 Marek Polacek <polacek@redhat.com>
21482 PR middle-end/81695
21483 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
21484 perform the computation in offset_int.
21486 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21488 PR tree-optimization/81136
21489 * tree-vectorizer.h: Include tree-hash-traits.h.
21490 (vec_base_alignments): New typedef.
21491 (vec_info): Add a base_alignments field.
21492 (vect_record_base_alignments): Declare.
21493 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
21495 (DR_IS_CONDITIONAL_IN_STMT): New macro.
21496 (create_data_ref): Add an is_conditional_in_stmt argument.
21497 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
21498 the is_conditional_in_stmt field.
21499 (data_ref_loc): Add an is_conditional_in_stmt field.
21500 (get_references_in_stmt): Set the is_conditional_in_stmt field.
21501 (find_data_references_in_stmt): Update call to create_data_ref.
21502 (graphite_find_data_references_in_stmt): Likewise.
21503 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
21504 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
21505 (vect_record_base_alignment): New function.
21506 (vect_record_base_alignments): Likewise.
21507 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
21508 for nested statements even if we fail to compute a misalignment.
21509 Use pooled base alignments for unconditional references.
21510 (vect_find_same_alignment_drs): Compare base addresses instead
21512 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
21513 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
21515 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21517 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
21518 Add an explicit name for the enum. Use auto_vec for slp_instances
21519 and grouped_stores.
21520 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
21522 (_bb_vec_info): Add a constructor and destructor.
21523 (vinfo_for_stmt): Return NULL for uids of -1 as well.
21524 (destroy_loop_vec_info): Delete.
21525 (vect_destroy_datarefs): Likewise.
21526 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
21527 (vec_info::vec_info): New function.
21528 (vec_info::~vec_info): Likewise.
21529 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
21530 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
21531 destroy_loop_vec_info.
21532 * tree-vect-loop.c (new_loop_vec_info): Replace with...
21533 (_loop_vec_info::_loop_vec_info): ...this.
21534 (destroy_loop_vec_info): Replace with...
21535 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
21536 the stmt_vec_infos. Leave handling of vec_info information to its
21537 destructor. Remove explicit vector releases.
21538 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
21539 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
21540 * tree-vect-slp.c (new_bb_vec_info): Replace with...
21541 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
21542 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
21543 (destroy_bb_vec_info): Replace with...
21544 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
21545 information to its destructor.
21546 (vect_slp_analyze_bb_1): Use new and delete instead of
21547 new_bb_vec_info and destroy_bb_vec_info.
21548 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
21551 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
21553 * tree-data-ref.h (subscript): Add access_fn field.
21554 (data_dependence_relation): Add could_be_independent_p.
21555 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
21556 (same_access_functions): Move to tree-data-ref.c.
21557 * tree-data-ref.c (ref_contains_union_access_p): New function.
21558 (access_fn_component_p): Likewise.
21559 (access_fn_components_comparable_p): Likewise.
21560 (dr_analyze_indices): Add a reference to access_fn_component_p.
21561 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
21563 (constant_access_functions): Likewise.
21564 (add_other_self_distances): Likewise.
21565 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
21566 (initialize_data_dependence_relation): Use XCNEW and remove
21567 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
21568 of access functions that have the same type. Allow the
21569 subsequence to end with different bases in some circumstances.
21570 Record the chosen access functions in SUB_ACCESS_FN.
21571 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
21572 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
21573 (subscript_dependence_tester_1): Likewise dra and drb.
21574 (build_classic_dist_vector): Update calls accordingly.
21575 (subscript_dependence_tester): Likewise.
21576 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
21577 DDR_COULD_BE_INDEPENDENT_P.
21578 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
21579 comp_alias_ddrs instead of may_alias_ddrs.
21580 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
21582 (vect_analyze_data_ref_dependence): Use it if
21583 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
21584 distance vectors if that fails.
21585 (dependence_distance_ge_vf): New function.
21586 (vect_prune_runtime_alias_test_list): Use it. Don't clear
21587 LOOP_VINFO_MAY_ALIAS_DDRS.
21589 2017-08-04 Richard Biener <rguenther@suse.de>
21591 PR middle-end/81705
21592 * fold-const.c (fold_binary_loc): Properly restrict
21593 minus_var0 && minus_var1 case when associating undefined overflow
21596 2017-08-04 Tom de Vries <tom@codesourcery.com>
21598 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
21600 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21602 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21603 Don't start diagnostic messages with a capital letter.
21604 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21606 (rs6000_invalid_builtin): Likewise.
21607 (rs6000_trampoline_init): Likewise.
21609 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21612 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
21613 after setting changeable df flags.
21615 2017-08-03 Richard Biener <rguenther@suse.de>
21617 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
21618 up if the use is in USE - X.
21620 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
21622 * toplev.c (dumpfile.h): New include.
21623 (internal_error_reentered): New static function. Use it...
21624 (internal_error_function): ...here to handle reentered internal_error.
21626 2017-08-03 Richard Biener <rguenther@suse.de>
21628 PR middle-end/81148
21629 * fold-const.c (split_tree): Add minus_var and minus_con
21630 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
21631 here but always use minus_*.
21632 (associate_trees): Assert we never associate with MINUS_EXPR
21633 and NULL first operand. Do not recurse for PLUS_EXPR operands
21634 when associating as MINUS_EXPR either.
21635 (fold_binary_loc): Track minus_var and minus_con.
21637 2017-08-03 Tom de Vries <tom@codesourcery.com>
21640 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
21641 ACCEL_COMPILER, apply finish_options on
21642 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
21644 2017-08-03 Tom de Vries <tom@codesourcery.com>
21647 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
21648 function_entry_patch_area_size > 0.
21650 2017-08-03 Jakub Jelinek <jakub@redhat.com>
21653 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
21654 instead of 10??LU, perform unit multiplication in wide_int,
21655 don't change alloc_object_size_limit if the limit is larger
21658 PR tree-optimization/81655
21659 PR tree-optimization/81588
21660 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
21661 the case when ranges[i].low and high are 1 for unsigned type with
21664 PR middle-end/81052
21665 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
21666 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
21668 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21670 * tree-vrp.h: Add include guard.
21672 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
21675 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
21676 (ud2): New insn pattern.
21677 * config/i386/i386.c (ix86_expand_epilogue):
21678 For naked functions, generate ud2 instead of trap insn.
21680 2017-08-02 Marek Polacek <polacek@redhat.com>
21683 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
21685 2017-08-02 Tom de Vries <tom@codesourcery.com>
21686 Cesar Philippidis <cesar@codesourcery.com>
21688 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
21689 Add missing edge probabilities.
21691 2017-08-02 Tamar Christina <tamar.christina@arm.com>
21693 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
21694 Correct endianness.
21696 2017-08-02 Jakub Jelinek <jakub@redhat.com>
21698 PR middle-end/79499
21699 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
21700 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
21701 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
21702 prologue_seq sequences - if any.
21704 2017-08-02 Richard Biener <rguenther@suse.de>
21706 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
21707 via vectors if supported, integer extracts via punning if supported
21708 or otherwise vector extracts.
21710 2017-08-02 Richard Biener <rguenther@suse.de>
21712 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
21714 (bitmap_insert_into_set): ... this.
21716 2017-08-02 Richard Biener <rguenther@suse.de>
21718 PR tree-optimization/81633
21720 2015-08-17 Alan Hayward <alan.hayward@arm.com>
21722 PR tree-optimization/71752
21723 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
21725 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
21727 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
21728 (machine_function::call_ms2sysv_pad_out): Remove field.
21729 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
21730 (ix86_compute_frame_layout): Likewise.
21732 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
21735 * config/i386/i386.c (ix86_set_func_type): Disallow naked
21736 attribute with interrupt attribute.
21738 2017-08-01 Andrew Pinski <apinski@cavium.com>
21740 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
21741 BIT_INSERT_EXPR's operand 1
21742 to see if the types precision matches.
21744 2017-08-01 Martin Liska <mliska@suse.cz>
21746 PR middle-end/70140
21747 * builtins.c (expand_builtin_memcpy_args): Remove.
21748 (expand_builtin_memcpy): Call newly added function
21749 expand_builtin_memory_copy_args.
21750 (expand_builtin_memcpy_with_bounds): Likewise.
21751 (expand_builtin_mempcpy): Remove last argument.
21752 (expand_builtin_mempcpy_with_bounds): Likewise.
21753 (expand_builtin_memory_copy_args): New function created from
21754 expand_builtin_mempcpy_args with small modifications.
21755 (expand_builtin_mempcpy_args): Remove.
21756 (expand_builtin_stpcpy): Remove unused argument.
21757 (expand_builtin): Likewise.
21758 (expand_builtin_with_bounds): Likewise.
21760 2017-08-01 Martin Liska <mliska@suse.cz>
21763 Make mempcpy more optimal (PR middle-end/70140).
21765 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21768 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
21769 __builtin_vec_cmpne verify both arguments are compatible vectors
21770 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
21771 verify arg1_type is a pointer or array type. For __builtin_vec_st,
21772 move computation of aligned to after checking the argument types.
21776 * config/rs6000/vsx.md (vextract_fp_from_shorth,
21777 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
21780 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
21781 Doug Rupp <rupp@adacore.com>
21782 Olivier Hainque <hainque@adacore.com>
21784 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
21785 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
21787 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
21788 for TARGET_OS_CPP_BUILTIN.
21789 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
21790 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
21792 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
21793 passing required abi options to the assembler for EABI configurations.
21794 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
21795 of .text.hot and .text.unlikely sections for kernel modules when
21796 using ARM style exceptions.
21797 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
21798 options. Add EXTRA_CC1_SPEC.
21799 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
21801 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
21803 (ARM_TARGET2_DWARF_FORMAT): Define.
21804 * config/arm/t-vxworks: Adjust multilib control to removal of the
21805 Diab command line options.
21807 2017-08-01 Martin Liska <mliska@suse.cz>
21809 PR gcov-profile/81561
21810 * gcov.c (unblock): Make unblocking safe as we need to preserve
21811 index correspondence of blocks and block_lists.
21813 2017-08-01 Richard Biener <rguenther@suse.de>
21815 PR tree-optimization/81181
21816 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
21817 (compute_antic): ... end of iteration here.
21819 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
21821 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
21822 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
21823 (ftree-slp-vectorize): Likewise.
21824 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
21825 can no longer be set independent of flag_tree_loop_vectorize.
21826 * omp-general.c (emp_max_vf): Likewise.
21827 * opts.c (enable_fdo_optimizations): Remove references to
21828 flag_tree_vectorize, these are now implicit.
21829 (common_handle_option): Remove handling for OPT_ftree_vectorize,
21830 and leave it for the options machinery.
21832 2017-08-01 Martin Liska <mliska@suse.cz>
21834 PR middle-end/70140
21835 * builtins.c (expand_builtin_memcpy_args): Remove.
21836 (expand_builtin_memcpy): Call newly added function
21837 expand_builtin_memory_copy_args.
21838 (expand_builtin_memcpy_with_bounds): Likewise.
21839 (expand_builtin_mempcpy): Remove last argument.
21840 (expand_builtin_mempcpy_with_bounds): Likewise.
21841 (expand_builtin_memory_copy_args): New function created from
21842 expand_builtin_mempcpy_args with small modifications.
21843 (expand_builtin_mempcpy_args): Remove.
21844 (expand_builtin_stpcpy): Remove unused argument.
21845 (expand_builtin): Likewise.
21846 (expand_builtin_with_bounds): Likewise.
21848 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21851 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
21852 print "ds:" only for immediates in generic address space.
21854 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
21857 * config/i386/i386.c (ix86_funciton_naked): New prototype.
21858 (ix86_function_ok_for_sibcall): Return false for naked functions.
21860 2017-08-01 Richard Biener <rguenther@suse.de>
21862 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
21863 (compute_antic): Seed worklist with exit block predecessors.
21864 * cfganal.c (dfs_find_deadend): For a cycle return the source
21865 of the edge closing it.
21867 2017-08-01 Tamar Christina <tamar.christina@arm.com>
21869 * config/aarch64/aarch64.c
21870 (aarch64_can_const_movi_rtx_p): Move 0 check.
21872 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21874 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
21875 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
21877 * match.pd: Ditto in address comparison pattern.
21879 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21881 PR tree-optimization/81627
21882 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
21883 closed ssa form for store-store chain.
21885 2017-08-01 Bin Cheng <bin.cheng@arm.com>
21887 PR tree-optimization/81620
21888 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
21889 for store-store chain.
21891 2017-08-01 Jakub Jelinek <jakub@redhat.com>
21893 PR tree-optimization/81588
21894 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
21895 ranges[i].in_p, invert comparison code ccode. For >/>=,
21896 swap rhs1 and rhs2 and comparison code unconditionally,
21897 for </<= don't do that. Don't swap rhs1/rhs2 again if
21898 ranges[i].in_p, instead invert comparison code ccode if
21899 opcode or oe->rank is BIT_IOR_EXPR.
21902 * optabs.def (vec_extract_optab, vec_init_optab): Change from
21903 a direct optab to conversion optab.
21904 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
21905 with GET_MODE_INNER as last argument instead of optab_handler.
21906 * expmed.c (extract_bit_field_1): Likewise. Use vector from
21907 vector extraction if possible and optab is available.
21908 * expr.c (store_constructor): Use convert_optab_handler instead
21909 of optab_handler. Use vector initialization from smaller
21910 vectors if possible and optab is available.
21911 * tree-vect-stmts.c (vectorizable_load): Likewise.
21912 * doc/md.texi (vec_extract, vec_init): Document that the optabs
21913 now have two modes.
21914 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
21915 of vec_init from half-sized vectors with the same element mode.
21916 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
21917 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
21918 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
21919 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
21920 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
21921 after mode in gen_vec_extract* calls.
21922 (vec_extract<mode>): Renamed to ...
21923 (vec_extract<mode><ssescalarmodelower>): ... this.
21924 (vec_extract<mode><ssehalfvecmodelower>): New expander.
21925 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
21926 element mode after mode in gen_vec_init* calls.
21927 (VEC_INIT_HALF_MODE): New mode iterator.
21928 (vec_init<mode>): Renamed to ...
21929 (vec_init<mode><ssescalarmodelower>): ... this.
21930 (vec_init<mode><ssehalfvecmodelower>): New expander.
21931 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
21932 (vec_extractv2sfsf): ... this.
21933 (vec_initv2sf): Renamed to ...
21934 (vec_initv2sfsf): ... this.
21935 (vec_extractv2si): Renamed to ...
21936 (vec_extractv2sisi): ... this.
21937 (vec_initv2si): Renamed to ...
21938 (vec_initv2sisi): ... this.
21939 (vec_extractv4hi): Renamed to ...
21940 (vec_extractv4hihi): ... this.
21941 (vec_initv4hi): Renamed to ...
21942 (vec_initv4hihi): ... this.
21943 (vec_extractv8qi): Renamed to ...
21944 (vec_extractv8qiqi): ... this.
21945 (vec_initv8qi): Renamed to ...
21946 (vec_initv8qiqi): ... this.
21947 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
21948 (vec_init<mode>): Renamed to ...
21949 (vec_init<mode><VEC_base_l>): ... this.
21950 (vec_extract<mode>): Renamed to ...
21951 (vec_extract<mode><VEC_base_l>): ... this.
21952 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
21953 (vec_initv2sfsf): ... this.
21954 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
21955 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
21956 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
21957 element mode after mode in gen_vec_init* calls.
21958 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
21959 (vec_init<mode><Vel>): ... this.
21960 (vec_extract<mode>): Renamed to ...
21961 (vec_extract<mode><Vel>): ... this.
21962 * config/aarch64/iterators.md (Vel): New mode attribute.
21963 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
21964 Add element mode after mode in gen_vec_extract* calls.
21965 * config/s390/vector.md (non_vec_l): New mode attribute.
21966 (vec_extract<mode>): Renamed to ...
21967 (vec_extract<mode><non_vec_l>): ... this.
21968 (vec_init<mode>): Renamed to ...
21969 (vec_init<mode><non_vec_l>): ... this.
21970 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
21971 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
21973 * config/arm/iterators.md (V_elem_l): New mode attribute.
21974 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
21975 (vec_extract<mode><V_elem_l>): ... this.
21976 (vec_extractv2di): Renamed to ...
21977 (vec_extractv2didi): ... this.
21978 (vec_init<mode>): Renamed to ...
21979 (vec_init<mode><V_elem_l>): ... this.
21980 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
21981 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
21982 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
21983 Add element mode after gen_vec_extract* calls.
21984 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
21985 (vec_init<mode><unitmode>): ... this.
21986 (vec_extract<mode>): Renamed to ...
21987 (vec_extract<mode><unitmode>): ... this.
21988 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
21989 (vec_init<mode><unitmode>): ... this.
21990 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
21991 (vec_initv2sfsf): ... this.
21992 (vec_extractv2sf): Renamed to ...
21993 (vec_extractv2sfsf): ... this.
21994 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
21995 Add element mode after gen_vec_extract* calls.
21996 * config/mips/mips.md (unitmode): New mode iterator.
21997 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
21998 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
21999 * config/spu/spu.md (inner_l): New mode attribute.
22000 (vec_init<mode>): Renamed to ...
22001 (vec_init<mode><inner_l>): ... this.
22002 (vec_extract<mode>): Renamed to ...
22003 (vec_extract<mode><inner_l>): ... this.
22004 * config/sparc/sparc.md (veltmode): New mode iterator.
22005 (vec_init<VMALL:mode>): Renamed to ...
22006 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22007 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22008 (vec_initv2sisi): ... this.
22009 (vec_initv2sf): Renamed to ...
22010 (vec_initv2sfsf): ... this.
22011 (vec_extractv2sf): Renamed to ...
22012 (vec_extractv2sfsf): ... this.
22013 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22014 (vec_init<mode>): Renamed to ...
22015 (vec_init<mode><VEC_base_l>): ... this.
22016 (vec_extract<mode>): Renamed to ...
22017 (vec_extract<mode><VEC_base_l>): ... this.
22018 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22019 (vec_initv2sfsf): ... this.
22020 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22021 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22022 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22023 gen_vec_init* calls.
22025 2017-08-01 Richard Biener <rguenther@suse.de>
22027 PR tree-optimization/81297
22028 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22029 TREE_OVERFLOW from INTEGER_CSTs.
22031 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22033 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22035 2017-07-31 Carl Love <cel@us.ibm.com>
22037 * config/rs6000/rs6000-c: Add support for built-in functions
22038 vector signed char vec_xl_be (signed long long, signed char *);
22039 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22040 vector signed int vec_xl_be (signed long long, signed int *);
22041 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22042 vector signed long long vec_xl_be (signed long long, signed long long *);
22043 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22044 vector signed short vec_xl_be (signed long long, signed short *);
22045 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22046 vector double vec_xl_be (signed long long, double *);
22047 vector float vec_xl_be (signed long long, float *);
22048 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22049 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22050 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22052 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22053 (altivec_expand_builtin): Add switch statement to call
22054 altivec_expand_xl_be for each builtin.
22055 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22056 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22057 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22058 __builtin_vsx_le_be_v16qi.
22059 * doc/extend.texi: Update the built-in documentation file for the
22060 new built-in functions.
22062 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22065 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22067 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22069 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22071 * config.gcc: Add z14.
22072 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22073 CPU model numbers for z13s and z14.
22074 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22076 * config/s390/s390-opts.h (enum processor_type): Rename
22077 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22078 * config/s390/s390.c (processor_table): Add field for CPU name to
22079 be passed to Binutils.
22080 (s390_asm_output_machine_for_arch): Use the new field in
22081 processor_table for Binutils.
22082 (s390_expand_builtin): Replace arch12 with z14.
22083 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22084 (s390_get_sched_attrmask): Likewise.
22085 (s390_get_unit_mask): Likewise.
22086 * config/s390/s390.opt: Add z14 to processor_type enum.
22088 2017-07-31 Martin Jambor <mjambor@suse.cz>
22091 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22092 regardless of optimization level.
22094 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22095 Martin Liska <mliska@suse.cz>
22097 * predict.def: Remove old comment and adjust probability.
22098 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22099 PREDICT statements.
22101 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22104 * config/i386/i386.c (ix86_function_naked): New function.
22105 (ix86_can_use_return_insn_p): Return false for naked functions.
22106 (ix86_expand_prologue): Skip prologue for naked functions.
22107 (ix86_expand_epilogue): Skip epilogue for naked functions
22108 and emit trap instruction.
22109 (ix86_warn_func_return): New function.
22110 (ix86_attribute_table): Add "naked" attribute specification.
22111 (TARGET_WARN_FUNC_RETURN): Define.
22112 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22114 2017-07-31 Martin Liska <mliska@suse.cz>
22116 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22117 (dump_gimple_bb_header): Always dump BB info.
22118 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22120 2017-07-31 Martin Liska <mliska@suse.cz>
22123 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22124 also with current_function_decl non-null equality.
22126 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22129 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22130 change type to the element type, instead add eltype variable and
22131 use it where we are interested in the element type.
22133 PR tree-optimization/81603
22134 * ipa-polymorphic-call.c
22135 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22136 offset arithmetic in offset_int, bail out if the resulting bit offset
22137 doesn't fit into shwi.
22139 2017-07-31 Martin Liska <mliska@suse.cz>
22141 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22142 (gimplify_save_expr): Fix comment.
22144 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22147 * config/i386/i386.c (ix86_function_arg): Update arguments for
22149 (ix86_compute_frame_layout): Set the initial stack offset to
22150 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22151 INCOMING_FRAME_SP_OFFSET.
22152 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22153 stack before exception handler returns.
22154 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22155 the 'ERROR_CODE' for exception handler.
22157 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22159 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22160 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22161 (ASM_OUTPUT_REG_POP): Ditto.
22162 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22163 instead of asm_fprintf to output pure string.
22165 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22167 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22168 to imported_module_or_decl hook.
22169 (debug_nothing_tree_tree_tree_bool): Remove.
22170 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22171 * debug.c (do_nothing_debug_hooks): Use
22172 debug_nothing_tree_tree_tree_bool_bool instead of
22173 debug_nothing_tree_tree_tree_bool.
22174 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22175 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22176 * sdbout.c (sdb_debug_hooks): Likewise.
22177 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22178 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22180 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22181 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22182 attribute, don't add anything.
22184 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22186 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22187 (fold_build2_stat_loc): Likewise.
22188 (fold_build3_stat_loc): Likewise.
22189 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22190 (fold_build1_loc): Remove macro.
22191 (fold_build2_loc): Likewise.
22192 (fold_build3_loc): Likewise.
22194 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22196 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22197 (gimple_build_debug_bind_source_stat): Likewise.
22198 * gimple.h (gimple_build_debug_bind): Remove macro.
22199 (gimple_build_debug_bind_source): Likewise.
22201 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22203 * bitmap.c (bitmap_alloc): Adjust.
22204 (bitmap_gc_alloc): Likewise.
22205 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22207 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22209 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22210 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22211 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22212 (bitmap_gc_alloc_stat): Likewise.
22213 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22215 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22217 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22218 * rtl.h (shallow_copy_rtx): Remove macro.
22220 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22222 * emit-rtl.c (gen_raw_REG): Adjust.
22223 * gengenrtl.c (gendef): Likewise.
22224 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22225 * rtl.h (rtx_alloc): Remove macro.
22227 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22229 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22230 (build_tree_list_stat): Likewise.
22231 * tree.h (build_tree_list): Remove macro.
22232 (build_tree_list_vec): Likewise.
22234 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22236 * tree.c (make_vector_stat): Remove _stat from name.
22237 (build_vector_stat): Likewise.
22238 * tree.h (make_vector_stat): Remove macro.
22239 (build_vector_stat): Likewise.
22241 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22243 * tree.h (build_var_debug_value): Remove prototype.
22245 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22247 * tree.c (tree_cons_stat): Remove _stat from name.
22248 * tree.h (tree_cons): Remove macro.
22250 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22252 * tree.c (build_vl_exp_stat): Remove _stat from name.
22253 * tree.h (build_vl_exp): Remove macro.
22255 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22257 * tree.c (build_decl_stat): Remove _stat from name.
22258 * tree.h (build_decl): Remove macro.
22260 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22262 * gimple.c (gimple_build_with_ops_stat): Adjust.
22263 (gimple_alloc_stat): Remove _stat from name.
22264 * gimple.h (gimple_alloc): Remove macro.
22266 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22268 * tree.c (make_tree_vec_stat): Remove _stat from name.
22269 (grow_tree_vec_stat): Likewise.
22270 * tree.h (make_tree_vec_stat): Adjust prototype.
22271 (grow_tree_vec_stat): Likewise.
22272 (make_tree_vec): Remove macro.
22273 (grow_tree_vec): Likewise.
22275 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22277 * fold-const.c (fold_build1_stat_loc): Adjust.
22278 (fold_build2_stat_loc): Likewise.
22279 (fold_build3_stat_loc): Likewise.
22280 * tree.c (build0_stat): Remove _stat from name.
22281 (build1_stat): Likewise.
22282 (build2_stat): Likewise.
22283 (build3_stat): Likewise.
22284 (build4_stat): Likewise.
22285 (build5_stat): Likewise.
22286 * tree.h (build1_loc): Remove macro, and rename _stat function
22288 (build2_loc): Likewise.
22289 (build3_loc): Likewise.
22290 (build4_loc): Likewise.
22291 (build5_loc): Likewise.
22293 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22295 * tree.c (make_int_cst_stat): Remove _stat from name.
22296 * tree.h (make_int_cst_stat): Adjust prototype.
22297 (make_int_cst): Remove macro.
22299 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22301 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22302 * tree.h (make_tree_binfo_stat): Adjust prototype.
22303 (make_tree_binfo): Remove.
22305 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22307 * tree.c (copy_node_stat): Rename to copy_node.
22308 (build_distinct_type_copy): Adjust.
22309 * tree.h (copy_node_stat): Adjust prototype.
22310 (copy_node): Remove macro.
22312 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22314 * tree.c (make_node_stat): rename to make_node.
22315 (build_tree_list_stat): Adjust.
22316 (build0_stat): Likewise.
22317 (build2_stat): Likewise.
22318 (build3_stat): Likewise.
22319 (build4_stat): Likewise.
22320 (build5_stat): Likewise.
22321 (build_decl_stat): Likewise.
22322 * tree.h (make_node_stat): Adjust prototype.
22323 (make_node): remove macro.
22325 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22327 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22328 (PPC_FEATURE2_SCV): Likewise.
22329 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22331 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22333 * config/aarch64/aarch64.c
22334 (aarch64_internal_mov_immediate): Add new special pattern.
22335 * config/aarch64/aarch64.md (*movdi_aarch64):
22336 Add reg/32bit const mov case.
22338 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22339 Richard Sandiford <richard.sandiford@linaro.org>
22341 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22342 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22343 Add integer and movi cases.
22344 (movi-split-hf-df-sf split, fp16): New.
22345 (enabled): Added TARGET_FP_F16INST.
22346 * config/aarch64/iterators.md (GPF_HF): New.
22347 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22349 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22351 * config/aarch64/aarch64.c
22352 (aarch64_simd_container_mode): Add prototype.
22353 (aarch64_expand_mov_immediate): Add HI support.
22354 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22355 (aarch64_can_const_movi_rtx_p): New.
22356 (aarch64_preferred_reload_class):
22357 Remove restrictions of using FP registers for certain SIMD operations.
22358 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22359 (aarch64_valid_floating_const): Add integer move validation.
22360 (aarch64_simd_imm_scalar_p): Remove.
22361 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22362 (aarch64_legitimate_constant_p): Expand list of supported cases.
22363 * config/aarch64/aarch64-protos.h
22364 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22365 (aarch64_reinterpret_float_as_int): New.
22366 (aarch64_simd_imm_scalar_p): Remove.
22367 * config/aarch64/constraints.md (Uvi): New.
22368 (Dd): Split into Ds and new Dd.
22369 * config/aarch64/aarch64.md (*movsi_aarch64):
22371 (*movdi_aarch64): Add SIMD mov case.
22373 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22375 * tree-predcom.c: (struct chain): Handle store-store chain in which
22376 stores for elimination only store loop invariant values.
22377 (execute_pred_commoning_chain): Ditto.
22378 (prepare_initializers_chain_store_elim): Ditto.
22379 (prepare_finalizers): Ditto.
22380 (is_inv_store_elimination_chain): New function.
22381 (initialize_root_vars_store_elim_1): New function.
22383 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22385 * tree-predcom.c: Revise general description of the pass.
22386 (enum chain_type): New enum type for store elimination.
22387 (struct chain): New field supporting store elimination.
22388 (struct component): Ditto.
22389 (dump_chain): Dump store-stores chain.
22390 (release_chain): Release resources.
22391 (split_data_refs_to_components): Compute and create component
22392 contains only stores for elimination.
22393 (get_chain_last_ref_at): New function.
22394 (make_invariant_chain): Initialization.
22395 (make_rooted_chain): Specify chain type in parameter and record it.
22396 (add_looparound_copies): Skip for store-stores chain.
22397 (determine_roots_comp): Compute type of chain and pass it to
22399 (initialize_root_vars_store_elim_2): New function.
22400 (finalize_eliminated_stores): New function.
22401 (remove_stmt): Handle store for elimination.
22402 (execute_pred_commoning_chain): Execute predictive commoning on
22403 store-store chains.
22404 (determine_unroll_factor): Skip unroll for store-stores chain.
22405 (prepare_initializers_chain_store_elim): New function.
22406 (prepare_initializers_chain): Hanlde store-store chain.
22407 (prepare_finalizers_chain, prepare_finalizers): New function.
22408 (tree_predictive_commoning_loop): Return integer value indicating
22409 if loop is unrolled or lcssa form is corrupted.
22410 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
22412 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22414 * tree-predcom.c (initialize_root): Delete.
22415 (execute_pred_commoning_chain): Initialize root vars and replace
22416 reference of non-combined chain directly, rather than call above
22419 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22421 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
22422 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
22424 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22426 * tree-predcom.c (struct chain): New field init_seq.
22427 (release_chain): Release init_seq.
22428 (prepare_initializers_chain): Record intialization stmts in above
22430 (insert_init_seqs): New function.
22431 (tree_predictive_commoning_loop): Call insert_init_seqs.
22433 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22435 * tree-predcom.c (determine_roots_comp): Skip trivial components.
22437 2017-07-28 Richard Biener <rguenther@suse.de>
22439 * match.pd: Remove superfluous :c.
22440 * genmatch.c (simplify::id): Add member.
22441 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
22443 (current_id): New global.
22444 (dt_node::parent): Move from ...
22445 (dt_operand::parent): ... here. Add for_id member.
22446 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
22447 (decision_tree::find_node): Relax order requirement when
22448 merging DT_TRUE nodes to ones inbetween the current simplify
22449 and the one we try to merge with. Add diagnostic whenever
22450 we need to enforce pattern order by not merging.
22451 (decision_tree::insert): Set current_id.
22452 (decision_tree::print_node): Dump parent node and for_id.
22453 (parser::last_id): Add member.
22454 (parser::push_simplify): Assign unique id.
22455 (parser::parser): Initialize last_id.
22457 2017-07-28 Martin Liska <mliska@suse.cz>
22460 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
22461 gimple_build_debug_bind.
22463 2017-07-28 Richard Biener <rguenther@suse.de>
22465 PR tree-optimization/81502
22466 * match.pd: Add pattern combining BIT_INSERT_EXPR with
22468 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
22470 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
22471 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
22472 for BIT_FIELD_REF args.
22473 * fold-const.c (make_bit_field_ref): Likewise.
22474 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
22476 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22479 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
22480 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
22481 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
22482 Or it into SANITIZER_UNDEFINED.
22483 * ubsan.c: Include gimple-fold.h and varasm.h.
22484 (ubsan_expand_ptr_ifn): New function.
22485 (instrument_pointer_overflow): New function.
22486 (maybe_instrument_pointer_overflow): New function.
22487 (instrument_object_size): Formatting fix.
22488 (pass_ubsan::execute): Call instrument_pointer_overflow
22489 and maybe_instrument_pointer_overflow.
22490 * internal-fn.c (expand_UBSAN_PTR): New function.
22491 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
22492 * sanitizer.def (__ubsan_handle_pointer_overflow,
22493 __ubsan_handle_pointer_overflow_abort): New builtins.
22494 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
22495 * internal-fn.def (UBSAN_PTR): New internal function.
22496 * opts.c (sanitizer_opts): Add pointer-overflow.
22497 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
22498 * fold-const.c (build_range_check): Compute pointer range check in
22499 integral type if pointer arithmetics would be needed. Formatting
22502 2017-07-28 Martin Liska <mliska@suse.cz>
22505 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
22506 parameters that are of a variable-length.
22508 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
22510 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
22512 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
22513 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
22514 (CRT_CALL_STATIC_FUNCTION): Likewise.
22515 (ASM_DEFAULT_SPEC): New define.
22516 (ASM_SPEC32): Likewise.
22517 (ASM_SPEC64): Likewise.
22518 (ASM_SPEC_COMMON): Likewise.
22519 (ASM_SPEC): Likewise.
22520 (INVALID_64BIT): Likewise.
22521 (LINK_OS_DEFAULT_SPEC): Likewise.
22522 (LINK_OS_SPEC32): Likewise.
22523 (LINK_OS_SPEC64): Likewise.
22524 (POWERPC_LINUX): Likewise.
22525 (PTRDIFF_TYPE): Likewise.
22526 (RESTORE_FP_PREFIX): Likewise.
22527 (RESTORE_FP_SUFFIX): Likewise.
22528 (SAVE_FP_PREFIX): Likewise.
22529 (SAVE_FP_SUFFIX): Likewise.
22530 (SIZE_TYPE): Likewise.
22531 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
22532 (TARGET_64BIT): Likewise.
22533 (TARGET_64BIT): Likewise.
22534 (TARGET_AIX): Likewise.
22535 (WCHAR_TYPE_SIZE): Likewise.
22536 (WCHAR_TYPE): Undefine.
22537 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
22538 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
22539 (CPP_OS_RTEMS_SPEC): Delete.
22540 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
22541 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
22543 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
22545 2017-07-28 Jakub Jelinek <jakub@redhat.com>
22547 PR tree-optimization/81578
22548 * tree-parloops.c (build_new_reduction): Bail out if
22549 reduction_code isn't one of the standard OpenMP reductions.
22550 Move the details printing after that decision.
22552 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22554 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
22555 related to reload_in_progress.
22556 (splat_input_operand): Likewise.
22557 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
22559 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
22561 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
22562 (TARGET_INSTANTIATE_DECLS): Likewise.
22563 (legitimate_indexed_address_p): Delete reload_in_progress code.
22564 (rs6000_debug_legitimate_address_p): Likewise.
22565 (rs6000_eliminate_indexed_memrefs): Likewise.
22566 (rs6000_emit_le_vsx_store): Likewise.
22567 (rs6000_emit_move_si_sf_subreg): Likewise.
22568 (rs6000_emit_move): Likewise.
22569 (register_to_reg_type): Likewise.
22570 (rs6000_pre_atomic_barrier): Likewise.
22571 (rs6000_machopic_legitimize_pic_address): Likewise.
22572 (rs6000_allocate_stack_temp): Likewise.
22573 (rs6000_address_for_fpconvert): Likewise.
22574 (rs6000_address_for_altivec): Likewise.
22575 (rs6000_secondary_memory_needed_rtx): Delete function.
22576 (rs6000_check_sdmode): Likewise.
22577 (rs6000_alloc_sdmode_stack_slot): Likewise.
22578 (rs6000_instantiate_decls): Likewise.
22579 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
22580 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
22581 Delete reload_in_progress.
22582 (*vec_reload_and_plus_<mptrsize>): Likewise.
22583 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
22584 (vsx_div_v2di): Likewise.
22585 (vsx_udiv_v2di): Likewise.
22587 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22589 * config/rs6000/rs6000.opt (mlra): Replace with stub.
22590 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
22591 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
22592 (rs6000_debug_reg_global): Delete print of LRA status.
22593 (rs6000_option_override_internal): Delete dead LRA related code.
22594 (rs6000_lra_p): Delete function.
22595 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
22597 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
22599 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
22600 * config/riscv/rtems.h: New file.
22602 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22603 Sudakshina Das <sudi.das@arm.com>
22605 * config/aarch64/aarch64.md
22606 (define_split for and<mode>3nr_compare): Move
22607 non aarch64_logical_operand to a register.
22608 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
22609 register immediate operand to a register.
22610 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
22612 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
22614 PR middle-end/81564
22615 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
22617 2017-07-27 Richard Biener <rguenther@suse.de>
22619 PR tree-optimization/81573
22620 PR tree-optimization/81494
22621 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
22622 multi defuse cycle case.
22624 2017-07-27 Richard Biener <rguenther@suse.de>
22626 PR tree-optimization/81571
22627 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
22630 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
22632 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
22633 earlier and only if MASK_FPU is set. Adjust formatting.
22635 2017-07-27 Martin Liska <mliska@suse.cz>
22637 * opt-functions.awk: Add validation of value of Init.
22638 * optc-gen.awk: Pass new argument.
22640 2017-07-27 Martin Liska <mliska@suse.cz>
22642 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
22643 Fix wrong condition.
22645 2017-07-27 Martin Liska <mliska@suse.cz>
22647 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
22648 BBs and edges seen by autoFDO.
22650 2017-07-27 Richard Biener <rguenther@suse.de>
22652 PR tree-optimization/81502
22653 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
22654 with incompatible but same sized type.
22655 (execute_update_addresses_taken): Likewise.
22657 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
22659 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
22660 flag_tree_loop_vectorize rather than flag_tree_vectorize.
22662 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22665 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
22666 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
22667 Change s_operand to memory_operand.
22669 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
22671 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
22672 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
22673 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
22674 Emit instructions rather than returning an expression. Handle TFmode
22675 and KFmode by casting to TImode.
22676 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
22677 (rs6000_emit_le_vsx_store): Likewise.
22678 * config/rs6000/vsx.md (VSX_TI): New iterator.
22679 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
22680 (*vsx_le_undo_permute_<mode>): Likewise.
22681 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
22682 emit the split sequence.
22683 (*vsx_le_perm_store_<mode>): Likewise.
22685 2017-07-27 Jakub Jelinek <jakub@redhat.com>
22687 PR tree-optimization/81555
22688 PR tree-optimization/81556
22689 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
22690 if true, force CHANGED for the recursive invocation.
22691 (reassociate_bb): Remember original length of ops array, pass
22692 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
22694 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
22695 attributes for noipa attribute. For naked attribute use
22696 lookup_attribute first before lookup_attribute_spec.
22697 * final.c (rest_of_handle_final): Disable IPA RA for functions with
22699 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
22700 for functions with noipa attribute.
22701 (cgraph_externally_visible_p): Return true for functions with noipa
22703 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
22704 for functions with noipa attribute.
22705 * doc/extend.texi: Document noipa function attribute.
22706 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
22707 also for functions with noipa attribute.
22708 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
22710 2017-07-26 Andrew Pinski <apinski@cavium.com>
22712 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
22713 vec_unalign_load_cost and vec_unalign_store_cost.
22715 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
22717 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
22718 -mvsx-small-integer option.
22719 (ISA_3_0_MASKS_IEEE): Likewise.
22720 (OTHER_VSX_VECTOR_MASKS): Likewise.
22721 (POWERPC_MASKS): Likewise.
22722 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
22723 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
22724 code, only testing for DImode being allowed in non-VSX floating
22726 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
22727 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
22729 (rs6000_option_override_internal): Delete -mvsx-small-integer.
22730 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
22731 TARGET_P8_VECTOR test.
22732 (rs6000_secondary_reload_simple_move): Likewise.
22733 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
22734 since TARGET_P9_VECTOR was already tested.
22735 (rs6000_opt_masks): Remove -mvsx-small-integer.
22736 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
22737 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22739 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
22740 test for TARGET_VEXTRACTUB was used, and that uses
22742 (p9 extract splitter): Likewise.
22743 (vsx_extract_<mode>_di_p9): Likewise.
22744 (vsx_extract_<mode>_store_p9): Likewise.
22745 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
22746 for TARGET_P9_VECTOR was used. Delete code that is now dead with
22747 the elimination of TARGET_VSX_SMALL_INTEGER.
22748 (vsx_extract_<mode>_p8): Likewise.
22749 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
22750 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
22751 (vsx_set_<mode>_p9): Likewise.
22752 (vsx_set_v4sf_p9): Likewise.
22753 (vsx_set_v4sf_p9_zero): Likewise.
22754 (vsx_insert_extract_v4sf_p9): Likewise.
22755 (vsx_insert_extract_v4sf_p9_2): Likewise.
22756 * config/rs6000/rs6000.md (sign extend splitter): Change
22757 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
22758 (floatsi<mode>2_lfiwax_mem): Likewise.
22759 (floatunssi<mode>2_lfiwzx_mem): Likewise.
22760 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
22761 since a test for TARGET_P9_VECTOR was used.
22762 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22763 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
22764 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22765 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22766 TARGET_P8_VECTOR test.
22767 (fix_trunc<mode>si2_stfiwx): Likewise.
22768 (fix_trunc<mode>si2_internal): Likewise.
22769 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
22770 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22772 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22773 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
22774 TARGET_P8_VECTOR test.
22775 (fixuns_trunc<mode>si2_stfiwx): Likewise.
22776 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
22777 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
22779 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
22780 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
22781 since a test for TARGET_P9_VECTOR was used.
22782 (splitter for loading small constants): Likewise.
22784 2017-07-26 Andrew Pinski <apinski@cavium.com>
22786 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
22789 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
22792 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
22793 (fp_valid_at): Likewise.
22795 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22797 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
22798 (qdf24xx_addrcost_table): Likewise.
22799 (cortexa57_tunings): Update to use generic_branch_cost.
22800 (cortexa72_tunings): Likewise.
22801 (cortexa73_tunings): Likewise.
22802 (qdf24xx_tunings): Likewise.
22804 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
22806 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
22807 (thunderx2t99_branch_cost): Likewise.
22808 (cortexa35_tunings): Update to use generic_branch_cost.
22809 (cortexa53_tunings): Likewise.
22810 (cortexa57_tunings): Likewise.
22811 (cortexa72_tunings): Likewise.
22812 (cortexa73_tunings): Likewise.
22813 (thunderx2t99_tunings): Likewise.
22815 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22817 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
22818 (sparc_option_override): Honour MASK_FSMULD.
22819 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
22820 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
22821 * config/sparc/sparc.opt (mfsmuld): New option.
22822 * doc/invoke.texi (mfsmuld): Document option.
22824 2017-07-26 Marek Polacek <polacek@redhat.com>
22826 PR middle-end/70992
22827 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
22829 2017-07-26 Richard Biener <rguenther@suse.de>
22831 * gimple-match-head.c (do_valueize): Return OP if valueize
22833 (get_def): New helper to get at the def stmt of a SSA name
22834 if valueize allows.
22835 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
22836 do_valueize to get at the def stmt.
22837 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
22839 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
22841 PR middle-end/46932
22842 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
22844 2017-07-26 Martin Liska <mliska@suse.cz>
22847 * function.c (expand_function_start): Make expansion of
22848 nonlocal_goto_save_area after parm_birth_insn.
22850 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
22852 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
22853 from all CPU target flags enable members.
22855 2017-07-26 Richard Biener <rguenther@suse.de>
22857 * genmatch.c (dt_simplify::gen): Make iterator vars const.
22858 (decision_tree::gen): Make 'type' const.
22859 (write_predicate): Likewise.
22861 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
22863 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
22864 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
22865 (rs6000_option_override_internal): Likewise.
22866 (rs6000_expand_vector_set): Likewise.
22867 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
22868 (TARGET_UPPER_REGS_SF): Likewise.
22869 (TARGET_UPPER_REGS_DI): Likewise.
22870 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
22871 (TARGET_DIRECT_MOVE_64BIT): Likewise.
22872 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
22873 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
22874 (Splitters for DI constants in Altivec registers): Likewise.
22875 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
22876 (vsx_set_v4sf_p9): Likewise.
22877 (vsx_set_v4sf_p9_zero): Likewise.
22878 (vsx_insert_extract_v4sf_p9): Likewise.
22879 (vsx_insert_extract_v4sf_p9_2): Likewise.
22881 2017-07-25 Carl Love <cel@us.ibm.com>
22883 * doc/extend.texi: Update the built-in documentation file for the
22884 existing built-in functions
22885 vector signed char vec_cnttz (vector signed char);
22886 vector unsigned char vec_cnttz (vector unsigned char);
22887 vector signed short vec_cnttz (vector signed short);
22888 vector unsigned short vec_cnttz (vector unsigned short);
22889 vector signed int vec_cnttz (vector signed int);
22890 vector unsigned int vec_cnttz (vector unsigned int);
22891 vector signed long long vec_cnttz (vector signed long long);
22892 vector unsigned long long vec_cnttz (vector unsigned long long);
22894 2017-07-25 Andrew Pinski <apinski@cavium.com>
22896 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
22897 accesses where the use is for the first operand of a BIT_INSERT.
22899 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22902 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
22903 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
22905 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
22907 * config/i386/gstabs.h: Delete.
22908 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
22910 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
22912 * config/i386/i386.c (ix86_decompose_address): Do not check for
22913 register RTX when looking at index_reg or base_reg.
22914 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
22916 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
22918 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
22919 to update EH info here.
22921 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22923 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
22925 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
22927 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
22929 2017-07-25 Torsten Duwe <duwe@suse.de>
22931 * common.opt: Introduce -fpatchable-function-entry
22932 command line option, and its variables function_entry_patch_area_size
22933 and function_entry_patch_area_start.
22934 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
22935 including a two-value parser.
22936 * target.def (print_patchable_function_entry): New target hook.
22937 * targhooks.h (default_print_patchable_function_entry): New function.
22938 * targhooks.c (default_print_patchable_function_entry): Likewise.
22939 * toplev.c (process_options): Switch off IPA-RA if
22940 patchable function entries are being generated.
22941 * varasm.c (assemble_start_function): Look at the
22942 patchable-function-entry command line switch and current
22943 function attributes and maybe generate NOP instructions by
22944 calling the print_patchable_function_entry hook.
22945 * doc/extend.texi: Document patchable_function_entry attribute.
22946 * doc/invoke.texi: Document -fpatchable_function_entry
22947 command line option.
22948 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
22950 * doc/tm.texi: Re-generate.
22952 2017-07-25 Jakub Jelinek <jakub@redhat.com>
22955 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
22956 TARGET_AVX512DQ rather than TARGET_AVX512BW.
22958 2017-07-25 Tamar Christina <tamar.christina@arm.com>
22960 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
22962 2017-07-25 Richard Biener <rguenther@suse.de>
22964 PR tree-optimization/81455
22965 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
22966 not walk in cycles when looking for guards.
22968 2017-07-25 Richard Biener <rguenther@suse.de>
22970 PR tree-optimization/81529
22971 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
22972 when optimizing backedge uses.
22974 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
22976 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
22978 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
22979 to dl_section_ref. On AIX, append an expression to subtract
22980 the size of the section length to dl_section_ref.
22982 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
22984 * configure.ac: If any of the config.* scripts fail, exit 1.
22985 * configure: Regenerate.
22987 2017-07-25 Richard Biener <rguenther@suse.de>
22989 PR middle-end/81546
22990 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
22991 of immediate uses, be more verbose on errors.
22993 2017-07-25 Richard Biener <rguenther@suse.de>
22995 PR tree-optimization/81510
22996 * tree-vect-loop.c (vect_is_simple_reduction): When the
22997 reduction stmt is not inside the loop bail out.
22999 2017-07-25 Richard Biener <rguenther@suse.de>
23001 PR tree-optimization/81303
23002 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23003 profitability check against LOOP_VINFO_NITERSM1.
23005 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23007 * domwalk.c (cmp_bb_postorder): Simplify.
23008 (sort_bbs_postorder): New function. Use it...
23009 (dom_walker::walk): ...here to optimize common cases.
23011 2017-07-25 Martin Liska <mliska@suse.cz>
23014 * ipa-visibility.c (function_and_variable_visibility): Make the
23015 redirection just on target that supports aliasing.
23016 Fix GNU coding style.
23018 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23021 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23023 * config/arm/rtems.h: Likewise.
23024 * config/bfin/rtems.h: Likewise.
23025 * config/i386/rtemself.h: Likewise.
23026 * config/lm32/rtems.h: Likewise.
23027 * config/m32c/rtems.h: Likewise.
23028 * config/m68k/rtemself.h: Likewise.
23029 * config/microblaze/rtems.h: Likewise.
23030 * config/mips/rtems.h: Likewise.
23031 * config/moxie/rtems.h: Likewise.
23032 * config/nios2/rtems.h: Likewise.
23033 * config/powerpcspe/rtems.h: Likewise.
23034 * config/rs6000/rtems.h: Likewise.
23035 * config/rtems.h: Likewise.
23036 * config/sh/rtems.h: Likewise.
23037 * config/sh/rtemself.h: Likewise.
23038 * config/sparc/rtemself.h: Likewise.
23040 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23043 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23044 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23045 * tree-ssa-structalias.c (alias_get_name): Same.
23047 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23050 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23051 instructions if no du chain is found.
23053 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23055 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23057 2017-07-25 Richard Biener <rguenther@suse.de>
23059 PR middle-end/81505
23060 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23063 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23065 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23066 upper-regs options.
23067 (ISA_2_7_MASKS_SERVER): Likewise.
23068 (ISA_3_0_MASKS_IEEE): Likewise.
23069 (OTHER_P8_VECTOR_MASKS): Likewise.
23070 (OTHER_VSX_VECTOR_MASKS): Likewise.
23071 (POWERPC_MASKS): Likewise.
23072 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23073 duplicate list of options.
23074 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23075 explicit -mupper-regs options.
23076 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23077 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23078 alias for -mupper-regs-df.
23079 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23080 (rs6000_init_hard_regno_mode_ok): Likewise.
23081 (rs6000_option_override_internal): Likewise.
23082 (rs6000_opt_masks): Likewise.
23083 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23084 options in terms of whether -mvsx or -mpower8-vector was used.
23085 (TARGET_UPPER_REGS_DI): Likewise.
23086 (TARGET_UPPER_REGS_SF): Likewise.
23087 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23088 -mupper-regs-* options.
23090 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23092 * passes.c (emergency_dump_function): Print some empty lines and a
23093 header before the RTL dump.
23095 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23097 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23099 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23102 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23103 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23105 2017-07-24 Carl Love <cel@us.ibm.com>
23107 * config/rs6000/rs6000-c.c: Add support for built-in functions
23108 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23109 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23110 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23111 vec_extract_fp_from_shortl): Add defines for the two builtins.
23112 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23113 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23115 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23116 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23117 * doc/extend.texi: Update the built-in documentation file for the
23118 new built-in function.
23120 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23123 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23125 * doc/generic.texi: Likewise.
23126 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23127 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23129 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23131 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23132 (aarch64_mls_elt_merge<mode>): Likewise.
23134 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23136 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23137 having __cxa_atexit.
23139 2017-07-23 Michael Collison <michael.collison@arm.com>
23141 * config/arm/arm.c (arm_option_override): Deprecate
23142 use of -mstructure-size-boundary.
23143 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23144 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23146 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23149 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23150 Reduce cost estimate for direct moves.
23152 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23155 * config/i386/i386.c (ix86_option_override_internal): Disable
23156 BMI, BMI2 and TBM instructions for -m16.
23158 2017-07-21 Carl Love <cel@us.ibm.com>
23160 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23161 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23162 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23163 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23164 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23165 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23166 VMULOSW): New enum "unspec" values.
23167 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23168 altivec_vmulosw): New patterns.
23169 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23170 VMULOSW): Add definitions.
23172 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23174 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23175 (qdf24xx): Likewise.
23176 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23177 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23178 (AARCH64_FL_V8_1): Renumber.
23179 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23180 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23181 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23182 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23183 rdma to feature modifiers list.
23185 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23187 PR middle-end/56727
23188 * ipa-visibility (function_and_variable_visibility): Convert
23189 recursive PLT call to direct call if appropriate.
23191 2017-07-21 Andrew Pinski <apinski@cavium.com>
23193 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23194 operand 1 to see if the types precision matches.
23195 * fold-const.c (operand_equal_p): Likewise.
23197 2017-07-21 Richard Biener <rguenther@suse.de>
23199 PR tree-optimization/81303
23200 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23201 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23202 (vect_peeling_hash_get_lowest_cost): Adjust.
23203 (vect_enhance_data_refs_alignment): Likewise. Use
23204 vect_get_peeling_costs_all_drs to compute the penalty for no
23205 peeling to match up costs.
23207 2017-07-21 Richard Biener <rguenther@suse.de>
23209 PR tree-optimization/81500
23210 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23211 we didn't identify a reduction path.
23213 2017-07-21 Tom de Vries <tom@codesourcery.com>
23214 Cesar Philippidis <cesar@codesourcery.com>
23216 PR gcov-profile/81442
23217 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23220 2017-07-21 Tom de Vries <tom@codesourcery.com>
23223 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23225 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23226 nvptx_override_options_after_change.
23228 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23230 * dwarf2out.c (output_file_names): Avoid double testing for
23231 dwarf_version >= 5.
23233 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23235 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23237 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23239 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23241 (try_crossjump_to_edge): Do not punt on partitioned functions.
23243 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23245 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23246 Put all BBs reachable only via paths crossing cold region to cold
23248 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23250 2016-07-21 Richard Biener <rguenther@suse.de>
23252 PR tree-optimization/81303
23253 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23254 into account prologue and epilogue iterations when raising
23255 min_profitable_iters to sth at least covering one vector iteration.
23257 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23259 * config/arm/arm.c (arm_test_cpu_arch_dat):
23262 2017-07-20 Nathan Sidwell <nathan@acm.org>
23264 Remove TYPE_METHODS.
23265 * tree.h (TYPE_METHODS): Delete.
23266 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23267 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23268 (dbxout_type_methods): Scan TYPE_FIELDS.
23269 (dbxout_type): Don't check TYPE_METHODS here.
23270 * function.c (use_register_for_decl): Always ignore register for
23271 class types when not optimizing.
23272 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23273 * tree.c (free_lang_data_in_type): Stitch out member functions and
23274 templates from TYPE_FIELDS.
23275 (build_distinct_type_copy, verify_type_variant,
23276 verify_type): Member fns are on TYPE_FIELDS.
23277 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23278 * tree-pretty-print.c (dump_generic_node): Likewise.
23280 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23283 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23284 V2TImode and V4TImode.
23285 (ix86_expand_vector_extract): Likewise.
23286 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23287 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23288 (ssescalarmode): Handle V4TImode and V2TImode.
23289 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23290 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23291 (VEXTRACTI128_MODE): New mode iterator.
23292 (splitter for *vec_extractv?ti first element): New.
23293 (VEC_INIT_MODE): New mode iterator.
23294 (vec_init<mode>): Consolidate 3 expanders into one using
23295 VEC_INIT_MODE mode iterator.
23297 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23299 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23300 non_spilled_static_chain_regno_p.
23302 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23304 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23306 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23308 * bb-reorder.c (connect_traces): Allow copying of blocks within
23311 2017-07-20 Richard Biener <rguenther@suse.de>
23313 * gimple.h (gimple_phi_result): Add gphi * overload.
23314 (gimple_phi_result_ptr): Likewise.
23315 (gimple_phi_arg): Likewise. Adjust index assert to only
23316 allow actual argument accesses rather than all slots available
23318 (gimple_phi_arg_def): Add gphi * overload.
23319 * tree-phinodes.c (make_phi_node): Initialize only actual
23321 (resize_phi_node): Clear memory not covered by old node,
23322 do not initialize excess argument slots.
23323 (reserve_phi_args_for_new_edge): Initialize new argument slot
23326 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23328 PR tree-optimization/81388
23330 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23332 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23333 by removing computation of may_be_zero.
23335 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23336 Tom de Vries <tom@codesourcery.com>
23338 PR middle-end/81030
23339 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23340 when gimple level profile disagrees with what RTL expander did.
23342 2017-07-20 Richard Biener <rguenther@suse.de>
23344 PR tree-optimization/61171
23345 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23346 (vect_analyze_stmt): Add slp instance parameter.
23347 (vectorizable_reduction): Likewise.
23348 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23349 (vect_is_simple_reduction): Deal with chains not detected
23350 as SLP reduction chain, specifically not properly associated
23351 chains containing a mix of plus/minus.
23352 (get_reduction_op): Remove.
23353 (get_initial_defs_for_reduction): Simplify, pass in whether
23354 this is a reduction chain, pass in the SLP node for the PHIs.
23355 (vect_create_epilog_for_reduction): Get the SLP instance as
23357 (vectorizable_reduction): Get the SLP instance as arg.
23358 During analysis remember the SLP node with the PHIs in the
23359 instance. Simplify getting at the vectorized reduction PHIs.
23360 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23361 through SLP instance.
23362 (vect_slp_analyze_operations): Likewise.
23363 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23364 (vect_transform_stmt): Likewise.
23366 2017-07-20 Tom de Vries <tom@codesourcery.com>
23368 PR tree-optimization/81489
23369 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23370 read of phi arg location to before loop that modifies phi.
23372 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23374 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23377 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23379 PR middle-end/81331
23380 * except.c (execute): Fix ordering issue.
23382 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23384 PR rtl-optimization/81423
23385 * combine.c (make_compound_operation_int): Don't try to optimize
23386 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23388 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23390 PR rtl-optimization/81423
23391 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23392 with a constant that is -1 in the truncated to mode.
23394 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23396 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
23397 (determine_unlikely_bbs): ... here.
23398 * predict.h (propagate_unlikely_bbs_forward): Declare.
23399 * cfgexpand.c (pass_expand::execute): Use it.
23400 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
23402 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
23403 propagate_unlikely_bbs_forward.
23405 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23407 PR middle-end/81331
23408 * except.c (maybe_add_nop_after_section_switch): New function.
23411 2017-07-19 Tom de Vries <tom@codesourcery.com>
23413 * gimple.h (gimple_phi_set_arg): Make assert more strict.
23415 2017-07-19 Tom de Vries <tom@codesourcery.com>
23417 * gimple.h (gimple_phi_arg): Make assert more strict.
23419 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
23421 * config.gcc (powerpc*-*-*): Add mmintrin.h.
23422 * config/rs6000/mmintrin.h: New file.
23423 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
23425 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23427 PR tree-optimization/81346
23428 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
23430 2017-07-19 Tom de Vries <tom@codesourcery.com>
23432 * config/nvptx/nvptx.md (VECIM): Add V2DI.
23434 2017-07-19 Tom de Vries <tom@codesourcery.com>
23436 * config/nvptx/nvptx-modes.def: Add V2DImode.
23437 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
23438 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
23439 (nvptx_output_mov_insn): Handle lack of mov.b128.
23440 (nvptx_print_operand): Handle 'H' and 'L' codes.
23441 (nvptx_vector_mode_supported): Allow V2DImode.
23442 (nvptx_preferred_simd_mode): New function.
23443 (nvptx_data_alignment): New function.
23444 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
23445 nvptx_preferred_simd_mode.
23446 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
23448 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
23450 2017-07-19 Tom de Vries <tom@codesourcery.com>
23452 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
23453 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
23454 (nvptx_vector_mode_supported): New function. Allow V2SImode.
23455 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
23456 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
23457 (mov<VECIM>_insn): New define_insn.
23458 (define_expand "mov<VECIM>): New define_expand.
23460 2017-07-19 Tom de Vries <tom@codesourcery.com>
23462 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
23464 2017-07-19 Jakub Jelinek <jakub@redhat.com>
23466 PR tree-optimization/81346
23467 * fold-const.h (fold_div_compare, range_check_type): Declare.
23468 * fold-const.c (range_check_type): New function.
23469 (build_range_check): Use range_check_type.
23470 (fold_div_compare): No longer static, rewritten into
23471 a match.pd helper function.
23472 (fold_comparison): Don't call fold_div_compare here.
23473 * match.pd (X / C1 op C2): New optimization using fold_div_compare
23474 as helper function.
23476 2017-07-19 Nathan Sidwell <nathan@acm.org>
23478 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
23479 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
23480 * tree.c (find_decls_types_r, verify_type): Use
23481 TYPE_{MIN,MAX}_VALUE_RAW.
23482 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
23483 (hash_tree): Likewise.
23484 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
23486 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
23489 2017-07-18 Tom de Vries <tom@codesourcery.com>
23491 PR middle-end/81464
23492 * omp-expand.c (expand_omp_for_static_chunk): Handle
23493 equal-argument loop exit phi.
23495 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
23498 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
23499 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
23500 operand 2 predicate.
23501 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
23502 operand 2 predicate.
23503 (ror,rol -> rorx splitters): Use const_int_operand as
23504 operand 2 predicate.
23506 2017-06-18 Richard Biener <rguenther@suse.de>
23508 PR tree-optimization/81410
23509 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
23510 the gap in the ! slp_perm SLP case after each group.
23512 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23514 PR middle-end/81463
23515 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
23518 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23520 PR middle-end/81462
23521 * predict.c (set_even_probabilities): Cleanup; do not affect
23522 probabilities that are already known.
23523 (combine_predictions_for_bb): Call even when count is set.
23525 2017-07-18 Nathan Sidwell <nathan@acm.org>
23527 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
23530 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23533 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
23534 optimization for loop niter analysis.
23536 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
23539 * config/avr/avr.c (avr_optimize_casesi): Don't use
23540 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
23542 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
23544 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
23545 body_cost_vec from _vect_peel_extended_info.
23546 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
23547 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
23550 2017-07-18 Bin Cheng <bin.cheng@arm.com>
23552 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
23554 2017-07-18 Richard Biener <rguenther@suse.de>
23556 PR tree-optimization/80620
23557 PR tree-optimization/81403
23558 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
23559 info when re-using a VN table entry.
23561 2017-07-18 Richard Biener <rguenther@suse.de>
23563 PR tree-optimization/81418
23564 * tree-vect-loop.c (vectorizable_reduction): Properly compute
23565 vectype_in. Verify that with lane-reducing reduction operations
23566 we have a single def-use cycle.
23568 2017-07-17 Carl Love <cel@us.ibm.com>
23570 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
23572 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23573 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23574 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23575 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23576 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23577 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23578 VMULOSW): New enum "unspec" values.
23579 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
23580 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
23581 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23582 altivec_vmulosw): New patterns.
23583 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23584 VMULOSW): Add definitions.
23586 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
23588 * config/alpha/alpha.c: Include predict.h.
23590 2017-07-17 Yury Gribov <tetra2005@gmail.com>
23592 * tree-vrp.c (compare_assert_loc): Fix comparison function
23593 to return predictable results.
23595 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23597 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
23599 (subdi3): Likewise.
23600 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
23601 * doc/invoke.texi (mexpand-adddi): Update text.
23603 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23605 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
23606 that also clobbers the CC register. The old expand code is moved
23608 (*arc_clzsi2): ... here.
23609 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
23610 the CC register. The old expand code is moved to ...
23611 (arc_ctzsi2): ... here.
23613 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23615 * config/arc/arc.opt (mindexed-loads): Use initial value
23616 TARGET_INDEXED_LOADS_DEFAULT.
23617 (mauto-modify-reg): Use initial value
23618 TARGET_AUTO_MODIFY_REG_DEFAULT.
23619 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23620 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23621 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
23622 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
23624 2017-07-17 Martin Liska <mliska@suse.cz>
23627 * opts.c (finish_options): Do not allow -fgnu-tm
23628 w/ -fsanitize={kernel-,}address. Say sorry.
23630 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23633 * tree-loop-distribution.c (classify_partition): Only assert on
23634 numer of iterations.
23635 (merge_dep_scc_partitions): Delete prameter. Update function call.
23636 (distribute_loop): Remove code handling loop with unknown niters.
23637 (pass_loop_distribution::execute): Skip loop with unknown niters.
23639 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23642 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
23643 function sort_partitions_by_post_order.
23645 2017-07-17 Bin Cheng <bin.cheng@arm.com>
23647 PR tree-optimization/81374
23648 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
23649 the max index of basic blocks, rather than number of basic blocks.
23651 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23653 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
23655 (arc_legitimate_pic_operand_p): Likewise.
23656 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
23658 (arc_needs_pcl_p): Likewise.
23659 (arc_legitimate_pc_offset_p): Likewise.
23660 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
23661 function is also used in constrains.md.
23662 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
23663 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
23664 PLUS. Only return true/false in known cases, otherwise assert.
23665 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
23666 is already called in arc_legitimate_constant_p.
23667 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
23669 (LEGITIMATE_PIC_OPERAND_P): Use
23670 arc_raw_symbolic_reference_mentioned_p function.
23671 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
23676 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
23677 Andrew Burgess <andrew.burgess@embecosm.com>
23679 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
23680 (arc_return_address_register): New function.
23681 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
23682 (arc_handle_fndecl_attribute): Add naked attribute.
23683 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
23684 (TARGET_WARN_FUNC_RETURN): Likewise.
23685 (arc_allocate_stack_slots_for_args): New function.
23686 (arc_warn_func_return): Likewise.
23687 (machine_function): Change type fn_type.
23688 (arc_compute_function_type): Consider new naked function type,
23689 change function return type.
23690 (arc_must_save_register): Adapt to handle new
23691 arc_compute_function_type's return type.
23692 (arc_expand_prologue): Likewise.
23693 (arc_expand_epilogue): Likewise.
23694 (arc_return_address_regs): Delete.
23695 (arc_return_address_register): New function.
23696 (arc_epilogue_uses): Use above function.
23697 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
23698 (arc_function_type): Change encoding, add naked type.
23699 (ARC_INTERRUPT_P): Change to handle the new encoding.
23700 (ARC_FAST_INTERRUPT_P): Likewise.
23701 (ARC_NORMAL_P): Define.
23702 (ARC_NAKED_P): Likewise.
23703 (arc_compute_function_type): Delete prototype.
23704 * config/arc/arc.md (in_ret_delay_slot): Use
23705 arc_return_address_register function.
23706 (simple_return): Likewise.
23707 (p_return_i): Likewise.
23709 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23711 PR tree-optimization/81428
23712 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
23713 can't be built for those types.
23715 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23717 Remove stuff dead since r239246.
23719 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
23720 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
23721 (avr_inform_devices): Remove dead stuff.
23723 2017-07-17 Tamar Christina <tamar.christina@arm.com>
23725 * config/arm/arm_neon.h: Fix softp typo.
23727 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23729 PR tree-optimization/81365
23730 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
23731 aggregate moves onto bb predecessor edges, make sure there are no
23732 loads that could alias the lhs in between the start of bb and the
23735 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
23738 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
23739 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
23740 [LSHIFTRT, outer_code = TRUNCATE]: Same.
23742 2017-07-17 Jakub Jelinek <jakub@redhat.com>
23744 PR tree-optimization/81396
23745 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
23746 (init_symbolic_number): Initialize it to 1.
23747 (perform_symbolic_merge): Add n_ops from both operands into the new
23749 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
23750 without base_addr as useless if they need more than one operation.
23751 (bswap_replace): Handle !bswap case for NULL base_addr.
23753 2017-07-17 Tom de Vries <tom@codesourcery.com>
23756 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
23759 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
23761 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
23762 conditional builtin define __FIX_LEON3FT_B2BST.
23764 2017-07-17 Daniel Cederman <cederman@gaisler.com>
23766 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
23767 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
23770 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23772 PR rtl-optimization/81424
23773 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
23774 to remove potential trapping from operands if -fnon-call-exceptions.
23776 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23778 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
23779 profile_proability for scalling.
23780 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
23782 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23784 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
23786 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23788 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
23789 fixpoint arithmetics.
23791 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23793 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
23794 fixpoint arithmetics.
23796 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23798 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
23799 fixpoint arithmetics.
23801 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
23803 * profile-count.h (profile_probability::from_reg_br_prob_note,
23804 profile_probability::to_reg_br_prob_note): New functions.
23805 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
23806 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
23807 * predict.c (probability_reliable_p): Update.
23808 (edge_probability_reliable_p): Update.
23809 (br_prob_note_reliable_p): Update.
23810 (invert_br_probabilities): Update.
23811 (add_reg_br_prob_note): New function.
23812 (combine_predictions_for_insn): Update.
23813 * asan.c (asan_clear_shadow): Update.
23814 * cfgbuild.c (compute_outgoing_frequencies): Update.
23815 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
23816 (update_br_prob_note): Update.
23817 (rtl_verify_edges): Update.
23818 (purge_dead_edges): Update.
23819 (fixup_reorder_chain): Update.
23820 * emit-rtl.c (try_split): Update.
23821 * ifcvt.c (cond_exec_process_insns): Update.
23822 (cond_exec_process_if_block): Update.
23823 (dead_or_predicable): Update.
23824 * internal-fn.c (expand_addsub_overflow): Update.
23825 (expand_neg_overflow): Update.
23826 (expand_mul_overflow): Update.
23827 * loop-doloop.c (doloop_modify): Update.
23828 * loop-unroll.c (compare_and_jump_seq): Update.
23829 * optabs.c (emit_cmp_and_jump_insn_1): Update.
23830 * predict.h: Update.
23831 * reorg.c (mostly_true_jump): Update.
23833 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
23834 * config/alpha/alpha.c (emit_unlikely_jump): Update.
23835 * config/arc/arc.c: (emit_unlikely_jump): Update.
23836 * config/arm/arm.c: (emit_unlikely_jump): Update.
23837 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
23838 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
23839 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
23840 (ix86_print_operand): Update.
23841 (ix86_split_fp_branch): Update.
23842 (predict_jump): Update.
23843 * config/ia64/ia64.c (ia64_print_operand): Update.
23844 * config/mmix/mmix.c (mmix_print_operand): Update.
23845 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
23846 (rs6000_expand_split_stack_prologue): Update.
23847 * config/rs6000/rs6000.c: Update.
23848 * config/s390/s390.c (s390_expand_vec_strlen): Update.
23849 (s390_expand_vec_movstr): Update.
23850 (s390_expand_cs_tdsi): Update.
23851 (s390_expand_split_stack_prologue): Update.
23852 * config/sh/sh.c (sh_print_operand): Update.
23853 (expand_cbranchsi4): Update.
23854 (expand_cbranchdi4): Update.
23855 * config/sparc/sparc.c (output_v9branch): Update.
23856 * config/spu/spu.c (get_branch_target): Update.
23857 (ea_load_store_inline): Update.
23858 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
23859 * config/tilepro/tilepro.c: Update.
23861 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
23863 * gimplify.c (mostly_copy_tree_r): Revert latest change.
23864 (gimplify_save_expr): Likewise.
23866 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23868 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
23870 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23872 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
23874 * timevar.def (TV_IPA_FNSUMMARY): Define.
23876 2017-07-16 Daniel Cederman <cederman@gaisler.com>
23878 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
23879 to back store errata sensitive sequence from being generated.
23880 (sqrtdf2_fix): Likewise.
23882 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
23884 * tree-ssa-threadupdate.c (compute_path_counts,
23885 update_joiner_offpath_counts): Use profile_probability.
23887 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
23890 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23892 * config/arm/arm-c.c (arm_cpu_builtins): Define
23893 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23895 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23897 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23898 array entries to represent __ieee128 versions of the
23899 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
23900 scalar_extract_sig, and scalar_insert_exp built-in functions.
23901 (altivec_resolve_overloaded_builtin): Add special case handling
23902 for the __builtin_scalar_insert_exp function, as represented by
23903 the P9V_BUILTIN_VEC_VSIEDP constant.
23904 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
23905 exponent support for __ieee128 argument.
23906 (VSESQP): Add scalar extract signature support for __ieee128
23908 (VSTDCNQP): Add scalar test negative support for __ieee128
23910 (VSIEQP): Add scalar insert exponent support for __int128 argument
23911 with __ieee128 result.
23912 (VSIEQPF): Add scalar insert exponent support for __ieee128
23913 argument with __ieee128 result.
23914 (VSTDCQP): Add scalar test data class support for __ieee128
23916 (VSTDCNQP): Add overload support for scalar test negative with
23917 __ieee128 argument.
23918 (VSTDCQP): Add overload support for scalar test data class
23919 __ieee128 argument.
23920 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
23921 UNSPEC_VSX_SXSIGDP.
23922 (UNSPEC_VSX_SIEXPQP): New constant.
23923 (xsxexpqp): New insn for VSX scalar extract exponent quad
23925 (xsxsigqp): New insn for VSX scalar extract significand quad
23927 (xsiexpqpf): New insn for VSX scalar insert exponent quad
23928 precision with floating point argument.
23929 (xststdcqp): New expand for VSX scalar test data class quad
23931 (xststdcnegqp): New expand for VSX scalar test negative quad
23933 (xststdcqp): New insn to match expansions for VSX scalar test data
23934 class quad precision and VSX scalar test negative quad precision.
23935 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
23936 special case operand checking to enforce that second operand of
23937 VSX scalar test data class with quad precision argument is a 7-bit
23939 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23940 prototypes and descriptions of __ieee128 versions of
23941 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
23942 scalar_test_data_class, and scalar_test_neg built-in functions.
23944 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23946 PR tree-optimization/81162
23947 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
23948 replace a negate with an add.
23950 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
23952 * doc/invoke.texi (arm/-mcpu): Document +crypto.
23954 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23956 * config/arm/arm-c.c (arm_cpu_builtins): Define
23957 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
23959 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23961 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
23962 (armv8-r): Set ARM Cortex-R52 as default CPU.
23963 * config/arm/arm-tables.opt: Regenerate.
23964 * config/arm/arm-tune.md: Regenerate.
23965 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23967 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
23968 extension for -mcpu=cortex-r52.
23970 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
23972 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
23973 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
23974 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
23975 (fp-armv8): Define it as FP_ARMv8 only.
23976 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
23977 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
23979 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
23980 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
23981 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
23982 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
23983 than TARGET_FPU_ARMV8.
23984 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
23985 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
23986 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
23988 * config/arm/neon.md (neon_vrint): Likewise.
23989 (neon_vcvt): Likewise.
23990 (neon_<fmaxmin_op><mode>): Likewise.
23991 (<fmaxmin><mode>3): Likewise.
23992 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
23993 * config/arm/predicates.md (arm_cond_move_operator): Check against
23994 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
23996 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
23998 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
23999 to top of function.
24001 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24003 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24004 loop in comment with memset.
24006 2017-07-14 Martin Liska <mliska@suse.cz>
24008 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24009 * dwarf2out.c (is_java): Remove the function.
24010 (output_pubname): Remove usage of the function.
24011 (lower_bound_default): Remove usage of DW_LANG_Java.
24012 (gen_compile_unit_die): Likewise.
24013 * gcc.c: Remove compiler defaults for .java and .zip files.
24014 * gimple-expr.c (remove_suffix): Change as there's no longer
24015 extension than 4-letter one.
24016 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24017 (gimplify_save_expr): Likewise.
24018 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24019 as it's possible even for other languages than Java.
24020 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24021 * lto-opts.c (lto_write_options): Remove reference to Java.
24022 * opts.c (strip_off_ending): Update file extension handling.
24023 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24024 * tree-eh.c (lower_resx): Likewise.
24025 * tree.c (free_lang_data_in_type): Remove dead code.
24026 (find_decls_types_r): Likewise.
24027 (build_common_builtin_nodes): Remove Java from a comment.
24028 (verify_type): Remove dead code.
24029 * varasm.c (assemble_external): Remove Java from a comment.
24031 2017-07-14 Martin Liska <mliska@suse.cz>
24033 * opts.c (finish_options): Add quotes.
24034 (common_handle_option): Likewise.
24036 2017-07-14 Martin Liska <mliska@suse.cz>
24038 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24039 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24040 Remove N_SO_PASCAL.
24041 * dwarf2out.c (lower_bound_default): Do not handle
24043 (gen_compile_unit_die): Likewise.
24044 * gcc.c: Remove default extension binding for GNU Pascal.
24045 * stmt.c: Remove Pascal language from a comment.
24046 * xcoffout.c: Likewise.
24048 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24051 * diagnostic-show-locus.c (fixit_cmp): New function.
24052 (layout::layout): Sort m_fixit_hints.
24053 (column_range::column_range): Assert that the values are valid.
24054 (struct char_span): New struct.
24055 (correction::overwrite): New method.
24056 (struct source_line): New struct.
24057 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24058 calls in terms of classes source_line and char_span, and
24059 correction::overwrite.
24060 (selftest::test_overlapped_fixit_printing_2): New function.
24061 (selftest::diagnostic_show_locus_c_tests): Call it.
24063 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24065 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24066 early if there is no lhs.
24068 2017-07-13 Martin Liska <mliska@suse.cz>
24070 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24071 (gen_reference_type_die): Likewise.
24072 * stor-layout.c: Remove Pascal-related comment.
24074 2017-07-13 Martin Liska <mliska@suse.cz>
24076 * opts.c (finish_options): Add quotes to error messages.
24077 (parse_sanitizer_options): Likewise.
24079 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24081 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24083 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24085 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24087 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24089 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24091 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24093 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24096 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24097 provides the hardware capability bits, define the macro
24098 __BUILTIN_CPU_SUPPORTS__.
24099 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24100 if GLIBC does not provide the hardware capability bits. Add a
24101 gcc_unreachable call if the built-in cpu function is neither
24102 __builtin_cpu_is nor __builtin_cpu_supports.
24103 (rs6000_get_function_versions_dispatcher): Change the warning
24104 that an old GLIBC is used which does not export the capability
24105 bits to be an error.
24106 * doc/extend.texi (target_clones attribute): Document the
24107 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24108 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24109 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24110 the macros defined by GCC if the newer GLIBC is available.
24112 2017-07-12 Jeff Law <law@redhat.com>
24114 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24115 remaining includes slightly.
24116 * config/riscv/riscv-builtins.c: Include profile-count.h.
24118 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24121 * config/avr/avr.c (avr_set_current_function): In diagnostic
24122 messages: Quote keywords and (parts of) identifiers.
24123 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24126 2017-07-12 Carl Love <cel@us.ibm.com>
24128 * config/rs6000/rs6000-c.c: Add support for built-in functions
24129 vector bool char vec_revb (vector bool char);
24130 vector bool short vec_revb (vector short char);
24131 vector bool int vec_revb (vector bool int);
24132 vector bool long long vec_revb (vector bool long long);
24133 * doc/extend.texi: Update the built-in documentation file for the
24134 new built-in functions.
24136 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24138 * config/s390/s390.md: Remove movcc splitter.
24140 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24142 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24143 load/store on condition.
24145 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24148 * config/avr/avr.c (avr_encode_section_info)
24149 [progmem && !TREE_READONLY]: Error if progmem object needs
24152 2017-07-11 Michael Collison <michael.collison@arm.com>
24154 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24157 2017-07-11 Carl Love <cel@us.ibm.com>
24159 * config/rs6000/rs6000-c.c: Add support for builtins
24160 vector unsigned int vec_parity_lsbb (vector signed int);
24161 vector unsigned int vec_parity_lsbb (vector unsigned int);
24162 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24163 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24164 vector unsigned long long vec_parity_lsbb (vector signed long long);
24165 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24166 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24167 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24168 * doc/extend.texi: Update the built-in documentation file for the
24169 new built-in functions.
24171 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24173 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24174 (layout::m_primary_loc): New field.
24175 (layout::layout): Initialize new field. Move location filtering
24176 logic from here to...
24177 (layout::maybe_add_location_range): ...this new method. Add
24178 support for filtering to just the lines already specified by other
24180 (layout::will_show_line_p): New method.
24181 (gcc_rich_location::add_location_if_nearby): New method.
24182 (selftest::test_add_location_if_nearby): New test function.
24183 (selftest::diagnostic_show_locus_c_tests): Call it.
24184 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24187 2017-07-11 Tom de Vries <tom@codesourcery.com>
24189 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24190 (bb_first_real_insn): New function.
24191 (nvptx_single): Add extra initialization of broadcasted condition
24194 2017-07-11 Nathan Sidwell <nathan@acm.org>
24196 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24198 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24200 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24201 Binutils doc as TEXI will mess them up.
24202 * doc/invoke.texi (AVR Options): Same here.
24204 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24206 * config/sparc/sparc.opt (mfix-ut700): New option.
24207 (mfix-gr712rc): Likewise.
24208 (sparc_fix_b2bst): New variable.
24209 * doc/invoke.texi (SPARC options): Document them.
24210 (ARM options): Fix warnings.
24211 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24212 instructions to prevent sequences that can trigger the store-store
24213 errata for certain LEON3FT processors.
24214 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24215 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24216 * config/sparc/sparc.md (fix_b2bst): New attribute.
24217 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24219 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24222 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24224 (*rsqrtsf2_sse): Ditto.
24226 (div<mode>3): Macroize insn from divdf3 and divsf3
24227 using MODEF mode iterator.
24229 2017-07-10 Martin Sebor <msebor@redhat.com>
24231 PR tree-optimization/80397
24232 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24233 instead of testing for equality to INTEGER_TYPE.
24235 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24237 * config.gcc: Remove uclibc from arc target spec.
24239 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24241 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24243 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24246 * lto-wrapper.c (remove_option): New function.
24247 (merge_and_complain): Merge PIC/PIE options more realistically.
24249 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24251 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24255 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24256 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24257 * config.in: Regenerate.
24258 * configure: Regenerate.
24259 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24260 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24261 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24262 (TARGET_GASISR_PROLOGUES): ...target mask.
24263 * common/config/avr/avr-common.c
24264 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24265 Set -mgas-isr-prologues.
24266 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24267 INSERT_PASS_BEFORE for it.
24268 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24269 * config/avr/avr.c (avr_option_override)
24270 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24271 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24272 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24273 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24274 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24276 (make_avr_pass_pre_proep): New function.
24277 (emit_push_sfr) <treg>: Add argument to function and use it
24278 instead of TMP_REG.
24279 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24280 and set machine->gasisr.yes.
24281 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24282 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24283 __gcc_isr.n_pushed to .L__stack_usage.
24284 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24285 (avr_asm_final_postscan_insn): ...this new static function.
24286 * config/avr/avr.h (machine_function)
24287 <is_no_gccisr, use_L__stack_usage>: New fields.
24288 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24289 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24290 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24291 (gasisr, *gasisr): New expander and insn.
24292 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24293 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24294 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24296 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24298 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24301 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24303 Move jump-tables out of .text again.
24306 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24307 (ASM_OUTPUT_ADDR_VEC): New function.
24308 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24309 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24310 INSN_ADDRESSes as asm comment.
24311 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24312 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24313 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24314 * config/avr/avr.md (*tablejump): Adjust comment.
24315 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24316 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24318 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24319 (avr_output_addr_vec): New proto.
24320 (avr_log_t) <insn_addresses>: New field.
24322 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24325 * config/i386/i386.c (ix86_function_arg_advance): Set
24326 outgoing_args_on_stack to true if there are outgoing arguments
24328 (ix86_function_arg): Likewise.
24329 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24330 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24331 * config/i386/i386.h (machine_function): Add
24332 outgoing_args_on_stack.
24334 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24336 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24337 supporting pthreds.
24338 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24340 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24342 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24343 (REAL_H): Remove $(MACHMODE_H).
24344 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24346 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24347 $(MACHMODE_H) and double-int.h.
24348 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24350 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24353 2017-07-07 Andrew Pinski <apinski@cavium.com>
24355 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24356 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24358 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24360 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24361 Add warning if GCC was not configured to link against a GLIBC that
24362 exports the hardware capability bits.
24363 (make_resolver_func): Make resolver function private and not a
24364 COMDAT function. Create the name with clone_function_name instead
24365 of make_unique_name.
24368 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24369 correct operand in doing the split.
24371 2017-07-07 Carl Love <cel@us.ibm.com>
24373 * config/rs6000/rs6000-c: Add support for built-in function
24374 vector unsigned short vec_pack_to_short_fp32 (vector float,
24376 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24377 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24378 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24379 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24380 (convert_4f32_8i16): Add define_expand.
24381 * doc/extend.texi: Update the built-in documentation file for the
24382 new built-in function.
24384 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24386 * config/sparc/m8.md: New file.
24387 * config/sparc/sparc.md: Include m8.md.
24389 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24391 * config/sparc/sparc.opt: New option -mvis4b.
24392 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
24393 (sparc_option_override): Handle VIS4B.
24394 (enum sparc_builtins): Define
24395 SPARC_BUILTIN_DICTUNPACK{8,16,32},
24396 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
24397 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
24398 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
24399 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
24400 (check_constant_argument): New function.
24401 (sparc_vis_init_builtins): Define builtins
24402 __builtin_vis_dictunpack{8,16,32},
24403 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
24404 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
24405 __builtin_vis_fpcmpde{8,16,32}shl and
24406 __builtin_vis_fpcmpur{8,16,32}shl.
24407 (sparc_expand_builtin): Check that the constant operands to
24408 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
24409 constant and in range.
24410 * config/sparc/sparc-c.c (sparc_target_macros): Handle
24412 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
24413 (SPARC_IMM5_P): Likewise.
24414 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
24415 (enabled): Handle vis4b.
24416 (UNSPEC_DICTUNPACK): New unspec.
24417 (UNSPEC_FPCMPSHL): Likewise.
24418 (UNSPEC_FPUCMPSHL): Likewise.
24419 (UNSPEC_FPCMPDESHL): Likewise.
24420 (UNSPEC_FPCMPURSHL): Likewise.
24421 (cpu_feature): New CPU feature `vis4b'.
24422 (dictunpack{8,16,32}): New insns.
24423 (FPCSMODE): New mode iterator.
24424 (fpcscond): New code iterator.
24425 (fpcsucond): Likewise.
24426 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
24427 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
24428 (fpcmpde{8,16,32}{si,di}shl): Likewise.
24429 (fpcmpur{8,16,32}{si,di}shl): Likewise.
24430 * config/sparc/constraints.md: Define constraints `q' for unsigned
24431 2-bit integer constants and `t' for unsigned 5-bit integer
24433 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
24435 (imm5_operand_dictunpack16): Likewise.
24436 (imm5_operand_dictunpack32): Likewise.
24437 (imm2_operand): Likewise.
24438 * doc/invoke.texi (SPARC Options): Document -mvis4b.
24439 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
24440 ditunpack* and fpcmp*shl builtins.
24442 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24444 * config.gcc: Handle m8 in --with-{cpu,tune} options.
24445 * config.in: Add HAVE_AS_SPARC6 define.
24446 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
24448 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
24450 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
24451 (CPP_CPU_SPEC): Handle m8.
24452 (ASM_CPU_SPEC): Likewise.
24453 * config/sparc/sparc-opts.h (enum processor_type): Add
24455 * config/sparc/sparc.c (m8_costs): New struct.
24456 (sparc_option_override): Handle TARGET_CPU_m8.
24457 (sparc32_initialize_trampoline): Likewise.
24458 (sparc64_initialize_trampoline): Likewise.
24459 (sparc_issue_rate): Likewise.
24460 (sparc_register_move_cost): Likewise.
24461 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
24462 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
24463 (ASM_CPU64_DEFAULT_SPEC): Likewise.
24464 (CPP_CPU_SPEC): Handle M8.
24465 (ASM_CPU_SPEC): Likewise.
24466 (AS_M8_FLAG): Define.
24467 * config/sparc/sparc.md: Add m8 to the cpu attribute.
24468 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
24469 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
24471 * configure: Regenerate.
24472 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
24475 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24477 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
24479 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
24480 ("*movdi_insn_sp32"): Do not set v3pipe.
24481 ("*movsi_insn"): Likewise.
24482 ("*movdi_insn_sp64"): Likewise.
24483 ("*movsf_insn"): Likewise.
24484 ("*movdf_insn_sp32"): Likewise.
24485 ("*movdf_insn_sp64"): Likewise.
24486 ("*zero_extendsidi2_insn_sp64"): Likewise.
24487 ("*sign_extendsidi2_insn"): Likewise.
24488 ("*mov<VM32:mode>_insn"): Likewise.
24489 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24490 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24491 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24492 ("<vlop:code><VL:mode>3"): Likewise.
24493 ("*not_<vlop:code><VL:mode>3"): Likewise.
24494 ("*nand<VL:mode>_vis"): Likewise.
24495 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
24496 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
24497 ("one_cmpl<VL:mode>2"): Likewise.
24498 ("faligndata<VM64:mode>_vis"): Likewise.
24499 ("alignaddrsi_vis"): Likewise.
24500 ("alignaddrdi_vis"): Likweise.
24501 ("alignaddrlsi_vis"): Likewise.
24502 ("alignaddrldi_vis"): Likewise.
24503 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24504 ("bmaskdi_vis"): Likewise.
24505 ("bmasksi_vis"): Likewise.
24506 ("bshuffle<VM64:mode>_vis"): Likewise.
24507 ("cmask8<P:mode>_vis"): Likewise.
24508 ("cmask16<P:mode>_vis"): Likewise.
24509 ("cmask32<P:mode>_vis"): Likewise.
24510 ("pdistn<P:mode>_vis"): Likewise.
24511 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24513 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24515 * config/sparc/sparc.md ("subtype"): New insn attribute.
24516 ("*wrgsr_sp64"): Set insn subtype.
24517 ("*rdgsr_sp64"): Likewise.
24518 ("alignaddrsi_vis"): Likewise.
24519 ("alignaddrdi_vis"): Likewise.
24520 ("alignaddrlsi_vis"): Likewise.
24521 ("alignaddrldi_vis"): Likewise.
24522 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
24523 ("fexpand_vis"): Likewise.
24524 ("fpmerge_vis"): Likewise.
24525 ("faligndata<VM64:mode>_vis"): Likewise.
24526 ("bshuffle<VM64:mode>_vis"): Likewise.
24527 ("cmask8<P:mode>_vis"): Likewise.
24528 ("cmask16<P:mode>_vis"): Likewise.
24529 ("cmask32<P:mode>_vis"): Likewise.
24530 ("fchksm16_vis"): Likewise.
24531 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
24532 ("fmean16_vis"): Likewise.
24533 ("fp<plusminus_insn>64_vis"): Likewise.
24534 ("<plusminus_insn>v8qi3"): Likewise.
24535 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
24536 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
24537 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
24538 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
24539 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
24540 ("*movqi_insn"): Likewise.
24541 ("*movhi_insn"): Likewise.
24542 ("*movsi_insn"): Likewise.
24543 ("movsi_pic_gotdata_op"): Likewise.
24544 ("*movdi_insn_sp32"): Likewise.
24545 ("*movdi_insn_sp64"): Likewise.
24546 ("movdi_pic_gotdata_op"): Likewise.
24547 ("*movsf_insn"): Likewise.
24548 ("*movdf_insn_sp32"): Likewise.
24549 ("*movdf_insn_sp64"): Likewise.
24550 ("*zero_extendhisi2_insn"): Likewise.
24551 ("*zero_extendqihi2_insn"): Likewise.
24552 ("*zero_extendqisi2_insn"): Likewise.
24553 ("*zero_extendqidi2_insn"): Likewise.
24554 ("*zero_extendhidi2_insn"): Likewise.
24555 ("*zero_extendsidi2_insn_sp64"): Likewise.
24556 ("ldfsr"): Likewise.
24557 ("prefetch_64"): Likewise.
24558 ("prefetch_32"): Likewise.
24559 ("tie_ld32"): Likewise.
24560 ("tie_ld64"): Likewise.
24561 ("*tldo_ldub_sp32"): Likewise.
24562 ("*tldo_ldub1_sp32"): Likewise.
24563 ("*tldo_ldub2_sp32"): Likewise.
24564 ("*tldo_ldub_sp64"): Likewise.
24565 ("*tldo_ldub1_sp64"): Likewise.
24566 ("*tldo_ldub2_sp64"): Likewise.
24567 ("*tldo_ldub3_sp64"): Likewise.
24568 ("*tldo_lduh_sp32"): Likewise.
24569 ("*tldo_lduh1_sp32"): Likewise.
24570 ("*tldo_lduh_sp64"): Likewise.
24571 ("*tldo_lduh1_sp64"): Likewise.
24572 ("*tldo_lduh2_sp64"): Likewise.
24573 ("*tldo_lduw_sp32"): Likewise.
24574 ("*tldo_lduw_sp64"): Likewise.
24575 ("*tldo_lduw1_sp64"): Likewise.
24576 ("*tldo_ldx_sp64"): Likewise.
24577 ("*mov<VM32:mode>_insn"): Likewise.
24578 ("*mov<VM64:mode>_insn_sp64"): Likewise.
24579 ("*mov<VM64:mode>_insn_sp32"): Likewise.
24581 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24583 * config/sparc/sparc.md ("type"): New insn type viscmp.
24584 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
24586 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
24587 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
24588 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
24589 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
24591 ("n7_vis_logical_11cycle"): Likewise.
24592 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
24593 * config/sparc/niagara2.md ("niag3_vis": Likewise.
24594 * config/sparc/niagara.md ("niag_vis"): Likewise.
24595 * config/sparc/ultra3.md ("us3_fga"): Likewise.
24596 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
24598 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24600 * config/sparc/sparc.md: New instruction type `bmask'.
24601 (bmaskdi_vis): Use the `bmask' type.
24602 (bmasksi_vis): Likewise.
24603 * config/sparc/ultra3.md (us3_array): Likewise.
24604 * config/sparc/niagara7.md (n7_array): Likewise.
24605 * config/sparc/niagara4.md (n4_array): Likewise.
24606 * config/sparc/niagara2.md (niag2_vis): Likewise.
24607 (niag3_vis): Likewise.
24608 * config/sparc/niagara.md (niag_vis): Likewise.
24610 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24612 * ipa-comdats.c: Remove optimize check from gate.
24613 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
24614 for functions not optimized.
24615 (ipa_fn_summary_read): Skip optimize check.
24616 (ipa_fn_summary_write): Likewise.
24617 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
24619 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
24621 (can_inline_edge_p): Check flag_pcc_struct_return for match.
24622 (check_callers): Give up on caller which is not optimized.
24623 (inline_small_functions): Likewise.
24624 (ipa_inline): Do not give up when not optimizing.
24625 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
24626 away unoptimizes cdtors.
24627 (whole_program_function_and_variable_visibility): Do
24628 ipa_discover_readonly_nonaddressable_vars in LTO mode.
24629 * ipa.c (process_references): Do not check optimize.
24630 (symbol_table::remove_unreachable_nodes): Update optimize check.
24631 (set_writeonly_bit): Update optimize check.
24632 (pass_ipa_cdtor_merge::gate): Do not check optimize.
24633 (pass_ipa_single_use::gate): Remove.
24635 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24637 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
24638 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
24639 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
24640 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
24641 permute_load, permute_store, adjust_extract, adjust_splat,
24642 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
24643 replace_swap_with_copy, dump_swap_insn_table,
24644 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
24645 recombine_lvx_pattern, recombine_stvx_pattern,
24646 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
24647 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
24648 to file rs6000-p8swap.c.
24649 * config/rs6000/rs6000-p8swap.c: New file.
24650 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
24651 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
24652 and rs6000*-*-* targets.
24654 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24656 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
24658 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24660 * lto-wrapper.c (merge_and_complain): Do not merge
24661 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
24662 fsigned_zeros, ftrapping_math, fwrapv.
24663 (append_compiler_options): Do not track these options.
24664 (append_linker_options): Likewie
24666 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24668 * cgraphunit.c (cgraph_node::finalize_function): When
24669 !flag_toplevel_reorde set no_reorder flag.
24670 (varpool_node::finalize_decl): Likewise.
24671 (symbol_table::compile): Drop no toplevel reorder path.
24673 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24675 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
24676 edges; zero probability is not better than uninitialized.
24678 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24680 * asan.h (asan_sanitize_allocas_p): Declare.
24681 * asan.c (asan_sanitize_allocas_p): New function.
24682 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
24683 (handle_builtin_alloca): Likewise.
24684 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
24685 if !asan_sanitize_allocas_p.
24686 * params.def (asan-instrument-allocas): Add new option.
24687 * params.h (ASAN_PROTECT_ALLOCAS): Define.
24688 * opts.c (common_handle_option): Disable allocas sanitization for
24691 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
24693 * asan.c: Include gimple-fold.h.
24694 (get_last_alloca_addr): New function.
24695 (handle_builtin_stackrestore): Likewise.
24696 (handle_builtin_alloca): Likewise.
24697 (asan_emit_allocas_unpoison): Likewise.
24698 (get_mem_refs_of_builtin_call): Add new parameter, remove const
24699 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
24700 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
24701 (instrument_builtin_call): Pass gimple iterator to
24702 get_mem_refs_of_builtin_call.
24703 (last_alloca_addr): New global.
24704 * asan.h (asan_emit_allocas_unpoison): Declare.
24705 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
24706 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
24707 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
24708 if function calls alloca.
24709 * gimple-fold.c (replace_call_with_value): Remove static keyword.
24710 * gimple-fold.h (replace_call_with_value): Declare.
24711 * internal-fn.c: Include asan.h.
24712 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
24713 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
24715 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24717 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
24718 (C_SELFTEST_FLAGS): New.
24719 (CPP_SELFTEST_FLAGS): New.
24720 (SELFTEST_DEPS): New, from deps of s-selftest.
24721 (C_SELFTEST_DEPS): New, from deps of s-selftest.
24722 (CPP_SELFTEST_DEPS): New.
24723 (selftest): Add dependency on s-selftest-c++.
24724 (s-selftest): Rename to...
24725 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
24726 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
24727 than SELFTEST_FLAGS.
24728 (selftest-gdb): Rename to...
24729 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
24731 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
24732 (selftest-valgrind): Rename to...
24733 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
24735 (selftest-valgrind): Reintroduce as an alias for
24736 selftest-c-valgrind.
24737 (s-selftest-c++): New.
24738 (selftest-c++-gdb): New.
24739 (selftest-c++-valgrind): New.
24741 2017-07-06 Olivier Hainque <hainque@adacore.com>
24743 * gcc.c (process_command): When deciding if undefined variables
24744 should be ignored when processing specs, accept "gcc -v" as well.
24746 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
24748 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
24749 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
24751 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
24753 * config/arm/arm-cpus.in (armv8-r): Add new entry.
24754 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
24755 * config/arm/arm-tables.opt: Regenerate.
24756 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
24758 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
24760 2017-07-06 Carl Love <cel@us.ibm.com>
24762 * ChangeLog: Clean up from mid air collision
24764 2017-07-06 Carl Love <cel@us.ibm.com>
24766 * config/rs6000/rs6000-c.c: Add support for built-in functions
24767 vector signed int vec_subc (vector signed int, vector signed int);
24768 vector signed __int128 vec_subc (vector signed __int128,
24769 vector signed __int128);
24770 vector unsigned __int128 vec_subc (vector unsigned __int128,
24771 vector unsigned __int128);
24772 vector signed int vec_sube (vector signed int, vector signed int,
24773 vector signed int);
24774 vector unsigned int vec_sube (vector unsigned int,
24775 vector unsigned int,
24776 vector unsigned int);
24777 vector signed __int128 vec_sube (vector signed __int128,
24778 vector signed __int128,
24779 vector signed__int128);
24780 vector unsigned __int128 vec_sube (vector unsigned __int128,
24781 vector unsigned __int128,
24782 vector unsigned __int128);
24783 vector signed int vec_subec (vector signed int, vector signed int,
24784 vector signed int);
24785 vector unsigned int vec_subec (vector unsigned int,
24786 vector unsigned int,
24787 vector unsigned int);
24788 vector signed __int128 vec_subec (vector signed __int128,
24789 vector signed __int128,
24790 vector signed__int128);
24791 vector unsigned __int128 vec_subec (vector unsigned __int128,
24792 vector unsigned __int128,
24793 vector unsigned __int128);
24794 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
24795 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
24796 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
24797 BU_ALTIVEC_OVERLOAD_X definitions.
24798 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
24799 * doc/extend.texi: Update the built-in documentation file for the new
24800 built-in functions.
24802 2017-07-06 David Malcolm <dmalcolm@redhat.com>
24805 * diagnostic-show-locus.c (layout::layout): Use start and finish
24806 spelling location for the start and finish of each range.
24807 * genmatch.c (linemap_client_expand_location_to_spelling_point):
24808 Add unused aspect param.
24809 * input.c (expand_location_1): Add "aspect" param, and use it
24810 to access the correct part of the location.
24811 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
24813 (expand_location_to_spelling_point): Likewise.
24814 (linemap_client_expand_location_to_spelling_point): Add "aspect"
24815 param, and pass it to expand_location_1.
24817 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
24819 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
24820 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
24821 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
24822 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
24823 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
24824 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
24825 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
24826 _mm_maskz_getmant_ss): New intrinsics.
24827 (__builtin_ia32_getexpss128_mask): Changed to ...
24828 __builtin_ia32_getexpss128_round ... this.
24829 (__builtin_ia32_getexpsd128_mask): Changed to ...
24830 __builtin_ia32_getexpsd128_round ... this.
24831 * config/i386/i386-builtin-types.def
24832 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
24833 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
24834 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
24835 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
24836 __builtin_ia32_getmantss_mask_round): New builtins.
24837 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
24838 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
24839 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
24840 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
24841 * config/i386/sse.md
24842 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
24843 avx512f_sgetexp<mode><mask_scalar_name>
24844 <round_saeonly_scalar_name> ... this.
24845 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
24846 %0, %1, %2<round_saeonly_op3>}): Changed to ...
24847 vgetexp<ssescalarmodesuffix>
24848 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24849 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
24850 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
24851 avx512f_vgetmant<mode><mask_scalar_name>
24852 <round_saeonly_scalar_name> ... this.
24853 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24854 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
24855 vgetmant<ssescalarmodesuffix>
24856 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
24857 %0<mask_scalar_operand4>, %1, %2
24858 <round_saeonly_scalar_mask_op4>, %3} ... this.
24859 * config/i386/subst.md (mask_scalar_operand4,
24860 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
24861 round_saeonly_scalar_nimm_predicate): New subst attributes.
24863 2017-07-06 Julia Koval <julia.koval@intel.com>
24865 * config/i386/i386.c (ix86_erase_embedded_rounding):
24866 Remove code for old rounding pattern.
24868 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
24870 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
24872 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
24874 * doc/sourcebuild.texi (Test Directives, Variants of
24875 dg-require-support): Add documentation for dg-require-stack-check.
24877 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
24879 * config/i386/subst.md (mask_scalar, round_scalar,
24880 round_saeonly_scalar): New meta-templates.
24881 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
24882 round_scalar_mask_operand3, round_scalar_mask_op3,
24883 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
24884 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
24885 round_saeonly_scalar_constraint,
24886 round_saeonly_scalar_prefix): New subst attribute.
24887 * config/i386/sse.md
24888 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
24889 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
24890 <round_scalar_name> ... this.
24891 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
24892 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
24893 <round_scalar_name> ... this.
24894 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
24895 <sse>_vm<code><mode>3<mask_scalar_name>
24896 <round_saeonly_scalar_name> ... this.
24897 (v<plusminus_mnemonic><ssescalarmodesuffix>
24898 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24899 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24900 v<plusminus_mnemonic><ssescalarmodesuffix>
24901 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24902 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24903 (v<multdiv_mnemonic><ssescalarmodesuffix>
24904 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
24905 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
24906 v<multdiv_mnemonic><ssescalarmodesuffix>
24907 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24908 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
24909 (v<maxmin_float><ssescalarmodesuffix>
24910 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
24911 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
24912 v<maxmin_float><ssescalarmodesuffix>
24913 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
24914 %0<mask_scalar_operand3>, %1, %<iptr>2
24915 <round_saeonly_scalar_mask_op3>} ... this.
24917 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
24919 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
24920 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
24922 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
24923 Alan Hayward <alan.hayward@arm.com>
24924 David Sherwood <david.sherwood@arm.com>
24926 * combine.c (simplify_if_then_else): Remove "enum" before
24928 * compare-elim.c (can_eliminate_compare): Likewise.
24929 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
24931 (aarch64_lookup_simd_builtin_type): Likewise.
24932 (aarch64_simd_builtin_type): Likewise.
24933 (aarch64_init_simd_builtin_types): Likewise.
24934 (aarch64_simd_expand_args): Likewise.
24935 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
24937 (aarch64_reverse_mask): Likewise.
24938 (aarch64_simd_emit_reg_reg_move): Likewise.
24939 (aarch64_gen_adjusted_ldpstp): Likewise.
24940 (aarch64_ccmp_mode_to_code): Likewise.
24941 (aarch64_operands_ok_for_ldpstp): Likewise.
24942 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24943 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
24945 (aarch64_min_divisions_for_recip_mul): Likewise.
24946 (aarch64_reassociation_width): Likewise.
24947 (aarch64_get_condition_code_1): Likewise.
24948 (aarch64_simd_emit_reg_reg_move): Likewise.
24949 (aarch64_simd_attr_length_rglist): Likewise.
24950 (aarch64_reverse_mask): Likewise.
24951 (aarch64_operands_ok_for_ldpstp): Likewise.
24952 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
24953 (aarch64_gen_adjusted_ldpstp): Likewise.
24954 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
24956 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
24957 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
24958 (arm_lookup_simd_builtin_type): Likewise.
24959 (arm_simd_builtin_type): Likewise.
24960 (arm_init_simd_builtin_types): Likewise.
24961 (arm_expand_builtin_args): Likewise.
24962 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
24963 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
24964 (ft32_setup_incoming_varargs): Likewise.
24965 (ft32_function_arg): Likewise.
24966 (ft32_function_arg_advance): Likewise.
24967 (ft32_pass_by_reference): Likewise.
24968 (ft32_arg_partial_bytes): Likewise.
24969 (ft32_valid_pointer_mode): Likewise.
24970 (ft32_addr_space_pointer_mode): Likewise.
24971 (ft32_addr_space_legitimate_address_p): Likewise.
24972 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
24974 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
24975 (ix86_emit_outlined_ms2sysv_restore): Likewise.
24976 (iamcu_alignment): Likewise.
24977 (canonicalize_vector_int_perm): Likewise.
24978 (ix86_noce_conversion_profitable_p): Likewise.
24979 (ix86_mpx_bound_mode): Likewise.
24980 (ix86_operands_ok_for_move_multiple): Likewise.
24981 * config/microblaze/microblaze-protos.h
24982 (microblaze_expand_conditional_branch_reg): Likewise.
24983 * config/microblaze/microblaze.c
24984 (microblaze_expand_conditional_branch_reg): Likewise.
24985 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
24987 (rs6000_reassociation_width): Likewise.
24988 (rs6000_invalid_binary_op): Likewise.
24989 (fusion_p9_p): Likewise.
24990 (emit_fusion_p9_load): Likewise.
24991 (emit_fusion_p9_store): Likewise.
24992 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
24994 (riscv_hard_regno_mode_ok_p): Likewise.
24995 (riscv_address_insns): Likewise.
24996 (riscv_split_symbol): Likewise.
24997 (riscv_legitimize_move): Likewise.
24998 (riscv_function_value): Likewise.
24999 (riscv_hard_regno_nregs): Likewise.
25000 (riscv_expand_builtin): Likewise.
25001 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25002 (riscv_build_integer): Likewise.
25003 (riscv_split_integer): Likewise.
25004 (riscv_legitimate_constant_p): Likewise.
25005 (riscv_cannot_force_const_mem): Likewise.
25006 (riscv_regno_mode_ok_for_base_p): Likewise.
25007 (riscv_valid_base_register_p): Likewise.
25008 (riscv_valid_offset_p): Likewise.
25009 (riscv_valid_lo_sum_p): Likewise.
25010 (riscv_classify_address): Likewise.
25011 (riscv_legitimate_address_p): Likewise.
25012 (riscv_address_insns): Likewise.
25013 (riscv_load_store_insns): Likewise.
25014 (riscv_force_binary): Likewise.
25015 (riscv_split_symbol): Likewise.
25016 (riscv_force_address): Likewise.
25017 (riscv_legitimize_address): Likewise.
25018 (riscv_move_integer): Likewise.
25019 (riscv_legitimize_const_move): Likewise.
25020 (riscv_legitimize_move): Likewise.
25021 (riscv_address_cost): Likewise.
25022 (riscv_subword): Likewise.
25023 (riscv_output_move): Likewise.
25024 (riscv_canonicalize_int_order_test): Likewise.
25025 (riscv_emit_int_order_test): Likewise.
25026 (riscv_function_arg_boundary): Likewise.
25027 (riscv_pass_mode_in_fpr_p): Likewise.
25028 (riscv_pass_fpr_single): Likewise.
25029 (riscv_pass_fpr_pair): Likewise.
25030 (riscv_get_arg_info): Likewise.
25031 (riscv_function_arg): Likewise.
25032 (riscv_function_arg_advance): Likewise.
25033 (riscv_arg_partial_bytes): Likewise.
25034 (riscv_function_value): Likewise.
25035 (riscv_pass_by_reference): Likewise.
25036 (riscv_setup_incoming_varargs): Likewise.
25037 (riscv_print_operand): Likewise.
25038 (riscv_elf_select_rtx_section): Likewise.
25039 (riscv_save_restore_reg): Likewise.
25040 (riscv_for_each_saved_reg): Likewise.
25041 (riscv_register_move_cost): Likewise.
25042 (riscv_hard_regno_mode_ok_p): Likewise.
25043 (riscv_hard_regno_nregs): Likewise.
25044 (riscv_class_max_nregs): Likewise.
25045 (riscv_memory_move_cost): Likewise.
25046 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25047 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25048 (rl78_addr_space_address_mode): Likewise.
25049 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25051 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25052 (rs6000_reassociation_width): Likewise.
25053 (rs6000_invalid_binary_op): Likewise.
25054 (fusion_p9_p): Likewise.
25055 (emit_fusion_p9_load): Likewise.
25056 (emit_fusion_p9_store): Likewise.
25057 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25058 (ok_for_simple_move_operands): Likewise.
25059 (ok_for_simple_move_strict_operands): Likewise.
25060 (ok_for_simple_arith_logic_operands): Likewise.
25061 (visium_legitimize_reload_address): Likewise.
25062 (visium_select_cc_mode): Likewise.
25063 (output_cbranch): Likewise.
25064 (visium_split_double_move): Likewise.
25065 (visium_expand_copysign): Likewise.
25066 (visium_expand_int_cstore): Likewise.
25067 (visium_expand_fp_cstore): Likewise.
25068 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25069 (visium_function_arg): Likewise.
25070 (visium_function_arg_advance): Likewise.
25071 (visium_libcall_value): Likewise.
25072 (visium_setup_incoming_varargs): Likewise.
25073 (visium_legitimate_constant_p): Likewise.
25074 (visium_legitimate_address_p): Likewise.
25075 (visium_legitimize_address): Likewise.
25076 (visium_secondary_reload): Likewise.
25077 (visium_register_move_cost): Likewise.
25078 (visium_memory_move_cost): Likewise.
25079 (prepare_move_operands): Likewise.
25080 (ok_for_simple_move_operands): Likewise.
25081 (ok_for_simple_move_strict_operands): Likewise.
25082 (ok_for_simple_arith_logic_operands): Likewise.
25083 (visium_function_value_1): Likewise.
25084 (rtx_ok_for_offset_p): Likewise.
25085 (visium_legitimize_reload_address): Likewise.
25086 (visium_split_double_move): Likewise.
25087 (visium_expand_copysign): Likewise.
25088 (visium_expand_int_cstore): Likewise.
25089 (visium_expand_fp_cstore): Likewise.
25090 (visium_split_cstore): Likewise.
25091 (visium_select_cc_mode): Likewise.
25092 (visium_split_cbranch): Likewise.
25093 (output_cbranch): Likewise.
25094 (visium_print_operand_address): Likewise.
25095 * expmed.c (flip_storage_order): Likewise.
25096 * expmed.h (emit_cstore): Likewise.
25097 (flip_storage_order): Likewise.
25098 * genrecog.c (validate_pattern): Likewise.
25099 * hsa-gen.c (gen_hsa_addr): Likewise.
25100 * internal-fn.c (expand_arith_overflow): Likewise.
25101 * ira-color.c (allocno_copy_cost_saving): Likewise.
25102 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25103 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25104 (process_invariant_for_inheritance): Likewise.
25105 * lra-eliminations.c (move_plus_up): Likewise.
25106 * omp-low.c (lower_oacc_reductions): Likewise.
25107 * simplify-rtx.c (simplify_subreg): Likewise.
25108 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25109 (TARGET_CHKP_BOUND_MODE): Likewise..
25110 * targhooks.c (default_chkp_bound_mode): Likewise.
25111 (default_setup_incoming_vararg_bounds): Likewise.
25112 * targhooks.h (default_chkp_bound_mode): Likewise.
25113 (default_setup_incoming_vararg_bounds): Likewise.
25114 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25115 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25116 (have_whole_vector_shift): Likewise.
25117 * tree-vect-stmts.c (vectorizable_load): Likewise.
25118 * doc/tm.texi: Regenerate.
25120 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25122 Graceful degrade if Binutils PR21472 is not available.
25125 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25126 .rodata in flash test fails.
25127 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25128 * confgure: Regenerate.
25129 * config.in: Regenerate.
25130 * config/avr/avr.c (avr_asm_named_section)
25131 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25132 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25133 (avr_asm_init_sections): Same.
25135 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25137 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25138 (fma<VH:mode>4_intrinsic): Likewise.
25139 (*fmsub<VCVTF:mode>4): Likewise.
25140 (*fmsub<VH:mode>4_intrinsic): Likewise.
25142 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25145 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25146 Don't depend on "optimize > 0".
25147 (out_movhi_r_mr, out_movqi_mr_r): Same.
25148 (out_movhi_mr_r, out_movqi_r_mr): Same.
25149 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25150 io_address_operand on "optimize > 0".
25152 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25154 * tree-loop-distribution.c: Add general explanantion on the pass.
25155 (generate_loops_for_partition): Mark distributed loop.
25156 (pg_add_dependence_edges): New parameter. Handle alias data
25157 dependence specially and record it in the parameter if asked.
25158 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25159 (init_partition_graph_vertices, add_partition_graph_edge): New.
25160 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25161 (free_partition_graph_vdata, build_partition_graph): New.
25162 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25163 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25164 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25165 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25166 (version_for_distribution_p, finalize_partitions): New.
25167 (distribute_loop): Handle alias data dependence specially. Factor
25168 out loop fusion code as functions and call these functions.
25170 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25172 * tree-loop-distribution.c (classify_partition): New parameter and
25173 better handle reduction statement.
25174 (rdg_build_partitions): Revise comment.
25175 (distribute_loop): Compute statements in all partitions and pass it
25176 to classify_partition.
25178 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25180 * tree-loop-distribution.c (enum partition_type): New.
25181 (struct partition): New field type.
25182 (partition_merge_into): Add parameter. Update partition type.
25183 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25184 (build_rdg_partition_for_vertex): Compute partition type.
25185 (rdg_build_partitions): Dump partition type.
25186 (distribute_loop): Update calls to partition_merge_into.
25188 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25190 * tree-loop-distribution.c (struct ddr_hasher): New.
25191 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25193 (classify_partition): Call get_data_dependence.
25194 (pg_add_dependence_edges): Ditto.
25195 (distribute_loop): Release data dependence hash table.
25197 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25199 * tree-loop-distribution.c (ref_base_address): Delete.
25200 (similar_memory_accesses): Rename ...
25201 (share_memory_accesses): ... to this. Check if partitions access
25202 the same memory reference.
25203 (distribute_loop): Call share_memory_accesses.
25205 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25207 * tree-loop-distribution.c (struct partition): New field recording
25208 its data reference.
25209 (partition_alloc, partition_free): Init and release data refs.
25210 (partition_merge_into): Merge data refs.
25211 (build_rdg_partition_for_vertex): Collect data refs for partition.
25212 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25214 (distribute_loop): Remve data refs from vertice data of partition
25217 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25219 * tree-loop-distribution.c (params.h): Include header file.
25220 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25221 (datarefs_vec): New global var.
25222 (create_rdg_vertices): Use datarefs_vec directly.
25223 (free_rdg): Don't free data references.
25224 (build_rdg): Update use. Don't free data references.
25225 (distribute_loop): Compute global variable for data references.
25226 Bail out if there are too many data references.
25228 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25230 * tree-loop-distribution.c (loop_nest): New global var.
25231 (build_rdg): Use loop directly, rather than loop nest.
25232 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25234 (distribute_loop): Compute global variable loop nest. Update use.
25236 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25238 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25239 (partition_merge_into): New parameter. Dump reason for fusion.
25240 (distribute_loop): Update use of partition_merge_into.
25242 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25244 * tree-loop-distribution.c (bb_top_order_index): New.
25245 (bb_top_order_index_size, bb_top_order_cmp): New.
25246 (stmts_from_loop): Use topological order.
25247 (pass_loop_distribution::execute): Compute and release topological
25248 order for basic blocks.
25250 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25252 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25255 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25257 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25258 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25259 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25260 * internal-fn.def (LOOP_DIST_ALIAS): New.
25261 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25262 (fold_loop_internal_call): ... this.
25263 (vect_loop_dist_alias_call): New function.
25264 (set_uid_loop_bbs): Call fold_loop_internal_call.
25265 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25268 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25271 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25272 Require dead FLAGS_REG at the beginning of a peephole.
25274 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25277 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25278 arguments in the call to __builtin_ia32_sbb_u32.
25279 (_subborrow_u64): Swap _X and _Y arguments in the call to
25280 __builtin_ia32_sbb_u64.
25282 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25285 * tree-vrp.c (compare_assert_loc): Turn into a function template
25286 with stable template parameter. Only test if a->e is NULL,
25287 !a->e == !b->e has been verified already. Use e == NULL or
25288 e != NULL instead of e or ! e tests. If stable is true, don't use
25289 iterative_hash_expr, on the other side allow a or b or both NULL
25290 and sort the NULLs last.
25291 (process_assert_insertions): Sort using compare_assert_loc<false>
25292 instead of compare_assert_loc, later sort using
25293 compare_assert_loc<true> before calling process_assert_insertions_for
25294 in a loop. Use break instead of continue once seen NULL pointer.
25296 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25298 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25299 Cortex-R7 and Cortex-R8 processors.
25301 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25303 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25304 uninitialized while src is not.
25306 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25308 * common/config/arm/arm-common.c: Adjust include path for
25310 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25311 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25312 (arm-cpu-data.h): Likewise.
25313 (arm-cpu-cdata.h): Likewise.
25314 * config/arm/arm-cpu.h: Delete.
25315 * config/arm/arm-cpu-cdata.h: Delete.
25316 * config/arm/arm-cpu-data.h: Delete.
25318 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25320 * config/arm/arm-cpus.in (cortex-a55): New.
25321 (cortex-a75): Likewise.
25322 (cortex-a75.cortex-a55): Likewise.
25323 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25325 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25326 * config/arm/arm-cpu-cdata.h: Regenerate.
25327 * config/arm/arm-cpu-data.h: Regenerate.
25328 * config/arm/arm-cpu.h: Regenerate.
25329 * config/arm/arm-tables.opt: Regenerate.
25330 * config/arm/arm-tune.md: Regenerate.
25332 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25334 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25336 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25338 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25341 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25343 PR tree-optimization/81292
25344 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25345 full_string_p, also call adjust_related_strinfos if the adjustment
25346 is simple, otherwise invalidate related strinfos.
25348 2017-07-04 Martin Liska <mliska@suse.cz>
25351 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25352 newly created variable as DECL_IGNORED_P.
25354 2017-07-04 Martin Liska <mliska@suse.cz>
25357 * ipa-inline.c (inline_small_functions):
25358 Use xstrdup_for_dump.
25360 2017-07-04 Tom de Vries <tom@codesourcery.com>
25362 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25364 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25367 * config/darwin.c (darwin_function_switched_text_sections):
25370 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25372 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25374 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25376 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25378 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25380 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25381 min_profitable_iters, and th as inclusive lower bounds.
25382 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25383 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25384 for min_profitable_iters and min_profitable_estimate.
25385 (vect_transform_loop): Treat th as an inclusive lower bound.
25386 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25388 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25391 * config/darwin.c (darwin_function_switched_text_sections):
25392 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
25393 in two pieces, and suppress the use of buf.
25395 2017-07-03 Nathan Sidwell <nathan@acm.org>
25397 * hash-table.h (hash_table_mod1): Fix indentation.
25399 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25401 PR middle-end/81290
25402 * predict.c (force_edge_cold): Be more careful about propagation
25404 * profile-count.h (profile_probability::guessed,
25405 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
25407 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
25409 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
25411 * doc/invoke.texi (rcpc architecture extension): Document it.
25413 2017-07-03 Richard Biener <rguenther@suse.de>
25415 PR tree-optimization/60510
25416 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
25417 the scalar reduction PHI and use it.
25418 (vectorizable_reduction): Properly guard the single_defuse_cycle
25419 path for non-SLP reduction chains where we cannot use it.
25420 Rework reduc_def/index and vector type deduction. Rework
25421 vector operand gathering during reduction op code-gen.
25422 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
25423 chains dissolve the chain and leave it to non-SLP reduction
25426 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25428 * tree-data-ref.h (dr_alignment): Declare.
25429 * tree-data-ref.c (dr_alignment): New function.
25430 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
25431 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
25433 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
25435 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25437 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
25438 and base_misalignment fields.
25439 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
25440 * tree-data-ref.c: Include builtins.h.
25441 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
25442 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
25443 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
25444 * tree-vect-data-refs.c: Include tree-cfg.h.
25445 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
25446 fields instead of calculating an alignment here.
25447 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
25448 innermost_loop_behavior fields.
25450 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25452 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
25454 (DR_STEP_ALIGNMENT): New macro.
25455 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
25456 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
25457 (create_data_ref): Print it.
25458 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
25459 to tell whether the step preserves vector (mis)alignment.
25460 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25461 Move the check for an integer step and generalise to all INTEGER_CST.
25462 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
25463 Print the outer step alignment.
25465 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25467 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
25468 with offset_alignment.
25469 (DR_ALIGNED_TO): Delete.
25470 (DR_OFFSET_ALIGNMENT): New macro.
25471 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
25472 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
25473 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
25474 (create_data_ref): Likewise.
25475 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
25476 (vect_analyze_data_refs): Likewise.
25477 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
25478 creating dummy innermost behavior.
25480 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25482 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
25483 with a "innermost_loop_behavior *" and refeence tree.
25484 * tree-data-ref.c (dr_analyze_innermost): Likewise.
25485 (create_data_ref): Update call accordingly.
25486 * tree-predcom.c (find_looparound_phi): Likewise.
25488 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25490 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
25491 fields with dr_wrt_vec_loop.
25492 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
25493 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
25494 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
25495 (vect_dr_behavior): New function.
25496 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25497 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25498 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
25499 track whether the step preserves the misalignment.
25500 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
25501 Use vect_dr_behavior.
25502 (vect_setup_realignment): Update call accordingly.
25503 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
25504 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
25505 call to vect_create_addr_base_for_vector_ref.
25506 (vect_create_cond_for_align_checks): Likewise.
25507 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
25508 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
25509 (vect_recog_mask_conversion_pattern): Likewise.
25510 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
25511 (new_stmt_vec_info): Remove redundant zeroing.
25513 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25515 * common/config/arm/arm-common.c (arm_be8_option): New function.
25516 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
25517 (ISA_ARMv6): Add isa_bit_be8.
25518 * config/arm/arm.h (arm_be8_option): Add prototype.
25519 (BE8_SPEC_FUNCTION): New define.
25520 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
25521 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
25522 (mlittle-endian): Similarly.
25523 (mbe8, mbe32): New options.
25524 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
25525 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
25527 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25529 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
25531 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25533 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
25534 (cleanup_tree_cfg_bb): Use it.
25535 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
25537 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
25539 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25542 * loop-doloop.c (add_test): Update profile.
25544 2017-07-03 Martin Liska <mliska@suse.cz>
25547 * sanopt.c (rewrite_usage_of_param): New function.
25548 (sanitize_rewrite_addressable_params): Likewise.
25549 (pass_sanopt::execute): Call rewrite_usage_of_param.
25551 2017-07-03 Richard Biener <rguenther@suse.de>
25553 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
25554 back to using VIEW_CONVERT_EXPR.
25556 2017-07-03 Martin Liska <mliska@suse.cz>
25559 * doc/extend.texi: Document when a resolver function is
25560 generated for target_clones.
25562 2017-07-03 Martin Liska <mliska@suse.cz>
25564 * asan.c (asan_emit_stack_protection): Unpoison just red zones
25565 and shadow memory of auto variables which are subject of
25566 use-after-scope sanitization.
25567 (asan_expand_mark_ifn): Add do set only when is_poison.
25569 2016-07-03 Richard Biener <rguenther@suse.de>
25571 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
25573 (vect_force_simple_reduction): Record reduction def -> phi mapping.
25574 (vectorizable_reduction): Perform reduction PHI creation when
25575 visiting a reduction PHI and adjust and simplify code generation
25576 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
25577 (vect_transform_loop): Visit reduction PHIs.
25578 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
25579 defs into the SLP tree.
25580 (vect_build_slp_tree): Reduction defs terminate the recursion.
25581 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
25583 (vect_get_vec_defs_for_stmt_copy): Export.
25584 (vect_get_vec_defs): Likewise.
25585 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
25587 (vect_get_vec_defs_for_stmt_copy): Declare.
25588 (vect_get_vec_defs): Likewise.
25590 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25592 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
25593 parameter with a "loop" parameter and use it instead of the
25594 loop containing DR_STMT. Don't check simple_iv when doing
25595 BB analysis. Describe the two analysis modes in the comment.
25597 2017-07-03 Tom de Vries <tom@codesourcery.com>
25599 PR tree-optimization/69468
25600 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
25601 (find_same_succ_bb): Handle ignore_edge_flags.
25603 2017-07-03 Tom de Vries <tom@codesourcery.com>
25605 PR tree-optimization/81192
25606 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
25608 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
25610 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
25612 2017-07-03 Tom de Vries <tom@codesourcery.com>
25614 PR tree-optimization/81192
25615 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
25616 BB_SAME_SUCC (bb) == NULL.
25618 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25620 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
25623 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25625 * dumpfile.c: Include profile-count.h
25626 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
25628 (insert_cond_bb): Update profile.
25629 * tree-cfg.h (insert_cond_bb): Update prototype.
25630 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
25631 * tree-dump.c: Do not include tree-cfg.
25633 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25635 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
25637 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25639 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
25642 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
25644 * tree-complex.c (expand_complex_div_wide): update profile.
25646 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25647 Alan Hayward <alan.hayward@arm.com>
25648 David Sherwood <david.sherwood@arm.com>
25650 * Makefile.in (MACHMODE_H): Remove insn-modes.h
25651 (CORETYPES_H): New define.
25652 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
25653 (insn-modes-inline.h, s-modes-inline-h): New rules.
25654 (generated_files): Add insn-modes-inline.h.
25655 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
25656 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
25657 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
25658 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
25659 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
25660 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
25661 (build/gencodes.o, build/genconditions.o): Likewise.
25662 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
25663 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
25664 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
25665 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
25666 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
25667 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
25668 * coretypes.h: Include everything up to real.h for generators.
25669 Include insn-modes.h first. Include wide-int-print.h after
25670 wide-int.h. Include insn-modes-inline.h and then machmode.h.
25671 * machmode.h: Don't include insn-modes.h here.
25672 * function-tests.c: Remove includes of signop.h, machmode.h,
25673 double-int.h and wide-int.h.
25675 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
25677 * optc-save-gen.awk: Likewise.
25678 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
25679 * godump.c: Remove include of wide-int-print.h.
25680 * pretty-print.h: Likewise.
25681 * wide-int-print.cc: Likewise.
25682 * wide-int.cc: Likewise.
25683 * hash-map-tests.c: Remove include of signop.h.
25684 * hash-set-tests.c: Likewise.
25685 * rtl-tests.c: Likewise.
25686 * mkconfig.sh: Remove include of machmode.h.
25687 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
25689 (emit_insn_modes_inline_h): ...this new function. Emit the code
25690 into an insn-modes-inline.h header file, adding appropriate
25691 include guards and end comments.
25692 (emit_insn_modes_c_header): Remove include of machmode.h.
25693 (emit_min_insn_modes_c_header): Include coretypes.h rather than
25695 (main): Handle -i flag and call emit_insn_modes_inline_h when
25698 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25700 * tree-ssa-strlen.c (strinfo): Rename the length field to
25701 nonzero_chars. Add a full_string_p field.
25702 (compare_nonzero_chars, zero_length_string_p): New functions.
25703 (get_addr_stridx): Add an offset_out parameter.
25704 Use compare_nonzero_chars.
25705 (get_stridx): Update accordingly. Use compare_nonzero_chars.
25706 (new_strinfo): Update after above changes to strinfo.
25707 (set_endptr_and_length): Set full_string_p.
25708 (get_string_length): Update after above changes to strinfo.
25709 (unshare_strinfo): Update call to new_strinfo.
25710 (maybe_invalidate): Likewise.
25711 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
25712 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
25713 as a uhwi instead of an shwi. Update after above changes to
25714 strinfo and new_strinfo.
25715 (zero_length_string): Assert that chainsi contains full strings.
25716 Use zero_length_string_p. Update call to new_strinfo.
25717 (adjust_related_strinfos): Update after above changes to strinfo.
25718 Copy full_string_p from origsi.
25719 (adjust_last_stmt): Use zero_length_string_p.
25720 (handle_builtin_strlen): Update after above changes to strinfo and
25721 new_strinfo. Install the lhs as the string length if the previous
25722 entry didn't describe a full string.
25723 (handle_builtin_strchr): Update after above changes to strinfo
25725 (handle_builtin_strcpy): Likewise.
25726 (handle_builtin_strcat): Likewise.
25727 (handle_builtin_malloc): Likewise.
25728 (handle_pointer_plus): Likewise.
25729 (handle_builtin_memcpy): Likewise. Track nonzero characters
25730 that aren't necessarily followed by a nul terminator.
25731 (handle_char_store): Likewise.
25733 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25735 PR tree-optimization/80769
25736 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
25737 for malloc and calloc. Document the new invariant that all related
25738 strinfos have delayed lengths or none do.
25739 (verify_related_strinfos): Move earlier in file.
25740 (set_endptr_and_length): New function, split out from...
25741 (get_string_length): ...here. Also set the lengths of related
25743 (zero_length_string): Assert that chainsi has known (rather than
25745 (adjust_related_strinfos): Likewise.
25747 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
25749 PR tree-optimization/81136
25750 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
25751 assert that two references with the same misalignment have the same
25752 compile-time misalignment if those compile-time misalignments
25755 2017-07-01 Andi Kleen <ak@linux.intel.com>
25757 * print-tree.c (print_node): Print all attributes.
25759 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25761 * cfg.c (scale_bbs_frequencies): New function.
25762 * cfg.h (scale_bbs_frequencies): Declare it.
25763 * cfgloopanal.c (single_likely_exit): Cleanup.
25764 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
25766 (scale_loop_profile): Likewise.
25767 (loop_version): Likewise.
25768 (create_empty_loop_on_edge): Update.
25769 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
25770 scale_loop_frequencies, scale_loop_profile, loopify,
25771 loop_version): Update prototypes.
25772 * modulo-sched.c (sms_schedule): Update.
25773 * predict.c (unlikely_executed_edge_p): Also check probability.
25774 (probably_never_executed_edge_p): Fix typo.
25775 * tree-if-conv.c (version_loop_for_if_conversion): Update.
25776 * tree-parloops.c (gen_parallel_loop): Update.
25777 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
25778 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
25779 * tree-ssa-loop-split.c (split_loop): Update.
25780 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
25781 * tree-vect-loop-manip.c (vect_do_peeling): Update.
25782 (vect_loop_versioning): Update.
25783 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
25785 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25787 * trans-mem.c (split_bb_make_tm_edge): Update profile.
25789 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25791 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
25792 to keep profile consistent.
25794 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25796 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
25797 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
25798 * profile-count.h (max_safe_multiplier): Make unsigned.
25799 (profile_count::guessed_zero): New.
25801 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25803 * bb-reorder.c (fix_up_crossing_landing_pad,
25804 fix_crossing_conditional_branches): Use make_single_succ_edge
25805 to keep profile consistent.
25807 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
25809 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
25812 2017-07-01 Jakub Jelinek <jakub@redhat.com>
25815 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
25816 the right scopes, make sure cond_jump isn't preserved between multiple
25817 iterations. Search for fallthru edge whenever there are 3+ edges and
25818 use find_fallthru_edge for it.
25820 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25822 Patch by Alexander Monakov <amonakov@ispras.ru>
25823 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
25824 probabilities consistently.
25826 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25828 * pa.c (pa_expand_compare_and_swap_loop): Update call of
25829 emit_cmp_and_jump_insns.
25831 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25834 * tree-inline.c (expand_call_inline): Combine profile statuses.
25836 2017-06-30 Andrew Pinski <apinski@cavium.com>
25838 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
25839 fold_stmt returned true.
25841 2017-06-30 Nathan Sidwell <nathan@acm.org>
25843 * ggc.h (empty_string): Delete.
25844 * cfgexpand.c (expand_asm_stmt): Use plain "".
25845 * optabs.c (expand_asm_memory_barrier): Likewise.
25846 * stringpool.c (empty_string): Delete.
25847 (digit_vector, digit_string): Delete.
25848 (ggc_alloc_string): Use plain "", don't optimize single digit
25849 strings. Use ggc_alloc_atomic.
25851 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
25853 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
25854 comparison set and one other set, use the cost of the non-comparison
25857 2017-06-30 Nathan Sidwell <nathan@acm.org>
25859 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
25862 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
25864 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
25865 loops. Remove now unneeded calls to gimple_switch_set_label() that
25866 just set removed labels to NULL_TREE.
25868 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
25870 * tree-ssanames.c (set_range_info_raw): Abstract from ...
25871 (set_range_info): ...here. Only call set_range_info_raw if domain
25873 (set_nonzero_bits): Call set_range_info_raw.
25874 * tree-ssanames.h (set_range_info_raw): New.
25876 2017-06-30 Jakub Jelinek <jakub@redhat.com>
25879 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
25880 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
25881 of nonimmediate_operand and <store_mask_constraint> instead of m
25882 for the input operand. For V8FI iterator, always split if input
25883 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
25884 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
25885 <store_mask_predicate> instead of register_operand and
25886 <store_mask_constraint> instead of v for the input operand. Make
25887 sure both operands aren't MEMs for if not <mask_applied>.
25889 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
25891 * lto-wrapper.c (copy_file) Close both file descriptors before
25894 2017-06-30 Martin Liska <mliska@suse.cz>
25897 * multiple_target.c (create_dispatcher_calls): Make ifunc
25898 also for function that don't have calls or are not referenced.
25900 2017-06-30 Richard Biener <rguenther@suse.de>
25902 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
25903 analyze the first scalar stmt. Move vector type computation
25904 for the BB case here from ...
25905 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
25906 live operation processing in the SLP case properly.
25908 2017-06-30 Richard Biener <rguenther@suse.de>
25910 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
25912 2017-06-30 Martin Liska <mliska@suse.cz>
25915 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
25916 before BUILT_IN_UNWIND_RESUME when ASAN is used.
25918 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
25920 * doc/invoke.texi (AArch64): Add missing options and remove redundant
25923 2017-06-30 Richard Biener <rguenther@suse.de>
25925 PR tree-optimization/81249
25926 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
25927 condition reduction result to original scalar type.
25929 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25931 * profile-count.h (enum profile_quality): Fix typos and whitespace
25934 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25936 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
25937 type for branch probabilities.
25939 2017-06-29 Julian Brown <julian@codesourcery.com>
25940 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25942 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
25943 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
25944 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
25945 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
25947 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25949 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
25950 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
25951 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
25952 CC usage from generic code to here.
25953 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
25954 CC usage into the target macros.
25956 2017-06-29 Maya Rashish <coypu@sdf.org>
25958 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
25961 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
25963 * arm/arm-builtins.c: Include profile-count.h
25964 * except.c (sjlj_emit_function_enter): Use
25965 profile_probability::unlikely.
25967 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25969 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
25970 and tocrel_offset be pointer args rather than implicitly using
25972 (legitimate_constant_pool_address_p, rs6000_emit_move,
25973 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
25974 tocrel_offset and use in toc_relative_expr_p call.
25975 (print_operand, print_operand_address): Use static tocrel_base_oac
25976 and tocrel_offset_oac.
25977 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
25980 2017-06-29 Maya Rashish <coypu@sdf.org>
25982 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
25984 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
25986 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
25987 objects, take into account only the alignment of 'op0' and 'mode1' if
25990 2017-06-29 Steve Ellcey <sellcey@cavium.com>
25992 * ccmp.c (ccmp_tree_comparison_p): New function.
25993 (ccmp_candidate_p): Update to use above function.
25994 (get_compare_parts): New function.
25995 (expand_ccmp_next): Update to use new functions.
25996 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
25998 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
25999 take mode as argument.
26000 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26001 * expr.c (expand_expr_real_1): Pass mode as argument.
26003 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26005 * combine.c (combine_instructions): Print insns to dump_file, together
26008 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26010 * asan.c (asan_emit_stack_protection): Update.
26011 (create_cond_insert_point): Update.
26012 * auto-profile.c (afdo_propagate_circuit): Update.
26013 * basic-block.h (struct edge_def): Turn probability to
26014 profile_probability.
26015 (EDGE_FREQUENCY): Update.
26016 * bb-reorder.c (find_traces_1_round): Update.
26017 (better_edge_p): Update.
26018 (sanitize_hot_paths): Update.
26019 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26020 (make_single_succ_edge): Update.
26021 (check_bb_profile): Update.
26022 (dump_edge_info): Update.
26023 (update_bb_profile_for_threading): Update.
26024 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26025 probabilitycount to 0.
26026 * cfgbuild.c (compute_outgoing_frequencies): Update.
26027 * cfgcleanup.c (try_forward_edges): Update.
26028 (outgoing_edges_match): Update.
26029 (try_crossjump_to_edge): Update.
26030 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26031 (expand_gimple_tailcall): Update.
26032 (construct_init_block): Use make_single_succ_edge.
26033 (construct_exit_block): Use make_single_succ_edge.
26034 * cfghooks.c (verify_flow_info): Update.
26035 (redirect_edge_succ_nodup): Update.
26036 (split_edge): Update.
26037 (account_profile_record): Update.
26038 * cfgloopanal.c (single_likely_exit): Update.
26039 * cfgloopmanip.c (scale_loop_profile): Update.
26040 (set_zero_probability): Remove.
26041 (duplicate_loop_to_header_edge): Update.
26042 * cfgloopmanip.h (loop_version): Update prototype.
26043 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26044 (force_nonfallthru_and_redirect): Update.
26045 (update_br_prob_note): Update.
26046 (rtl_verify_edges): Update.
26047 (purge_dead_edges): Update.
26048 (rtl_lv_add_condition_to_bb): Update.
26049 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26050 * cgraphunit.c (init_lowered_empty_function): Update.
26051 (cgraph_node::expand_thunk): Update.
26052 * cilk-common.c: Include profile-count.h
26053 * dojump.c (inv): Remove.
26054 (jumpifnot): Update.
26055 (jumpifnot_1): Update.
26056 (do_jump_1): Update.
26058 (do_jump_by_parts_greater_rtx): Update.
26059 (do_compare_rtx_and_jump): Update.
26060 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26061 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26062 * dwarf2cfi.c: Include profile-count.h
26063 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26064 (sjlj_emit_dispatch_table): Likewise.
26065 * explow.c: Include profile-count.h
26066 * expmed.c (emit_store_flag_force): Update.
26067 (do_cmp_and_jump): Update.
26068 * expr.c (compare_by_pieces_d::generate): Update.
26069 (compare_by_pieces_d::finish_mode): Update.
26070 (emit_block_move_via_loop): Update.
26071 (store_expr_with_bounds): Update.
26072 (store_constructor): Update.
26073 (expand_expr_real_2): Update.
26074 (expand_expr_real_1): Update.
26075 * expr.h (try_casesi, try_tablejump): Update prototypes.
26076 * gimple-pretty-print.c (dump_probability): Update.
26077 (dump_profile): New.
26078 (dump_gimple_label): Update.
26079 (dump_gimple_bb_header): Update.
26080 * graph.c (draw_cfg_node_succ_edges): Update.
26081 * hsa-gen.c (convert_switch_statements): Update.
26082 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26083 (find_if_case_1): Update.
26084 (find_if_case_2): Update.
26085 * internal-fn.c (expand_arith_overflow_result_store): Update.
26086 (expand_addsub_overflow): Update.
26087 (expand_neg_overflow): Update.
26088 (expand_mul_overflow): Update.
26089 (expand_vector_ubsan_overflow): Update.
26090 * ipa-cp.c (good_cloning_opportunity_p): Update.
26091 * ipa-split.c (split_function): Use make_single_succ_edge.
26092 * ipa-utils.c (ipa_merge_profiles): Update.
26093 * loop-doloop.c (add_test): Update.
26094 (doloop_modify): Update.
26095 * loop-unroll.c (compare_and_jump_seq): Update.
26096 (unroll_loop_runtime_iterations): Update.
26097 * lra-constraints.c (lra_inheritance): Update.
26098 * lto-streamer-in.c (input_cfg): Update.
26099 * lto-streamer-out.c (output_cfg): Update.
26100 * mcf.c (adjust_cfg_counts): Update.
26101 * modulo-sched.c (sms_schedule): Update.
26102 * omp-expand.c (expand_omp_for_init_counts): Update.
26103 (extract_omp_for_update_vars): Update.
26104 (expand_omp_ordered_sink): Update.
26105 (expand_omp_for_ordered_loops): Update.
26106 (expand_omp_for_generic): Update.
26107 (expand_omp_for_static_nochunk): Update.
26108 (expand_omp_for_static_chunk): Update.
26109 (expand_cilk_for): Update.
26110 (expand_omp_simd): Update.
26111 (expand_omp_taskloop_for_outer): Update.
26112 (expand_omp_taskloop_for_inner): Update.
26113 * omp-simd-clone.c (simd_clone_adjust): Update.
26114 * optabs.c (expand_doubleword_shift): Update.
26115 (expand_abs): Update.
26116 (emit_cmp_and_jump_insn_1): Update.
26117 (expand_compare_and_swap_loop): Update.
26118 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26119 * predict.c (predictable_edge_p): Update.
26120 (edge_probability_reliable_p): Update.
26121 (set_even_probabilities): Update.
26122 (combine_predictions_for_insn): Update.
26123 (combine_predictions_for_bb): Update.
26124 (propagate_freq): Update.
26125 (estimate_bb_frequencies): Update.
26126 (force_edge_cold): Update.
26127 * profile-count.c (profile_count::dump): Add missing space into dump.
26128 (profile_count::debug): Add newline.
26129 (profile_count::differs_from_p): Explicitly convert to unsigned.
26130 (profile_count::stream_in): Update.
26131 (profile_probability::dump): New member function.
26132 (profile_probability::debug): New member function.
26133 (profile_probability::differs_from_p): New member function.
26134 (profile_probability::differs_lot_from_p): New member function.
26135 (profile_probability::stream_in): New member function.
26136 (profile_probability::stream_out): New member function.
26137 * profile-count.h (profile_count_quality): Rename to ...
26138 (profile_quality): ... this one.
26139 (profile_probability): New.
26140 (profile_count): Update.
26141 * profile.c (compute_branch_probabilities): Update.
26142 * recog.c (peep2_attempt): Update.
26143 * sched-ebb.c (schedule_ebbs): Update.
26144 * sched-rgn.c (find_single_block_region): Update.
26145 (compute_dom_prob_ps): Update.
26146 (schedule_region): Update.
26147 * sel-sched-ir.c (compute_succs_info): Update.
26148 * stmt.c (struct case_node): Update.
26149 (do_jump_if_equal): Update.
26150 (get_outgoing_edge_probs): Update.
26151 (conditional_probability): Update.
26152 (emit_case_dispatch_table): Update.
26153 (expand_case): Update.
26154 (expand_sjlj_dispatch_table): Update.
26155 (emit_case_nodes): Update.
26156 * targhooks.c: Update.
26157 * tracer.c (better_p): Update.
26158 (find_best_successor): Update.
26159 * trans-mem.c (expand_transaction): Update.
26160 * tree-call-cdce.c: Update.
26161 * tree-cfg.c (gimple_split_edge): Upate.
26162 (move_sese_region_to_fn): Upate.
26163 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26164 * tree-eh.c (lower_resx): Upate.
26165 (cleanup_empty_eh_move_lp): Upate.
26166 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26167 * tree-inline.c (copy_edges_for_bb): Update.
26168 (copy_cfg_body): Update.
26169 * tree-parloops.c (gen_parallel_loop): Update.
26170 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26171 (gimple_gen_time_profiler): Update.
26172 * tree-ssa-dce.c (remove_dead_stmt): Update.
26173 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26174 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26175 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26176 (unloop_loops): Update.
26177 (try_peel_loop): Update.
26178 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26179 * tree-ssa-loop-split.c (connect_loops): Update.
26180 (split_loop): Update.
26181 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26182 (hoist_guard): Update.
26183 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26184 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26185 (value_replacement): Update.
26186 * tree-ssa-reassoc.c (branch_fixup): Update.
26187 * tree-ssa-tail-merge.c (replace_block_by): Update.
26188 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26189 (create_edge_and_update_destination_phis): Update.
26190 (compute_path_counts): Update.
26191 (recompute_probabilities): Update.
26192 (update_joiner_offpath_counts): Update.
26193 (freqs_to_counts_path): Update.
26194 (duplicate_thread_path): Update.
26195 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26196 (struct switch_conv_info): Update.
26197 (gen_inbound_check): Update.
26198 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26199 (vect_do_peeling): Update.
26200 (vect_loop_versioning): Update.
26201 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26202 (optimize_mask_stores): Update.
26203 * ubsan.c (ubsan_expand_null_ifn): Update.
26204 * value-prof.c (gimple_divmod_fixed_value): Update.
26205 (gimple_divmod_fixed_value_transform): Update.
26206 (gimple_mod_pow2): Update.
26207 (gimple_mod_pow2_value_transform): Update.
26208 (gimple_mod_subtract): Update.
26209 (gimple_mod_subtract_transform): Update.
26210 (gimple_ic): Update.
26211 (gimple_stringop_fixed_value): Update.
26212 (gimple_stringops_transform): Update.
26213 * value-prof.h: Update.
26215 2017-06-29 Carl Love <cel@us.ibm.com>
26217 * config/rs6000/rs6000-c.c: Add support for built-in functions
26218 vector signed int vec_signed (vector float);
26219 vector signed long long vec_signed (vector double);
26220 vector signed int vec_signed2 (vector double, vector double);
26221 vector signed int vec_signede (vector double);
26222 vector signed int vec_signedo (vector double);
26223 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26224 instruction generator.
26225 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26226 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26227 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26229 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26230 vunsignede_v2df): Add define_expands.
26231 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26232 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26233 VEC_UNSIGNEDO): Add definitions.
26234 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26235 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26236 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26237 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26238 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26239 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26240 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26241 vec_unsignede, vec_unsignedo): Add builtin defines.
26242 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26244 * doc/extend.texi: Update the built-in documentation file for the
26245 new built-in functions.
26247 2017-06-29 Richard Biener <rguenther@suse.de>
26249 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26250 reduction chains to LOOP_VINFO_REDUCTIONS.
26251 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26252 SLP reductions after processing reduction chains.
26254 2017-06-29 Nathan Sidwell <nathan@acm.org>
26256 * builtins.c (fold_builtin_FUNCTION): Use
26257 lang_hooks.decl_printable_name.
26259 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26261 PR middle-end/81194
26262 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26263 with only one label.
26264 * stmt.c (expand_case): Assert NCASES is greater than one.
26266 2017-06-29 Richard Biener <rguenther@suse.de>
26268 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26270 (group_case_labels): Likewise.
26271 (find_taken_edge): Push sanity checking on val to workers...
26272 (find_taken_edge_cond_expr): ... here
26273 (find_taken_edge_switch_expr): ... and here, handle cases
26274 with just a default label.
26275 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26276 (group_case_labels): Likewise.
26277 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26278 group_case_labels does anything cleanup the CFG again.
26280 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26282 PR tree-optimization/81196
26283 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26284 exit condition comparing two IVs.
26286 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26288 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26289 profile to the dummy entry at the end of the list of architectures.
26290 * config/arm/arm-cpu-cdata.h: Regenerated.
26292 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26293 Michael Collison <michael.collison@arm.com>
26296 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26298 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26299 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26300 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26301 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26302 of shift when the shift amount is masked with constant equal to
26303 the size of the mode.
26304 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26307 2017-06-29 Martin Liska <mliska@suse.cz>
26309 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26311 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26314 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26317 (edge_growth_cache_entry::edge_growth_cache_entry): New
26319 (reset_edge_growth_cache): Update to use constructor.
26321 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26323 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26324 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26325 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26327 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26329 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26330 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26332 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26334 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26335 (*-linux-uclibc*): Add t-uclibc tmake_file.
26336 * config/t-musl: New.
26337 * config/t-uclibc: New.
26339 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26341 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26343 (gen_comm_data): Emit architectural setting of arch_prof.
26344 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26346 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26347 (armv8-m.base, armv8-m.main): Likewise.
26348 * arm-protos.h (arm_build_target): Add profile field.
26349 (arch_option): Likewise.
26350 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26352 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26353 arm_active_target.profile.
26355 2017-06-28 Richard Biener <rguenther@suse.de>
26357 PR middle-end/81227
26358 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26359 TYPE_OVERFLOW_WRAPS.
26360 * match.pd (negate_expr_p): Likewise.
26361 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26362 fold_build2, not fold_binary.
26364 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26366 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26367 Convert memory address to Pmode.
26368 (aarch64_print_operand): Assert MEM operands are always Pmode.
26370 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26373 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26374 Remove redundant if.
26375 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26376 * config/arm/aarch-common-protos.h
26377 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26378 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26380 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26383 * multiple_target.c (create_dispatcher_calls): Set the default
26384 clone to be static, not public.
26386 2017-06-28 Richard Biener <rguenther@suse.de>
26388 * tree-vect-loop.c (vectorizable_reduction): Move special
26389 cond reduction IV var creation ...
26390 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26391 parameter. Use STMT_VINFO_VECTYPE.
26392 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
26395 2017-06-28 Martin Liska <mliska@suse.cz>
26398 * ipa-visibility.c (non_local_p): Handle visibility.
26400 2017-06-28 Martin Liska <mliska@suse.cz>
26403 * common.opt: Add IntegerRange to various options.
26404 * opt-functions.awk (integer_range_info): New function.
26405 * optc-gen.awk: Add integer_range_info to cl_options struct.
26406 * opts-common.c (decode_cmdline_option): Handle
26407 CL_ERR_INT_RANGE_ARG.
26408 (cmdline_handle_error): Likewise.
26409 * opts.c (print_filtered_help): Show valid interval in
26410 when --help is provided.
26411 * opts.h (struct cl_option): Add range_min and range_max fields.
26412 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
26414 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
26416 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
26417 (x * C EQ/NE y * C): New transformation.
26419 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
26421 * genmultilib (combination_space): Accept '+' in option names.
26423 2017-06-28 Martin Liska <mliska@suse.cz>
26426 * asan.c (instrument_derefs): Bail out inner references
26427 that are hard register variables.
26429 2017-06-28 Jakub Jelinek <jakub@redhat.com>
26432 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
26433 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
26435 2017-06-28 Richard Biener <rguenther@suse.de>
26437 * tree-vectorizer.h (vect_get_vec_defs): Remove.
26438 (vect_get_slp_defs): Adjust.
26439 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
26441 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
26443 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
26444 get_initial_defs_for_reduction instead of vect_get_vec_defs.
26445 (vectorizable_reduction): Adjust.
26446 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
26448 (vect_get_slp_defs): Likewise.
26449 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
26450 (vectorizable_bswap): Adjust.
26451 (vectorizable_call): Likewise.
26452 (vectorizable_conversion): Likewise.
26453 (vectorizable_assignment): Likewise.
26454 (vectorizable_shift): Likewise.
26455 (vectorizable_operation): Likewise.
26456 (vectorizable_store): Likewise.
26457 (vectorizable_condition): Likewise.
26458 (vectorizable_comparison): Likewise.
26460 2017-06-28 Michael Collison <michael.collison@arm.com>
26463 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
26465 (arm_gen_movmemqi): Removed unused variable 'i'.
26466 Convert 'for' loop into 'while' loop.
26467 (arm_expand_prologue): Remove last unnecessary set of insn.
26468 (thumb_pop): Remove unused variable 'pushed_words'.
26469 (thumb_exit): Remove last unnecessary set of regs_to_pop.
26471 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26473 * config/s390/predicates.md: Use s390_rel_address_ok_p.
26474 * config/s390/s390-protos.h: Add prototype of
26475 s390_rel_address_ok_p.
26476 * config/s390/s390.c (s390_got_symbol): New function.
26477 (s390_rel_address_ok_p): New function.
26478 (legitimize_pic_address): Use s390_rel_address_ok_p.
26479 (s390_load_got): Use s390_got_symbol.
26480 (s390_option_override): Issue error if
26481 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
26482 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
26484 * config/s390/s390.opt: New option mpic-data-is-text-relative.
26486 2017-06-27 Andrew Pinski <apinski@cavium.com>
26488 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
26489 (X * copysign (1.0, X)): New pattern.
26490 (X * copysign (1.0, -X)): New pattern.
26491 (copysign (-1.0, CST)): New pattern.
26493 2017-06-27 Joseph Myers <joseph@codesourcery.com>
26495 * genmultilib (combination_space): Remove variable.
26496 Validate reuse rules against regular expression for any sequence
26497 of multilib options in any order.
26499 2017-06-27 Michael Collison <michael.collison@arm.com>
26501 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
26502 call aarch64_split_simd_combine.
26503 * (aarch64_combine_internal<mode>): Delete pattern.
26504 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
26505 Allow register and subreg operands.
26507 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26509 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
26510 specific need, just fallback on defaults.
26511 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
26513 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26514 Olivier Hainque <hainque@adacore.com>
26516 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
26518 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
26519 targets. Pick a default if no particular attempt applied.
26520 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
26523 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26525 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
26526 (x86_64-wrs-vxworks7): Likewise.
26528 2017-06-27 Marek Polacek <polacek@redhat.com>
26531 * ubsan.c (instrument_null): Check get_base_address's result for null.
26533 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26535 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
26537 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
26539 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
26540 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
26541 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
26542 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
26543 New function types.
26544 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
26545 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
26546 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
26547 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
26548 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
26549 BUILT_IN_FEUPDATEENV): New builtins.
26550 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
26551 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
26552 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
26553 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
26555 (builtin_structptr_types): Adjust size.
26556 * tree.c (builtin_structptr_types): Add four entries.
26558 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26559 Olivier Hainque <hainque@adacore.com>
26561 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
26562 (TLS_SYM): New local macro, forcing reference to __tls__ on
26563 link command lines for VxWorks 7 RTPs, triggering initialization
26565 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
26566 OS features TLS support, true for RTPs on VxWorks 7.
26567 * config/vxworks.c (vxworks_override_options): Setup emutls
26570 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26572 * predict.c (test_prediction_value_range): Use -1U instead of -1
26573 to avoid narrowing conversion warning.
26574 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
26575 to avoid narrowing conversion warning.
26576 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
26578 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
26580 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
26582 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
26583 64bit configurations.
26584 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
26585 (SIZE_TYPE): Likewise.
26586 * config/vxworks.c (vxworks_emutls_var_fields): Use
26587 long_unsigned_type_node instead of unsigned_type_node as the offset
26588 field type, which is "pointer" mode in emutls.c.
26590 2017-06-27 Jakub Jelinek <jakub@redhat.com>
26593 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
26595 PR middle-end/81207
26596 * gimple-fold.c (replace_call_with_call_and_fold): Handle
26597 gimple_vuse copying separately from gimple_vdef copying.
26599 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26601 * value-prof.c (free_hist): Remove call to memset and the enclosing if
26604 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26605 Olivier Hainque <hainque@adacore.com>
26607 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
26608 for all vxworks7 targets.
26609 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
26610 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
26611 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
26612 variations for VX6/VX7 and 32/64bits later on in ...
26613 (VXWORKS_LIB_SPEC): Leverage new macros.
26614 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
26615 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
26617 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
26619 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
26620 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
26622 2017-06-26 Carl Love <cel@us.ibm.com>
26624 * config/rs6000/rs6000-c.c: Add support for built-in functions
26625 vector bool char vec_reve (vector bool char);
26626 vector signed char vec_reve (vector signed char);
26627 vector unsigned char vec_reve (vector unsigned char);
26628 vector bool int vec_reve (vector bool int);
26629 vector signed int vec_reve (vector signed int);
26630 vector unsigned int vec_reve (vector unsigned int);
26631 vector bool long long vec_reve (vector bool long long);
26632 vector signed long long vec_reve (vector signed long long);
26633 vector unsigned long long vec_reve (vector unsigned long long);
26634 vector bool short vec_reve (vector bool short);
26635 vector signed short vec_reve (vector signed short);
26636 vector double vec_reve (vector double);
26637 vector float vec_reve (vector float);
26638 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
26639 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
26640 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
26641 (altivec_vreve): New pattern.
26642 * config/rs6000/altivec.h (vec_reve): New define.
26643 * doc/extend.texi (vec_rev): Update the built-in documentation file
26644 for the new built-in functions.
26646 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26648 PR tree-optimization/71815
26649 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
26651 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
26653 (slsr_process_phi): Likewise.
26654 (replace_uncond_cands_and_profitable_phis): Don't replace a
26655 multiply candidate with a stride of 1 (copy or cast).
26656 (phi_incr_cost): Call uses_consumed_by_stmt rather than
26658 (lowest_cost_path): Likewise.
26659 (total_savings): Likewise.
26661 2017-06-26 Richard Biener <rguenther@suse.de>
26664 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
26665 Use def_builtin_pure for all gather builtins.
26667 2017-06-26 Richard Biener <rguenther@suse.de>
26669 PR tree-optimization/81203
26670 * tree-tailcall.c (find_tail_calls): Do not move stmts into
26671 non-dominating BBs.
26673 2017-06-26 Marek Polacek <polacek@redhat.com>
26676 * doc/invoke.texi: Document -Wmultistatement-macros.
26678 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
26680 * doc/sourcebuild.texi (ARM-specific attributes): Document new
26681 arm_neon_ok_no_float_abi effective target.
26683 2017-06-26 Richard Biener <rguenther@suse.de>
26685 PR tree-optimization/80928
26686 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
26687 (copy_bbs): Set BB_DUPLICATED flag early.
26688 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
26690 (execute_on_shrinking_pred): Likewise.
26691 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
26692 BB_DUPLICATED blocks.
26693 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
26694 iterate over all PHIs considering removal of *gsi.
26696 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
26698 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
26701 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26703 * config/rs6000/rs6000-string.c: (expand_block_clear,
26704 do_load_for_compare, select_block_compare_mode,
26705 compute_current_alignment, expand_block_compare,
26706 expand_strncmp_align_check, expand_strn_compare,
26707 expand_block_move, rs6000_output_load_multiple)
26708 Move functions related to string/block move/compare
26709 to a separate file.
26710 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
26711 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
26712 for this function which is now used in two files.
26713 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
26714 * config.gcc: Add rs6000-string.o to extra_objs for
26715 targets powerpc*-*-* and rs6000*-*-*.
26717 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
26720 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
26721 32-bit, since indexed is not valid for DImode.
26722 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
26723 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
26724 (define_peephole2 for Altivec d-form load): Add 32-bit support.
26725 (define_peephole2 for Altivec d-form store): Likewise.
26728 * multiple_target.c (create_dispatcher_calls): Only create the
26729 dispatcher call if the function is the default clone of a
26730 versioned function.
26732 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
26734 PR middle-end/80902
26735 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
26736 a call, force the call to not be a tail call.
26738 2017-06-23 Jeff Law <law@redhat.com>
26740 * doc/contrib.texi: Add entry for Steven Pemberton's work on
26743 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
26745 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
26746 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
26747 handling for early expansion of vector shifts (sl,sr,sra,rl).
26748 (builtin_function_type): Add vector shift right instructions
26749 to the unsigned argument list.
26751 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
26753 rtl-optimizatoin/79286
26754 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
26755 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
26756 trap. PIC register plus a const unspec without offset can never trap.
26758 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
26760 * tree.h (builtin_structptr_type): New type.
26761 (builtin_structptr_types): Declare new array.
26762 * tree.c (builtin_structptr_types): New array.
26763 (free_lang_data, build_common_tree_nodes): Use it.
26765 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
26768 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
26771 2017-06-22 Matt Turner <mattst88@gmail.com>
26773 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
26774 Lake models to skylake case. Assume skylake for unknown
26775 models with clflushopt.
26777 2017-06-22 Jeff Law <law@redhat.com>
26779 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
26780 frame sizes that do not satisfy aarch64_uimm12_shift.
26782 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
26784 * profile-count.h (apply_probability,
26785 apply_scale, probability_in): Fix checks for zero.
26787 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26789 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
26790 * doc/cppdiropts.texi (-I @var{dir}): Document it.
26792 2016-06-22 Richard Biener <rguenther@suse.de>
26794 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26795 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
26796 REDUC_MAX_EXPR support.
26797 (vectorizable_reduction): Likewise.
26798 (vect_create_epilog_for_reduction): Likewise.
26800 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
26802 * match.pd (A / (1 << B) -> A >> B): New.
26803 * generic-match-head.c: Include optabs-tree.h.
26804 * gimple-match-head.c: Likewise.
26805 * optabs-tree.h (target_supports_op_p): New.
26806 * optabs-tree.c (target_supports_op_p): New.
26808 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26810 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
26811 $gcc_cv_ld --help output.
26812 (gcc_cv_ld_demangle): Likewise.
26813 (gcc_cv_ld_eh_frame_hdr): Likewise.
26814 (gcc_cv_ld_pie): Likewise.
26815 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
26816 (gcc_cv_ld_buildid): Likewise.
26817 (gcc_cv_ld_sysroot): Likewise.
26818 (ld_bndplt_support): Likewise.
26819 (ld_pushpopstate_support): Likewise.
26820 * configure: Regenerate.
26821 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
26823 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26826 * config/i386/sse.md (round<mode>2): Renumber match_dup and
26827 operands indexes to avoid gap between operands and match_dups.
26829 2017-06-21 Andrew Pinski <apinski@cavium.com>
26831 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
26832 Increment Arith_shift and Arith_shift_reg by 1.
26833 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
26835 * config/aarch64/aarch64.c (thunderx_tunings): Enable
26836 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
26837 (aarch64_strip_extend): Add new argument and test for it.
26838 (aarch64_cheap_mult_shift_p): New function.
26839 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
26840 add a cost if it is true.
26841 Update calls to aarch64_strip_extend.
26842 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
26844 2017-06-21 Andrew Pinski <apinski@cavium.com>
26846 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
26848 (thunderxt88): Likewise.
26849 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
26850 (thunderx_prefetch_tune): New variable.
26851 (thunderx2t99_prefetch_tune): Update for the correct values.
26852 (thunderxt88_tunings): New variable.
26853 (thunderx_tunings): Use thunderx_prefetch_tune instead of
26854 generic_prefetch_tune.
26855 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
26857 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26859 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
26860 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
26861 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
26862 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
26863 (aarch64_atomic_cas<mode>, GPI): Likewise.
26865 2017-06-21 Martin Liska <mliska@suse.cz>
26867 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
26868 statements on cold and hot labels.
26869 * predict.c (tree_estimate_probability_bb): Remove the
26870 prediction from this place.
26872 2017-06-21 Martin Liska <mliska@suse.cz>
26874 PR tree-optimization/79489
26875 * gimplify.c (maybe_add_early_return_predict_stmt): New
26877 (gimplify_return_expr): Call the function.
26878 * predict.c (tree_estimate_probability_bb): Remove handling
26880 * predict.def: Update comment about early return predictor.
26881 * gimple-predict.h (is_gimple_predict): New function.
26882 * predict.def: Change default value of early return to 66.
26883 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
26885 * passes.def: Put pass_strip_predict_hints to the beginning of
26888 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
26890 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
26891 FUNCTION_DECL declarations.
26892 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
26894 (dwaf2out_decl): Likewise.
26895 * godump.c (go_early_global_decl): Skip call to the real debug hook
26896 for FUNCTION_DECL declarations.
26897 * passes.c (rest_of_decl_compilation): Skip call to the
26898 early_global_decl debug hook for FUNCTION_DECL declarations, unless
26899 -fdump-go-spec is passed.
26901 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26903 * config/i386/i386.c (struct builtin_isa): New field pure_p.
26904 Reorder for compactness.
26905 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
26906 (def_builtin_pure, def_builtin_pure2): New functions.
26907 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
26909 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
26911 * match.pd (nop_convert): New predicate.
26912 ((A +- CST1) +- CST2): Allow some NOP conversions.
26914 2017-06-21 Jakub Jelinek <jakub@redhat.com>
26917 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
26918 with ctors/dtors if GOVD_SHARED is set.
26920 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26922 * config/aarch64/aarch64.md (movti_aarch64):
26923 Emit mov rather than orr.
26924 (movtf_aarch64): Likewise.
26925 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
26926 Emit mov rather than orr.
26928 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26930 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
26931 Swap alternatives, make integer dup more expensive.
26933 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
26935 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26936 Return true for non-tls symbols.
26938 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
26940 * config/aarch64/aarch64-cores.def (cortex-a55): New.
26941 (cortex-a75): Likewise.
26942 (cortex-a75.cortex-a55): Likewise.
26943 * config/aarch64/aarch64-tune.md: Regenerate.
26944 * doc/invoke.texi (-mtune): Document new values for -mtune.
26946 2017-06-21 Tom de Vries <tom@codesourcery.com>
26948 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
26949 stack_size feature.
26950 (Effective-Target Keywords, Other attributes): Suggest using
26951 dg-add-options stack_size feature to get stack limit in stack_size
26952 effective target documentation.
26954 2017-06-21 Julian Brown <julian@codesourcery.com>
26955 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26957 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
26958 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
26959 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
26961 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
26962 attribute type list for neon_multiply.
26963 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
26964 attribute type list for neon_multiply.
26965 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
26966 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
26967 attribute type list for neon_multiply.
26968 * config/arm/types.md (crypto_pmull): Add.
26969 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
26970 attribute type list.
26972 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
26974 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
26977 2017-06-20 Jakub Jelinek <jakub@redhat.com>
26979 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
26980 to make sure not to dereference a NULL cost_classes_ptr pointer.
26982 2017-06-20 Carl Love <cel@us.ibm.com>
26984 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
26985 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
26986 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
26987 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
26988 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
26989 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
26990 VMULOSW): New enum "unspec" values.
26991 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
26992 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
26993 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
26994 altivec_vmulosw): New patterns.
26995 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
26996 VMULOSW): Add definitions.
26998 2017-06-20 Julia Koval <julia.koval@intel.com>
27000 * config/i386/i386.c: Fix rounding expand for new pattern.
27001 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27003 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27005 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27006 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27008 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27010 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27013 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27015 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27016 architecture, then by implementer ID.
27017 * config/aarch64/aarch64-tune.md: Regenerate.
27019 2017-06-20 Richard Biener <rguenther@suse.de>
27021 PR middle-end/81097
27022 * fold-const.c (split_tree): Fold to type before negating.
27024 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27026 * diagnostic-show-locus.c
27027 (selftest::test_fixit_deletion_affecting_newline): New function.
27028 (selftest::diagnostic_show_locus_c_tests): Call it.
27030 2017-06-20 Andreas Schwab <schwab@suse.de>
27033 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27036 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27038 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27039 __ARM_FEATURE_COPROC according to support.
27041 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27043 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27044 Rewritten to avoid overflow for > 32-bit pointers.
27047 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27048 by removing enum keyword.
27049 (ubsan_type_descriptor): Likewise. Formatting fix.
27052 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27053 splitter): Require TARGET_SSE2 in the condition.
27055 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27058 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27059 for doing vector set of SFmode on ISA 3.0.
27060 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27061 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27063 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27064 SFmode value into a V4SF variable that was extracted from another
27065 V4SF variable without converting the element to double precision
27066 and back to single precision vector format.
27067 (vsx_insert_extract_v4sf_p9_2): Likewise.
27069 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27071 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27072 in UWHI to avoid undefined overflow.
27075 * ubsan.h (enum ubsan_encode_value_phase): New.
27076 (ubsan_encode_value): Change second argument to
27077 enum ubsan_encode_value_phase with default value of
27078 UBSAN_ENCODE_VALUE_GENERIC.
27079 * ubsan.c (ubsan_encode_value): Change second argument to
27080 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27081 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27082 create_tmp_var_raw instead of create_tmp_var and use a
27084 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27085 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27086 ubsan_encode_value callers.
27089 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27090 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27091 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27093 2017-06-19 Richard Biener <rguenther@suse.de>
27095 PR middle-end/81118
27096 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27097 estimates if we changed anything.
27099 2017-06-19 Richard Biener <rguenther@suse.de>
27101 PR tree-optimization/80887
27102 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27103 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27104 simplified lookups, then reset mprts_hook.
27105 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27107 (try_to_simplify): Likewise.
27109 2017-06-19 Martin Liska <mliska@suse.cz>
27112 * gimplify.c (gimplify_switch_expr):
27113 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27115 2017-06-19 Martin Liska <mliska@suse.cz>
27117 * doc/install.texi: Document that PGO runs in 4 stages.
27119 2017-06-19 Martin Liska <mliska@suse.cz>
27122 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27123 to dispatcher function name.
27124 * multiple_target.c (replace_function_decl): New function.
27125 (create_dispatcher_calls): Redirect both edges and references.
27127 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27129 * profile-count.c (profile_count::dump): Dump quality.
27130 (profile_count::differs_from_p): Update for unsigned val.
27131 * profile-count.h (profile_count_quality): New enum.
27132 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27134 2017-06-19 Richard Biener <rguenther@suse.de>
27136 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27137 struct function as arg.
27138 (estimate_numbers_of_iterations): Export overload with loop arg.
27139 (free_numbers_of_iterations_estimates_loop): Use an overload of
27140 free_numbers_of_iterations_estimates instead.
27141 * tree-cfg.c (remove_bb): Adjust.
27142 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27143 * tree-parloops.c (gen_parallel_loop): Likewise.
27144 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27146 (tree_unroll_loops_completely): Likewise.
27147 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27148 Use an overload instead and export.
27149 (estimated_loop_iterations): Adjust.
27150 (max_loop_iterations): Likewise.
27151 (likely_max_loop_iterations): Likewise.
27152 (estimate_numbers_of_iterations): Take struct function as arg
27154 (loop_exits_before_overflow): Adjust.
27155 (free_numbers_of_iterations_estimates_loop): Use an overload.
27156 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27157 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27159 2017-06-19 Richard Biener <rguenther@suse.de>
27162 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27163 RANGE_EXPR conservatively.
27165 2017-06-16 Carl Love <cel@us.ibm.com>
27167 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27168 definitions for vec_float, vec_float2, vec_floato,
27169 vec_floate built-ins.
27170 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27171 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27173 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27174 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27175 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27176 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27177 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27178 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27179 vec_floato): Add builtin defines.
27180 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27181 Update the built-in documentation file for the new built-in
27184 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27186 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27187 (mthumb): Mark as the negative of -marm.
27189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27191 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27193 (ARM Options, -mtune): Document that this accepts the same
27194 extension options as -mcpu.
27195 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27197 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27199 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27200 permitted extensions.
27202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27204 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27205 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27206 (armv8-m.main): Add option +nodsp.
27207 * config/arm/arm-cpu-cdata.h: Regenerated.
27209 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27211 * config/arm/t-fuchsia: New file.
27212 * config.gcc (arm*-*-fuchsia*): Use it.
27214 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27216 * config/arm/t-symbian: Rewrite for new option infrastructure.
27218 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27220 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27221 (MULTILIB_REQUIRED): Likewise.
27223 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27225 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27226 (MULTILIB_RESUE): Likewise.
27227 (MULTILIB_MATCHES): Likewise.
27228 (MULTLIB_REQUIRED): Likewise.
27230 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27232 * config/arm/t-rtems: Rewrite for new option framework.
27234 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27236 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27237 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27238 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27239 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27240 * config/arm/t-multilib: ... here.
27241 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27242 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27243 armv7-a and armv8*-a when A-profile libraries have not been built.
27244 * config/arm/t-rmprofile: Rewrite.
27246 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27248 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27249 with a backslash. Remove the backslash after substituting unescaped
27251 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27253 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27255 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27256 the driver to pass through the default setting of -mfloat-abi.
27257 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27259 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27260 (all_feat_combs): New rule.
27261 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27263 * config/arm/t-aprofile: Rewrite.
27265 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27267 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27268 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27269 fpu is not specified by the user/command-line.
27270 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27271 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27272 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27273 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27274 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27275 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27277 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27279 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27280 * config/arm/t-arm-elf: Rewritten.
27282 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27284 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27285 have some floating-point instructions.
27286 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27287 (TARGET_MAYBE_HARD_FLOAT): New macro.
27288 * config/arm/arm-builtins.c (arm_init_builtins): Use
27289 TARGET_MAYBE_HARD_FLOAT.
27290 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27292 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27294 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27295 (configargs.h): Include it.
27296 (arm_print_hint_for_fpu_option): New function.
27297 (arm_parse_fpu_option): New function.
27298 (candidate_extension): New class.
27299 (arm_canon_for_multilib): New function.
27300 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27301 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27302 (ARCH_CANONICAL_SPECS): New macro.
27303 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27305 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27307 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27308 are set after handling multilib fragments. Set target_cpu_default2
27311 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27313 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27315 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27316 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27318 (check_arch): Likewise.
27319 * config/arm/arm.c (arm_configure_build_target): Handle
27320 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27321 options in the default.
27323 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27325 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27326 when an option is an alias of another.
27327 * config/arm/parsecpu.awk (optalias): New parser token.
27328 (gen_comm_data): Mark non-alias options as such. Emit entries
27329 for extension aliases.
27330 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27331 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27332 (armv6kz, armv6zk, armv6t2): Likewise.
27333 (armv7): Make vfpv3-d16 an alias.
27334 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27336 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27337 Sort in canonical order.
27338 (armv8-a): Sort in canonical order.
27339 (armv8.1-a, armv8.2-a): Likewise.
27340 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27342 (cortex-a9): Sort in canonical order.
27343 * config/arm/arm.c (selftests.h): Include it.
27344 (arm_test_cpu_arch_data): New function.
27345 (arm_run_self_tests): New function.
27346 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27347 (targetm): Move declaration to the end of the file.
27348 * arm-cpu-cdata.h: Regenerated.
27350 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27352 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27353 call to target_mode_check describing the type of option passed.
27354 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27355 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27356 arm_parse_cpu_option_name to match parameters against list of
27358 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27359 arm_arch_core_flags data structure.
27360 * config/arm/arm-cpu_cdata.h: Regenerated.
27362 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27364 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27366 (arm_print_hint_for_cpu_option): Likewise.
27367 (arm_print_hint_for_arch_option): Likewise.
27368 (arm_parse_cpu_option_name): Likewise.
27369 (arm_parse_arch_option_name): Likewise.
27370 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27371 of entries in the all_fpus list.
27372 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27373 (arm_parse_cpu_option_name): Declare.
27374 (arm_parse_arch_option_name): Declare.
27375 (arm_parse_option_features): Declare.
27376 (arm_intialize_isa): Declare.
27377 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27379 (gen_comm_data): ... here. Make definitions non-static.
27380 * config/arm/arm-cpu-data.h: Regenerated.
27381 * config/arm/arm-cpu-cdata.h: Regenerated.
27383 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27385 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27386 (cpu_arch_extension): New structure.
27387 (cpu_arch_option, arch_option, cpu_option): New structures.
27388 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27389 architecture types.
27390 (gen_data): Generate new format data tables.
27391 * config/arm/arm.c (cpu_tune): New structure.
27392 (cpu_option, processors): Delete.
27393 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
27394 (arm_print_hint_for_cpu_option): ... this and ...
27395 (arm_print_hint_for_arch_option): ... this.
27396 (arm_parse_arch_cpu_name): Delete. Replace with ...
27397 (arm_parse_cpu_option_name): ... this and ...
27398 (arm_parse_arch_option_name): ... this.
27399 (arm_unrecognized_feature): Change type of target parameter to
27401 (arm_parse_arch_cpu_features): Delete. Replace with ...
27402 (arm_parse_option_features): ... this.
27403 (arm_configure_build_target): Rework to use new configuration data
27405 (arm_print_tune_info): Rework for new configuration data tables.
27406 * config/arm/arm-cpu-data.h: Regenerated.
27407 * config/arm/arm-cpu.h: Regenerated.
27409 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27411 * Makefile.in (OBJS): Move sbitmap.o from here ...
27412 (OBJS-libcommon): ... to here.
27414 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27416 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
27417 (ISA_ALL_CRYPTO): New macro.
27418 (ISA_ALL_SIMD): New macro
27419 (ISA_ALL_FP): New macro.
27420 * config/arm/arm.c (fpu_bitlist): Update initializer.
27421 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
27423 (arm9e): Add fpu. Add option for nofp
27424 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
27425 (arm926ej-s, arm1026ej-s): Likewise.
27426 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
27427 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
27428 neon-fp16, neon-vfpv4, nofp and nosimd.
27429 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
27430 (cortex-a8): Add fpu. Add option for nofp.
27431 (cortex-a9): Add fpu. Add options for nosimd and nofp.
27432 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
27433 (cortex-r4f): Add fpu.
27434 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
27435 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
27437 (cortex-r8): Likewise.
27438 (cortex-m4): Add fpu. Add option for nofp.
27439 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
27440 (cortex-a17.cortex-a7): Likewise.
27441 (cortex-a32): Add fpu. Add options for crypto and nofp.
27442 (cortex-a35, cortex-a53): Likewise.
27443 (cortex-a57): Add fpu. Add option for crypto.
27444 (cortex-a72, cortex-a73): Likewise.
27445 (exynos-m1): Likewise.
27446 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
27447 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
27448 (cortex-m33): Add fpu. Add option for nofp.
27449 * config/arm/arm-cpu-cdata.h: Regenerated
27450 * config/arm/arm-cpu-data.h: Regenerated.
27452 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27454 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
27455 (armv5te, armv5tej): Likewise.
27456 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
27457 (armv7): Add options fp and vfpv3-d16.
27458 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
27459 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
27461 (armv7ve): Likewise.
27462 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
27463 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
27464 (armv8-a): Add nocrypto option.
27465 (armv8.1-a, armv8.2-a): Likewise.
27466 (armv8-m.main): add options fp, fp.dp and nofp.
27468 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27470 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
27472 (armv8-a+crc): Delete.
27473 (armv8.1-a): Add options simd, crypto and nofp.
27474 (armv8.2-a): Add options fp16, simd, crypto and nofp.
27475 (armv8.2-a+fp16): Delete.
27476 (armv8-m.main): Add option dsp.
27477 (armv8-m.main+dsp): Delete.
27478 (cortex-a8): Add fpu. Add nofp option.
27479 (cortex-a9): Add fpu. Add nofp and nosimd options.
27480 * config/arm/parsecpu.awk (gen_data): Generate option tables and
27481 link to main cpu and architecture data structures.
27482 (gen_comm_data): Only put isa attributes from the main architecture
27484 (option): New statement for architecture and CPU entries.
27485 * arm.c (struct cpu_option): New structure.
27486 (struct processors): Add entry for options.
27487 (arm_unrecognized_feature): New function.
27488 (arm_parse_arch_cpu_name): Ignore any characters after the first
27490 (arm_parse_arch_cpu_feature): New function.
27491 (arm_configure_build_target): Separate out any CPU and architecture
27492 features and parse separately. Don't error out if -mfpu=auto is
27493 used with only an architecture string.
27494 (arm_print_asm_arch_directives): New function.
27495 (arm_file_start): Call it.
27496 * config/arm/arm-cpu-cdata.h: Regenerated.
27497 * config/arm/arm-cpu-data.h: Likewise.
27498 * config/arm/arm-tables.opt: Likewise.
27500 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27502 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
27503 assembler when it is not -mfpu=auto.
27505 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27507 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
27508 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
27509 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
27510 (ASM_CPU_SPEC): Rewrite.
27511 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
27512 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
27513 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
27514 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
27515 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
27516 copied string is NUL-terminated. Also strip any characters prefixed
27518 (arm_rewrite_selected_arch): New function.
27519 (arm_rewrite_march): New function.
27521 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27523 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
27524 (x_arm_cpu_string, x_arm_tune_string): Likewise.
27525 (march, mcpu, mtune): Convert to string-based options.
27526 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
27527 (arm_parse_arch_cpu_name): New function.
27528 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
27529 identify selected architecture or CPU.
27530 (arm_option_save): New function.
27531 (TARGET_OPTION_SAVE): Redefine.
27532 (arm_option_restore): Restore string options.
27533 (arm_option_print): Print string options.
27535 2017-06-16 Martin Sebor <msebor@redhat.com>
27537 PR tree-optimization/80933
27538 PR tree-optimization/80934
27539 * builtins.c (fold_builtin_3): Do not handle bcmp here.
27540 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
27541 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
27542 (gimple_fold_builtin): Call them.
27544 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27546 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
27547 as unlikely; update profile.
27549 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27551 * predict.c (force_edge_cold): Handle declaring edges impossible
27554 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27556 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
27558 (try_unroll_loop_completely): Fix reporting.
27560 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27562 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
27564 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
27567 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
27568 if given a non-constant argument for an intrinsic which requires a
27571 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27573 * profile.c (compare_freqs): New function.
27574 (branch_prob): Sort edge list.
27575 (find_spanning_tree): Assume that the list is priority sorted.
27577 2017-06-16 Richard Biener <rguenther@suse.de>
27579 PR tree-optimization/81090
27580 * passes.def (pass_record_bounds): Remove.
27581 * tree-pass.h (make_pass_record_bounds): Likewise.
27582 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
27583 make_pass_record_bounds): Likewise.
27584 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
27585 not free niter estimates at the beginning but at the end.
27586 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
27588 2017-06-16 Richard Biener <rguenther@suse.de>
27590 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
27591 initializer to workaround ICE in host GCC 4.8.
27593 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
27595 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
27597 (clone_inlined_nodes): Update.
27599 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27601 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
27602 prefetch settings, and enable prefetching by default at -O3.
27604 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27606 * config/aarch64/aarch64.c (aarch64_override_options_internal):
27607 Set flag_prefetch_loop_arrays according to tuning data.
27609 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27611 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
27612 New tune structure.
27613 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
27614 [Unrelated to main purpose of the patch] Place the pointer field last
27615 to enable type checking errors when tune structure are wrongly merged.
27616 * config/aarch64/aarch64.c (generic_prefetch_tune,)
27617 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
27618 (thunderx2t99_prefetch_tune): New tune constants.
27619 (tune_params *_tunings): Update all tunings (no functional change).
27620 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
27621 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
27622 from tunings structures.
27624 2017-06-16 Jakub Jelinek <jakub@redhat.com>
27627 * ubsan.c (instrument_null): Add T argument, use it instead
27628 of computing it based on IS_LHS.
27629 (instrument_object_size): Likewise.
27630 (pass_ubsan::execute): Adjust instrument_null and
27631 instrument_object_size callers to pass gimple_get_lhs or
27632 gimple_assign_rhs1 result to it. Use instrument_null instead of
27633 calling get_base_address and instrument_mem_ref. Handle
27634 aggregate call arguments for object-size sanitization.
27636 2017-06-16 Yury Gribov <tetra2005@gmail.com>
27638 PR tree-optimization/81089
27639 * tree-vrp.c (is_masked_range_test): Validate operands of
27642 2017-06-15 Martin Sebor <msebor@redhat.com>
27645 * dumpfile.c (dump_register): Avoid calling memset to initialize
27646 a class with a default ctor.
27647 * gcc.c (struct compiler): Remove const qualification.
27648 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
27649 * hash-table.h: Ditto.
27650 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
27652 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
27653 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
27655 * params.h (struct param_info): Make struct members non-const.
27656 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
27657 with default initialization.
27658 * vec.h (vec_copy_construct, vec_default_construct): New helper
27660 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
27661 with vec_copy_construct.
27662 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
27663 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
27664 * doc/invoke.texi (-Wclass-memaccess): Document.
27666 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27668 * emit-rtl.h (is_leaf): Update comment about local
27669 register allocator.
27671 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27674 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
27675 for a variable to have a section before checking if the section has a
27677 Set section to.persistent if persistent attribute is set.
27678 Warn if .persistent attribute is used on an automatic variable.
27680 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
27682 PR rtl-optimization/80474
27683 * reorg.c (update_block): Do not ignore instructions in a delay slot.
27685 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
27687 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
27690 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
27692 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
27693 (casesi): Emit bounds checking as RTL.
27694 (casesi_internal_mips16_<mode>): Remove bounds checking.
27696 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
27698 * config/xtensa/xtensa.c (xtensa_option_override): Append
27699 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
27700 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
27701 xtensa_doloop_hooks): Define unconditionally.
27702 (xtensa_reorg_loops): Only call reorg_loops in the presence of
27704 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
27705 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
27706 for it in xtensa_option_override.
27707 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
27708 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
27710 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
27712 * doc/cppopts.texi: Document '-' special value to -MF.
27714 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
27716 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
27717 (cortex_a53_fconst): Likewise.
27718 (cortex_a53_fpmul): Likewise.
27719 (cortex_a53_f_load_64): Likewise.
27720 (cortex_a53_f_load_many): Likewise.
27721 (cortex_a53_advsimd_alu): Likewise.
27722 (cortex_a53_advsimd_alu_q): Likewise.
27723 (cortex_a53_advsimd_mul): Likewise.
27724 (cortex_a53_advsimd_mul_q): Likewise.
27725 (fpmac bypass): Add new bypass for fpmac-fpmac case.
27726 Add missing fmul, r2f_cvt and fconst cases.
27728 2017-06-14 Richard Biener <rguenther@suse.de>
27730 PR middle-end/81088
27731 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
27733 (fold_binary_loc): When associating do not treat pre-existing
27734 TREE_OVERFLOW on literal constants as a reason to allow
27735 TREE_OVERFLOW on associated literal constants.
27737 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27739 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
27740 (MASK_FEATURES): New macro.
27741 * config/sparc/sparc.c (sparc_option_override): Remove the special
27742 handling of -mfpu and generalize it to all MASK_FEATURES switches.
27744 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27746 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
27747 a division of 0 if non-call exceptions are enabled.
27749 2017-06-14 Andrew Pinski <apinski@cavium.com>
27750 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27753 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
27754 Improve vector initialization code gen for only variable case.
27756 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
27758 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
27760 2017-06-14 Richard Biener <rguenther@suse.de>
27762 PR tree-optimization/81083
27763 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
27766 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27768 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
27769 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
27770 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
27771 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
27772 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
27773 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
27775 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27777 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
27778 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
27780 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27782 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
27784 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27786 * config/rs6000/t-rtems: Don't handle SPE.
27788 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27790 * config/rs6000/t-linux: Don't handle SPE.
27792 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27794 * config/rs6000/eabispe.h: Delete file.
27796 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27798 * config/rs6000/t-spe: Delete file.
27800 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27802 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
27803 (rs6000_legitimate_offset_address_p): Return false for anything in
27804 V2SImode or V2SFmode.
27806 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
27808 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
27809 except V2SF and V2SI. Rearrange the vector modes, and add comments.
27810 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
27812 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
27813 (rs6000_legitimate_offset_address_p): Ditto.
27814 (rs6000_emit_move): Ditto.
27815 (rs6000_init_builtins): Remove V4HI_type_node.
27817 2017-06-13 Martin Liska <mliska@suse.cz>
27820 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
27821 (gate_asan): Likewise.
27822 * asan.h (asan_no_sanitize_address_p): Remove the function.
27823 (sanitize_flags_p): New function.
27824 * builtins.def: Fix coding style.
27825 * common.opt: Use renamed enum value.
27826 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
27827 * doc/extend.texi: Document no_sanitize attribute.
27828 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
27829 to SANITIZE_UNDEFINED_NONDEFAULT.
27830 * gcc.c (sanitize_spec_function): Use the renamed enum value.
27831 * gimple-fold.c (optimize_atomic_compare_exchange_p):
27832 Use sanitize_flags_p.
27833 * gimplify.c (gimplify_function_tree): Likewise.
27834 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
27835 * opts.c (parse_no_sanitize_attribute): New function.
27836 (common_handle_option): Use renamed enum value.
27837 * opts.h (parse_no_sanitize_attribute): Declare.
27838 * tree.c (sanitize_flags_p): New function.
27839 * tree.h: Declared here.
27840 * tsan.c: Use sanitize_flags_p.
27841 * ubsan.c (ubsan_expand_null_ifn): Likewise.
27842 (instrument_mem_ref): Likewise.
27843 (instrument_bool_enum_load): Likewise.
27844 (do_ubsan_in_current_function): Remove the function.
27845 (pass_ubsan::execute): Use sanitize_flags_p.
27846 * ubsan.h: Remove do_ubsan_in_current_function
27847 * tree-cfg.c (print_no_sanitize_attr_value): New function.
27848 (dump_function_to_file): Use it here.
27850 2017-06-13 Martin Jambor <mjambor@suse.cz>
27852 PR tree-optimization/80803
27853 PR tree-optimization/81063
27854 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
27855 (propagate_subaccesses_across_link): Enqueue subtree whenever
27856 necessary instead of relying on the caller.
27858 2017-06-13 Martin Jambor <mjambor@suse.cz>
27860 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
27861 that have a first_link.
27862 (sort_and_splice_var_accesses): Do not check first_link before
27864 (subtree_mark_written_and_enqueue): Likewise.
27865 (propagate_all_subaccesses): Likewise and do not stop at first
27866 parent with a first_link.
27868 2017-06-13 Martin Jambor <mjambor@suse.cz>
27870 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
27873 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27875 * match.pd: New pattern.
27877 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27879 * tree-vrp.c (is_masked_range_test): New function.
27880 (register_edge_assert_for): Determine ranges for
27883 2017-06-13 Yury Gribov <tetra2005@gmail.com>
27885 PR tree-optimization/67328
27886 * fold-const.c (maskable_range_p): New function.
27887 (build_range_check): Generate bittests if possible.
27889 2017-06-13 Martin Liska <mliska@suse.cz>
27891 * gimple-pretty-print.c (dump_probability): Add new argument.
27892 (dump_edge_probability): Dump both probability and count.
27893 (dump_gimple_label): Likewise.
27894 (dump_gimple_bb_header): Likewise.
27896 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
27899 * config/avr/avr-devices.c: Fix indentation.
27900 * config/avr/gen-avr-mmcu-specs.c: Dito.
27902 2017-06-13 Richard Biener <rguenther@suse.de>
27904 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
27905 instead get vector type from stmt_info.
27906 (vectorizable_reduction): Adjust. Remove dead code.
27908 2017-06-13 Richard Biener <rguenther@suse.de>
27910 PR middle-end/81065
27911 * fold-const.c (extract_muldiv_1): Remove bogus distribution
27912 case of C * (x * C2 + C3).
27913 (fold_addr_of_array_ref_difference): Properly fold index difference.
27915 2017-06-12 David S. Miller <davem@davemloft.net>
27918 * config/sparc/sparc.md (return expander): Emit frame blockage if
27919 function uses alloca.
27921 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
27923 * combine.c (make_field_assignment): Check len rather than the mode
27924 precision when calling force_to_mode.
27926 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
27928 Support multilibs and devices that see flash in RAM address range.
27931 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
27932 (avr_mcu_t) <flash_pm_offset>: New field.
27933 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
27934 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
27935 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
27936 (AVR_TINY_PM_OFFSET): Remove macro.
27937 * config/avr/avr.opt (-mshort-calls): New option.
27938 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
27939 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
27940 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
27941 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
27942 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
27943 instead of avr_arch->have_jmp_call.
27944 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
27945 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
27946 avr_arch->flash_pm_offset to define.
27947 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
27948 new field flash_pm_offset. Add entry for avrxmega3.
27949 (avr_texinfo): Add entry for avrxmega3.
27950 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
27951 attiny212, attiny214,
27952 attiny412, attiny414, attiny416, attiny417,
27953 attiny814, attiny816, attiny817,
27954 attiny1614, attiny1616, attiny1617,
27955 attiny3214, attiny3216, attiny3217.
27956 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
27957 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
27958 (avr_print_operand_address) [AVR_TINY]: Same.
27959 (avr_asm_init_sections) <readonly_data_section>: Only patch
27960 callback if avr_arch->flash_pm_offset = 0.
27961 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
27962 for rodata if avr_arch->flash_pm_offset != 0.
27963 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
27964 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
27965 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
27966 (m_options): Append opt_rcall.
27967 (m_dirnames): Append dir_rcall.
27968 * config/avr/t-multilib: Regenerate.
27970 * configure.ac [target=avr]: Check whether avrxmega3 default
27971 linker description file works as needed.
27972 * configure: Regenerate.
27973 * doc/avr-mmcu.texi: Regenerate.
27974 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
27975 <__AVR_ARCH__>: Document avrxmega3 and 103.
27976 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
27977 <__AVR_SHORT_CALLS__>: Document it.
27978 <__AVR_PM_BASE_ADDRESS__>: Document it.
27979 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
27980 (AVR Variable Attributes) <progmem>: Document this is
27981 not needed for avrxmega3.
27982 (AVR Named Address Spaces) <__flash>: Dito.
27984 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
27986 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
27988 2017-06-12 Doug Rupp <rupp@adacore.com>
27990 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
27991 Append vxworks-stdint.h to the tm_file list.
27992 * config/vxworks-stdint.h: New file.
27994 2017-06-12 Martin Liska <mliska@suse.cz>
27996 PR tree-optimization/81041
27997 * tree-profile.c (gimple_gen_ic_func_profiler):
27998 Create an extra BB in profile-generate
27999 (gimple_gen_time_profiler): Likewise.
28001 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28003 PR tree-optimization/81003
28004 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28005 (update_range_test): Use it instead of force_gimple_operand_gsi.
28007 2017-06-12 Richard Biener <rguenther@suse.de>
28009 PR tree-optimization/81053
28010 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28011 with backedge value not defined in loop. Simplify def stmt
28014 2017-06-11 Tom de Vries <tom@codesourcery.com>
28017 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28019 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28020 nvptx_cannot_force_const_mem.
28022 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28024 * opts.c (finish_options): Move test for flag_split_stack after
28025 it has been initialized.
28027 2017-06-11 Jason Merrill <jason@redhat.com>
28029 * tree.h (id_equal): New.
28030 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28031 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28032 instead of strcmp of IDENTIFIER_POINTER.
28034 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28036 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28037 (mark_all_inlined_calls_cdtor): Fix formating.
28038 (inline_transform): Rescale profile before inlining.
28040 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28042 * cgraph.h (cgraph_edge::clone): Update prototype.
28043 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28044 (cgraph_node::create_clone): Update.
28045 (cgraph_node::create_version_clone): Update.
28046 * tree-inline.c (copy_bb): Update.
28047 (expand_call_inline): Update.
28049 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28051 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28052 factored out from ...
28053 (rs6000_emit_prologue): ... here.
28055 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28057 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28058 factored out from ...
28059 (rs6000_emit_prologue): ... here.
28061 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28063 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28065 (handle_missing_profiles): Fix computation of tp_first_run.
28066 (counts_to_freqs): Do not touch freqs when count is 0.
28068 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28070 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28073 2017-06-10 Tom de Vries <tom@codesourcery.com>
28075 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28076 attributes): Document signal effective target.
28078 2017-06-10 Tom de Vries <tom@codesourcery.com>
28080 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28081 Document effective target stack_size.
28083 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28085 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28086 to the edit_context if they can be auto-applied.
28088 2017-06-9 Ian Lance Taylor <iant@golang.org>
28090 * opts.c (finish_options): If -fsplit-stack, disable implicit
28091 -forder-blocks-and-partition.
28092 * doc/invoke.texi (Optimize Options): Document that when using
28093 -fsplit-stack -forder-blocks-and-partition is not implicitly
28096 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28098 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28099 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28100 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28101 * builtins.def (abort, trap, unreachable): Declare cold.
28102 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28103 * tree-core.h (ECF_COLD): New.
28104 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28105 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28107 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28109 * predict.c (unlikely_executed_stmt_p): Cleanup.
28111 2017-06-09 Richard Biener <rguenther@suse.de>
28113 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28114 model if the ref is always written to.
28116 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28118 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28120 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28122 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28125 2017-06-09 Tom de Vries <tom@codesourcery.com>
28128 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28129 "target cannot support label values" when encountering LABEL_REF.
28131 2017-06-09 Martin Liska <mliska@suse.cz>
28133 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28134 (gimple_gen_ic_func_profiler): Emit direct comparison
28135 of __gcov_indirect_call_callee with NULL.
28136 (gimple_gen_time_profiler): Change probability from
28137 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28139 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28141 * profile.c (edge_gcov_counts): Turn to pointer.
28142 (compute_branch_probabilities, compute_branch_probabilities): Update.
28143 (branch_prob): Do not clear edge_gcov_count.
28144 * profile.h (edge_gcov_counts): Turn to pointer.
28145 (edge_gcov_count): Update.
28147 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28149 * gimple.h (gimple_check_failed): Mark cold.
28151 2017-06-09 Richard Biener <rguenther@suse.de>
28153 PR tree-optimization/66623
28154 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28155 refactor check_reduction into two parts, properly computing
28156 whether we have to check reduction validity for outer loop
28159 2017-06-09 Richard Biener <rguenther@suse.de>
28161 PR tree-optimization/79483
28162 * graphite-scop-detection.c (order): New global.
28163 (get_order): Compute bb to order mapping that satisfies code
28164 generation constraints.
28165 (cmp_pbbs): New helper.
28166 (build_scops): Start domwalk at entry block, sort generated
28169 2017-06-09 Richard Biener <rguenther@suse.de>
28171 PR middle-end/81007
28172 * ipa-polymorphic-call.c
28173 (ipa_polymorphic_call_context::restrict_to_inner_class):
28174 Skip FIELD_DECLs with error_mark_node type.
28175 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28178 2017-06-09 Martin Liska <mliska@suse.cz>
28180 * predict.c (struct branch_predictor): New struct.
28181 (test_prediction_value_range): New test.
28182 (predict_c_tests): New function.
28183 * selftest-run-tests.c (selftest::run_tests): Run the function.
28184 * selftest.h: Declare new tests.
28186 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28189 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28190 gen_add3_insn did not fail.
28191 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28192 r0, construct that number in a temporary reg and add that reg to r0.
28193 If asked to put the result in r0 as well, fail.
28195 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28197 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28198 for early expansion of vec_eqv.
28200 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28202 PR middle-end/81005
28203 * ubsan.c (instrument_null): Avoid pointless code temporary.
28204 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28206 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28210 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28213 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28216 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28218 * predict.c (unlikely_executed_edge_p): Move ahead.
28219 (probably_never_executed_edge_p): Use it.
28221 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28223 PR middle-end/79988
28224 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28225 gimple_call_builtin_p call.
28227 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28229 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28230 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28231 rtl_check_failed_type2, rtl_check_failed_code1,
28232 rtl_check_failed_code2, rtl_check_failed_code_mode,
28233 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28234 rtvec_check_failed_bounds, rtl_check_failed_flag,
28235 _fatal_insn_not_found, _fatal_insn): Likewise.
28236 * tree.h (tree_contains_struct_check_failed,
28237 tree_check_failed, tree_not_check_failed,
28238 tree_class_check_failed, tree_range_check_failed,
28239 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28240 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28241 tree_operand_check_failed, omp_clause_check_failed,
28242 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28245 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28247 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28248 flag_branch_probabilities.
28249 * ipa-inline.c (edge_badness): Likewise.
28250 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28251 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28252 * predict.c (maybe_hot_frequency_p): Likewise.
28253 (probably_never_executed): Likewise.
28254 * sched-ebb.c (schedule_ebbs): Likewise.
28255 * sched-rgn.c (find_single_block_region): Likewise.
28256 * tracer.c (tail_duplicate): Likewise.
28258 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28260 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28261 longer requires x_flag_profile_use.
28263 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28265 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28266 instead of flag_reorder_blocks_and_partition.
28267 * dbxout.c (dbxout_function_end): Likewise.
28268 * dwarf2out.c (gen_subprogram_die): Likewise.
28269 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28270 * hw-doloop.c (reorg_loops): Likewise.
28271 * varasm.c (assemble_start_function,
28272 assemble_end_function): Likewise.
28273 (decide_function_section): Do not check for
28274 flag_reorder_blocks_and_partition.
28276 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28278 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28280 (chkp_get_hard_register_fake_addr_expr): Ditto.
28281 (chkp_build_addr_expr): Add check for hard reg case.
28282 (chkp_parse_array_and_component_ref): Ditto.
28283 (chkp_find_bounds_1): Ditto.
28284 (chkp_process_stmt): Don't generate bounds store for
28287 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28289 * predict.c (maybe_hot_bb_p): Do not check profile status.
28290 (maybe_hot_edge_p): Likewise.
28291 (probably_never_executed): Check for zero counts even if profile
28293 (unlikely_executed_edge_p): New function.
28294 (unlikely_executed_stmt_p): New function.
28295 (unlikely_executed_bb_p): New function.
28296 (set_even_probabilities): Use unlikely predicates.
28297 (combine_predictions_for_bb): Likewise.
28298 (predict_paths_for_bb): Likewise.
28299 (predict_paths_leading_to_edge): Likewise.
28300 (determine_unlikely_bbs): New function.
28301 (estimate_bb_frequencies): Use it.
28302 (compute_function_frequency): Use zero counts even if profile is
28304 * profile-count.h: Fix typo.
28306 2017-08-08 Julia Koval <julia.koval@intel.com>
28308 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28309 _mm512_mask_cvtsepi16_storeu_epi8,
28310 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28311 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28312 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28313 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28314 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28315 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28316 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28317 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28318 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28319 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28320 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28321 __builtin_ia32_pmovuswb256mem_mask,
28322 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28323 __builtin_ia32_pmovwb512mem_mask): New builtins.
28325 2017-08-08 Julia Koval <julia.koval@intel.com>
28327 PR target/73350,80862
28328 * config/i386/subst.md (round): Fix round pattern.
28329 * config/i386/i386.c (ix86_erase_embedded_rounding):
28330 Fix erasing rounding for the fixed pattern.
28332 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28334 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28336 2017-06-08 Martin Liska <mliska@suse.cz>
28338 PR gcov-profile/80911
28339 * gcov.c (block_info::block_info): New constructor.
28341 2017-06-07 Carl Love <cel@us.ibm.com>
28343 * config/rs6000/rs6000-c: The return type of the following
28344 built-in functions was implemented as int not long long. Fix sign
28345 of return value for the unsigned version of vec_mulo and vec_mule.
28346 vector unsigned long long vec_bperm (vector unsigned long long,
28347 vector unsigned char)
28348 vector signed long long vec_mule (vector signed int,
28350 vector unsigned long long vec_mule (vector unsigned int,
28351 vector unsigned int)
28352 vector signed long long vec_mulo (vector signed int,
28354 vector unsigned long long vec_mulo (vector unsigned int,
28355 vector unsigned int)
28356 * doc/extend.texi: Fix the documentation for the built-in
28359 2017-06-07 Carl Love <cel@us.ibm.com>
28362 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28365 2017-06-07 Carl Love <cel@us.ibm.com>
28367 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28368 support, Generate doublehv for signed int/float for BE case only.
28370 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28372 * doc/invoke.texi (mcx16): Rewrite.
28374 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28376 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28377 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28378 *mov<mode>_softfloat, and an anonymous splitter): Use
28379 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28381 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28383 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28385 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28386 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28387 (rs6000_debug_reg_global): Adjust.
28388 (rs6000_init_hard_regno_mode_ok): Adjust.
28389 (rs6000_dbx_register_number): Adjust.
28390 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28391 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28392 Remove SPE_ACC and SPEFSCR.
28393 (REG_ALLOC_ORDER): Ditto.
28394 (FRAME_POINTER_REGNUM): Change to 111.
28395 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
28396 (REG_CLASS_NAMES): Ditto.
28397 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
28398 (REGISTER_NAMES): Ditto.
28399 (ADDITIONAL_REG_NAMES): Ditto.
28400 (rs6000_reg_names): Ditto.
28401 * config/rs6000/rs6000.md: Renumber some register number
28404 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28406 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
28408 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28409 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
28411 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
28412 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28413 Delete the SPE high registers.
28414 (REG_ALLOC_ORDER): Ditto.
28415 (enum reg_class): Remove SPE_HIGH_REGS.
28416 (REG_CLASS_NAMES): Ditto.
28417 (REG_CLASS_CONTENTS): Delete the SPE high registers.
28418 (REGISTER_NAMES): Ditto.
28419 (rs6000_reg_names): Ditto.
28420 * doc/tm.texi.in: Remove SPE as example.
28421 * doc/tm.texi: Regenerate.
28423 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28425 * config/rs6000/8540.md (ppc8540_brinc): Delete.
28426 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
28427 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
28428 * config/rs6000/rs6000.md (type): Remove "brinc".
28430 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28432 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
28433 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
28434 * config/rs6000/linuxspe.h: Delete file.
28435 * config/rs6000/rs6000.md: Don't include spe.md.
28436 * config/rs6000/spe.h: Delete file.
28437 * config/rs6000/spe.md: Delete file.
28438 * config/rs6000/t-rs6000: Remove spe.md.
28440 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28442 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
28443 (reg_or_none500mem_operand): Delete.
28444 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
28445 instead of reg_or_none500mem_operand.
28447 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28449 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28450 handling of SPE flags.
28451 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
28453 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28455 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
28457 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
28458 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
28459 paired_divv2sf3): Similar.
28460 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
28461 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
28462 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
28464 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
28465 Rename the paired_* instruction patterns.
28466 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
28468 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
28469 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
28470 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
28471 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
28472 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
28473 PAIRED_VECTOR_MODE.
28474 (struct machine_function): Delete field spe_insn_chain_scanned_p.
28475 (spe_func_has_64bit_regs_p): Delete.
28476 (spe_expand_predicate_builtin): Delete.
28477 (spe_expand_evsel_builtin): Delete.
28478 (TARGET_DWARF_REGISTER_SPAN): Do not define.
28479 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
28480 (invalid_e500_subreg): Delete.
28481 (rs6000_legitimize_address): Always force_reg op2 as well, for
28482 paired single memory accesses.
28483 (rs6000_member_type_forces_blk): Delete.
28484 (rs6000_spe_function_arg): Delete.
28485 (rs6000_expand_unop_builtin): Delete SPE handling.
28486 (rs6000_expand_binop_builtin): Ditto.
28487 (spe_expand_stv_builtin): Delete.
28488 (bdesc_2arg_spe): Delete.
28489 (spe_expand_builtin): Delete.
28490 (spe_expand_predicate_builtin): Delete.
28491 (spe_expand_evsel_builtin): Delete.
28492 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
28493 (spe_init_builtins): Delete.
28494 (spe_func_has_64bit_regs_p): Delete.
28495 (savres_routine_name): Delete "info" parameter. Adjust callers.
28496 (rs6000_emit_stack_reset): Ditto.
28497 (rs6000_dwarf_register_span): Delete.
28498 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
28499 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
28500 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
28502 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
28504 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
28505 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
28506 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
28507 mulv2sf3, divv2sf3): Delete expanders.
28509 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28511 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
28513 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28515 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
28516 * config/rs6000/rs6000.c: Ditto.
28518 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28520 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
28521 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
28522 comparison_operator.
28524 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28526 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
28527 * config/rs6000/rs6000.opt: Ditto.
28528 * config/rs6000/t-rtems: Ditto.
28530 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28532 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
28533 TARGET_E500_SINGLE by 0, simplify.
28534 * config/rs6000/rs6000.c: Ditto.
28535 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
28536 (spe_build_register_parallel): Delete.
28537 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
28538 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
28539 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
28540 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
28541 (E500_CONVERT): Delete.
28542 * config/rs6000/spe.md: Remove many patterns and all define_constants.
28544 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28546 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
28547 * config/rs6000/dfp.md: Ditto.
28548 (negdd2, *negdd2_fpr): Merge.
28549 (absdd2, *absdd2_fpr): Merge.
28550 (negtd2, *negtd2_fpr): Merge.
28551 (abstd2, *abstd2_fpr): Merge.
28552 * config/rs6000/e500.h: Delete file.
28553 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
28554 TARGET_FPRS by 1 and simplify.
28555 * config/rs6000/rs6000-c.c: Ditto.
28556 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
28557 TARGET_DF_SPE by 0.
28558 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
28560 * config/rs6000/rs6000.md: Ditto.
28561 (floatdidf2, *floatdidf2_fpr): Merge.
28562 (move_from_CR_gt_bit): Delete.
28563 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
28564 (E500_CR_IOR_COMPARE): Delete.
28565 (All patterns that require !TARGET_FPRS): Delete.
28566 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
28568 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28570 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
28572 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28574 * graphds.c (add_edge): Intitialize edge's attached data.
28575 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
28576 pointer parameter. Call pointed function on each edge during
28577 graph traversing. Skip traversing the edge when the function
28579 (graphds_dfs, graphds_scc): Ditto.
28580 (for_each_edge): New parameter. Pass the new parameter to callback
28582 * graphds.h (skip_edge_callback): New function pointer type.
28583 (graphds_dfs, graphds_scc): New function pointer parameter.
28584 (graphds_edge_callback, for_each_edge): New parameter.
28586 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28588 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
28589 out code checking if runtime alias check is possible to below ...
28590 Call the new function.
28591 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
28592 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
28594 2017-06-07 Marek Polacek <polacek@redhat.com>
28597 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
28598 TYPE_OVERFLOW_WRAPS check.
28600 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28602 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
28603 if versioning is required.
28604 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
28605 peeling with the check for versioning.
28607 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28609 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
28610 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
28611 Set true to new parameter if new ssa variable is defined.
28612 (vect_gen_vector_loop_niters): Refactor. Set range information
28613 for the new vector loop bound variable.
28614 (vect_do_peeling): Ditto.
28616 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28618 * tree-affine.c (ssa.h): Include header file.
28619 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
28620 has wrapping overflow behavior.
28622 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28624 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
28626 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28628 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
28629 (tree_to_aff_combination): ... here.
28631 2017-06-07 Bin Cheng <bin.cheng@arm.com>
28633 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
28634 reg_pressure model function.
28635 (ivopts_global_cost_for_size): Delete.
28636 (determine_set_costs, iv_ca_recount_cost): Call new model function
28637 ivopts_estimate_reg_pressure.
28639 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28641 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
28642 expensive than udiv. Remove floating point cases from mod.
28644 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28646 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
28647 Increase idiv cost.
28649 2017-06-07 Tamar Christina <tamar.christina@arm.com>
28651 * config/aarch64/aarch64.md
28652 (copysignsf3): Fix mask generation.
28654 2017-06-07 Jakub Jelinek <jakub@redhat.com>
28656 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
28658 (class dump_manager): Add register_dumps method.
28659 * dumpfile.c: Include langhooks.h.
28660 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
28661 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
28662 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
28663 (dump_manager::dump_register): Start with 512 entries instead of 32.
28664 (dump_manager::register_dumps): New method.
28665 * toplev.c (general_init): Instead of invoking register_dumps
28666 langhook, invoke register_dumps method on the dump manager.
28667 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
28670 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
28672 * doc/md.texi: Clarify the restrictions on a define_insn condition.
28673 Say that # requires an associated define_split to exist, and that
28674 the define_split must be suitable for use after register allocation.
28676 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28678 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
28679 (compute_outgoing_frequencies): Also initialize zero counts.
28680 (find_many_sub_basic_blocks): Do not produce uninitialized profile
28681 around loops; preserve more of profile when nothing changes.
28683 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
28685 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
28687 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
28688 * config/arm/arm-cpu-cdata.h: Regenerate.
28689 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
28690 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
28691 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
28692 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
28693 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
28695 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
28697 * config/arm/t-rmprofile: Likewise.
28698 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
28700 2017-06-06 David S. Miller <davem@davemloft.net>
28703 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
28704 blockage if function uses alloca.
28706 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28708 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
28709 New "uid" fields to hold pretty-print IDs of group and ref.
28710 Memory references are now identified as <group_id>:<ref_id>
28711 instead of using [random] addresses.
28712 (dump_mem_details): Simplify, no functional change.
28713 (dump_mem_ref): Simplify and make output more concise.
28714 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
28715 (find_or_create_group): Initialize group uid.
28716 (record_ref): Initialize ref uid. Improve debug output.
28717 (prune_group_by_reuse, should_issue_prefetch_p,)
28718 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
28719 (mark_nontemporal_store, determine_loop_nest_reuse):
28720 Improve debug output.
28722 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28724 * dbgcnt.def (prefetch): New debug counter.
28725 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
28726 (schedule_prefetches): Stop issueing prefetches if debug counter
28729 2017-06-06 Tom de Vries <tom@codesourcery.com>
28731 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
28732 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
28734 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28736 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
28737 Use aarch64_reg_or_zero predicate for operand 4.
28738 (aarch64_compare_and_swap<mode> define_insn_and_split):
28739 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
28740 (aarch64_store_exclusive<mode>): Likewise for operand 2.
28742 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
28744 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
28745 (arm_compute_save_core_reg_mask): This.
28746 (thumb1_compute_save_reg_mask): Rename into ...
28747 (thumb1_compute_save_core_reg_mask): This.
28748 (arm_compute_save_reg0_reg12_mask): Adapt comment.
28749 (arm_compute_frame_layout): Likewise.
28751 2017-06-06 Richard Biener <rguenther@suse.de>
28753 PR tree-optimization/80974
28754 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
28755 keep or clear leaders SSA info.
28757 2017-06-06 Tom de Vries <tom@codesourcery.com>
28759 * config/nvptx/nvptx.c (split_mode_p): New function.
28760 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
28762 2017-06-06 Tom de Vries <tom@codesourcery.com>
28764 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
28766 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28769 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
28772 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28774 * shrink-wrap.c (handle_simple_exit): Update profile.
28775 (try_shrink_wrapping): Upate profile.
28777 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28779 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
28780 (tree_guess_outgoing_edge_probabilities): New.
28781 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
28782 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
28784 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28786 * ipa-split.c (split_function): Initialize return bb profile.
28788 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
28790 * profile.c (compute_branch_probabilities): Also initialize
28791 EXIT_BLOCK profile.
28793 2017-06-06 Richard Biener <rguenther@suse.de>
28795 PR tree-optimization/80928
28796 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
28797 (vect_analyze_loop_operations): Properly guard analysis for
28799 (vect_transform_loop): Likewise.
28800 (vect_analyze_loop_2): Also reset SLP type on PHIs.
28801 (vect_model_induction_cost): Do not cost for pure SLP.
28802 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
28803 of induction in inner loop vectorization.
28804 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
28805 (vect_get_and_check_slp_defs): Handle vect_induction_def.
28806 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
28808 (vect_analyze_slp_cost_1): Cost induction.
28809 (vect_detect_hybrid_slp_stmts): Handle PHIs.
28810 (vect_get_slp_vect_defs): Likewise.
28811 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
28812 (vect_transform_stmt): Handle SLP reductions.
28813 * tree-vectorizer.h (vectorizable_induction): Adjust.
28815 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28817 * config/rs6000/rs6000.c (make_resolver_func): Update
28818 init_lowered_empty_function call.
28820 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
28822 * doc/invoke.texi: Document the -fprofile-abs-path option.
28823 * common.opt (fprofile-abs-path): New option.
28824 * gcov-io.h (gcov_write_filename): Declare.
28825 * gcov-io.c (gcov_write_filename): New function.
28826 * coverage.c (coverage_begin_function): Use gcov_write_filename.
28827 * profile.c (output_location): Likewise.
28829 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28831 * shring-wrap.c: Revert accidental commit.
28833 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
28835 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
28837 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28839 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
28841 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
28842 profile in callgraph edge.
28843 * profile-count.h (apply_probability): If THIS is 0, then result is 0
28844 (apply_scale): Likewise.
28845 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
28846 Also scale profile when inlining function with zero profile.
28847 (initialize_cfun): Update exit block profile even when it is zero.
28848 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
28849 when profile is read.
28851 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
28853 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
28854 (CLONE_*): New constants to define the processors we can generate
28855 code for with the target_clone attribute.
28856 (rs6000_clone_map): New array to identify which clone processors
28857 the current program is running on.
28858 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
28859 target_clone attribute.
28860 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
28861 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
28862 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
28863 (cpu_expand_builtin): Add support for target_clone attribute.
28864 (rs6000_valid_attribute_p): Allow "default" attribute.
28865 (get_decl_name): New debug function to simplify printing the
28866 current function name in debugging statements.
28867 (rs6000_clone_priority): New functions to support the target_clone
28868 attribute, and be able to generate code to switch between ISA 2.05
28869 through ISA 3.0 (power6 through power9).
28870 (rs6000_compare_version_priority): Likewise.
28871 (rs6000_get_function_versions_dispatcher): Likewise.
28872 (make_resolver_func): Likewise.
28873 (add_condition_to_bb): Likewise.
28874 (dispatch_function_versions): Likewise.
28875 (rs6000_generate_version_dispatcher_body): Likewise.
28876 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
28877 (fusion_gpr_load_p): Fix a spacing issue.
28878 * doc/extend.texi (Common Function Attributes): Document that the
28879 PowerPC supports the target_clone attribute.
28881 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
28883 * config/arm/arm.h: explain F symbol found in description of ARM
28884 register allocation in its legend.
28886 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28888 * config/mips/frame-header-opt.c: Include profile-count.h.
28889 * config/riscv/riscv.c: Include profile-count.h
28891 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
28893 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
28895 (sm_set_flag_if_changed): Add bbs field.
28896 (execute_sm_if_changed_flag_set): Pass BBS.
28897 (execute_sm): Update.
28899 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28901 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
28904 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28906 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
28907 (peephole2): New peephole2 to emit the above.
28908 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
28910 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28912 * config/aarch64/aarch64.c (define_peephole2 above
28913 *sub_<shift>_<mode>): New peephole.
28915 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28917 * config/i386/i386.c (make_resolver_func): Update.
28918 * Makefile.in: Add profile-count.h and profile-count.o
28919 * auto-profile.c (afdo_indirect_call): Update to new API.
28920 (afdo_set_bb_count): Update.
28921 (afdo_propagate_edge): Update.
28922 (afdo_propagate_circuit): Update.
28923 (afdo_calculate_branch_prob): Update.
28924 (afdo_annotate_cfg): Update.
28925 * basic-block.h: Include profile-count.h
28926 (struct edge_def): Turn count to profile_count.
28927 (struct basic_block_def): Likewie.
28928 (REG_BR_PROB_BASE): Move to profile-count.h
28929 (RDIV): Move to profile-count.h
28930 * bb-reorder.c (max_entry_count): Turn to profile_count.
28931 (find_traces): Update.
28932 (rotate_loop):Update.
28933 (connect_traces):Update.
28934 (sanitize_hot_paths):Update.
28935 * bt-load.c (migrate_btr_defs): Update.
28936 * cfg.c (RDIV): Remove.
28937 (init_flow): Use alloc_block.
28938 (alloc_block): Uninitialize count.
28939 (unchecked_make_edge): Uninitialize count.
28940 (check_bb_profile): Update.
28941 (dump_edge_info): Update.
28942 (dump_bb_info): Update.
28943 (update_bb_profile_for_threading): Update.
28944 (scale_bbs_frequencies_int): Update.
28945 (scale_bbs_frequencies_gcov_type): Update.
28946 (scale_bbs_frequencies_profile_count): New.
28947 * cfg.h (update_bb_profile_for_threading): Update.
28948 (scale_bbs_frequencies_profile_count): Declare.
28949 * cfgbuild.c (compute_outgoing_frequencies): Update.
28950 (find_many_sub_basic_blocks): Update.
28951 * cfgcleanup.c (try_forward_edges): Update.
28952 (try_crossjump_to_edge): Update.
28953 * cfgexpand.c (expand_gimple_tailcall): Update.
28954 (construct_exit_block): Update.
28955 * cfghooks.c (verify_flow_info): Update.
28956 (dump_bb_for_graph): Update.
28957 (split_edge): Update.
28958 (make_forwarder_block): Update.
28959 (duplicate_block): Update.
28960 (account_profile_record): Update.
28961 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
28962 (get_estimated_loop_iterations): Update.
28963 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
28964 (single_likely_exit): Update.
28965 * cfgloopmanip.c (scale_loop_profile): Update.
28967 (set_zero_probability): Update.
28968 (lv_adjust_loop_entry_edge): Update.
28969 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
28970 (purge_dead_edges): Update.
28971 (rtl_account_profile_record): Update.
28972 * cgraph.c (cgraph_node::create): Uninitialize count.
28973 (symbol_table::create_edge): Uninitialize count.
28974 (cgraph_update_edges_for_call_stmt_node): Update.
28975 (cgraph_edge::dump_edge_flags): Update.
28976 (cgraph_node::dump): Update.
28977 (cgraph_edge::maybe_hot_p): Update.
28978 * cgraph.h: Include profile-count.h
28979 (create_clone), create_edge, create_indirect_edge): Update.
28980 (cgraph_node): Turn count to profile_count.
28981 (cgraph_edge0: Likewise.
28982 (make_speculative, clone): Update.
28983 (create_edge): Update.
28984 (init_lowered_empty_function): Update.
28985 * cgraphclones.c (cgraph_edge::clone): Update.
28986 (duplicate_thunk_for_node): Update.
28987 (cgraph_node::create_clone): Update.
28988 * cgraphunit.c (cgraph_node::analyze): Update.
28989 (cgraph_node::expand_thunk): Update.
28990 * final.c (dump_basic_block_info): Update.
28991 * gimple-streamer-in.c (input_bb): Update.
28992 * gimple-streamer-out.c (output_bb): Update.
28993 * graphite.c (print_global_statistics): Update.
28994 (print_graphite_scop_statistics): Update.
28995 * hsa-brig.c: Include basic-block.h.
28996 * hsa-dump.c: Include basic-block.h.
28997 * hsa-gen.c (T sum_slice): Update.
28998 (convert_switch_statements):Update.
28999 * hsa-regalloc.c: Include basic-block.h.
29000 * ipa-chkp.c (chkp_produce_thunks): Update.
29001 * ipa-cp.c (struct caller_statistics): Update.
29002 (init_caller_stats): Update.
29003 (gather_caller_stats): Update.
29004 (ipcp_cloning_candidate_p): Update.
29005 (good_cloning_opportunity_p): Update.
29006 (get_info_about_necessary_edges): Update.
29007 (dump_profile_updates): Update.
29008 (update_profiling_info): Update.
29009 (update_specialized_profile): Update.
29010 (perhaps_add_new_callers): Update.
29011 (decide_about_value): Update.
29012 (ipa_cp_c_finalize): Update.
29013 * ipa-devirt.c (struct odr_type_warn_count): Update.
29014 (struct decl_warn_count): Update.
29015 (struct final_warning_record): Update.
29016 (possible_polymorphic_call_targets): Update.
29017 (ipa_devirt): Update.
29018 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29019 * ipa-icf.c (sem_function::merge): Update.
29020 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29021 * ipa-inline.c (compute_uninlined_call_time): Update.
29022 (compute_inlined_call_time): Update.
29023 (want_inline_small_function_p): Update.
29024 (want_inline_self_recursive_call_p): Update.
29025 (edge_badness): Update.
29026 (lookup_recursive_calls): Update.
29027 (recursive_inlining): Update.
29028 (inline_small_functions): Update.
29029 (dump_overall_stats): Update.
29030 (dump_inline_stats): Update.
29031 * ipa-profile.c (ipa_profile_generate_summary): Update.
29032 (ipa_propagate_frequency): Update.
29033 (ipa_profile): Update.
29034 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29035 * ipa-utils.c (ipa_merge_profiles): Update.
29036 * loop-doloop.c (doloop_modify): Update.
29037 * loop-unroll.c (report_unroll): Update.
29038 (unroll_loop_runtime_iterations): Update.
29039 * lto-cgraph.c (lto_output_edge): Update.
29040 (lto_output_node): Update.
29041 (input_node): Update.
29042 (input_edge): Update.
29043 (merge_profile_summaries): Update.
29044 * lto-streamer-in.c (input_cfg): Update.
29045 * lto-streamer-out.c (output_cfg): Update.
29046 * mcf.c (create_fixup_graph): Update.
29047 (adjust_cfg_counts): Update.
29048 (sum_edge_counts): Update.
29049 * modulo-sched.c (sms_schedule): Update.
29050 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29051 * predict.c (maybe_hot_count_p): Update.
29052 (probably_never_executed): Update.
29053 (dump_prediction): Update.
29054 (combine_predictions_for_bb): Update.
29055 (propagate_freq): Update.
29056 (handle_missing_profiles): Update.
29057 (counts_to_freqs): Update.
29058 (rebuild_frequencies): Update.
29059 (force_edge_cold): Update.
29060 * predict.h: Include profile-count.h
29061 (maybe_hot_count_p, counts_to_freqs): UPdate.
29062 * print-rtl-function.c: Do not include cfg.h
29063 * print-rtl.c: Include basic-block.h
29064 * profile-count.c: New file.
29065 * profile-count.h: New file.
29066 * profile.c (is_edge_inconsistent): Update.
29067 (correct_negative_edge_counts): Update.
29068 (is_inconsistent): Update.
29069 (set_bb_counts): Update.
29070 (read_profile_edge_counts): Update.
29071 (compute_frequency_overlap): Update.
29072 (compute_branch_probabilities): Update; Initialize and deinitialize
29074 (branch_prob): Update.
29075 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29076 (edge_gcov_count): New.
29077 (bb_gcov_count): New.
29078 * shrink-wrap.c (try_shrink_wrapping): Update.
29079 * tracer.c (better_p): Update.
29080 * trans-mem.c (expand_transaction): Update.
29081 (ipa_tm_insert_irr_call): Update.
29082 (ipa_tm_insert_gettmclone_call): Update.
29083 * tree-call-cdce.c: Update.
29084 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29085 (gimple_duplicate_sese_tail): Update.
29086 (gimple_account_profile_record): Update.
29087 (execute_fixup_cfg): Update.
29088 * tree-inline.c (copy_bb): Update.
29089 (copy_edges_for_bb): Update.
29090 (initialize_cfun): Update.
29091 (freqs_to_counts): Update.
29092 (copy_cfg_body): Update.
29093 (expand_call_inline): Update.
29094 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29095 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29096 (try_unroll_loop_completely): Update.
29097 (try_peel_loop): Update.
29098 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29099 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29100 * tree-ssa-loop-split.c (connect_loops): Update.
29101 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29102 * tree-ssa-reassoc.c (branch_fixup): Update.
29103 * tree-ssa-tail-merge.c (replace_block_by): Update.
29104 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29105 (compute_path_counts): Update.
29106 (update_profile): Update.
29107 (recompute_probabilities): Update.
29108 (update_joiner_offpath_counts): Update.
29109 (estimated_freqs_path): Update.
29110 (freqs_to_counts_path): Update.
29111 (clear_counts_path): Update.
29112 (ssa_fix_duplicate_block_edges): Update.
29113 (duplicate_thread_path): Update.
29114 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29115 (struct switch_conv_info): Update.
29116 * tree-tailcall.c (decrease_profile): Update.
29117 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29118 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29119 * value-prof.c (check_counter): Update.
29120 (gimple_divmod_fixed_value): Update.
29121 (gimple_mod_pow2): Update.
29122 (gimple_mod_subtract): Update.
29123 (gimple_ic_transform): Update.
29124 (gimple_stringop_fixed_value): Update.
29125 * value-prof.h (gimple_ic): Update.
29127 2017-06-02 Carl Love <cel@us.ibm.com>
29129 * config/rs6000/rs6000-c: Add support for built-in functions
29130 vector double vec_doublee (vector signed int);
29131 vector double vec_doublee (vector unsigned int);
29132 vector double vec_doublee (vector float);
29133 vector double vec_doubleh (vector signed int);
29134 vector double vec_doubleh (vector unsigned int);
29135 vector double vec_doubleh (vector float);
29136 vector double vec_doublel (vector signed int);
29137 vector double vec_doublel (vector unsigned int);
29138 vector double vec_doublel (vector float);
29139 vector double vec_doubleo (vector signed int);
29140 vector double vec_doubleo (vector unsigned int);
29141 vector double vec_doubleo (vector float);.
29142 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29143 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29145 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29146 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29147 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29149 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29150 vec_doublel, vec_doubleh.
29151 * doc/extend.texi: Update the built-in documentation file for the
29152 new built-in functions.
29154 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29157 * ipa-inline-analysis.c (free_growth_caches): Set
29158 edge_removal_hook_holder to NULL after removing it.
29160 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29162 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29163 comparision with zero.
29165 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29166 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29167 for early expansion of vec_min and vec_max builtins.
29168 (builtin_function_type): Add min/max unsigned variants to those
29169 identified as having unsigned arguments.
29171 2017-06-02 Olivier Hainque <hainque@adacore.com>
29173 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29175 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29177 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29178 Use VALL_F16 iterator rather than VALL.
29180 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29182 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29183 Emit CBNZ inside loop when doing a strong exchange and comparing
29184 against zero. Generate the CC flags after the loop.
29186 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29188 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29189 (dl_section_ref): New.
29190 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29191 On AIX, append an expression to subtract the size of the
29192 section length to dl_section_ref.
29194 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29196 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29197 for early expansion of vector absolute builtins.
29199 2017-06-02 Richard Biener <rguenther@suse.de>
29201 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29202 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29204 2017-06-02 Richard Biener <rguenther@suse.de>
29206 PR tree-optimization/80948
29207 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29208 stmt order as well.
29210 2017-06-02 Richard Biener <rguenther@suse.de>
29212 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29214 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29215 for inductions as relevant.
29217 2017-06-02 Richard Biener <rguenther@suse.de>
29219 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29220 (vectorizable_induction): ... this. Remove dead code.
29222 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29224 * builtins. (expand_builtin_alloca): Remove second parameter and
29225 infer its value from the first parameter instead.
29226 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29228 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29230 PR rtl-optimization/80903
29231 * loop-doloop.c (add_test): Unshare sequence.
29233 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29235 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29237 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29239 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29241 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29242 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29243 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29244 (xlogue_layout::get_stub_rtx): Make static.
29245 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29246 (xlogue_layout::compute_stub_managed_regs): Rename to...
29247 (xlogue_layout::count_stub_managed_regs): ...this.
29248 (xlogue_layout::is_stub_managed_reg): New function.
29249 (xlogue_layout::m_stub_names): Rename to...
29250 (xlogue_layout::s_stub_names): ...this, make static.
29251 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29252 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29253 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29254 xlogue_layout::s_stub_names): Instantiate statics.
29255 (stub_managed_regs): Remove.
29256 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29257 (disable_call_ms2sysv_xlogues): Rename to...
29258 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29259 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29261 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29262 change after reload_completed.
29263 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29265 (ix86_expand_prologue): Likewise.
29266 (ix86_expand_epilogue): Likewise.
29267 (ix86_expand_split_stack_prologue): Likewise.
29268 (ix86_compute_frame_layout): Remove frame parameter ...
29269 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29270 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29272 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29273 (ix86_frame): Move from here ...
29274 * config/i386/i386.h (ix86_frame): ... to here.
29275 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29276 complete ix86_frame data structure instead. Remove some_ld_name.
29278 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29280 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29281 symbols that hold a DECL_VALUE_EXPR.
29283 2017-06-01 Martin Jambor <mjambor@suse.cz>
29285 PR tree-optimization/80898
29286 * tree-sra.c (process_subtree_disqualification): Removed.
29287 (disqualify_candidate): Do not acll
29288 process_subtree_disqualification.
29289 (subtree_mark_written_and_enqueue): New function.
29290 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29291 RHS has been disqualified and re-queue LHS if necessary. Apart
29292 from that, ignore disqualified RHS.
29294 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29296 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29297 address fetch for z10 or later.
29299 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29301 * config/arc/arc.md (tst_movb): Add guard when splitting.
29303 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29305 * config/arc/arc.c (arc_can_eliminate): Test against
29306 arc_frame_pointer_needed.
29308 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29310 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29311 to prevent store reordering.
29312 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29313 (type): Add block type.
29314 (stack_tie): Define special instruction to be used in
29317 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29319 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29320 constraint. It is not valid for the pattern.
29321 (noncommutative_binary_comparison): Likewise.
29323 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29325 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29328 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29330 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29331 be used by the reg-alloc.
29333 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29335 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29336 reg-alloc when having mul64 or mul32x16 instructions.
29337 (mulsidi3): Likewise.
29338 (umulsidi3): Likewise.
29339 (mulsi32x16): New pattern.
29340 (mulsi64): Likewise.
29341 (mulsidi64): Likewise.
29342 (umulsidi64): Likewise.
29343 (MUL32x16_REG): Define.
29344 (mul64_600): Use MUL32x16_REG.
29345 (mac64_600): Likewise.
29346 (umul64_600): Likewise.
29347 (umac64_600): Likewise.
29349 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29351 * config/arc/arc.md (mulsi3_700): Make it commutative.
29353 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29355 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29357 (*sign_extendsidi2_insn): Likewise for movstosw.
29359 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29361 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29362 the type of the input discriminant value. Convert the
29363 discriminant value of signedness vary.
29365 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29367 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29368 Add to -Wall section.
29370 2017-06-01 Richard Biener <rguenther@suse.de>
29372 PR middle-end/66313
29373 * fold-const.c (fold_plusminus_mult_expr): If the factored
29374 factor may be zero use a wrapping type for the inner operation.
29375 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29376 and handle moved defs.
29377 (process_assignment): Properly guard the unary op case. Return a
29378 tri-state indicating that moving the stmt before the call may allow
29379 to continue. Pass through to_move.
29380 (find_tail_calls): Handle moving unrelated defs before
29383 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29386 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29387 splitter result in the canonical way.
29389 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29391 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29392 also for 32bit target. Update insn attributes.
29393 (zero-extendsidi2 splitter): Allow all registers for operand 1.
29395 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
29397 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
29398 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
29399 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
29400 (_mm_maskz_min_ss): New intrinsics.
29402 2017-05-31 Martin Liska <mliska@suse.cz>
29404 * tree-vect-loop.c (vect_create_epilog_for_reduction):
29405 Change comment style to one we normally use.
29406 (vectorizable_reduction): Likewise.
29407 (vectorizable_induction): Likewise.
29408 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
29409 (vectorizable_call): Likewise.
29410 (vectorizable_simd_clone_call): Likewise.
29411 (vectorizable_conversion): Likewise.
29412 (vectorizable_assignment): Likewise.
29413 (vectorizable_shift): Likewise.
29414 (vectorizable_operation): Likewise.
29415 (vectorizable_store): Likewise.
29416 (vectorizable_load): Likewise.
29417 * tree-vectorizer.h: Likewise.
29419 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
29421 * passes.c (emergency_dump_function): New.
29422 * tree-pass.h (emergency_dump_function): Declare.
29423 * plugin.c (plugins_internal_error_function): Remove.
29424 * plugin.h (plugins_internal_error_function): Remove declaration.
29425 * toplev.c (internal_error_function): New static function. Use it...
29426 (general_init): ...here.
29428 2017-05-31 Graham Markall <graham.markall@embecosm.com>
29430 * config/arc/arc.c (arc_print_operand): Handle constant operands.
29431 (arc_rtx_costs): Add costs for new patterns.
29432 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
29433 * config/arc/predicates.md: Add _1_2_3_operand predicate.
29435 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29437 * tree-ssa-strlen.c (get_next_strinfo): New function.
29438 (get_stridx_plus_constant): Use it.
29439 (zero_length_string): Likewise.
29440 (adjust_related_strinfos): Likewise.
29441 (adjust_last_stmt): Likewise.
29443 2017-05-31 Richard Biener <rguenther@suse.de>
29446 * config/i386/i386.c (ix86_expand_builtin): Remove assert
29447 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
29449 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
29451 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
29452 loop_vinfo argument and use of dependence distance vectors.
29453 Check instead whether the two references differ only in their
29454 initial value and assume that they have the same alignment if the
29455 difference is a multiple of the vector alignment.
29456 (vect_analyze_data_refs_alignment): Update call accordingly.
29458 2017-05-31 Martin Liska <mliska@suse.cz>
29461 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
29463 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29465 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
29466 (create_intersect_range_checks): Move from ...
29467 * tree-data-ref.c (create_intersect_range_checks_index)
29468 (create_intersect_range_checks): ... to here.
29469 (create_runtime_alias_checks): New function factored from ...
29470 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
29471 here. Call above function.
29472 * tree-data-ref.h (create_runtime_alias_checks): New function.
29474 2017-05-31 Bin Cheng <bin.cheng@arm.com>
29476 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
29477 segment length for dr_b and compute it in wide_int.
29479 2017-05-31 Richard Biener <rguenther@suse.de>
29481 PR tree-optimization/80906
29482 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
29483 and pass through iv_map.
29484 (copy_bb_and_scalar_dependences): Adjust.
29485 (translate_pending_phi_nodes): Likewise.
29486 (copy_loop_close_phi_args): Handle code-generating IVs instead
29489 2017-05-30 David Malcolm <dmalcolm@redhat.com>
29491 * diagnostic-color.c (color_dict): Add "type-diff".
29492 (parse_gcc_colors): Update comment.
29493 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
29494 -fdiagnostics-show-template-tree and -fno-elide-type.
29495 (GCC_COLORS): Add type-diff to example.
29497 (-fdiagnostics-show-template-tree): New.
29498 (-fno-elide-type): New.
29499 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
29500 the pp_format_decoder callback. Call any m_format_postprocessor's
29502 (pretty_printer::pretty_printer): Initialize
29503 m_format_postprocessor.
29504 (pretty_printer::~pretty_printer): Delete any
29505 m_format_postprocessor.
29506 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
29507 (class format_postprocessor): New class.
29508 (struct pretty_printer::format_decoder): Document the new parameters.
29509 (struct pretty_printer::m_format_postprocessor): New field.
29510 * tree-diagnostic.c (default_tree_printer): Update for new
29511 bool and const char ** params.
29512 * tree-diagnostic.h (default_tree_printer): Likewise.
29514 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
29516 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
29517 (lwa_operand): Delete rs6000_gen_cell_microcode test.
29518 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29519 rs6000_gen_cell_microcode code.
29520 (rs6000_final_prescan_insn): Delete.
29521 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
29522 "warn-cell-microcode" entries.
29523 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
29524 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
29525 throughout. Change cc_reg_not_micro_cr0_operand to
29526 cc_reg_not_cr0_operand throughout.
29527 (*extendhi<mode>2_noload): Delete.
29528 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
29529 (mwarn-cell-microcode): Delete.
29530 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
29531 -mgen-cell-microcode and -mwarn-cell-microcode.
29533 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
29536 * config/i386/constraints.md (Yd): New constraint.
29538 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
29539 and (?Yd, r) alternatives. Update insn attributes.
29540 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
29541 and (?*Yd, r) alternatives. Update insn attributes.
29542 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
29544 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
29546 * gimplify.c (gimplify_modify_expr): Don't create a
29547 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
29550 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
29552 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
29554 2017-05-30 Richard Biener <rguenther@suse.de>
29556 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
29557 and reduc_def fields.
29558 (STMT_VINFO_REDUC_TYPE): New define.
29559 (STMT_VINFO_REDUC_DEF): Likewise.
29560 (vect_force_simple_reduction): Adjust prototype.
29561 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
29562 (vect_is_simple_reduction): Remove check_reduction argument.
29563 (vect_force_simple_reduction): Adjust and set
29564 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29565 (vectorizable_reduction): Do not re-do reduction analysis
29566 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
29567 * tree-parloops.c (gather_scalar_reductions): Adjust.
29569 2017-05-30 Richard Biener <rguenther@suse.de>
29571 PR middle-end/80901
29572 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
29575 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
29577 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
29578 Introduce unknown_misalignment parameter and remove vf.
29579 (vect_peeling_hash_get_lowest_cost):
29580 Pass unknown_misalignment parameter.
29581 (vect_enhance_data_refs_alignment):
29582 Fix unsupportable data ref treatment.
29584 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29586 * tree-vect-data-refs.c (vect_get_data_access_cost):
29587 Workaround for SLP handling.
29588 (vect_enhance_data_refs_alignment):
29589 Compute costs for doing no peeling at all, compare to the best
29590 peeling costs so far and avoid peeling if cheaper.
29592 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29594 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
29595 Return peeling info and set costs to zero for unlimited cost
29597 (vect_enhance_data_refs_alignment): Also inspect all datarefs
29598 with unknown misalignment. Compute and costs for unknown
29599 misalignment, compare them to the costs for known misalignment
29600 and choose the cheapest for peeling.
29602 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29604 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
29605 (vect_get_peeling_costs_all_drs): Create function.
29606 (vect_peeling_hash_get_lowest_cost):
29607 Use vect_get_peeling_costs_all_drs.
29608 (vect_peeling_supportable): Create function.
29609 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
29611 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
29613 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
29614 DR_HAS_NEGATIVE_STEP.
29615 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
29616 (vect_enhance_data_refs_alignment): Use.
29617 (vect_duplicate_ssa_name_ptr_info): Use.
29618 * tree-vectorizer.h (dr_misalignment): Use.
29619 (known_alignment_for_access_p): Use.
29621 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
29624 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
29626 (has_section_name): New function.
29628 2017-05-30 Martin Liska <mliska@suse.cz>
29631 * auto-profile.c (get_function_decl_from_block): Fix
29634 2017-05-30 Richard Biener <rguenther@suse.de>
29636 PR middle-end/80876
29637 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
29639 2017-05-30 Martin Liska <mliska@suse.cz>
29641 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
29642 * dumpfile.h (struct dump_file_info): Remove ctors.
29644 2017-05-30 Martin Liska <mliska@suse.cz>
29646 * predict.def: Fix GNU coding style.
29648 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29650 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29651 Mark 'to' argument with ATTRIBUTE_UNUSED.
29653 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
29655 * config/xtensa/xtensa.c (xtensa_emit_call): Use
29656 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
29657 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
29660 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
29662 * doc/install.texi (Options specification): Restore entry of
29663 --enable-sjlj-exceptions.
29665 2017-05-27 Michael Eager <eager@eagercon.com>
29668 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
29670 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
29672 * config/microblaze/microblaze.h
29673 (FIXED_REGISTERS): Update in macro.
29674 (CALL_USED_REGISTERS): Update in macro.
29676 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
29678 * doc/install.texi: Add links to macOS binary distributions.
29680 2017-05-27 Jakub Jelinek <jakub@redhat.com>
29684 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29686 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29688 2017-05-26 Martin Liska <mliska@suse.cz>
29690 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
29692 2017-05-26 Martin Liska <mliska@suse.cz>
29694 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
29695 always leading ';; '.
29696 (dump_bb_info): Likewise.
29697 (brief_dump_cfg): Likewise.
29698 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
29699 * dumpfile.c: Remove usage of TDF_VERBOSE.
29700 * dumpfile.h (enum dump_kind): Likewise.
29701 (dump_gimple_bb_header): Do not use TDF_COMMENT.
29702 * print-tree.c (debug_verbose): Remove.
29703 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
29704 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
29705 * tree-diagnostic.c (default_tree_printer): Replace
29706 TDF_DIAGNOSTIC with TDF_SLIM.
29708 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29710 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
29711 in parameter loop, rather than loop_vinfo.
29712 (create_intersect_range_checks): Ditto.
29713 (vect_create_cond_for_alias_checks): Update call to above functions.
29715 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29717 PR tree-optimization/80815
29718 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
29719 for merging runtime alias checks. Handle negative DR_STEPs.
29721 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29723 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
29725 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
29726 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
29727 out code pruning runtime alias checks.
29728 * tree-data-ref.c (prune_runtime_alias_test_list): New function
29729 factored out from above.
29730 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29732 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
29734 (prune_runtime_alias_test_list): New decalaration.
29736 2017-05-26 Bin Cheng <bin.cheng@arm.com>
29738 * tree-vect-data-refs.c (compare_tree): Rename and move ...
29739 * tree-data-ref.c (data_ref_compare_tree): ... to here.
29740 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
29741 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
29742 (operator==, comp_dr_with_seg_len_pair): Ditto.
29743 (vect_prune_runtime_alias_test_list): Ditto.
29745 2017-05-26 Martin Liska <mliska@suse.cz>
29748 * params.def: Bound partial-inlining-entry-probability param.
29750 2017-05-26 Marek Polacek <polacek@redhat.com>
29753 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
29756 2017-05-26 Richard Biener <rguenther@suse.de>
29758 PR tree-optimization/80842
29759 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
29762 2017-05-26 Richard Biener <rguenther@suse.de>
29764 PR tree-optimization/80844
29765 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
29767 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
29769 * doc/md.texi (Machine Constraints): Update x86 family
29770 machine constraints section to match 'config/i386/constraints.md'.
29772 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
29774 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
29776 2017-05-25 Nathan Sidwell <nathan@acm.org>
29778 * doc/invoke.texi (--enable-languages): Update documentation.
29780 2017-05-25 Martin Liska <mliska@suse.cz>
29782 * dumpfile.c: Add TDF_FOLDING.
29783 * dumpfile.h (enum dump_kind): Likewise.
29784 * genmatch.c (dt_simplify::gen_1): Use it.
29786 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29788 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
29790 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29792 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
29793 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
29795 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
29797 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
29798 * match.pd (X == C): Rewrite it here.
29799 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
29800 with_certain_nonzero_bits2): New predicates.
29801 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
29803 2017-05-24 Nathan Sidwell <nathan@acm.org>
29805 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
29808 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
29811 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29813 * config/powerpcspe: New port. Files are copied from the rs6000
29814 port, with "rs6000" in filenames replaced by "powerpcspe".
29816 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
29818 PR rtl-optimization/80754
29819 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
29821 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
29823 * config/sparc/sparc.md (length): Return the correct value for -mflat
29824 sibcalls to match output_sibcall.
29826 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
29830 * config/rs6000/rs6000.c (struct machine_function): Add new field
29832 (rs6000_get_separate_components): Init that field, use it.
29833 (rs6000_components_for_bb): Use the field.
29835 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
29837 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
29839 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
29841 PR middle-end/80823
29842 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
29844 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29847 * config/s390/s390.c (s390_check_qrst_address): Check incoming
29848 address against address_operand predicate.
29849 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
29851 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
29853 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
29854 parameters passed indirectly.
29856 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
29858 * config/i386/i386.md (*movdi_internal): Remove SSE4
29859 alternative 18 (?r, *v). Update insn attributes.
29860 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
29861 Update insn attributes.
29862 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
29863 Update insn attributes.
29864 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
29865 alternative 1 (r, v). Remove isa attribute.
29866 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
29867 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
29868 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
29870 2017-05-23 Tom de Vries <tom@codesourcery.com>
29872 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
29875 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29877 * cgraphunit.c (symbol_table::process_new_functions): Update.
29878 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
29879 (inline_generate_summary): Rename to ...
29880 (ipa_fn_summary_generate): ... this one.
29881 (inline_read_summary): Rename to ...
29882 (ipa_fn_summary_read): ... this one.
29883 (inline_write_summary): Rename to ...
29884 (ipa_fn_summary_write): ... this one.
29885 (inline_free_summary): Rename to ...
29886 (ipa_free_fn_summary): ... this one.
29887 (pass_data_local_fn_summary, pass_local_fn_summary,
29888 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
29889 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
29890 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
29891 make_pass_ipa_fn_summary): New.
29892 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
29893 inline_write_summary, inline_free_summary): Remove.
29894 (ipa_free_fn_summary) : New.
29895 * ipa-inline.c (ipa_inline): Update.
29896 (pass_ipa_inline): Do not generate summaries.
29897 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
29899 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
29900 and add pass_ipa_fn_summary.
29901 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
29903 (make_pass_inline_parameters): Remove.
29905 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
29907 * omp-low.c (struct omp_context): Remove "default_kind" member.
29910 * omp-offload.c (execute_oacc_device_lower): Remove the
29911 parallelism dimensions function attributes for unparallelized
29912 OpenACC kernels constructs.
29914 2017-05-23 Martin Liska <mliska@suse.cz>
29916 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
29918 (cgraph_edge::make_speculative): Likewise.
29919 (cgraph_edge::resolve_speculation): Likewise.
29920 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29921 (cgraph_node::dump): Likewise.
29922 * cgraph.h: Likewise.
29923 * cgraphunit.c (analyze_functions): Likewise.
29924 (symbol_table::compile): Likewise.
29925 * ipa-cp.c (print_all_lattices): Likewise.
29926 (determine_versionability): Likewise.
29927 (initialize_node_lattices): Likewise.
29928 (ipcp_verify_propagated_values): Likewise.
29929 (estimate_local_effects): Likewise.
29930 (update_profiling_info): Likewise.
29931 (create_specialized_node): Likewise.
29932 (perhaps_add_new_callers): Likewise.
29933 (decide_about_value): Likewise.
29934 (decide_whether_version_node): Likewise.
29935 (identify_dead_nodes): Likewise.
29936 (ipcp_store_bits_results): Likewise.
29937 * ipa-devirt.c (dump_targets): Likewise.
29938 (ipa_devirt): Likewise.
29939 * ipa-icf.c (sem_item::dump): Likewise.
29940 (sem_function::equals): Likewise.
29941 (sem_variable::equals): Likewise.
29942 (sem_item_optimizer::read_section): Likewise.
29943 (sem_item_optimizer::execute): Likewise.
29944 (congruence_class::dump): Likewise.
29945 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
29946 (dump_inline_summary): Likewise.
29947 (estimate_node_size_and_time): Likewise.
29948 (inline_analyze_function): Likewise.
29949 * ipa-inline-transform.c (inline_call): Likewise.
29950 * ipa-inline.c (report_inline_failed_reason): Likewise.
29951 (want_early_inline_function_p): Likewise.
29952 (edge_badness): Likewise.
29953 (update_edge_key): Likewise.
29954 (inline_small_functions): Likewise.
29955 * ipa-profile.c (ipa_profile): Likewise.
29956 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
29957 (ipa_make_edge_direct_to_target): Likewise.
29958 (remove_described_reference): Likewise.
29959 (ipa_impossible_devirt_target): Likewise.
29960 (propagate_controlled_uses): Likewise.
29961 (ipa_print_node_params): Likewise.
29962 (ipcp_transform_function): Likewise.
29963 * ipa-pure-const.c (pure_const_read_summary): Likewise.
29964 (propagate_pure_const): Likewise.
29965 * ipa-reference.c (generate_summary): Likewise.
29966 (read_write_all_from_decl): Likewise.
29967 (propagate): Likewise.
29968 (ipa_reference_read_optimization_summary): Likewise.
29969 * ipa-utils.c (ipa_merge_profiles): Likewise.
29970 * ipa.c (walk_polymorphic_call_targets): Likewise.
29971 (symbol_table::remove_unreachable_nodes): Likewise.
29972 (ipa_single_use): Likewise.
29973 * passes.c (execute_todo): Likewise.
29974 * predict.c (drop_profile): Likewise.
29975 * symtab.c (symtab_node::get_dump_name): New function.
29976 (symtab_node::dump_name): Likewise.
29977 (symtab_node::dump_asm_name): Likewise.
29978 (symtab_node::dump_references): Likewise.
29979 (symtab_node::dump_referring): Likewise.
29980 (symtab_node::dump_base): Likewise.
29981 (symtab_node::debug_symtab): Likewise.
29982 * tree-sra.c (convert_callers_for_node): Likewise.
29983 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
29984 * value-prof.c (init_node_map): Likewise.
29986 2017-05-23 Martin Liska <mliska@suse.cz>
29988 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
29989 and symtab_node::debug_symtab to symbol_table::debug.
29990 * cgraphunit.c (analyze_functions): Use the renamed function.
29991 (symbol_table::compile): Likewise.
29992 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
29993 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
29994 * passes.c (execute_todo): Likewise.
29995 * symtab.c (symbol_table::dump): New function.
29996 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
29998 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30000 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30001 that nonconst implies exec.
30003 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30005 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30006 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30007 (inline_edge_summary_vec): Turn into ...
30008 (ipa_call_summaries): ... this one.
30009 (redirect_to_unreachable, edge_set_predicate,
30010 evaluate_properties_for_edge, inline_summary_alloc,
30011 reset_ipa_call_summary, reset_inline_summary,
30012 inline_summary_t::duplicate): Update.
30013 (inline_edge_duplication_hook): Turn to ...
30014 (ipa_call_summary_t::duplicate): ... this one.
30015 (inline_edge_removal_hook): Turn to ...
30016 (ipa_call_summary_t::remove): ... this one.
30017 (dump_inline_edge_summary): Turn to ...
30018 (dump_ipa_call_summary): ... this one.
30019 (estimate_function_body_sizes): Update.
30020 (inline_update_callee_summaries): Update.
30021 (remap_edge_change_prob): Update.
30022 (remap_edge_summaries): Update.
30023 (inline_merge_summary): Update.
30024 (do_estimate_edge_time): Update.
30025 (inline_generate_summary): Update.
30026 (inline_read_section): Update.
30027 (inline_read_summary): Update.
30028 (inline_free_summary): Update.
30029 * ipa-inline.c (can_inline_edge_p): Update.
30030 (compute_inlined_call_time): Update.
30031 (want_inline_small_function_p): Update.
30032 (edge_badness): Update.
30033 (early_inliner): Update.
30034 * ipa-inline.h (inline_edge_summary): Turn to ...
30035 (ipa_call_summary): ... this one.
30036 (ipa_call_summary_t): New class.
30037 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30038 (ipa_call_summaries): New.
30039 (inline_edge_summary): Remove.
30040 (estimate_edge_growth): Update.
30041 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30042 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30043 * ipa-split.c (execute_split_functions): Update.
30044 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30046 2017-05-23 Tom de Vries <tom@codesourcery.com>
30048 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30049 attributes): Document rdrand effective target.
30051 2017-05-23 Tom de Vries <tom@codesourcery.com>
30053 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30054 attributes): Sort alphabetically.
30056 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30058 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30060 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30063 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30064 V2DF/V2DI splat into two separate patterns, one that handles
30065 registers, and the other that only handles memory. Drop support
30066 for splatting from a GPR on ISA 2.07 and then splitting the
30067 splat into direct move and splat.
30068 (vsx_splat_<mode>_reg): Likewise.
30069 (vsx_splat_<mode>_mem): Likewise.
30071 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30073 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30075 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30077 PR middle-end/80809
30078 * omp-low.c (finish_taskreg_remap): New function.
30079 (finish_taskreg_scan): If unit size of ctx->record_type
30080 is non-constant, unshare the size expression and replace
30081 decls in it with possible outer var refs.
30083 PR middle-end/80809
30084 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30085 GOVD_SHARED rather than GOVD_PRIVATE with it.
30086 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30087 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30089 PR middle-end/80853
30090 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30091 as last argument to build_outer_var_ref for pointer bases of array
30092 section reductions.
30094 2017-05-19 Martin Sebor <msebor@redhat.com>
30096 * print-tree.c (print_node): Print DECL_READ_P flag.
30098 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30100 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30101 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30102 * cgraph.c: Likewise.
30103 * cgraphunit.c: Likewise.
30104 * gengtype.c: Likewise.
30105 * ipa-cp.c: Likewise.
30106 * ipa-devirt.c: Likewise.
30107 * ipa-icf.c: Likewise.
30108 * ipa-predicate.c: Likewise.
30109 * ipa-profile.c: Likewise.
30110 * ipa-prop.c: Likewise.
30111 * ipa-split.c: Likewise.
30113 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30114 edge_predicate_pool, dump_inline_hints,
30115 inline_summary::account_size_time, redirect_to_unreachable,
30116 edge_set_predicate, set_hint_predicate,
30117 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30118 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30119 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30120 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30121 ipa_call_summary_t::remove, initialize_growth_caches,
30122 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30123 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30124 mark_modified, unmodified_parm_1, unmodified_parm,
30125 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30126 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30127 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30128 will_be_nonconstant_predicate, record_modified_bb_info,
30129 get_minimal_bb, record_modified, param_change_prob,
30130 phi_result_unknown_predicate, predicate_for_phi_result,
30131 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30132 estimate_function_body_sizes, compute_inline_parameters,
30133 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30134 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30135 inline_update_callee_summaries, remap_edge_change_prob,
30136 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30137 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30138 inline_analyze_function, inline_summary_t::insert,
30139 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30140 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30141 inline_free_summary): Move to ipa-fnsummary.h
30142 (predicate_t): Remove.
30143 * ipa-fnsummary.c: New file.
30144 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30145 (enum inline_hints_vals, inline_hints, agg_position_info,
30146 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30147 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30148 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30149 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30150 inline_read_summary, inline_write_summary, inline_free_summary,
30151 inline_analyze_function, initialize_inline_failed,
30152 inline_merge_summary, inline_update_overall_summary,
30153 compute_inline_parameters): Move to ipa-fnsummary.h
30154 * ipa-fnsummary.h: New file.
30155 * ipa-inline-transform.h: Include ipa-inline.h.
30156 * ipa-inline.c: LIkewise.
30158 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30160 * ipa-inline.c (edge_badness): Use inlined_time instead of
30161 inline_summaries->get.
30163 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30165 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30167 2017-05-22 Nathan Sidwell <nathan@acm.org>
30169 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30170 (fdump-lang): Document 'raw' option.
30171 * dumpfile.h (TDI_tu): Delete.
30172 * dumpfile.c (dump_files): Remove translation-unit.
30173 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30175 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30177 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30178 command option from $(AWK) call.
30179 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30180 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30181 [FORMAT]: Remove handling of variable.
30182 * config/avr/t-multilib: Regenerate.
30184 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30186 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30188 (dump_inline_summary): Do not print self_time.
30189 (estimate_function_body_sizes): Do not set self_time.
30190 (compute_inline_parameters): Likewise.
30191 (inline_read_section, inline_write_summary): Do not stream self_time.
30192 * ipa-inline.h (inline_summary): Drop self_time.
30194 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30196 * ipa-inline-analysis.c (account_size_time): Rename to ...
30197 (inline_summary::account_size_time): ... this one.
30198 (reset_ipa_call_summary): Turn to ...
30199 (ipa_call_summary::reset): ... this one.
30200 (reset_inline_summary): Turn to ...
30201 (inline_summary::reset): ... this one.
30202 (inline_summary_t::remove): Update.
30203 (inline_summary_t::duplicate): Update.
30204 (ipa_call_summary_t::remove): Update.
30205 (dump_inline_summary): Update.
30206 (estimate_function_body_sizes): Update.
30207 (compute_inline_parameters): Update.
30208 (estimate_node_size_and_time): Update.
30209 (inline_merge_summary): Update.
30210 (inline_update_overall_summary): Update.
30211 (inline_read_section): Update.
30212 (inline_write_summary): Update.
30213 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30214 add account_size_time and reset member functions.
30215 (ipa_call_summary): Add reset function.
30216 * ipa-predicate.h (predicate::operator &): Constify.
30218 2017-05-22 Richard Biener <rguenther@suse.de>
30220 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30222 2017-05-19 Jason Merrill <jason@redhat.com>
30224 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30226 2017-05-19 Marek Polacek <polacek@redhat.com>
30229 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30230 TYPE_OVERFLOW_WRAPS checks.
30232 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30234 * tree-core.h (enum omp_clause_default_kind): Add
30235 "OMP_CLAUSE_DEFAULT_PRESENT".
30236 * tree-pretty-print.c (dump_omp_clause): Handle it.
30237 * gimplify.c (enum gimplify_omp_var_data): Add
30238 "GOVD_MAP_FORCE_PRESENT".
30239 (gimplify_adjust_omp_clauses_1): Map it to
30240 "GOMP_MAP_FORCE_PRESENT".
30241 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30243 * gimplify.c (oacc_default_clause): Clarify.
30245 2017-05-19 Nathan Sidwell <nathan@acm.org>
30247 LANG_HOOK_REGISTER_DUMPS
30248 * toplev.c (general_init): Call register dump lang hook.
30249 * doc/invoke.texi: Document -fdump-lang option family.
30250 * dumpfile.c (dump_files): Remove class dump here.
30251 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30252 * dumpfile.h (tree_dump_index): Remove TDI_class.
30253 * langhooks-def.h (lhd_register_dumps): Declare.
30254 (LANG_HOOKS_REGISTER_DUMPS): Define.
30255 (LANG_HOOKS_INITIALIZER): Add it.
30256 * langhooks.c (lhd_register_dumps): Define.
30257 * langhooks.h (struct lang_hooks): Add register_dumps.
30259 2017-05-19 Nathan Sidwell <nathan@acm.org>
30261 * context.h (context::set_passes): New.
30262 * context.c (context::context): Do not create pass manager.
30263 * toplev.c (general_init): Create pass manager here.
30265 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30267 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30268 use this splitter if two add or or instructions would also work for
30269 the constant we want to generate.
30271 2017-05-19 Richard Biener <rguenther@suse.de>
30274 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30275 predicate evaluation.
30277 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30279 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30281 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30284 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30286 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30287 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30288 (inline_edge_summary_vec): Turn into ...
30289 (ipa_call_summaries): ... this one.
30290 (redirect_to_unreachable, edge_set_predicate,
30291 evaluate_properties_for_edge, inline_summary_alloc,
30292 reset_ipa_call_summary, reset_inline_summary,
30293 inline_summary_t::duplicate): Update.
30294 (inline_edge_duplication_hook): Turn to ...
30295 (ipa_call_summary_t::duplicate): ... this one.
30296 (inline_edge_removal_hook): Turn to ...
30297 (ipa_call_summary_t::remove): ... this one.
30298 (dump_inline_edge_summary): Turn to ...
30299 (dump_ipa_call_summary): ... this one.
30300 (estimate_function_body_sizes): Update.
30301 (inline_update_callee_summaries): Update.
30302 (remap_edge_change_prob): Update.
30303 (remap_edge_summaries): Update.
30304 (inline_merge_summary): Update.
30305 (do_estimate_edge_time): Update.
30306 (inline_generate_summary): Update.
30307 (inline_read_section): Update.
30308 (inline_read_summary): Update.
30309 (inline_free_summary): Update.
30310 * ipa-inline.c (can_inline_edge_p): Update.
30311 (compute_inlined_call_time): Update.
30312 (want_inline_small_function_p): Update.
30313 (edge_badness): Update.
30314 (early_inliner): Update.
30315 * ipa-inline.h (inline_edge_summary): Turn to ...
30316 (ipa_call_summary): ... this one.
30317 (ipa_call_summary_t): New class.
30318 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30319 (ipa_call_summaries): New.
30320 (inline_edge_summary): Remove.
30321 (estimate_edge_growth): Update.
30322 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30323 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30324 * ipa-split.c (execute_split_functions): Update.
30325 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30327 2017-05-19 Richard Biener <rguenther@suse.de>
30329 PR middle-end/80764
30330 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30332 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30334 * config/rs6000/rs6000.c (struct machine_function): Add field
30335 fpr_is_wrapped_separately.
30336 (rs6000_get_separate_components): Use 64 components. Handle the
30337 new FPR components.
30338 (rs6000_components_for_bb): Handle the FPR components.
30339 (rs6000_emit_prologue_components): Handle the FPR components.
30340 (rs6000_emit_epilogue_components): Handle the FPR components.
30341 (rs6000_set_handled_components): Handle the FPR components.
30342 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30343 that are already separately shrink-wrapped.
30344 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30345 that are already separately shrink-wrapped.
30347 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30350 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30353 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30354 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30355 cases where the register allocator uses a move and an offsettable
30356 memory operation to/from a FPR register on ISA 2.06/2.07.
30357 (define_peephole2 for Altivec d-form store): Likewise.
30359 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30362 * config/i386/mmx.md (*mov<mode>_internal): Enable
30363 alternatives 11, 12, 13 and 14 also for 32bit targets.
30364 Remove alternatives 15, 16, 17 and 18.
30365 * config/i386/sse.md (vec_concatv2di): Change
30366 alternative (!x, *y) to (x, ?!*Yn).
30368 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30370 * dumpfile.h (enum dump_kind): Remove stray comma.
30372 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30374 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30375 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30376 predicate::num_conditions
30377 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30378 (CHANGED): turn into predicate::changed.
30379 (agg_position_info): Move to ipa-predicate.h
30380 (add_condition, predicate::add_clause, predicate::operator &=,
30381 predicate::or_with, predicate::evaluate, predicate::probability,
30382 dump_condition, dump_clause, predicate::dump,
30383 predicate::remap_after_duplication, predicate::remap_after_inlining,
30384 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30385 (evaluate_conditions_for_known_args): Update.
30386 (set_cond_stmt_execution_predicate): Update.
30387 * ipa-inline.h: Include ipa-predicate.h
30388 (condition, inline_param_summary, conditions, agg_position_info,
30389 predicate): Move to ipa-predicate.h
30390 * ipa-predicate.c: New file.
30391 * ipa-predicate.h: New file.
30393 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
30395 * final.c (leaf_function_p): Check we are not in a sequence.
30397 2017-05-18 Martin Liska <mliska@suse.cz>
30399 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
30400 * dumpfile.c (dump_register): Use new enum dump_kind.
30401 (get_dump_file_name): Likewise.
30402 (dump_enable_all): Likewise.
30403 (dump_switch_p_1): Likewise.
30404 (enable_rtl_dump_file): Remove usage of TDF_RTL.
30405 * dumpfile.h (enum dump_kind): New enum type.
30406 (struct dump_file_info): Create constructor and
30407 format fields and comments.
30408 * passes.c (pass_manager::register_one_dump_file):
30410 * statistics.c (statistics_early_init): Likewise.
30411 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
30412 TDF_TREE with TDF_SLIM.
30413 (gather_memory_references_ref): Likewise.
30415 2017-05-18 Martin Liska <mliska@suse.cz>
30417 * vec.h (struct vnull): Use it.
30419 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30421 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
30422 (true_predicate, false_predicate, true_predicate_p,
30423 false_predicate_p): Remove.
30424 (single_cond_predicate, not_inlined_predicate): Turn to member function
30426 (add_condition): Update.
30427 (add_clause): Turn to...
30428 (predicate::add_clause): ... this one; update; allow passing NULL
30430 (and_predicates): Turn to ...
30431 (predicate::operator &=): ... this one.
30432 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
30433 (or_predicates): Turn to ...
30434 (predicate::or_with): ... this one.
30435 (evaluate_predicate): Turn to ...
30436 (predicate::evaluate): ... this one.
30437 (predicate_probability): Turn to ...
30438 (predicate::probability): ... this one.
30439 (dump_condition): Update.
30440 (dump_predicate): Turn to ...
30441 (predicate::dump): ... this one.
30442 (account_size_time): Update.
30443 (edge_set_predicate): Update.
30444 (set_hint_predicate): UPdate.
30445 (evaluate_conditions_for_known_args): Update.
30446 (evaluate_properties_for_edge): Update.
30447 (remap_predicate_after_duplication): Turn to...
30448 (predicate::remap_after_duplication): ... this one.
30449 (remap_hint_predicate_after_duplication): Update.
30450 (inline_summary_t::duplicate): UPdate.
30451 (dump_inline_edge_summary): Update.
30452 (dump_inline_summary): Update.
30453 (set_cond_stmt_execution_predicate): Update.
30454 (set_switch_stmt_execution_predicate): Update.
30455 (compute_bb_predicates): Update.
30456 (will_be_nonconstant_expr_predicate): Update.
30457 (will_be_nonconstant_predicate): Update.
30458 (phi_result_unknown_predicate): Update.
30459 (predicate_for_phi_result): Update.
30460 (array_index_predicate): Update.
30461 (estimate_function_body_sizes): Update.
30462 (estimate_node_size_and_time): Update.
30463 (estimate_ipcp_clone_size_and_time): Update.
30464 (remap_predicate): Rename to ...
30465 (predicate::remap_after_inlining): ... this one.
30466 (remap_hint_predicate): Update.
30467 (inline_merge_summary): Update.
30468 (inline_update_overall_summary): Update.
30469 (estimate_size_after_inlining): Update.
30470 (read_predicate): Rename to ...
30471 (predicate::stream_in): ... this one.
30472 (read_inline_edge_summary): Update.
30473 (write_predicate): Rename to ...
30474 (predicate::stream_out): ... this one.
30475 (write_inline_edge_summary): Update.
30476 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
30477 (clause_t): Turn to uint32_t
30478 (predicate): Turn to class; implement constructor and operators
30480 (size_time_entry): Update.
30481 (inline_summary): Update.
30482 (inline_edge_summary): Update.
30484 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
30486 * fold-const.c (fold_binary_loc): Move transformation...
30487 * match.pd (C - X CMP X): ... here.
30489 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
30491 * config/sparc/sparc.c (sparc_option_override): Set function
30492 alignment for -mcpu=niagara7 to 64 to match the I$ line.
30493 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
30495 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
30497 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
30499 2017-05-18 Marek Polacek <polacek@redhat.com>
30502 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
30503 (pass_ubsan::execute): Call gimple_assign_single_p instead of
30504 gimple_assign_load_p.
30506 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
30508 PR middle-end/80692
30509 * real.c (do_compare): Give decimal_do_compare preference over
30510 comparing just the signs.
30512 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
30514 * doc/md.texi (Canonicalization of Instructions): Describe the
30515 canonical form of instructions that inherently set a condition
30518 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
30520 PR middle-end/80775
30521 * tree-cfg.c: Move deletion of unreachable case statements to after
30522 the merging of consecutive case labels.
30524 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
30526 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
30527 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
30528 restoring of callee-saved registers.
30530 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
30532 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
30533 * config/visium/visium.c (single_set_and_flags): Likewise.
30534 * config/visium/visium.md (Substitutions): Likewise.
30536 2017-05-17 Martin Liska <mliska@suse.cz>
30538 * cfg.c: Introduce dump_flags_t type and
30539 use it instead of int type.
30541 * cfghooks.c: Likewise.
30542 * cfghooks.h (struct cfg_hooks): Likewise.
30543 * cfgrtl.c: Likewise.
30544 * cfgrtl.h: Likewise.
30545 * cgraph.c (cgraph_node::get_body): Likewise.
30546 * coretypes.h: Likewise.
30547 * domwalk.c: Likewise.
30548 * domwalk.h: Likewise.
30549 * dumpfile.c (struct dump_option_value_info): Likewise.
30550 (dump_enable_all): Likewise.
30551 (dump_switch_p_1): Likewise.
30552 (opt_info_switch_p): Likewise.
30553 * dumpfile.h (enum tree_dump_index): Likewise.
30554 (struct dump_file_info): Likewise.
30555 * genemit.c: Likewise.
30556 * generic-match-head.c: Likewise.
30557 * gengtype.c (open_base_files): Likewise.
30558 * gimple-pretty-print.c: Likewise.
30559 * gimple-pretty-print.h: Likewise.
30560 * graph.c (print_graph_cfg): Likewise.
30561 * graphite-scop-detection.c (dot_all_sese): Likewise.
30562 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
30563 * loop-unroll.c (report_unroll): Likewise.
30564 * passes.c (pass_manager::register_one_dump_file): Likewise.
30565 * print-tree.c: Likewise.
30566 * statistics.c: Likewise.
30567 * tree-cfg.c: Likewise.
30568 * tree-cfg.h: Likewise.
30569 * tree-dfa.c: Likewise.
30570 * tree-dfa.h: Likewise.
30571 * tree-dump.c (dump_function): Likewise.
30572 * tree-dump.h (struct dump_info): Likewise.
30573 * tree-pretty-print.c: Likewise.
30574 * tree-pretty-print.h: Likewise.
30575 * tree-ssa-live.c: Likewise.
30576 * tree-ssa-live.h: Likewise.
30577 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
30578 * tree-vect-loop.c: Likewise.
30579 * tree-vect-slp.c: Likewise.
30581 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
30582 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30584 PR tree-optimization/80457
30585 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
30586 of all arguments to a statement as scalar_to_vec operations.
30587 (vectorizable_call): Adjust call to vect_model_simple_cost for
30589 (vectorizable_conversion): Likewise.
30590 (vectorizable_assignment): Likewise.
30591 (vectorizable_shift): Likewise.
30592 (vectorizable_operation): Likewise.
30593 (vectorizable_comparison): Likewise.
30594 (vect_is_simple_cond): Record the def types for operands.
30595 (vectorizable_condition): Likewise, call vect_model_simple_cost.
30596 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
30597 for statement argument count.
30599 2017-05-16 Carl Love <cel@us.ibm.com>
30601 * config/rs6000/rs6000-c: Add support for built-in functions
30602 vector unsigned long long vec_bperm (vector unsigned long long,
30603 vector unsigned char)
30604 vector signed long long vec_mule (vector signed int,
30606 vector unsigned long long vec_mule (vector unsigned int,
30607 vector unsigned int)
30608 vector signed long long vec_mulo (vector signed int,
30610 vector unsigned long long vec_mulo (vector unsigned int,
30611 vector unsigned int)
30612 vector signed char vec_sldw (vector signed char,
30613 vector signed char,
30615 vector unsigned char vec_sldw (vector unsigned char,
30616 vector unsigned char,
30618 vector signed short vec_sldw (vector signed short,
30619 vector signed short,
30621 vector unsigned short vec_sldw (vector unsigned short,
30622 vector unsigned short,
30624 vector signed int vec_sldw (vector signed int,
30627 vector unsigned int vec_sldw (vector unsigned int,
30628 vector unsigned int,
30630 vector signed long long vec_sldw (vector signed long long,
30631 vector signed long long,
30633 vector unsigned long long vec_sldw (vector unsigned long long,
30634 vector unsigned long long,
30636 * config/rs6000/rs6000-c: Add support for built-in functions
30637 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
30638 * config/rs6000/altivec.h: Add defintion for vec_sldw.
30639 * doc/extend.texi: Update the built-in documentation for the
30640 new built-in functions.
30642 2017-05-16 Marek Polacek <polacek@redhat.com>
30646 * tree.c (save_expr): Don't fold the expression.
30648 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
30650 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
30651 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
30652 and (?*y,m). Update insn attributes.
30654 2017-05-16 Martin Liska <mliska@suse.cz>
30656 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
30657 flags argument of print_gimple_stmt, print_gimple_expr,
30658 print_generic_stmt and print_generic_expr.
30659 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
30660 * coretypes.h: Likewise.
30661 * except.c (dump_eh_tree): Likewise.
30662 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
30663 * gimple-pretty-print.h: Likewise.
30664 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
30665 (backprop::push_to_worklist): Likewise.
30666 (backprop::pop_from_worklist): Likewise.
30667 (backprop::process_use): Likewise.
30668 (backprop::intersect_uses): Likewise.
30669 (note_replacement): Likewise.
30670 * gimple-ssa-store-merging.c
30671 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
30672 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
30673 (pass_store_merging::execute): Likewise.
30674 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
30675 (ssa_base_cand_dump_callback): Likewise.
30676 (dump_incr_vec): Likewise.
30677 (replace_refs): Likewise.
30678 (replace_mult_candidate): Likewise.
30679 (create_add_on_incoming_edge): Likewise.
30680 (create_phi_basis): Likewise.
30681 (insert_initializers): Likewise.
30682 (all_phi_incrs_profitable): Likewise.
30683 (introduce_cast_before_cand): Likewise.
30684 (replace_one_candidate): Likewise.
30685 * gimplify.c (gimplify_expr): Likewise.
30686 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
30687 (set_rename): Likewise.
30688 (rename_uses): Likewise.
30689 (copy_loop_phi_nodes): Likewise.
30690 (add_close_phis_to_merge_points): Likewise.
30691 (copy_loop_close_phi_args): Likewise.
30692 (copy_cond_phi_args): Likewise.
30693 (graphite_copy_stmts_from_block): Likewise.
30694 (translate_pending_phi_nodes): Likewise.
30695 * graphite-poly.c (print_pdr): Likewise.
30696 (dump_gbb_cases): Likewise.
30697 (dump_gbb_conditions): Likewise.
30698 (print_scop_params): Likewise.
30699 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
30700 (build_cross_bb_scalars_use): Likewise.
30701 (gather_bbs::before_dom_children): Likewise.
30702 * hsa-dump.c (dump_hsa_immed): Likewise.
30703 * ipa-cp.c (print_ipcp_constant_value): Likewise.
30704 (get_replacement_map): Likewise.
30705 * ipa-inline-analysis.c (dump_condition): Likewise.
30706 (estimate_function_body_sizes): Likewise.
30707 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
30708 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
30709 * ipa-prop.c (ipa_dump_param): Likewise.
30710 (ipa_print_node_jump_functions_for_edge): Likewise.
30711 (ipa_modify_call_arguments): Likewise.
30712 (ipa_modify_expr): Likewise.
30713 (ipa_dump_param_adjustments): Likewise.
30714 (ipa_dump_agg_replacement_values): Likewise.
30715 (ipcp_modif_dom_walker::before_dom_children): Likewise.
30716 * ipa-pure-const.c (check_stmt): Likewise.
30717 (pass_nothrow::execute): Likewise.
30718 * ipa-split.c (execute_split_functions): Likewise.
30719 * omp-offload.c (dump_oacc_loop_part): Likewise.
30720 (dump_oacc_loop): Likewise.
30721 * trans-mem.c (tm_log_emit): Likewise.
30722 (tm_memopt_accumulate_memops): Likewise.
30723 (dump_tm_memopt_set): Likewise.
30724 (dump_tm_memopt_transform): Likewise.
30725 * tree-cfg.c (gimple_verify_flow_info): Likewise.
30726 (print_loop): Likewise.
30727 * tree-chkp-opt.c (chkp_print_addr): Likewise.
30728 (chkp_gather_checks_info): Likewise.
30729 (chkp_get_check_result): Likewise.
30730 (chkp_remove_check_if_pass): Likewise.
30731 (chkp_use_outer_bounds_if_possible): Likewise.
30732 (chkp_reduce_bounds_lifetime): Likewise.
30733 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
30734 (chkp_mark_completed_bounds): Likewise.
30735 (chkp_register_incomplete_bounds): Likewise.
30736 (chkp_mark_invalid_bounds): Likewise.
30737 (chkp_maybe_copy_and_register_bounds): Likewise.
30738 (chkp_build_returned_bound): Likewise.
30739 (chkp_get_bound_for_parm): Likewise.
30740 (chkp_build_bndldx): Likewise.
30741 (chkp_get_bounds_by_definition): Likewise.
30742 (chkp_generate_extern_var_bounds): Likewise.
30743 (chkp_get_bounds_for_decl_addr): Likewise.
30744 * tree-chrec.c (chrec_apply): Likewise.
30745 * tree-data-ref.c (dump_data_reference): Likewise.
30746 (dump_subscript): Likewise.
30747 (dump_data_dependence_relation): Likewise.
30748 (analyze_overlapping_iterations): Likewise.
30749 * tree-inline.c (expand_call_inline): Likewise.
30750 (tree_function_versioning): Likewise.
30751 * tree-into-ssa.c (dump_defs_stack): Likewise.
30752 (dump_currdefs): Likewise.
30753 (dump_names_replaced_by): Likewise.
30754 (dump_update_ssa): Likewise.
30755 (update_ssa): Likewise.
30756 * tree-object-size.c (pass_object_sizes::execute): Likewise.
30757 * tree-parloops.c (build_new_reduction): Likewise.
30758 (try_create_reduction_list): Likewise.
30759 (ref_conflicts_with_region): Likewise.
30760 (oacc_entry_exit_ok_1): Likewise.
30761 (oacc_entry_exit_single_gang): Likewise.
30762 * tree-pretty-print.h: Likewise.
30763 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
30764 (get_scalar_evolution): Likewise.
30765 (add_to_evolution): Likewise.
30766 (get_loop_exit_condition): Likewise.
30767 (analyze_evolution_in_loop): Likewise.
30768 (analyze_initial_condition): Likewise.
30769 (analyze_scalar_evolution): Likewise.
30770 (instantiate_scev): Likewise.
30771 (number_of_latch_executions): Likewise.
30772 (gather_chrec_stats): Likewise.
30773 (final_value_replacement_loop): Likewise.
30774 (scev_const_prop): Likewise.
30775 * tree-sra.c (dump_access): Likewise.
30776 (disqualify_candidate): Likewise.
30777 (create_access): Likewise.
30778 (reject): Likewise.
30779 (maybe_add_sra_candidate): Likewise.
30780 (create_access_replacement): Likewise.
30781 (analyze_access_subtree): Likewise.
30782 (analyze_all_variable_accesses): Likewise.
30783 (sra_modify_assign): Likewise.
30784 (initialize_constant_pool_replacements): Likewise.
30785 (find_param_candidates): Likewise.
30786 (decide_one_param_reduction): Likewise.
30787 (replace_removed_params_ssa_names): Likewise.
30788 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
30789 * tree-ssa-copy.c (dump_copy_of): Likewise.
30790 (copy_prop_visit_cond_stmt): Likewise.
30791 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
30792 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
30793 (record_equivalences_from_stmt): Likewise.
30794 * tree-ssa-dse.c (compute_trims): Likewise.
30795 (delete_dead_call): Likewise.
30796 (delete_dead_assignment): Likewise.
30797 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
30798 (forward_propagate_into_cond): Likewise.
30799 (pass_forwprop::execute): Likewise.
30800 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
30801 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
30803 (move_computations_worker): Likewise.
30804 (execute_sm): Likewise.
30805 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
30806 (remove_exits_and_undefined_stmts): Likewise.
30807 (remove_redundant_iv_tests): Likewise.
30808 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
30809 (adjust_iv_update_pos): Likewise.
30810 * tree-ssa-math-opts.c (bswap_replace): Likewise.
30811 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
30812 (value_replacement): Likewise.
30813 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
30814 * tree-ssa-pre.c (print_pre_expr): Likewise.
30815 (get_representative_for): Likewise.
30816 (create_expression_by_pieces): Likewise.
30817 (insert_into_preds_of_block): Likewise.
30818 (eliminate_insert): Likewise.
30819 (eliminate_dom_walker::before_dom_children): Likewise.
30820 (eliminate): Likewise.
30821 (remove_dead_inserted_code): Likewise.
30822 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
30823 * tree-ssa-reassoc.c (get_rank): Likewise.
30824 (eliminate_duplicate_pair): Likewise.
30825 (eliminate_plus_minus_pair): Likewise.
30826 (eliminate_not_pairs): Likewise.
30827 (undistribute_ops_list): Likewise.
30828 (eliminate_redundant_comparison): Likewise.
30829 (update_range_test): Likewise.
30830 (optimize_range_tests_var_bound): Likewise.
30831 (optimize_vec_cond_expr): Likewise.
30832 (rewrite_expr_tree): Likewise.
30833 (rewrite_expr_tree_parallel): Likewise.
30834 (linearize_expr): Likewise.
30835 (break_up_subtract): Likewise.
30836 (linearize_expr_tree): Likewise.
30837 (attempt_builtin_powi): Likewise.
30838 (attempt_builtin_copysign): Likewise.
30839 (transform_stmt_to_copy): Likewise.
30840 (transform_stmt_to_multiply): Likewise.
30841 (dump_ops_vector): Likewise.
30842 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
30843 (print_scc): Likewise.
30844 (set_ssa_val_to): Likewise.
30845 (visit_reference_op_store): Likewise.
30846 (visit_use): Likewise.
30847 (sccvn_dom_walker::before_dom_children): Likewise.
30848 (run_scc_vn): Likewise.
30849 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
30851 (expr_hash_elt::print): Likewise.
30852 (const_and_copies::pop_to_marker): Likewise.
30853 (const_and_copies::record_const_or_copy_raw): Likewise.
30854 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
30855 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
30856 (dump_predicates): Likewise.
30857 (find_uninit_use): Likewise.
30858 (warn_uninitialized_phi): Likewise.
30859 (pass_late_warn_uninitialized::execute): Likewise.
30860 * tree-ssa.c (verify_vssa): Likewise.
30861 (verify_ssa): Likewise.
30862 (maybe_optimize_var): Likewise.
30863 * tree-vrp.c (dump_value_range): Likewise.
30864 (dump_all_value_ranges): Likewise.
30865 (dump_asserts_for): Likewise.
30866 (register_edge_assert_for_2): Likewise.
30867 (vrp_visit_cond_stmt): Likewise.
30868 (vrp_visit_switch_stmt): Likewise.
30869 (vrp_visit_stmt): Likewise.
30870 (vrp_visit_phi_node): Likewise.
30871 (simplify_cond_using_ranges_1): Likewise.
30872 (fold_predicate_in): Likewise.
30873 (evrp_dom_walker::before_dom_children): Likewise.
30874 (evrp_dom_walker::push_value_range): Likewise.
30875 (evrp_dom_walker::pop_value_range): Likewise.
30876 (execute_early_vrp): Likewise.
30878 2017-05-16 Richard Biener <rguenther@suse.de>
30880 * dwarf2out.c (loc_list_from_tree_1): Do not create
30881 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
30883 2017-05-16 Richard Biener <rguenther@suse.de>
30885 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
30887 (note_variable_value_in_expr): If we resolved the decl ref
30888 do not push to the stack.
30890 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
30892 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
30893 operations in fast-math mode.
30894 (vaddq_f16): Likewise.
30895 (vmul_f16): Likewise.
30896 (vmulq_f16): Likewise.
30897 (vsub_f16): Likewise.
30898 (vsubq_f16): Likewise.
30899 * config/arm/neon.md (add<mode>3): New.
30901 (fma:<VH:mode>3): New. Also remove outdated comment.
30904 2017-05-16 Martin Liska <mliska@suse.cz>
30908 * ipa-devirt.c (warn_types_mismatch): Fix typo.
30909 (odr_types_equivalent_p): Likewise.
30911 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
30913 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
30915 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
30918 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
30919 non-interunit SSE move alternatives with '?'.
30920 (zero-extendsidi peephole2): New peephole to skip intermediate
30921 general register in SSE zero-extend sequence.
30923 2017-05-15 Jeff Law <law@redhat.com>
30925 * reorg.c (relax_delay_slots): Create a new variable to hold
30926 the temporary target rather than clobbering TARGET_LABEL.
30928 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
30929 missing argument to extract_bit_field call.
30930 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
30932 2017-05-15 Martin Liska <mliska@suse.cz>
30935 * gcc.c (process_command): Do not allow empty argument of -o option.
30937 2017-05-15 Renlin Li <renlin.li@arm.com>
30939 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
30940 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
30941 * config/aarch64/constraints.md (Usf): Add long call check.
30942 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
30943 (call_value): Likewise.
30944 (sibcall): Likewise.
30945 (sibcall_value): Likewise.
30947 (call_value_insn): New.
30948 (sibcall_insn): Update rtx pattern.
30949 (sibcall_value_insn): Likewise.
30950 (call_internal): Remove.
30951 (call_value_internal): Likewise.
30952 (sibcall_internal): Likewise.
30953 (sibcall_value_internal): Likewise.
30954 (call_reg): Likewise.
30955 (call_symbol): Likewise.
30956 (call_value_reg): Likewise.
30957 (call_value_symbol): Likewise.
30959 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
30962 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
30964 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
30966 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
30967 compatible with CCGOCmode and with CCZmode.
30969 2017-05-14 Martin Sebor <msebor@redhat.com>
30971 PR middle-end/77671
30972 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
30973 (gimple_fold_builtin_snprintf): Same.
30974 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
30975 (gimple_fold_builtin_snprintf): Same.
30976 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
30977 of character types.
30978 (is_call_safe): New function.
30979 (try_substitute_return_value): Call it.
30980 (try_simplify_call): New function.
30981 (pass_sprintf_length::handle_gimple_call): Call it.
30983 2017-05-14 Martin Sebor <msebor@redhat.com>
30985 PR middle-end/80669
30986 * builtins.c (expand_builtin_stpncpy): Simplify.
30988 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
30990 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
30991 * config/i386/i386.h
30992 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
30993 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
30994 (struct machine_function): Add new members call_ms2sysv,
30995 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
30996 (struct machine_frame_state): New fields sp_realigned and
30997 sp_realigned_offset.
30998 * config/i386/i386.c
30999 (enum xlogue_stub): New enum.
31000 (enum xlogue_stub_sets): New enum.
31001 (class xlogue_layout): New class.
31002 (struct ix86_frame): New fields stack_realign_allocate_offset,
31003 stack_realign_offset and outlined_save_offset. Modify comments to
31004 detail stack layout when using out-of-line stubs.
31005 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31006 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31007 -mcall-ms2sysv-xlogues.
31008 (stub_managed_regs): New static variable.
31009 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31010 registers managed by out-of-line stub.
31011 (disable_call_ms2sysv_xlogues): New function.
31012 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31013 m->call_ms2sysv when appropriate and compute frame layout for
31015 (sp_valid_at, fp_valid_at): New inline functions.
31016 (choose_basereg): New function.
31017 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31019 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31020 Use align parameter of choose_baseaddr to generated aligned SSE movs
31022 (pro_epilogue_adjust_stack): Modify to track
31023 machine_frame_state::sp_realigned.
31024 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31025 (ix86_nsaved_sseregs): Likewise.
31026 (ix86_emit_save_regs): Likewise.
31027 (ix86_emit_save_regs_using_mov): Likewise.
31028 (ix86_emit_save_sse_regs_using_mov): Likewise.
31029 (get_scratch_register_on_entry): Likewise.
31030 (gen_frame_set): New function.
31031 (gen_frame_load): Likewise.
31032 (gen_frame_store): Likewise.
31033 (emit_outlined_ms2sysv_save): Likewise.
31034 (emit_outlined_ms2sysv_restore): Likewise.
31035 (ix86_expand_prologue): Modify stack re-alignment code and call
31036 emit_outlined_ms2sysv_save when appropriate.
31037 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31038 parameter rtx_insn *insn, which allows the function to be used to only
31039 generate the notes.
31040 (ix86_expand_epilogue): Modify validity checks of frame and stack
31041 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31042 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31043 * config/i386/predicates.md
31044 (save_multiple): New predicate.
31045 (restore_multiple): Likewise.
31046 * config/i386/sse.md
31047 (save_multiple<mode>): New pattern.
31048 (save_multiple_realign<mode>): Likewise.
31049 (restore_multiple<mode>): Likewise.
31050 (restore_multiple_and_return<mode>): Likewise.
31051 (restore_multiple_leave_return<mode>): Likewise.
31052 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31054 2017-05-14 Julia Koval <julia.koval@intel.com>
31056 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31057 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31058 (__builtin_ia32_xsetbv): New builtins.
31059 * config/i386/i386.c (ix86_expand_special_args_builtin):
31061 (ix86_expand_builtin): Special expand for new intrinsics.
31062 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31063 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31064 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31066 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31068 * cfganal.c (inverted_post_order_compute): Change argument type
31070 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31071 * df-core.c (rest_of_handle_df_initialize): Adjust.
31072 (rest_of_handle_df_finish): Likewise.
31073 (df_analyze_1): Likewise.
31074 (df_analyze): Likewise.
31075 (loop_inverted_post_order_compute): Change argument to be a vec *.
31076 (df_analyze_loop): Adjust.
31077 (df_get_n_blocks): Likewise.
31078 (df_get_postorder): Likewise.
31079 * df.h (struct df_d): Change field to be a vec.
31080 * lcm.c (compute_laterin): Adjust.
31081 (compute_available): Likewise.
31082 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31083 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31084 * tree-ssa-pre.c (compute_antic): Likewise.
31086 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31088 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31089 (depth_first_search::depth_first_search): Change structure init
31090 function to this constructor.
31091 (depth_first_search::add_bb): Rename function to this member.
31092 (depth_first_search::execute): Likewise.
31093 (flow_dfs_compute_reverse_finish): Adjust.
31095 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31097 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31098 (longest_simple_path): Likewise.
31099 * shrink-wrap.c (spread_components): Likewise.
31100 (disqualify_problematic_components): Likewise.
31101 (emit_common_heads_for_components): Likewise.
31102 (emit_common_tails_for_components): Likewise.
31103 (insert_prologue_epilogue_for_components): Likewise.
31105 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31107 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31110 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31112 * df-core.c (df_set_blocks): Start using auto_bitmap.
31113 (df_compact_blocks): Likewise.
31114 * df-problems.c (df_rd_confluence_n): Likewise.
31115 * df-scan.c (df_insn_rescan_all): Likewise.
31116 (df_process_deferred_rescans): Likewise.
31117 (df_update_entry_block_defs): Likewise.
31118 (df_update_exit_block_uses): Likewise.
31119 (df_entry_block_bitmap_verify): Likewise.
31120 (df_exit_block_bitmap_verify): Likewise.
31121 (df_scan_verify): Likewise.
31122 * lra-constraints.c (lra_constraints): Likewise.
31123 (undo_optional_reloads): Likewise.
31124 (lra_undo_inheritance): Likewise.
31125 * lra-remat.c (calculate_gen_cands): Likewise.
31126 (do_remat): Likewise.
31127 * lra-spills.c (assign_spill_hard_regs): Likewise.
31128 (spill_pseudos): Likewise.
31129 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31130 (bitmap_set_subtract_values): Likewise.
31132 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31134 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31135 management with auto_bitmap.
31136 (fix_inter_tick): Likewise.
31137 (fix_recovery_deps): Likewise.
31138 * ira.c (add_store_equivs): Likewise.
31139 (find_moveable_pseudos): Likewise.
31140 (split_live_ranges_for_shrink_wrap): Likewise.
31141 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31142 (rtx_reuse_manager::seen_def_p): Likewise.
31143 (rtx_reuse_manager::set_seen_def): Likewise.
31144 * print-rtl.h (class rtx_reuse_manager): Likewise.
31146 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31148 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31150 (migrate_btr_def): Likewise.
31151 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31152 * df-core.c (loop_post_order_compute): Likewise.
31153 (loop_inverted_post_order_compute): Likewise.
31154 * hsa-common.h: Likewise.
31155 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31156 * init-regs.c (initialize_uninitialized_regs): Likewise.
31157 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31158 (inline_small_functions): Likewise.
31159 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31160 * ira.c (combine_and_move_insns): Likewise.
31161 (build_insn_chain): Likewise.
31162 * loop-invariant.c (find_invariants): Likewise.
31163 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31164 * predict.c (tree_predict_by_opcode): Likewise.
31165 (predict_paths_leading_to): Likewise.
31166 (predict_paths_leading_to_edge): Likewise.
31167 (estimate_loops_at_level): Likewise.
31168 (estimate_loops): Likewise.
31169 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31170 (spread_components): Likewise.
31171 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31172 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31173 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31174 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31175 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31176 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31177 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31178 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31179 (mark_threaded_blocks): Likewise.
31180 (thread_through_all_blocks): Likewise.
31181 * tree-ssa.c (verify_ssa): Likewise.
31182 (execute_update_addresses_taken): Likewise.
31183 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31185 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31187 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31189 (post_order_compute): Likewise.
31190 (inverted_post_order_compute): Likewise.
31191 (pre_and_rev_post_order_compute_fn): Likewise.
31193 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31195 * genrecog.c (int_set::int_set): Explicitly construct our
31196 auto_vec base class.
31197 * vec.h (auto_vec::auto_vec): New constructor.
31199 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31201 * bitmap.h (class auto_bitmap): New constructor taking
31202 bitmap_obstack * argument.
31204 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31206 * bitmap.h (class auto_bitmap): Change type of m_bits to
31207 bitmap_head, and adjust ctor / dtor and member operators.
31209 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31211 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31212 when returned register mode doesn't match original mode.
31214 2017-05-12 Jeff Law <law@redhat.com>
31215 Jakub Jelinek <jakub@redhat.com>
31217 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31218 we look for cc setter after the compare-elim changes.
31219 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31220 within the vector to match what compare-elim now expects.
31221 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31222 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31224 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31225 after the compare-elim changes.
31226 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31227 the vector to match what compare-elim now expects.
31228 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31229 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31230 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31231 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31232 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31234 * config/visium/visium.c (single_set_and_flags): Fix where
31235 we look for cc setter after the compare-elim changes.
31236 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31237 with the vector to match what compare-elim now expects.
31238 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31239 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31240 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31241 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31242 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31243 (neg<mode>2_insn_set_overflow): Likewise.
31245 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31247 PR middle-end/79794
31248 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31249 maybe_expand_insn call, set ops[0].target. If still set after call,
31250 set alt_rtl. Add extra arg to recursive calls.
31251 (extract_bit_field): Add alt_rtl argument. Pass to
31253 * expmed.h (extract_bit_field): Fix prototype.
31254 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31255 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31256 to extract_bit_field_calls.
31257 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31258 Pass alt_rtl to extract_bit_field calls.
31259 * calls.c (store_unaligned_arguments_into_psuedos)
31260 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31261 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31263 * optabs.h (struct expand_operand): Add target bitfield.
31265 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31267 * compare-elim.c (try_eliminate_compare): Canonicalize
31268 operation with embedded compare to
31269 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31270 (set (reg) (operation)].
31272 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31274 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31277 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31278 cost of adding a carry flag for ADC instruction.
31279 [case MINUS]: Ignore the cost of subtracting a carry flag
31280 for SBB instruction.
31282 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31284 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31286 * config/rs6000/bmiintrin.h: New file.
31287 * config/rs6000/bmi2intrin.h: New file.
31288 * config/rs6000/x86intrin.h: New file.
31290 2017-05-12 Jeff Law <law@redhat.com>
31292 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31295 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31297 PR middle-end/80707
31298 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31300 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31302 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31303 early expansion of vector divide builtins.
31304 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31305 builtins identified as having unsigned arguments.
31307 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31309 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31310 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31311 expansion of vector logical operations (and, andc, or, xor,
31314 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31316 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31317 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31319 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31321 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31322 early GIMPLE expansion of vector multiplies.
31324 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31326 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31327 TARGET_HAVE_MOVT conditional.
31328 (movt splitter): Likewise.
31330 2017-05-12 Richard Biener <rguenther@suse.de>
31332 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31333 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31334 Fold all stmts not inplace.
31336 2017-05-12 Richard Biener <rguenther@suse.de>
31338 PR tree-optimization/80713
31339 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31340 inserted_exprs bit for not removed stmts.
31342 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31344 PR middle-end/69921
31345 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31346 parallelized" attribute for parallelized OpenACC kernels.
31347 * omp-offload.c (execute_oacc_device_lower): Use it.
31349 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31350 Set "oacc kernels" attribute.
31351 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31352 parameter. Adjust all users.
31353 (oacc_fn_attrib_kernels_p): Remove function.
31354 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31355 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31356 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31357 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31358 assert "oacc kernels" attribute is set.
31360 2017-05-11 Carl Love <cel@us.ibm.com>
31362 * config/rs6000/rs6000-c: Add support for built-in functions
31363 vector unsigned char vec_popcnt (vector signed char)
31364 vector unsigned char vec_popcnt (vector unsigned char)
31365 vector unsigned short vec_popcnt (vector signed short)
31366 vector unsigned short vec_popcnt (vector unsigned short)
31367 vector unsigned int vec_popcnt (vector signed int)
31368 vector unsigned int vec_popcnt (vector unsigned int)
31369 vector unsigned long long vec_popcnt (vector signed long long)
31370 vector unsigned long long vec_popcnt (vector unsigned long long)
31371 vector signed long long vec_slo (vector signed long long,
31372 vector signed char)
31373 vector signed long long vec_slo (vector signed long long,
31374 vector unsigned char)
31375 vector unsigned long long vec_slo (vector unsigned long long,
31376 vector signed char)
31377 vector unsigned long long vec_slo (vector unsigned long long,
31378 vector unsigned char)
31379 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31380 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31381 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31382 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31383 * doc/extend.texi: Update the built-in documentation file for the
31384 new built-in functions.
31386 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31388 * attribs.h (sorted_attr_string): Move machine independent
31389 functions for target clone support from the i386 port to common
31390 code. Rename ix86_function_versions to common_function_versions.
31391 Rename make_name to make_unique_name.
31392 (common_function_versions): Likewise.
31393 (make_unique_name): Likewise.
31394 (make_dispatcher_decl): Likewise.
31395 (is_function_default_version): Likewise.
31396 * attribs.c (attr_strcmp): Likewise.
31397 (sorted_attr_string): Likewise.
31398 (common_function_versions): Likewise.
31399 (make_unique_name): Likewise.
31400 (make_dispatcher_decl): Likewise.
31401 (is_function_default_version): Likewise.
31402 * config/i386/i386.c (attr_strcmp): Likewise.
31403 (sorted_attr_string): Likewise.
31404 (ix86_function_versions): Likewise.
31405 (make_name): Likewise.
31406 (make_dispatcher_decl): Likewise.
31407 (is_function_default_version): Likewise.
31408 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
31410 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31413 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
31414 Account for direct move costs for vec_construct of integer
31417 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
31420 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
31421 (UNSPEC_STX_ATOMIC): Ditto.
31422 (loaddi_via_sse): New insn.
31423 (storedi_via_sse): Ditto.
31424 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
31425 Update corresponding peephole2 patterns.
31426 (atomic_storedi_fpu): Ditto.
31428 2017-05-11 Julia Koval <julia.koval@intel.com>
31430 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
31431 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
31433 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
31434 (__builtin_ia32_rsqrt14ss_mask): New builtins.
31435 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
31437 2017-05-11 Nathan Sidwell <nathan@acm.org>
31439 * graphite-poly.c: Include dumpfile.h.
31441 * dumpfle.h (dump_function): Declare here ...
31442 * tree-dump.h (dump_function): ... not here.
31443 * dumpfile.c: #include tree-cfg.h.
31444 (dump_function): Move here from ...
31445 * tree-dump.c (dump_function): ... here.
31446 * gimplify.c: #include splay-tree.h, not tree-dump.h.
31447 * graphite-poly.c: Don't include tree-dump.h.
31448 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
31449 * print-tree.c: Likewise.
31450 * stor-layout.c: Likewise.
31451 * tree-nested.c: Likewise.
31453 * dumpfile.c (dump_start): Use TDF_FLAGS.
31454 (dump_enable_all): Fix TDF_KIND check thinko.
31456 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
31458 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31459 array entries to represent two legal parameterizations of the
31460 overloaded __builtin_cmpb function, as represented by the
31461 P6_OV_BUILTIN_CMPB constant.
31462 (altivec_resolve_overloaded_builtin): Add special case handling
31463 for the __builtin_cmpb function, as represented by the
31464 P6_OV_BUILTIN_CMPB constant.
31465 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
31466 (BU_P6_64BIT_2): New macro.
31467 (BU_P6_OVERLOAD_2): New macro
31468 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
31469 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
31470 (CMPB): Add overload support to represent both 32-bit and 64-bit
31471 compare-bytes function.
31472 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
31473 support for TARGET_CMPB.
31474 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
31475 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
31476 documentation of the __builtin_cmpb overloaded built-in function.
31478 2017-05-11 Richard Biener <rguenther@suse.de>
31480 PR tree-optimization/80705
31481 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
31482 bases are not vectorizable.
31484 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31486 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
31487 when counting register pressure.
31489 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31491 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
31492 (iv_ca_more_deps): Renamed to ...
31493 (iv_ca_compare_deps): ... this.
31494 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
31496 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31498 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
31500 (determine_group_iv_costs): ... here.
31501 (find_inv_vars_cb): Record inv var if it's not recorded before.
31503 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31505 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
31506 (get_shiftadd_cost): Ditto.
31508 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31510 * tree-ssa-address.c: Include header file.
31511 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
31513 (add_to_parts): Refactor.
31514 (addr_to_parts): New parameter. Update use of move_hint_to_base.
31515 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
31518 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31520 PR tree-optimization/53090
31521 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
31523 (extract_cond_operands): Detect condition with IV on both sides
31524 and return COMP_IV_EXPR_2.
31525 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
31526 (rewrite_use_compare): Simplify by removing call to function
31527 extract_cond_operands.
31529 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31531 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
31532 (extract_cond_operands): Detect condition comparing against non-
31533 invariant bound and return appropriate enum value.
31534 (find_interesting_uses_cond): Update use of extract_cond_operands.
31535 Handle its return value accordingly.
31536 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
31538 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31540 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
31541 nonlinear iv_use computation in loop invariant sensitive way.
31543 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31545 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
31546 (find_iv_candidates): Call relate_compare_use_with_all_cands.
31548 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31550 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
31551 (dump_cand): Support iv_cand.inv_exprs.
31552 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
31554 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
31557 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31559 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
31561 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
31562 as local function. Include necessary header files.
31563 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
31565 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31567 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
31569 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31571 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
31572 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
31573 RSHIFT_EXPR and BIT_NOT_EXPR.
31575 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31577 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
31578 (adjust_setup_cost): New parameter supporting round up adjustment.
31579 (struct address_cost_data): Delete.
31580 (force_expr_to_var_cost): Don't bound cost with spill_cost.
31581 (split_address_cost, ptr_difference_cost): Delete.
31582 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
31583 (struct ainc_cost_data): New struct.
31584 (get_address_cost_ainc): New function.
31585 (get_address_cost, get_computation_cost): Reimplement.
31586 (determine_group_iv_cost_address): Record inv_expr for all uses of
31588 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
31589 (iv_ca_has_deps): Reimplemented to ...
31590 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
31592 (iv_ca_extend): Call iv_ca_more_deps.
31594 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31596 * tree-ssa-address.c (struct mem_address): Move to header file.
31597 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
31598 * tree-ssa-address.h (struct mem_address): Move from C file.
31599 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
31601 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31603 * tree-affine.h (aff_combination_type): New interface.
31604 (aff_combination_zero_p): Remove static.
31605 (aff_combination_const_p): New interface.
31606 (aff_combination_singleton_var_p): New interfaces.
31608 2017-05-11 Richard Biener <rguenther@suse.de>
31610 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31611 Skip unreachable blocks and destinations.
31612 (eliminate): Move stmt removal and fixup ...
31613 (fini_eliminate): ... here. Skip inserted exprs.
31614 (pass_pre::execute): Move fini_pre after fini_eliminate.
31615 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
31616 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
31617 PRE to get rid of dead code that has invalid SSA form and
31618 split critical edges again.
31620 2017-05-11 Bin Cheng <bin.cheng@arm.com>
31622 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31624 2017-05-11 Richard Biener <rguenther@suse.de>
31626 * passes.c (execute_function_todo): Verify loops if they are
31627 said to be up-to-date.
31628 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
31629 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
31631 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
31634 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
31635 handle calling assemble_external ourself.
31638 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
31639 modes with zero size. Enhance comment.
31641 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31643 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
31644 built-ins for vec_xl and vec_xst with short and char pointer
31647 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31649 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
31650 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
31651 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
31652 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
31653 (_mm_maskz_min_round_ss): New intrinsics.
31654 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
31655 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
31656 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
31657 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
31658 (__builtin_ia32_minss_mask_round): New builtins.
31659 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31660 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31661 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
31663 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
31664 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
31666 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
31669 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
31671 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
31672 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
31673 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
31674 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
31675 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
31676 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
31677 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
31678 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31679 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31680 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
31681 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
31682 (__builtin_ia32_mulss_mask_round): New builtins.
31683 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31684 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31685 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
31687 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
31688 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31690 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31693 2017-05-10 Julia Koval <julia.koval@intel.com>
31695 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
31696 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
31697 (_mm256_setr_m128i): New intrinsics.
31699 2017-05-10 Julia Koval <julia.koval@intel.com>
31701 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
31702 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
31703 (_mm_maskz_rcp14_ss): New intrinsics.
31704 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
31705 (__builtin_ia32_rcp14ss_mask): New builtins.
31706 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
31708 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
31710 PR tree-optimization/51513
31711 * tree-cfg.c (gimple_seq_unreachable_p): New function.
31712 (assert_unreachable_fallthru_edge_p): Use it.
31713 (group_case_labels_stmt): Likewise.
31714 * tree-cfg.h: Prototype it.
31715 * stmt.c: Include cfghooks.h and tree-cfg.h.
31716 (emit_case_dispatch_table) <gap_label>: New local variable.
31717 Use it to fill dispatch table gaps.
31718 Test for default_label before updating probabilities.
31719 (expand_case) <default_label>: Remove unneeded initialization.
31720 Test for unreachable default case statement and remove its edge.
31721 Set default_label accordingly.
31722 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
31724 2017-05-10 Carl Love <cel@us.ibm.com>
31726 * config/rs6000/rs6000-c: Add support for built-in functions
31727 vector signed char vec_neg (vector signed char)
31728 vector signed short int vec_neg (vector short int)
31729 vector signed int vec_neg (vector signed int)
31730 vector signed long long vec_neg (vector signed long long)
31731 vector float vec_neg (vector float)
31732 vector double vec_neg (vector double)
31733 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
31735 * config/rs6000/altivec.h: Add define for vec_neg
31736 * doc/extend.texi: Update the built-in documentation for the
31737 new built-in functions.
31739 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31741 PR tree-optimization/77644
31742 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
31744 2017-05-10 Nathan Sidwell <nathan@acm.org>
31746 * dumpfile.h (TDI_lang_all): New.
31747 (TDF_KIND): New. Renumber others
31748 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
31750 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
31752 (get_dump_file_name): Adjust suffix generation.
31753 (dump_enable_all): Use TDF_KIND.
31754 * doc/invoke.texi (-fdump-lang-all): Document.
31756 * dumpfile.h: Tabify.
31758 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
31761 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
31762 Move member access before delete.
31764 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
31766 * tree-inline.c (expand_call_inline): Split block at stmt
31769 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
31772 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
31773 are now unused after splitting mov{sf,sd}_hardfloat.
31774 (f32_lr2): Likewise.
31775 (f32_lm): Likewise.
31776 (f32_lm2): Likewise.
31777 (f32_li): Likewise.
31778 (f32_li2): Likewise.
31779 (f32_lv): Likewise.
31780 (f32_sr): Likewise.
31781 (f32_sr2): Likewise.
31782 (f32_sm): Likewise.
31783 (f32_sm2): Likewise.
31784 (f32_si): Likewise.
31785 (f32_si2): Likewise.
31786 (f32_sv): Likewise.
31787 (f32_dm): Likewise.
31788 (f32_vsx): Likewise.
31789 (f32_av): Likewise.
31790 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
31791 For movsf, order stores so the VSX stores occur before the GPR
31792 store which encourages the register allocator to use a traditional
31793 FPR instead of a GPR. For movsd, order the stores so that the GPR
31794 store comes before the VSX stores to allow the power6 to work.
31795 This is due to the power6 not having a 32-bit integer store
31796 instruction from a FPR.
31797 (movsf_hardfloat): Likewise.
31798 (movsd_hardfloat): Likewise.
31800 2017-05-09 Martin Sebor <msebor@redhat.com>
31802 PR translation/80280
31803 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
31806 PR translation/80280
31807 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
31808 data member added in r247778.
31809 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
31811 2017-05-09 Nathan Sidwell <nathan@acm.org>
31813 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
31815 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
31818 2017-05-09 Marek Polacek <polacek@redhat.com>
31820 * doc/invoke.texi: Fix typo.
31822 2017-05-09 Richard Biener <rguenther@suse.de>
31824 * tree-vrp.c (vrp_val_is_max): Adjust comment.
31825 (vrp_val_is_min): Likewise.
31826 (set_value_range_to_value): Likewise.
31827 (set_value_range_to_nonnegative): Likewise.
31828 (gimple_assign_nonzero_p): Likewise.
31829 (gimple_stmt_nonzero_p): Likewise.
31830 (vrp_int_const_binop): Likewise. Remove unreachable case.
31831 (adjust_range_with_scev): Adjust comments.
31832 (compare_range_with_value): Likewise.
31833 (extract_range_from_phi_node): Likewise.
31834 (test_for_singularity): Likewise.
31836 2017-05-09 Richard Biener <rguenther@suse.de>
31838 * tree-vrp.c (get_single_symbol): Add assert that we don't
31839 get overflowed constants as invariant part.
31840 (compare_values_warnv): Add comment before the TREE_NO_WARNING
31841 checks. Use wi::cmp instead of recursing for integer constants.
31842 (compare_values): Just ignore whether we assumed undefined
31843 overflow instead of failing the compare.
31844 (extract_range_for_var_from_comparison_expr): Add comment before the
31845 TREE_NO_WARNING sets.
31846 (test_for_singularity): Likewise.
31847 (extract_range_from_comparison): Do not disable optimization
31848 when we assumed undefined overflow.
31849 (extract_range_basic): Remove init of unused var.
31851 2017-05-09 Richard Biener <rguenther@suse.de>
31853 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
31854 (extract_range_from_multiplicative_op_1): Adjust.
31855 (extract_range_from_binary_expr_1): Use int_const_binop.
31857 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
31860 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
31861 rs6000_store_data_bypass_p in seven define_bypass directives and
31862 in several comments.
31863 * config/rs6000/rs6000-protos.h: Add prototype for
31864 rs6000_store_data_bypass_p function.
31865 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
31866 function implements slightly different (rs6000-specific) semantics
31867 than store_data_bypass_p, returning false rather than aborting
31868 with assertion error when arguments do not satisfy the
31869 requirements of store data bypass.
31870 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
31871 rs6000_store_data_bypass_p.
31873 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
31875 * config/xtensa/xtensa-protos.h
31876 (xtensa_initial_elimination_offset): New declaration.
31877 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
31878 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
31879 macro definition, add case for FRAME_POINTER_REGNUM when
31880 FRAME_GROWS_DOWNWARD.
31881 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
31882 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
31883 xtensa_initial_elimination_offset.
31885 2017-05-08 Nathan Sidwell <nathan@acm.org>
31887 * doc/invoke.texi: Alphabetize -fdump options.
31889 2017-05-08 Martin Sebor <msebor@redhat.com>
31891 PR translation/80280
31892 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
31894 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
31896 * target.def (compute_frame_layout): New optional target hook.
31897 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
31898 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
31899 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
31901 * reload1.c (verify_initial_elim_offsets): Likewise.
31902 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
31903 (use_simple_return_p): Call arm_compute_frame_layout if needed.
31904 (arm_get_frame_offsets): Split up into this ...
31905 (arm_compute_frame_layout): ... and this function.
31907 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
31909 * config/aarch64/constraints.md (Usa): New constraint.
31910 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
31912 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
31914 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
31915 with_multilib_list after it has been checked.
31917 2017-05-08 Richard Biener <rguenther@suse.de>
31919 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
31920 (bitmap_set_subtract_values): Likewise.
31922 2017-05-08 Richard Biener <rguenther@suse.de>
31924 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
31925 (gimple_assign_nonzero): ... this and remove strict_overflow_p
31927 (gimple_stmt_nonzero_warnv_p): Rename to ...
31928 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
31930 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
31931 (extract_range_basic): Adjust, do not disable propagation on
31932 strict overflow sensitive simplification.
31933 (vrp_visit_cond_stmt): Likewise.
31935 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
31937 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
31938 body size unconditionally.
31940 2017-05-07 Jeff Law <law@redhat.com>
31943 2017-05-06 Jeff Law <law@redhat.com>
31944 PR tree-optimization/78496
31945 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31948 PR tree-optimization/78496
31949 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31950 (simplify_stmt_using_ranges): Call it.
31951 (vrp_dom_walker::before_dom_children): Extract equivalences
31952 from an ASSERT_EXPR with an equality comparison against a
31955 2017-05-06 Jeff Law <law@redhat.com>
31957 PR tree-optimization/78496
31958 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
31961 PR tree-optimization/78496
31962 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
31963 (simplify_stmt_using_ranges): Call it.
31964 (vrp_dom_walker::before_dom_children): Extract equivalences
31965 from an ASSERT_EXPR with an equality comparison against a
31968 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31970 * lra-constraints.c (lra_copy_reg_equiv): New function.
31971 (split_reg): Use it to copy equivalence information from the
31972 original register to the spill register.
31974 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
31976 PR rtl-optimization/75964
31977 * simplify-rtx.c (simplify_const_relational_operation): Remove
31978 invalid handling of comparisons of integer ABS.
31980 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
31982 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
31983 initialize to zero.
31984 (init_regs): Remove declaration.
31985 (function_arg_advance_32): Initialize error_p as boolean variable.
31987 2017-05-05 Nathan Sidwell <nathan@acm.org>
31989 * store-motion.c (remove_reachable_equiv_notes): Reformat long
31990 lines. Use for (;;).
31992 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31994 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
31995 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
31996 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
31997 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
31998 VF=2 that require versioning.
32000 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32002 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32005 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32007 * diagnostic.h (diagnostic_override_option_index): Convert from
32008 macro to inline function.
32010 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32012 * diagnostic.c (last_module_changed_p): New function.
32013 (set_last_module): New function.
32014 (diagnostic_report_current_module): Convert macro usage to
32015 the above functions.
32016 * diagnostic.h (diagnostic_context::last_module): Strengthen
32017 from const line_map * to const line_map_ordinary *.
32018 (diagnostic_last_module_changed): Delete macro.
32019 (diagnostic_set_last_module): Delete macro.
32021 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32023 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32024 with diagnostic_report_diagnostic.
32025 (diagnostic_n_impl_richloc): Likewise.
32026 * diagnostic.h (report_diagnostic): Delete macro.
32027 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32028 with diagnostic_report_diagnostic.
32029 * substring-locations.c (format_warning_va): Likewise.
32031 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32033 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32034 save/restor of format_spec. Move option-printing code to...
32035 (print_option_information): ...this new function, and
32036 reimplement by simply printing to the pretty_printer,
32037 rather than appending to the format string.
32039 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32041 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32042 handling logic into...
32043 (update_effective_level_from_pragmas): ...this new function.
32045 2017-05-04 Andrew Waterman <andrew@sifive.com>
32047 * config/riscv/riscv.opt (mstrict-align): New option.
32048 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32049 (SLOW_UNALIGNED_ACCESS): Define.
32050 (riscv_slow_unaligned_access): Declare.
32051 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32053 (riscv_slow_unaligned_access): New variable.
32054 (rocket_tune_info): Set slow_unaligned_access to true.
32055 (optimize_size_tune_info): Set slow_unaligned_access to false.
32056 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32057 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32058 (riscv_option_override): Set riscv_slow_unaligned_access.
32059 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32061 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32063 * config/riscv/riscv.md: Unify indentation.
32065 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32070 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32072 (extendsi<mode>2): Add support for doing sign extension via
32073 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32074 don't have ISA 3.0 instructions.
32075 (extendsi<mode>2 splitter): Likewise.
32076 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32077 generate the normal insns since SImode can now go in vector
32078 registers. Disallow the special UNSPECs needed for previous
32079 machines to hide SImode being used. Add new insns
32080 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32081 (fix_trunc<mode>si2_stfiwx): Likewise.
32082 (fix_trunc<mode>si2_internal): Likewise.
32083 (fixuns_trunc<mode>si2): Likewise.
32084 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32085 (fctiw<u>z_<mode>_smallint): Likewise.
32086 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32087 of floating point to 32-bit integer from doing a direct move to
32088 the GPR registers to do a store.
32089 (fctiwz_<mode>): Break long line.
32091 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32093 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32094 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32095 (addr_list, addr_offset_valid_p): New.
32096 (split_address_groups): Check offset validity with above function.
32097 (gt-tree-ssa-loop-ivopts.h): Include header file.
32099 2017-05-05 Nathan Sidwell <nathan@acm.org>
32101 * config.gcc (arm*-*-*): Add missing 'fi'.
32103 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32105 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32106 included in -fopt-info does not matter.
32107 * doc/optinfo.texi (-fopt-info): Fix description of default
32108 behavour. Explicitly say order of options included in -fopt-info
32111 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32113 * config.gcc: Allow combinations of aprofile and rmprofile values for
32114 --with-multilib-list.
32115 * config/arm/t-multilib: New file.
32116 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32117 variables. Remove setting of ISA and floating-point ABI in
32118 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32119 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32120 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32121 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32123 * config/arm/t-rmprofile: Likewise except for the matches changes.
32124 * doc/install.texi (--with-multilib-list): Document the combination of
32125 aprofile and rmprofile values and warn about pitfalls in doing that.
32127 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32129 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32130 (movdi_aarch64): Likewise.
32132 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32134 PR tree-optimization/80632
32135 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32137 (build_arrays): Initialize it for virtual phis.
32138 (fix_phi_nodes): Use it for virtual phis.
32140 PR tree-optimization/80558
32141 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32142 [x, y] op z into [x op, y op z] for op & or | if conditions
32145 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32146 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32149 * config/arm/arm.md (use_literal_pool): Remove.
32150 (64-bit immediate split): No longer takes cost into consideration
32151 if arm_disable_literal_pool is enabled.
32152 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32153 used when arm_disable_literal_pool is enabled.
32154 (arm_max_const_double_inline_cost): Remove use of
32155 arm_disable_literal_pool.
32156 (push_minipool_fix): Add assert.
32157 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32158 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32159 (no_literal_pool_sf_immediate): New.
32161 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32163 PR tree-optimization/80613
32164 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32165 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32167 2017-05-05 Richard Biener <rguenther@suse.de>
32169 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32171 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32173 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32174 of this flag from insn conditions due to removal from r247495.
32176 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32178 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32180 (arm_early_store_addr_dep_ptr): Likewise.
32181 * config/arm/aarch-common-protos.h
32182 (arm_early_load_addr_dep_ptr): Add prototype.
32183 (arm_early_store_addr_dep_ptr): Likewise.
32184 * config/arm/cortex-a53.md: Add new bypasses.
32186 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32188 * tree.c (next_type_uid): Change type to unsigned.
32189 (type_hash_canon): Decrement back next_type_uid if
32190 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32191 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32194 2017-05-04 Martin Sebor <msebor@redhat.com>
32196 * builtins.c: Fix a trivial typo in a comment.
32198 PR middle-end/79234
32199 * builtins.c (check_sizes): Adjust to handle reading past the end.
32200 Avoid printing excessive upper bound of ranges. Use %E to print
32201 tree nodes instead of converting them to %wu.
32202 (expand_builtin_memchr): New function.
32203 (compute_dest_size): Rename...
32204 (compute_objsize): ...to this.
32205 (expand_builtin_memcpy): Adjust.
32206 (expand_builtin_mempcpy): Adjust.
32207 (expand_builtin_strcat): Adjust.
32208 (expand_builtin_strcpy): Adjust.
32209 (check_strncat_sizes): Adjust.
32210 (expand_builtin_strncat): Adjust.
32211 (expand_builtin_strncpy): Adjust and simplify.
32212 (expand_builtin_memset): Adjust.
32213 (expand_builtin_bzero): Adjust.
32214 (expand_builtin_memcmp): Adjust.
32215 (expand_builtin): Handle memcmp.
32216 (maybe_emit_chk_warning): Check strncat just once.
32218 2017-05-04 Martin Sebor <msebor@redhat.com>
32220 PR preprocessor/79214
32221 PR middle-end/79222
32222 PR middle-end/79223
32223 * builtins.c (check_sizes): Add inlining context and issue
32224 warnings even when -Wno-system-headers is set.
32225 (check_strncat_sizes): Same.
32226 (expand_builtin_strncat): Same.
32227 (expand_builtin_memmove): New function.
32228 (expand_builtin_stpncpy): Same.
32229 (expand_builtin): Handle memmove and stpncpy.
32231 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32233 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32234 which is not used any more.
32236 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32238 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32240 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32242 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32243 (cortexa53_tunings): Likewise.
32244 (cortexa57_tunings): Likewise.
32245 (cortexa72_tunings): Likewise.
32246 (cortexa73_tunings): Likewise.
32248 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32250 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32251 Set loop alignment to 8.
32253 2017-05-04 Martin Sebor <msebor@redhat.com>
32255 PR translation/80280
32256 * builtins.c (expand_builtin_object_size): Add missing quoting to
32257 %D and like directives.
32258 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32259 (hsa_type_for_tree_type): Same.
32260 (verify_function_arguments): Same.
32261 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32262 * varasm.c (get_section): Same.
32265 2017-05-04 Martin Sebor <msebor@redhat.com>
32267 PR translation/80280
32268 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32270 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32272 * config/aarch64/aarch64.c (generic_addrcost_table):
32273 Change HI/TI mode setting.
32275 2017-05-04 Martin Jambor <mjambor@suse.cz>
32277 PR tree-optimization/80622
32278 * tree-sra.c (comes_initialized_p): New function.
32279 (build_accesses_from_assign): Only set write lazily when
32280 comes_initialized_p is false.
32281 (analyze_access_subtree): Use comes_initialized_p.
32282 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32283 instead of testing for PARM_DECL.
32285 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32287 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32288 constraint on operand 0 to allow more general addressing modes.
32289 Adjust output template.
32290 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32292 * config/aarch64/aarch64-protos.h
32293 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32294 * config/aarch64/constraints.md (Dp): New address constraint.
32295 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32298 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32300 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32301 update use of estimate_ipcp_clone_size_and_time.
32302 (estimate_local_effects): Update use of
32303 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32304 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32305 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32306 Return nonspecialized time.
32308 2017-05-04 Richard Biener <rguenther@suse.de>
32310 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32311 for the last VUSE which def dominates the PHI. Directly call
32313 (get_continuation_for_phi_1): Remove.
32315 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32317 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32318 to explain the use of truncating division. Cap the number of
32319 iterations to the maximum given by nb_iterations_upper_bound,
32322 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32324 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32325 * configure: Regenerate.
32326 * config.in: Regenerate.
32327 * config/i386/driver-mingw32.c: new file.
32328 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32329 * config.host: Link driver-mingw32.o on MinGW host.
32330 * doc/install.texi: Document new --enable-mingw-wildcard configure
32333 2017-05-04 Marek Polacek <polacek@redhat.com>
32335 PR tree-optimization/80612
32336 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32338 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32339 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32341 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32342 (movt splitter): Likewise.
32343 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32344 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32345 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32346 block for Thumb-1 with MOVT.
32347 (thumb2_legitimate_address_p): Move code block ...
32348 (can_avoid_literal_pool_for_label_p): ... into this new function.
32349 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32351 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32352 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32353 "M-profile targets with the MOVT instruction".
32355 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32357 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32358 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32359 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32361 2017-05-04 Martin Liska <mliska@suse.cz>
32363 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32364 variable cond_code.
32366 2017-05-04 Richard Biener <rguenther@suse.de>
32368 * tree.c (array_at_struct_end_p): Handle arrays at struct
32369 end with flexarrays more conservatively. Refactor and treat
32370 arrays of arrays or aggregates more strict. Fix
32371 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32372 * tree.c (array_at_struct_end_p): Adjust prototype.
32373 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32374 * gimple-fold.c (get_range_strlen): Likewise.
32375 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32377 2017-05-04 Richard Biener <rguenther@suse.de>
32379 PR tree-optimization/31130
32380 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32382 (supports_overflow_infinity): Likewise.
32383 (is_negative_overflow_infinity): Likewise.
32384 (is_positive_overflow_infinity): Likewise.
32385 (is_overflow_infinity): Likewise.
32386 (stmt_overflow_infinity): Likewise.
32387 (overflow_infinity_range_p): Likewise.
32388 (usable_range_p): Remove as always returning true.
32389 (make_overflow_infinity): Remove.
32390 (negative_overflow_infinity): Likewise.
32391 (positive_overflow_infinity): Likewise.
32392 (avoid_overflow_infinity): Likewise.
32393 (set_value_range): Adjust accordingly.
32394 (set_value_range_to_nonnegative): Likewise, remove now unused
32395 overflow_infinity arg.
32396 (vrp_operand_equal_p): Adjust.
32397 (update_value_range): Likewise.
32398 (range_int_cst_singleton_p): Likewise.
32399 (operand_less_p): Likewise.
32400 (compare_values_warnv): Likewise.
32401 (extract_range_for_var_from_comparison_expr): Likewise.
32402 (vrp_int_const_binop): Likewise.
32403 (zero_nonzero_bits_from_vr): Likewise.
32404 (extract_range_from_multiplicative_op_1): Likewise.
32405 (extract_range_from_binary_expr_1): Likewise.
32406 (extract_range_from_unary_expr): Likewise.
32407 (extract_range_from_comparison): Likewise.
32408 (extract_range_basic): Likewise.
32409 (adjust_range_with_scev): Likewise.
32410 (compare_ranges): Likewise.
32411 (compare_range_with_value): Likewise.
32412 (dump_value_range): Likewise.
32413 (test_for_singularity): Likewise, remove strict_overflow_p parameter
32415 (simplify_cond_using_ranges): Adjust.
32417 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32419 * brig-builtins.def: Added a builtin for class_f64.
32420 * builtin-types.def: Added a builtin type needed by class_f64.
32422 2017-05-03 Jason Merrill <jason@redhat.com>
32424 * timevar.def: Add TV_CONSTEXPR.
32426 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32428 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
32430 2017-05-03 Martin Jambor <mjambor@suse.cz>
32432 * ipa-prop.c (ipa_update_after_lto_read): Removed.
32433 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
32434 * ipa-cp.c (ipcp_propagate_stage): Do not call
32435 ipa_update_after_lto_read.
32436 * ipa-inline.c (ipa_inline): Likewise.
32438 2017-05-03 Martin Jambor <mjambor@suse.cz>
32440 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
32441 tag. Added a default constructor and a destructor.
32442 (ipa_edge_args_sum_t): New class;
32443 (ipa_edge_args_sum): Declare.
32444 (ipa_edge_args_vector): Remove declaration.
32445 (IPA_EDGE_REF): Use ipa_edge_args_sum.
32446 (ipa_free_edge_args_substructures): Remove declaration.
32447 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
32448 (ipa_edge_args_info_available_for_edge_p): Likewise.
32449 * ipa-prop.c (ipa_edge_args_vector): Removed.
32450 (edge_removal_hook_holder): Likewise.
32451 (edge_duplication_hook_holder): Likewise.
32452 (ipa_edge_args_sum): New variable.
32453 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
32454 ipa_edge_args_vector.
32455 (ipa_free_edge_args_substructures): Likewise.
32456 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
32457 ipa_edge_args_vector.
32458 (ipa_edge_removal_hook): Turned into method
32459 ipa_edge_args_sum_t::remove.
32460 (ipa_edge_duplication_hook): Turned into method
32461 ipa_edge_args_sum_t::duplicate.
32462 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
32463 registering edge hooks.
32464 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
32465 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
32466 ipa_edge_args_sum instead of ipa_edge_args_vector.
32467 * ipa-profile.c (ipa_profile): Likewise.
32469 2017-05-03 Martin Jambor <mjambor@suse.cz>
32471 * symbol-summary.h (function_summary): New method exists.
32472 (function_summary::symtab_removal): Deallocate through release.
32473 (call_summary): New class.
32474 (gt_ggc_mx): New overload.
32475 (gt_pch_nx): Likewise.
32476 (gt_pch_nx): Likewise.
32478 2017-05-03 Jeff Law <law@redhat.com>
32480 PR tree-optimization/78496
32481 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
32482 from simplify_cond_using_ranges. Split off code to walk
32483 backwards through casts into ...
32484 (simplify_cond_using_ranges_2): New function.
32485 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
32486 (execute_vrp): After identifying jump threads, call
32487 simplify_cond_using_ranges_2.
32489 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
32492 * ipa-inline.h (inline_summary): Add ctor.
32493 (create_ggc): Do not use ggc_cleared_alloc.
32495 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
32496 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32498 * gcc.c (handle_braces): Support escaping in switch matching
32500 * doc/invoke.texi (Spec Files): Document it.
32501 Remove superfluous @code markup in items.
32503 2017-05-03 David Malcolm <dmalcolm@redhat.com>
32505 * diagnostic-show-locus.c (struct column_range): New struct.
32506 (get_affected_columns): New function.
32507 (get_printed_columns): New function.
32508 (struct correction): New struct.
32509 (correction::ensure_capacity): New function.
32510 (correction::ensure_terminated): New function.
32511 (struct line_corrections): New struct.
32512 (line_corrections::~line_corrections): New dtor.
32513 (line_corrections::add_hint): New function.
32514 (layout::print_trailing_fixits): Reimplement in terms of the new
32516 (selftest::test_overlapped_fixit_printing): New function.
32517 (selftest::diagnostic_show_locus_c_tests): Call it.
32519 2017-05-03 Nathan Sidwell <nathan@acm.org>
32521 Canonicalize canonical type hashing
32522 * tree.h (type_hash_canon_hash): Declare.
32523 * tree.c (type_hash_list, attribute_hash_list): Move into
32524 type_hash_canon_hash.
32525 (build_type_attribute_qual_variant): Break out hash code calc into
32526 type_hash_canon_hash.
32527 (type_hash_canon_hash): New. Generic type hash computation.
32528 (build_range_type_1, build_array_type_1, build_function_type,
32529 build_method_type_directly, build_offset_type, build_complex_type,
32530 make_vector_type): Call it.
32532 2017-05-03 Richard Biener <rguenther@suse.de>
32534 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
32535 When all DRs have unknown misaligned do not always peel
32536 when there is a store but apply the same costing model as if
32537 there were only loads.
32539 2017-05-03 Richard Biener <rguenther@suse.de>
32542 PR tree-optimization/80492
32543 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32544 compare_base_decls returning dont-know properly.
32546 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
32548 * config/arm/iterators.md (CCSI): New mode iterator.
32549 (arch): New mode attribute.
32550 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
32551 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
32552 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
32553 code iterator for success result mode.
32554 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
32555 the corresponding new insn generators.
32557 2017-05-03 Bin Cheng <bin.cheng@arm.com>
32560 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32561 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32563 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
32565 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
32566 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
32567 (DDR_A): Wrap DDR argument in brackets.
32568 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
32569 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
32570 (DDR_REVERSED_P): Likewise.
32572 2017-05-03 Jakub Jelinek <jakub@redhat.com>
32574 PR tree-optimization/79472
32575 * tree-switch-conversion.c (struct switch_conv_info): Add
32576 contiguous_range and default_case_nonstandard fields.
32577 (collect_switch_conv_info): Compute contiguous_range and
32578 default_case_nonstandard fields, don't clear final_bb if
32579 contiguous_range and only the default case doesn't have the required
32581 (check_all_empty_except_final): Set default_case_nonstandard instead
32582 of failing if contiguous_range and the default case doesn't have empty
32584 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
32585 and only the default case doesn't have the required constants. Skip
32587 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
32588 if default_case_nonstandard.
32589 (build_constructors): Build constant 1 just once. Assert that default
32590 values aren't inserted in between cases if contiguous_range. Skip
32592 (build_arrays): Skip virtual phis.
32593 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
32594 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
32595 Handle virtual phis.
32596 (gen_inbound_check): Handle default_case_nonstandard case.
32597 (process_switch): Adjust check_final_bb caller. Call
32598 gather_default_values with the first non-default case instead of
32599 default case if default_case_nonstandard.
32601 2017-05-02 Nathan Sidwell <nathan@acm.org>
32603 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
32604 check. Fix formatting.
32606 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
32608 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
32609 errors when comparing specialized and unspecialized times.
32611 2017-05-02 David Malcolm <dmalcolm@redhat.com>
32613 * diagnostic-show-locus.c
32614 (layout::should_print_annotation_line_p): Make private.
32615 (layout::print_annotation_line): Make private.
32616 (layout::annotation_line_showed_range_p): Make private.
32617 (layout::show_ruler): Make private.
32618 (layout::print_source_line): Make private. Pass in line and
32619 line_width, rather than calling location_get_source_line. Drop
32621 (layout::print_leading_fixits): New method.
32622 (layout::print_any_fixits): Rename to...
32623 (layout::print_trailing_fixits): ...this, and make private.
32624 Don't print newline fixits.
32625 (diagnostic_show_locus): Move logic for printing one row into...
32626 (layout::print_line): ...this new function. Move the
32627 location_get_source_line call and error-handling from
32628 print_source_line to here. Call print_leading_fixits, and rename
32629 print_any_fixits to print_trailing_fixits.
32630 (selftest::test_fixit_insert_containing_newline): Update now that
32631 newlines are partially supported.
32632 (selftest::test_fixit_insert_containing_newline_2): New test.
32633 (selftest::test_fixit_replace_containing_newline): Update comments.
32634 (selftest::diagnostic_show_locus_c_tests): Call the new test.
32635 * edit-context.c (class added_line): New class.
32636 (class edited_line): Describe newline handling in comment.
32637 (edited_line::actually_edited_p): New method.
32638 (edited_line::print_content): Delete redundant decl.
32639 (edited_line::m_predecessors): New field.
32640 (edited_file::print_content): Call edited_line::print_content.
32641 (edited_file::print_diff): Update to support newlines.
32642 (edited_file::print_diff_hunk): Likewise.
32643 (edited_file::print_run_of_changed_lines): New function.
32644 (edited_file::print_diff_line): Convert to...
32645 (print_diff_line): ...this.
32646 (edited_file::get_effective_line_count): New function.
32647 (edited_line::edited_line): Initialize new field m_predecessors.
32648 (edited_line::~edited_line): Clean up m_predecessors.
32649 (edited_line::apply_fixit): Handle newlines.
32650 (edited_line::get_effective_line_count): New function.
32651 (edited_line::print_content): New function.
32652 (edited_line::print_diff_lines): New function.
32653 (selftest::test_applying_fixits_insert_containing_newline): New
32655 (selftest::test_applying_fixits_replace_containing_newline): New
32657 (selftest::insert_line): New function.
32658 (selftest::test_applying_fixits_multiple_lines): Add example of
32660 (selftest::edit_context_c_tests): Call the new tests.
32662 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32664 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
32665 parameter cand. Update dump information.
32666 (get_computation_cost): Update uses.
32668 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32670 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
32671 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
32672 (get_computation_at, rewrite_use_address): Update use of
32673 get_computation_aff.
32675 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32677 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
32678 (get_computation): Delete.
32679 (get_computation_cost): Implement like get_computation_cost_at.
32680 Use get_computation_at.
32681 (get_computation_cost_at): Delete.
32682 (rewrite_use_nonlinear_expr): Use get_computation_at.
32683 (rewrite_use_compare, remove_unused_ivs): Ditto.
32685 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32687 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
32689 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32691 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
32692 (ivopts_global_cost_for_size): Rename parameter and update uses.
32693 (iv_ca_recount_cost): Update uses.
32694 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
32695 candidates seperately in n_invs and n_cands.
32696 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
32698 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32700 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
32701 (find_inv_vars_cb): New.
32702 (find_depends): Renamed to ...
32703 (find_inv_vars): ... this.
32704 (add_candidate_1, force_var_cost): Call find_inv_vars.
32705 (split_address_cost, determine_group_iv_cost_cond): Ditto.
32707 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32709 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
32710 inv_vars. Add inv_exprs.
32711 (struct iv_cand): Rename depends_on to inv_vars.
32712 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
32713 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
32714 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
32715 (dump_cand): Dump inv_vars.
32716 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
32717 (record_invariant, find_depends, add_candidate_1): Ditto.
32718 (set_group_iv_cost, force_var_cost): Ditto.
32719 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
32720 (get_computation_cost_at, get_computation_cost): Ditto.
32721 (determine_group_iv_cost_generic): Ditto.
32722 (determine_group_iv_cost_address): Ditto.
32723 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
32724 (determine_group_iv_costs): Ditto.
32725 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
32726 (iv_ca_set_remove_invariants): Renamed to ...
32727 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
32728 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
32729 (iv_ca_set_add_invariants): Renamed to ...
32730 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
32731 (iv_ca_set_cp): Use iv_ca_set_add_invs.
32732 (iv_ca_has_deps): Support inv_vars and inv_exprs.
32733 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
32734 (create_new_ivs): Remove useless dump.
32736 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32738 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
32740 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
32741 (iv_ca_set_no_cp, create_new_iv): Ditto.
32743 2017-05-02 Bin Cheng <bin.cheng@arm.com>
32745 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32747 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32749 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
32750 function tree_check2.
32752 2017-05-02 Martin Liska <mliska@suse.cz>
32754 * doc/gcov.texi: Add missing preposition.
32755 * gcov.c (function_info::function_info): Properly fill up
32756 all member variables.
32758 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32760 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
32762 2017-05-02 Tamar Christina <tamar.christina@arm.com>
32764 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
32766 2017-05-02 Martin Liska <mliska@suse.cz>
32769 * lto-streamer-in.c (lto_read_tree_1): Remove
32770 LTO_STREAMER_DEBUG.
32771 * lto-streamer.c (struct tree_hash_entry): Likewise.
32772 (struct tree_entry_hasher): Likewise.
32773 (tree_entry_hasher::hash): Likewise.
32774 (tree_entry_hasher::equal): Likewise.
32775 (lto_streamer_init): Likewise.
32776 (lto_orig_address_map): Likewise.
32777 (lto_orig_address_get): Likewise.
32778 (lto_orig_address_remove): Likewise.
32779 * lto-streamer.h: Likewise.
32780 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
32781 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
32783 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
32785 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
32786 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
32787 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
32788 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
32789 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
32790 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
32791 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
32792 (mm_maskz_sub_ss): New intrinsics.
32793 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32794 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32795 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
32796 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
32797 (__builtin_ia32_subss_mask_round): New builtins.
32798 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32799 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32800 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
32802 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
32803 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32805 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32808 2017-05-02 Martin Jambor <mjambor@suse.cz>
32810 PR tree-optimization/78687
32811 * tree-sra.c (access): New field parent.
32812 (process_subtree_disqualification): New function.
32813 (disqualify_candidate): Call it.
32814 (build_accesses_from_assign): Reset write flag if creating an
32816 (build_access_subtree): Fill in parent field and also prpagate
32817 down grp_write flag.
32818 (create_artificial_child_access): New parameter set_grp_write, set
32819 grp_write to its value.
32820 (propagate_subaccesses_across_link): Also propagate grp_write flag
32822 (propagate_all_subaccesses): Push the closest parent back to work
32823 queue if add_access_to_work_queue returned true.
32825 2017-05-02 Richard Biener <rguenther@suse.de>
32827 * common.opt (fstrict-overflow): Alias negative to fwrapv.
32828 * doc/invoke.texi (fstrict-overflow): Remove all traces of
32829 -fstrict-overflow documentation.
32830 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
32831 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
32832 flag_strict_overflow.
32833 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
32834 * lto-opts.c (lto_write_options): Do not stream it.
32835 * lto-wrapper.c (merge_and_complain): Do not handle it.
32836 * opts.c (default_options_table): Do not set -fstrict-overflow.
32837 (finish_options): Likewise do not clear it when sanitizing.
32838 * simplify-rtx.c (simplify_const_relational_operation): Do not
32839 test flag_strict_overflow.
32841 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32843 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
32844 using enabled attribute.
32845 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
32846 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
32847 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
32848 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
32849 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
32850 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
32851 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
32852 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
32853 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
32854 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
32856 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
32858 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
32860 2017-05-02 Richard Biener <rguenther@suse.de>
32862 PR tree-optimization/80591
32864 2017-04-10 Richard Biener <rguenther@suse.de>
32866 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32869 2017-05-02 Richard Biener <rguenther@suse.de>
32871 PR tree-optimization/80549
32872 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
32873 (cleanup_tree_cfg_noloop): Create forwarders to known loop
32874 headers if they do not have a preheader.
32876 2017-05-02 Martin Liska <mliska@suse.cz>
32879 * common.opt: Fix typo.
32880 * doc/invoke.texi: Likewise.
32882 2017-05-01 Jan Beulich <jbeulich@suse.com>
32884 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
32885 swapping, add (x,x,m,x,n) alternative.
32887 2017-05-01 Nathan Sidwell <nathan@acm.org>
32889 * calls.c (combine_pending_stack_adjustment_and_call): Remove
32890 unnecessary unadjusted_alignment check.
32892 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
32895 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
32897 * gimplify.c (gimplify_cilk_detach): New function.
32898 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
32899 * tree-core.h: Document EXPR_CILK_SPAWN.
32900 * tree.h (EXPR_CILK_SPAWN): Define.
32902 2017-05-01 David Malcolm <dmalcolm@redhat.com>
32904 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
32905 to use new fixit_hint representation, using the "replace" logic.
32906 (get_line_span_for_fixit_hint): Likewise.
32907 (layout::print_any_fixits): Likewise.
32908 (selftest::test_one_liner_many_fixits): Rename to...
32909 (selftest::test_one_liner_many_fixits_1): ...this, and update
32910 comment and expected output to reflect that the multiple fix-it
32911 hints are now consolidated into one insertion.
32912 (selftest::test_one_liner_many_fixits_2): New test.
32913 (selftest::test_diagnostic_show_locus_one_liner): Update for
32915 (selftest::test_fixit_consolidation): Update for fix-it API
32917 * diagnostic.c (print_parseable_fixits): Likewise.
32918 * edit-context.c (edited_line::m_line_events): Convert from
32919 auto_vec <line_event *> to auto_vec <line_event>.
32920 (class line_event): Convert from abstract base class to a concrete
32921 class, taking over the role of replace_event.
32922 (class insert_event): Delete.
32923 (class replace_event): Rename to class line_event. Convert to
32925 (edit_context::add_fixits): Reimplement.
32926 (edit_context::apply_insert): Delete.
32927 (edit_context::apply_replace): Rename to...
32928 (edit_context::apply_fixit): ...this. Convert to half-open range.
32929 (edited_file::apply_insert): Delete.
32930 (edited_file::apply_replace): Rename to...
32931 (edited_file::apply_fixit): ...this.
32932 (edited_line::~edited_line): Drop deletion of events.
32933 (edited_line::apply_insert): Delete.
32934 (edited_line::apply_replace): Rename to...
32935 (edited_line::apply_fixit): ...this. Convert to half-open range.
32936 Update for change to type of m_line_events.
32937 * edit-context.h (edit_context::apply_insert): Delete.
32938 (edit_context::apply_replace): Rename to...
32939 (edit_context::apply_fixit): ...this.
32941 2017-05-01 Martin Sebor <msebor@redhat.com>
32943 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
32946 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
32949 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
32950 __get_cpuid_max returns 0.
32951 (__get_cpuid_count): Ditto.
32953 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
32955 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
32956 replacement expression is another instance of one of its arguments.
32958 2017-05-01 Jakub Jelinek <jakub@redhat.com>
32961 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
32962 check for stack push/pop autoinc.
32963 * config/i386/i386.c (ix86_agi_dependent): Return false
32964 if the only reason why modified_in_p returned true is that
32965 addr is SP based and set_insn is a push or pop.
32967 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32969 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
32972 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
32975 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
32976 (account_size_time): Use two predicates - exec_pred and
32978 (evaluate_conditions_for_known_args): Compute both clause and
32980 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
32981 (inline_summary_t::duplicate): Update.
32982 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
32984 (compute_inline_parameters): Likewise.
32985 (estimate_edge_size_and_time): Update caluclation of time.
32986 (estimate_node_size_and_time): Compute both time and nonspecialized
32988 (estimate_ipcp_clone_size_and_time): Update.
32989 (inline_merge_summary): Update.
32990 (do_estimate_edge_time): Update.
32991 (do_estimate_edge_size): Update.
32992 (do_estimate_edge_hints): Update.
32993 (inline_read_section, inline_write_summary): Stream both new predicates.
32994 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
32996 (compute_inlined_call_time): Cleanup.
32997 (big_speedup_p): Update.
32998 (edge_badness): Update.
32999 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33000 (size_time_entry): Replace predicate by exec_predicate and
33001 nonconst_predicate.
33002 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33003 (estimate_edge_time): Return also nonspec_time.
33004 (reset_edge_growth_cache): Update.
33006 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33008 PR rtl-optimization/80491
33009 * ifcvt.c (noce_process_if_block): When looking for x setter
33010 with missing else_bb, don't check only the insn right before
33011 cond_earliest, but look for the last insn that x is modified in
33012 within the same bb.
33014 PR rtl-optimization/80491
33015 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33017 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33019 PR tree-optimization/80487
33020 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33022 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33024 PR tree-optimization/79697
33025 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33026 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33027 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33029 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33030 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33032 2017-04-28 Martin Sebor <msebor@redhat.com>
33034 PR tree-optimization/80523
33035 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33036 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33038 (maybe_warn, format_directive, parse_directive): Use new functions.
33039 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33041 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33043 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33045 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33047 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33048 target_header_dir): Set correctly.
33049 * configure: Regenerated.
33050 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33051 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33052 instead of SYSTEM_HEADER_DIR.
33054 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33056 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33057 (estimate_local_effects): Likewise.
33058 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33059 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33060 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33061 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33062 do_estimate_edge_time, estimate_edge_time): Likewise.
33063 * ipa-inline-analysis.c (estimate_node_size_and_time,
33064 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33065 (estimate_time_after_inlining): Remove.
33067 2017-04-28 Martin Liska <mliska@suse.cz>
33069 * doc/gcov.texi: Enhance documentation of gcov.
33071 2017-04-28 Martin Liska <mliska@suse.cz>
33073 * doc/gcov.texi: Sort options in alphabetic order.
33074 * doc/gcov-dump.texi: Likewise.
33075 * doc/gcov-tool.texi: Likewise.
33076 * gcov.c (print_usage): Likewise.
33077 * gcov-dump.c (print_usage): Likewise.
33078 * gcov-tool.c (print_merge_usage_message): Likewise.
33079 (print_rewrite_usage_message): Likewise.
33080 (print_overlap_usage_message): Likewise.
33082 2017-04-28 Martin Liska <mliska@suse.cz>
33084 PR gcov-profile/53915
33085 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33087 2017-04-28 Martin Liska <mliska@suse.cz>
33089 PR gcov-profile/79891
33090 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33091 is marked by compiler as living on a line.
33092 (get_cycles_count): Remove usage of the union.
33093 (output_intermediate_file): Likewise.
33094 (find_source): Fix GNU coding style.
33095 (accumulate_line_counts): Remove old non-all block mode.
33096 (output_lines): Remove usage of the union.
33097 * profile.c (output_location): Include all BBs, even if
33098 belonging to a same line (and file) as a previous BB.
33100 2017-04-28 Martin Liska <mliska@suse.cz>
33102 * gcov.c (process_args): Handle new argument 'w'.
33103 (read_graph_file): Assign ID to BBs.
33104 (output_branch_count): Display BB # if verbose flag is set.
33105 (output_lines): Likewise for arcs.
33106 (print_usage): Add '--verbose' option help.
33107 * doc/gcov.texi: Document --verbose (-w) option.
33109 2017-04-28 Martin Liska <mliska@suse.cz>
33111 * gcov.c (struct block_location_info): New struct.
33112 (process_file): Fill up the new structure.
33113 (read_graph_file): Replace usage of encoding by the newly added
33115 (add_line_counts): Likewise.
33116 (accumulate_line_counts): Remove usage of the union.
33117 (function_info::function_info): New function.
33118 (function_info::~function_info): Likewise.
33119 (process_file): Call delete instead of release_function.
33120 (release_function): Release the function.
33121 (release_structures): Call delete instead of release_function.
33122 (solve_flow_graph): Replace usage of num_blocks.
33123 (find_exception_blocks): Likewise.
33124 (output_lines): Fix GNU coding style.
33126 2017-04-28 Martin Liska <mliska@suse.cz>
33129 * coverage.c (coverage_remove_note_file): New function.
33130 * coverage.h: Declare the function.
33131 * toplev.c (finalize): Clean if an error has been seen.
33133 2017-04-28 Martin Liska <mliska@suse.cz>
33135 PR gcov-profile/80031
33136 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33137 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33138 * gcov.c (read_graph_file): Read just number of blocks.
33139 * profile.c (branch_prob): Do not stream 0 flags per a basic
33142 2017-04-28 Martin Liska <mliska@suse.cz>
33144 * gcov-dump.c (tag_*): Add new argument to declarations.
33145 (dump_gcov_file): Likewise.
33146 (tag_blocks): Add and use new argument depth.
33147 (tag_arcs): Likewise.
33148 (tag_lines): Likewise.
33149 (tag_counters): Likewise.
33150 (tag_summary): Likewise.
33151 (dump_working_sets): Use depth to do a proper indentation.
33153 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33156 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33157 * symtab.c (symtab_node::debug_symtab): Move definition here.
33159 2017-04-28 Richard Biener <rguenther@suse.de>
33161 * lto-streamer.h (LTO_major_version): Bump to 7.
33163 2017-04-28 Richard Biener <rguenther@suse.de>
33165 * tree-vrp.c (assert_info): New struct.
33166 (add_assert_info): New helper.
33167 (register_edge_assert_for_2): Refactor to add asserts to a vector
33169 (register_edge_assert_for_1): Likewise.
33170 (register_edge_assert_for): Likewise.
33171 (finish_register_edge_assert_for): New helper actually registering
33172 asserts where live on edge.
33173 (find_conditional_asserts): Adjust.
33174 (find_switch_asserts): Likewise.
33175 (evrp_dom_walker::try_find_new_range): Generalize.
33176 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33178 2017-04-27 Marek Polacek <polacek@redhat.com>
33181 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33182 arg10 and arg11 to itype.
33184 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33186 * doc/extend.texi (Object Size Checking): Improve grammar.
33188 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33191 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33192 that the logic for permitting reciprocal estimates matches that
33195 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33198 * tree.c (type_cache_hasher::equal): Only compare
33199 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33200 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33201 non-aggregate element types.
33202 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33203 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33205 2017-04-27 Richard Biener <rguenther@suse.de>
33207 PR middle-end/80533
33208 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33209 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33210 keeping a reference to a trailing array.
33212 2017-04-27 Richard Biener <rguenther@suse.de>
33214 PR middle-end/80539
33215 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33216 being in loop-closed SSA form conservatively.
33217 (chrec_fold_multiply_poly_poly): Likewise.
33219 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33221 PR middle-end/79665
33222 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33223 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33225 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33228 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33229 (aarch64_function_arg_alignment): Return unsigned int again, but still
33230 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33231 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33232 Don't emit -Wpsabi note.
33233 (aarch64_function_arg_boundary): Likewise.
33234 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33237 2017-04-26 Nathan Sidwell <nathan@acm.org>
33239 * tree.h (crc32_unsigned_n): Declare.
33240 (crc32_unsigned, crc32_unsigned): Make inline.
33241 * tree.c (crc32_unsigned_bits): Replace with ...
33242 (crc32_unsigned_n): ... this.
33243 (crc32_unsigned, crc32_byte): Remove.
33244 (crc32_string): Remove unnecessary braces.
33246 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33248 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33249 * ipa-inline-analysis.c (MAX_TIME): Remove.
33250 (account_size_time): Use sreal for time.
33251 (dump_inline_summary): Update.
33252 (estimate_function_body_sizes): Update.
33253 (estimate_edge_size_and_time): Update.
33254 (estimate_calls_size_and_time): Update.
33255 (estimate_node_size_and_time): Update.
33256 (inline_merge_summary): Update.
33257 (inline_update_overall_summary): Update.
33258 (estimate_time_after_inlining): Update.
33259 (inline_read_section): Update.
33260 (inline_write_summary): Update.
33261 * ipa-inline.c (compute_uninlined_call_time): Update.
33262 (compute_inlined_call_time): Update.
33263 (recursive_inlining): Update.
33264 (inline_small_functions): Update.
33265 (dump_overall_stats): Update.
33266 * ipa-inline.h: Include sreal.h.
33267 (size_time_entry): Turn time to sreal.
33268 (inline_summary): Turn self_time nad time to sreal.
33270 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33272 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33274 (sreal::stream_out, sreal::stream_in): New.
33275 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33277 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33279 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33282 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33285 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33287 (dimode_scalar_chain::compute_convert_gain): Ditto.
33288 (dimode_scalar_chain::make_vector_copies): Ditto.
33289 (dimode_scalar_chain::convert_reg): Ditto.
33290 (dimode_scalar_chain::convert_insn): Ditto.
33291 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33292 (VI248_AVX512BW_1): New mode iterator.
33293 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33294 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33297 2017-04-25 Martin Sebor <msebor@redhat.com>
33299 PR tree-optimization/80497
33300 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33301 constants are representable in HOST_WIDE_INT.
33302 (parse_directive): Ditto.
33304 2017-04-25 Martin Sebor <msebor@redhat.com>
33307 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33308 (new_zero_array): Adjust signature.
33309 (dom_info::dom_init): Used unsigned rather that size_t.
33310 (dom_info::dom_info): Same.
33312 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33313 Jakub Jelinek <jakub@redhat.com>
33316 * config/arm/arm.c: Include gimple.h.
33317 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33318 returns negative, increment ncrn only if it returned positive.
33319 (arm_needs_doubleword_align): Return int instead of bool,
33320 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33321 members, but if there is any such non-FIELD_DECL
33322 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33323 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33324 returns negative, increment nregs only if it returned positive.
33325 (arm_setup_incoming_varargs): Likewise.
33326 (arm_function_arg_boundary): Emit -Wpsabi note if
33327 arm_needs_doubleword_align returns negative, return
33328 DOUBLEWORD_ALIGNMENT only if it returned positive.
33330 2017-04-25 Marek Polacek <polacek@redhat.com>
33333 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33334 first argument to type.
33336 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33339 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33340 type checks to test for compatibility instead of equality.
33342 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33343 Jakub Jelinek <jakub@redhat.com>
33346 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33348 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33349 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33350 the alignment computation, but return their maximum in warn_alignment.
33351 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33352 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33354 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33355 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33358 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33360 * config/arc/simdext.md (dmpyh): Fix typo.
33362 2017-04-25 Richard Biener <rguenther@suse.de>
33364 PR tree-optimization/80492
33365 * alias.c (compare_base_decls): Handle registers with asm
33366 specification conservatively.
33367 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33368 compare_base_decls returning dont-know properly.
33370 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33372 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33373 (legitimate_offset_address_p): New function.
33374 (arc_legitimate_address_p): Use above function.
33376 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33378 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33380 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33382 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33383 ACCH registers whenever they are available.
33385 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33387 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33388 double regs fix when not used.
33390 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33392 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
33394 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
33395 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
33397 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33399 * config/arc/arc.c (arc_output_addsi): Check for h-register class
33400 when emitting short ADD instructions.
33402 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33404 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
33406 (cmpsi_cc_c_insn): Likewise.
33407 (cbranchsi4_scratch): Compute proper instruction length using
33408 compact_hreg_operand.
33409 * config/arc/predicates.md (compact_hreg_operand): New predicate.
33411 2017-04-25 Richard Biener <rguenther@suse.de>
33413 PR middle-end/80509
33414 * passes.c (pass_manager::pass_manager): Initialize
33415 m_name_to_pass_map.
33417 2017-04-25 Richard Biener <rguenther@suse.de>
33419 PR tree-optimization/79201
33420 * tree-ssa-sink.c (statement_sink_location): Handle calls.
33422 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33425 * config/s390/vector.md: Split MEM->GPR vector moves for
33426 non-s_operand addresses.
33428 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33431 * config/s390/predicates.md (reload_const_wide_int_operand): New
33433 * config/s390/s390.md ("movti"): Remove d/P alternative.
33434 ("movti_bigconst"): New pattern definition.
33436 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33439 * s390-protos.h (s390_expand_cs_hqi): Removed.
33440 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
33441 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
33442 modes as well as CCZ1mode and CCZmode.
33443 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
33444 signature of s390_emit_compare_and_swap.
33445 (s390_expand_cs_hqi): Likewise, make static.
33446 (s390_expand_cs_tdsi): Generate an explicit compare before trying
33447 compare-and-swap, in some cases.
33448 (s390_expand_cs): Wrapper function.
33449 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
33451 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
33452 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
33453 patterns for small and large integers. Forbid symref memory operands.
33454 Move expander to s390.c. Require cc register.
33455 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
33456 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
33457 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
33458 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
33459 symref memory operands. Remove CC mode and call s390_match_ccmode
33461 ("atomic_exchange<mode>"): Allow and implement all integer modes.
33463 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33465 * config/s390/s390.md (define_peephole2): New peephole to help
33466 combining the load-and-test pattern with volatile memory.
33468 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
33470 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
33471 with CCZmode for TARGET_Z196.
33473 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33475 PR rtl-optimization/80501
33476 * combine.c (make_compound_operation_int): Set subreg_code to SET
33477 even for AND with mask of the sign bit of mode.
33479 PR rtl-optimization/80500
33480 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
33481 sum's initial value.
33483 2017-04-25 Julian Brown <julian@codesourcery.com>
33484 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33486 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
33488 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
33490 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
33492 2017-04-25 Julian Brown <julian@codesourcery.com>
33493 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33495 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
33496 (thunderx2t99_sha): New Reservation.
33498 2017-04-25 Julian Brown <julian@codesourcery.com>
33499 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
33501 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
33502 type for 1-element load.
33504 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
33506 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
33508 2017-04-24 Martin Jambor <mjambor@suse.cz>
33510 PR tree-optimization/80293
33511 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
33512 char arrays not totally scalarizable if it is false.
33513 (analyze_all_variable_accesses): Pass correct value in the new
33514 parameter. Add a statistics counter.
33516 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
33518 PR middle-end/79931
33519 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
33521 2017-04-24 Richard Biener <rguenther@suse.de>
33523 PR tree-optimization/80494
33524 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
33525 out for complex types.
33527 2017-04-24 Richard Biener <rguenther@suse.de>
33529 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
33530 * tree-ssa-sccvn.c (print_scc): Print SCC size.
33531 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
33532 (DFS): Adjust and never fail.
33533 (sccvn_dom_walker::fail): Remove.
33534 (sccvn_dom_walker::before_dom_children): Adjust.
33535 (run_scc_vn): Likewise and never fail.
33536 * tree-ssa-pre.c (pass_pre::execute): Adjust.
33537 (pass_fre::execute): Likewise.
33539 2017-04-24 Richard Biener <rguenther@suse.de>
33541 PR tree-optimization/79725
33542 * tree-ssa-sink.c (statement_sink_location): Return whether
33543 failure reason was zero uses. Move that check later.
33544 (sink_code_in_bb): Deal with zero uses by removing the stmt
33547 2017-04-24 Richard Biener <rguenther@suse.de>
33550 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
33551 pointer-based references.
33553 2017-04-24 Richard Biener <rguenther@suse.de>
33556 * pass_manager.h (pass_manager::operator new): Remove.
33557 (pass_manager::operator delete): Likewise.
33558 * passes.c (pass_manager::operator new): Remove.
33559 (pass_manager::operator delete): Likewise.
33560 (pass_manager::pass_manager): Zero individual pass members.
33562 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
33565 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
33566 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
33567 Check "XEXP (src, 1)" operand here.
33568 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
33569 Check "XEXP (src, 1)" operand here.
33570 (dimode_scalar_chain::make_vector_copies): Detect count register
33571 of a shift instruction. Zero extend count register from QImode
33572 to DImode to satisfy vector shift pattern count operand predicate.
33573 Substitute vector shift count operand with a DImode copy.
33574 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
33577 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
33579 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
33580 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33581 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
33582 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
33583 (UNSPEC_NOREX_MEM): Remove definition.
33585 2017-04-21 Richard Biener <rguenther@suse.de>
33587 PR tree-optimization/79547
33588 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
33589 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
33590 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
33591 without any constraints.
33593 2017-04-21 Richard Biener <rguenther@suse.de>
33595 PR tree-optimization/78847
33596 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
33598 2017-04-21 Richard Biener <rguenther@suse.de>
33600 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
33601 (build_distinct_type_copy): Likewise.
33602 (build_variant_type_copy): Likewise.
33603 * tree.c (build_qualified_type): Pass down mem-stat info.
33604 (build_distinct_type_copy): Likewise.
33605 (build_variant_type_copy): Likewise.
33607 2017-04-21 Richard Biener <rguenther@suse.de>
33609 PR tree-optimization/80237
33610 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
33612 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
33613 for a simplified result.
33615 2016-04-21 Richard Biener <rguenther@suse.de>
33617 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
33618 sth as strict as a simple_iv but a chrec without symbols and an
33619 operand defined in the loop we are peeling (and not some subloop).
33620 (propagate_constants_for_unrolling): Propagate all constants.
33622 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33625 * config/i386/i386.c (print_reg): Remove assert for disalowed
33626 regno values, call output_operand_lossage instead.
33628 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
33631 * config/i386/constraints.md (Yc): New register constraint.
33632 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
33633 Use Yc constraint for alternative 2 of operand 0. Remove
33634 preferred_for_speed attribute.
33636 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33638 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
33639 lastprivate clauses in SIMT case.
33641 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
33643 * doc/invoke.texi (-Wextra-semi): Document new warning option.
33645 2017-04-20 Richard Biener <rguenther@suse.de>
33647 PR tree-optimization/57796
33648 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
33649 as N scalar stores.
33650 (vect_model_load_cost): Cost gathers as N scalar loads.
33652 2017-04-20 Richard Biener <rguenther@suse.de>
33654 * ggc-page.c (ggc_allocated_p): Rename to ...
33655 (safe_lookup_page_table_entry): ... this and return the lookup
33657 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
33659 2017-04-20 Richard Biener <rguenther@suse.de>
33661 PR tree-optimization/80453
33662 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
33663 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
33664 from the conditions.
33665 (vn_phi_eq): Pass them down.
33666 (vn_phi_lookup): Record them.
33667 (vn_phi_insert): Likewise.
33669 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
33671 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
33672 uninitialized variable warning to avoid buffer overrun.
33674 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
33677 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
33678 is suppressed for '{ 0 }' in C.
33680 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33682 * BASE-VER: Set to 8.0.0.
33684 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
33686 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
33687 priority .init_array and .fini_array section with SECTION_NOTYPE
33690 2017-04-20 Jakub Jelinek <jakub@redhat.com>
33692 PR middle-end/80423
33693 * tree.h (build_array_type): Add typeless_storage default argument.
33694 * tree.c (type_cache_hasher::equal): Also compare
33695 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
33696 (build_array_type): Add typeless_storage argument, set
33697 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
33699 (build_nonshared_array_type): Adjust build_array_type_1 caller.
33700 (build_array_type): Likewise. Add typeless_storage argument.
33702 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33703 Jakub Jelinek <jakub@redhat.com>
33705 PR tree-optimization/80426
33706 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
33707 operation on symbolic operands, also compute the overflow for the
33708 invariant part when the operation degenerates into a negation.
33710 2017-04-19 Jakub Jelinek <jakub@redhat.com>
33713 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
33714 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
33717 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
33719 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
33722 * config/avr/avr.c (tree.h): Include it.
33723 (cgraph.h): Include it.
33724 (avr_encode_section_info): Don't warn for uninitialized progmem
33725 variable if it's just an alias.
33727 2017-04-19 Richard Biener <rguenther@suse.de>
33730 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
33731 when needed by AutoPGO.
33733 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
33736 * doc/lto.texi: Remove an extra 'that'.
33738 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
33740 PR rtl-optimization/80429
33741 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
33742 are only used in debug insns.
33744 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
33745 Vladimir Makarov <vmakarov@redhat.com>
33747 * config/sparc/predicates.md (input_operand): Add comment. Return
33748 true for any memory operand when LRA is in progress.
33749 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
33751 2017-04-18 Jeff Law <law@redhat.com>
33754 * mips.md ({return,simple_return}_internal): Do not overwrite
33757 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33759 PR tree-optimization/80443
33760 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
33761 instead of adding 1, subtract -1 and similarly instead of subtracting
33764 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
33766 PR rtl-optimization/80357
33767 * haifa-sched.c (tmp_bitmap): New variable.
33768 (model_recompute): Handle duplicate use records.
33769 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
33770 (free_global_sched_pressure_data): Free it.
33772 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
33775 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
33776 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33777 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33778 instead of SYSTEM_HEADER_DIR.
33780 2017-04-18 Jeff Law <law@redhat.com>
33782 PR middle-end/80422
33783 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
33784 predecessors after walking up the insn chain.
33786 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33789 * dwarf2out.c (modified_type_die): Try harder not to emit internal
33790 sizetype type into debug info.
33792 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33795 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
33796 unneeded test for TARGET_UPPER_REGS_SF.
33797 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
33799 2017-04-18 Jakub Jelinek <jakub@redhat.com>
33802 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
33803 instead of gsi_after_labels.
33805 2017-04-18 Jeff Law <law@redhat.com>
33807 * regcprop.c (maybe_mode_change): Avoid creating copies of the
33811 2017-04-13 Jeff Law <law@redhat.com>
33812 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33813 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33815 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
33818 * config/avr/avr.c (intl.h): Include it.
33819 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
33821 2017-04-18 Martin Liska <mliska@suse.cz>
33823 PR gcov-profile/78783
33824 * gcov-tool.c (gcov_output_files): Validate that destination
33825 file is either removed by the tool or by a user.
33827 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33828 Guy Benyei <guybe@mellanox.com>
33830 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
33831 block, and do not negate it, the stored id is already negative.
33833 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
33835 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33837 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33840 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
33841 masks of options that should be turned off if the VSX vector
33842 options are turned off.
33843 (OTHER_P8_VECTOR_MASKS): Likewise.
33844 (OTHER_VSX_VECTOR_MASKS): Likewise.
33845 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
33846 rs6000_disable_incompatible_switches to validate no type switches
33848 (rs6000_incompatible_switch): New function to disallow turning on
33849 other vector options if -mno-vsx, -mno-power8-vector, or
33850 -mno-power9-vector are specified.
33852 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33854 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
33856 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33858 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
33859 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
33860 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
33861 (ARG_POINTER_CFA_OFFSET): Likewise.
33863 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
33865 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
33866 conditions to take advantage of various optimizations.
33868 2017-04-13 Jeff Law <law@redhat.com>
33870 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
33871 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
33872 (zero_extendsidi2_dext): Likewise.
33874 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33877 * fold-const.c (fold_ternary_loc): Revert
33878 use op0 instead of fold_convert_loc (loc, type, arg0) part of
33881 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
33883 PR rtl-optimization/80343
33884 * lra-remat.c (update_scratch_ops): Assign original hard reg to
33885 new scratch pseudo.
33887 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
33890 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
33891 to ubsan_encode_value.
33893 2017-04-13 Jeff Law <law@redhat.com>
33895 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
33896 appearing in DEBUG_INSNs.
33898 2017-04-13 Martin Liska <mliska@suse.cz>
33900 PR gcov-profile/80413
33901 * gcov-io.c (gcov_write_string): Copy to buffer just when
33902 allocated size is greater than zero.
33904 2017-04-13 Jakub Jelinek <jakub@redhat.com>
33907 * dwarf2out.c (decls_for_scope): Ignore declarations of
33908 current_function_decl in BLOCK_NONLOCALIZED_VARS.
33910 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
33913 * ipa-visibility.c (non_local_p): Fix typos.
33914 (localize_node): When localizing symbol in same comdat group,
33915 dissolve the group only when we know external symbols are going
33917 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
33919 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33921 PR tree-optimization/79390
33922 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
33923 order does not result in usable sequence, retry with reversed operand
33929 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33930 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
33931 op0 instead of fold_convert_loc (loc, type, arg0).
33933 2017-04-12 Jeff Law <law@redhat.com>
33935 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
33936 has a delay slot in the generated code.
33938 * config/cris/cris.md (cris_preferred_reload_class): Return
33939 GENNONACR_REGS rather than GENERAL_REGS.
33941 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33944 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
33945 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
33946 signedness of the result type.
33948 2017-04-12 Richard Biener <rguenther@suse.de>
33949 Jeff Law <law@redhat.com>
33951 PR tree-optimization/80359
33952 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
33953 trim stores to TARGET_MEM_REFs.
33955 2017-04-12 Richard Biener <rguenther@suse.de>
33957 PR tree-optimization/79390
33958 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
33959 threading case even more.
33961 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
33964 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
33965 for quad_address_p for TImode, instead of just not indexed_address.
33967 2017-04-12 Richard Biener <rguenther@suse.de>
33968 Bernd Edlinger <bernd.edlinger@hotmail.de>
33970 PR middle-end/79671
33971 * alias.c (component_uses_parent_alias_set_from): Handle
33972 TYPE_TYPELESS_STORAGE.
33973 (get_alias_set): Likewise.
33974 * tree-core.h (tree_type_common): Add typeless_storage flag.
33975 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
33976 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
33977 for types containing members with TYPE_TYPELESS_STORAGE.
33978 (place_field): Likewise.
33979 (layout_type): Likewise for ARRAY_TYPE.
33980 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
33981 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
33982 TYPE_TYPELESS_STORAGE.
33983 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
33985 2017-04-12 Jakub Jelinek <jakub@redhat.com>
33988 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33989 first argument to type.
33991 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33995 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
33996 CONST0_RTX (mode) rather than const0_rtx where appropriate.
33997 (rs6000_expand_binop_builtin): Likewise.
33998 (rs6000_expand_ternop_builtin): Likewise; also add missing
33999 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34000 vshasigma built-ins.
34001 * doc/extend.texi: Document that vec_xxpermdi's third argument
34002 must be a constant.
34004 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34006 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34007 Use shift_const cost parameter when calculating gain of STV shifts.
34009 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34011 PR rtl-optimization/70478
34012 * lra-constraints.c (process_alt_operands): Check memory for
34013 disfavoring memory insn operand.
34015 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34017 PR middle-end/80100
34018 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34019 left shift in unsigned HOST_WIDE_INT type.
34021 PR rtl-optimization/80385
34022 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34023 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34026 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34027 if they have any depend clauses.
34029 2017-04-11 Martin Liska <mliska@suse.cz>
34032 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34033 * ipa-split.c (split_function): Create a local comdat symbol
34034 if caller is in a comdat group.
34036 2017-04-11 Martin Liska <mliska@suse.cz>
34039 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34042 2017-04-11 Martin Sebor <msebor@redhat.com>
34044 PR middle-end/80364
34045 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34046 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34048 (directive::set_width, directive::set_precision, format_character):
34050 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34053 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34056 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34057 conflict, set target->arch_name instead of target->cpu_name.
34059 2017-04-11 Richard Biener <rguenther@suse.de>
34061 PR tree-optimization/80374
34062 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34063 build_zero_cst, remove fold_convertible_p check again.
34065 2017-04-11 Martin Liska <mliska@suse.cz>
34068 * ubsan.c (instrument_object_size): Do not instrument register
34071 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34074 * config/i386/i386-builtin-types.def
34075 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34076 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34077 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34078 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34079 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34080 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34081 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34082 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34083 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34084 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34085 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34086 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34087 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34088 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34089 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34090 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34091 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34092 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34093 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34094 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34095 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34096 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34097 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34098 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34099 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34100 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34101 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34102 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34103 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34104 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34105 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34106 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34107 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34108 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34109 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34110 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34111 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34112 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34113 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34114 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34115 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34116 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34117 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34118 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34119 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34120 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34122 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34123 flag to second_arg_count, handle 4 argument function type _COUNT
34124 aliases, handle second_arg_count on second argument rather than last.
34126 2017-04-10 Jeff Law <law@redhat.com>
34128 PR tree-optimization/80374
34129 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34130 record anything if we can not convert integer_zero_node to the
34133 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34136 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34137 Enhance special handling given to the TARGET_P9_MINMAX option in
34138 relation to certain other options.
34140 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34142 PR tree-optimization/80153
34143 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34144 remove POINTER_PLUS_EXPR's base part directly, rather than through
34147 2017-04-10 Richard Biener <rguenther@suse.de>
34148 Bin Cheng <bin.cheng@arm.com>
34150 PR tree-optimization/80153
34151 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34152 the first element of pointer type aff_tree. Build result expr in
34154 (add_elt_to_tree): Convert to type unconditionally. Remove other
34155 fold_convert calls.
34156 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34157 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34159 2017-04-10 Richard Biener <rguenther@suse.de>
34161 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34164 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34166 PR rtl-optimization/70478
34167 * lra-constraints.c (curr_small_class_check): New.
34168 (update_and_check_small_class_inputs): New.
34169 (process_alt_operands): Update curr_small_class_check. Disfavor
34170 alternative insn memory operands. Check available regs for small
34173 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34176 * config/mips/mips.opt (-mvirt): Update description.
34177 * doc/invoke.texi (-mvirt): Likewise.
34179 2017-04-10 Richard Biener <rguenther@suse.de>
34181 PR middle-end/80362
34182 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34183 looking for NEGATE_EXPR in -A / -B to A / B folding.
34185 2017-04-10 Martin Liska <mliska@suse.cz>
34187 PR gcov-profile/80224
34188 * gcov.c (print_usage): Fix usage string.
34189 (get_gcov_intermediate_filename): Remove.
34190 (output_gcov_file): Use both for normal and intermediate format.
34191 (generate_results): Do not initialize special file for
34192 intermediate format.
34194 2017-04-10 Richard Biener <rguenther@suse.de>
34196 PR tree-optimization/80304
34197 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34200 2017-04-10 Nathan Sidwell <nathan@acm.org>
34203 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34204 (rs6000_init_builtins): Use it.
34206 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34208 * config/arm/arm.md (<mrc>): Add mode to SET source.
34209 (<mrrc>): Likewise.
34211 2017-04-10 Richard Biener <rguenther@suse.de>
34213 PR middle-end/80344
34214 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34216 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34219 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34220 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34221 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34222 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34223 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34224 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34225 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34226 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34227 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34228 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34229 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34230 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34231 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34232 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34233 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34234 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34235 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34236 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34237 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34238 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34239 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34240 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34241 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34243 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34245 PR rtl-optimization/70478
34246 * lra-constraints.c: Reverse the last patch.
34248 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34250 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34251 Add comment for WCHAR_T.
34253 2017-04-08 Martin Liska <mliska@suse.cz>
34256 2017-04-07 Martin Liska <mliska@suse.cz>
34259 * ipa-split.c (split_function): Add function part to a same comdat
34262 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34265 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34267 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34269 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34270 (vec_pairing): ...to this.
34271 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34272 (rs6000_sched_init): Adjust for name change.
34273 (struct rs6000_sched_context): Likewise.
34274 (rs6000_init_sched_context): Likewise.
34275 (rs6000_set_sched_context): Likewise.
34277 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34283 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34285 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34286 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34287 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34289 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34291 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34293 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34295 PR rtl-optimization/70703
34296 * ira-color.c (update_conflict_hard_regno_costs): Use
34297 int64_t instead of HOST_WIDE_INT.
34299 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34301 PR rtl-optimization/70478
34302 * lra-constraints.c (process_alt_operands): Disfavor alternative
34303 insn memory operands.
34305 2017-04-07 Jeff Law <law@redhat.com>
34307 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34308 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34310 2017-04-07 Martin Liska <mliska@suse.cz>
34313 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34314 Show error message instead of an ICE.
34316 2017-04-07 Martin Liska <mliska@suse.cz>
34319 * ipa-split.c (split_function): Add function part to a same comdat
34322 2017-04-07 Richard Biener <rguenther@suse.de>
34324 PR middle-end/80341
34325 * tree.c (get_unwidened): Also handle ! for_type case for
34327 * convert.c (do_narrow): Split out from ...
34328 (convert_to_integer_1): ... here. Do not pass final truncation
34329 type to get_unwidened for TRUNC_DIV_EXPR.
34331 2017-04-07 Richard Biener <rguenther@suse.de>
34333 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34335 (aff_combination_const): Adjust.
34336 (aff_combination_scale): Likewise.
34337 (aff_combination_add_elt): Likewise.
34338 (aff_combination_add_cst): Likewise.
34339 (aff_combination_convert): Likewise.
34340 (add_elt_to_tree): Likewise. Remove unused argument.
34341 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34343 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34345 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34347 * config/arm/arm.c (arm_default_short_enums): Use
34348 ARM_DEFAULT_SHORT_ENUMS.
34349 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34351 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34354 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34355 members with redundant out-of-class redeclaration.
34357 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34360 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34361 * config/i386/i386.md (*zero_extendsidi2):
34362 Add (?*x,*x) and (?*v,*v) alternatives.
34364 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34367 * config/i386/i386.c (ix86_expand_builtin)
34368 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34369 mode from insn data. Convert operands to insn operand mode.
34370 Copy operands that don't satisfy insn predicate to a register.
34372 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34374 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34377 2017-04-06 Richard Biener <rguenther@suse.de>
34379 PR tree-optimization/80334
34380 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34381 preserve alignment of accesses.
34383 2017-04-06 Richard Biener <rguenther@suse.de>
34385 PR tree-optimization/80262
34386 * tree-sra.c (build_ref_for_offset): Preserve address-space
34388 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34389 Drop useless address-space information on MEM_REF offsets.
34391 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
34393 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
34395 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34397 PR rtl-optimization/70703
34398 * ira-color.c (update_conflict_hard_regno_costs): Use
34399 HOST_WIDE_INT instead of long.
34401 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
34404 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
34405 not defined for x86_64 target. Add -mmmx target option when __SSE2__
34407 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
34408 for x86_64 target. Handle -m3dnowa option.
34410 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
34412 PR rtl-optimization/70703
34413 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
34414 (update_conflict_hard_regno_costs): Use long instead of unsigned
34415 arithmetic for cost calculation.
34417 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34418 Bernd Edlinger <bernd.edlinger@hotmail.de>
34421 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
34424 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
34427 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
34428 ptr_mode with Pmode throughout.
34429 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
34430 into probe_stack_range and use DImode.
34432 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
34435 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
34436 call_eh_return is true.
34438 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34440 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
34441 Initialize last_match_fntype_index.
34443 2017-04-05 Jakub Jelinek <jakub@redhat.com>
34446 * tree-nvr.c: Include internal-fn.h.
34447 (pass_return_slot::execute): Ignore internal calls without
34450 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34451 Richard Biener <rguenther@suse.de>
34454 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
34455 captures used multiple times, except for the last use.
34456 * generic-match-head.c: Include gimplify.h.
34458 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34460 PR tree-optimization/79390
34461 * target.h (struct noce_if_info): Declare.
34462 * targhooks.h (default_noce_conversion_profitable_p): Declare.
34463 * target.def (noce_conversion_profitable_p): New target hook.
34464 * ifcvt.h (struct noce_if_info): New type, moved from ...
34465 * ifcvt.c (struct noce_if_info): ... here.
34466 (noce_conversion_profitable_p): Renamed to ...
34467 (default_noce_conversion_profitable_p): ... this. No longer
34469 (noce_try_store_flag_constants, noce_try_addcc,
34470 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
34471 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
34472 instead of noce_conversion_profitable_p.
34473 * config/i386/i386.c: Include ifcvt.h.
34474 (ix86_option_override_internal): Don't override
34475 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
34476 (ix86_noce_conversion_profitable_p): New function.
34477 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
34478 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
34479 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
34480 * doc/tm.texi: Regenerated.
34482 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34484 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
34487 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
34490 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
34491 instructions for small multiply cores.
34493 2017-04-04 Jeff Law <law@redhat.com>
34495 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
34497 (mips_expand_vec_perm_const): Initialize elements in orig_perm
34498 that are not set by the loop over the elements.
34500 2017-04-04 Jakub Jelinek <jakub@redhat.com>
34503 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
34504 int mode, convert_modes it to mode as unsigned, otherwise use
34505 lowpart_subreg to mode rather than SImode.
34506 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
34507 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
34508 Use DImode instead of SImode for the shift count operand.
34509 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
34512 2017-04-04 Richard Biener <rguenther@suse.de>
34514 PR middle-end/80281
34515 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
34516 arithmetic done for the negate or the plus. Simplify.
34517 (A - (-B) -> A + B): Likewise.
34518 * fold-const.c (split_tree): Make sure to not negate pointers.
34520 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
34522 PR rtl-optimization/60818
34523 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
34524 a compare of comparisons with the thing compared if this results
34525 in a different machine mode.
34527 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
34529 * alias.c (base_alias_check): Fix typo in comment.
34530 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
34531 * cgraphunit.c (symbol_table::compile): Likewise.
34532 * collect2.c (maybe_run_lto_and_relink): Likewise.
34533 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
34534 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
34535 * config/avr/avr.c (avr_map_op_t): Likewise.
34536 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
34537 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
34538 * config/epiphany/epiphany.md (movcc): Likewise.
34539 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
34540 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
34542 * config/mips/mips.c (mips_save_restore_reg): Likewise.
34543 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
34544 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
34545 * config/sh/sh.c (sh_rtx_costs): Likewise.
34546 * fold-const.c (fold_truth_andor): Likewise.
34547 * genautomata.c (collapse_flag): Likewise.
34548 * gengtype.h (struct type::u::s): Likewise.
34549 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
34550 * input.c (FORMAT_AMOUNT): Likewise.
34551 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
34552 (known_aggs_to_agg_replacement_list): Likewise.
34553 * ipa-inline-analysis.c: Likewise.
34554 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
34555 * ipa-polymorphic-call.c
34556 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
34557 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
34558 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
34560 * modulo-sched.c (apply_reg_moves): Likewise.
34561 * omp-expand.c (build_omp_regions_1): Likewise.
34562 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
34563 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
34564 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
34565 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
34566 * value-prof.c: Likewise.
34567 * var-tracking.c (val_reset): Likewise.
34569 2017-04-03 Richard Biener <rguenther@suse.de>
34571 PR tree-optimization/80275
34572 * fold-const.c (split_address_to_core_and_offset): Handle
34575 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
34577 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
34578 descriptors is at least equal to that of functions.
34580 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34582 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
34584 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
34587 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
34588 (mov<IMOD4:mode>): New expander.
34589 (*mov<IMOD4:mode>_internal): New insn and split pattern.
34591 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
34593 PR rtl-optimization/79405
34594 * fwprop.c (propagations_left): New variable.
34595 (forward_propagate_into): Decrement it.
34596 (fwprop_init): Initialize it.
34597 (fw_prop): If the variable has reached zero, stop propagating.
34598 (fwprop_addr): Ditto.
34600 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34603 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
34604 a FUNCTION_DECL, pass it as decl instead of origin to
34607 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
34609 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
34612 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
34615 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
34616 TARGET_VSX_SMALL_INTEGER.
34618 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34620 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
34621 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
34623 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34625 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
34626 extraction from odd-numbered MSA register.
34628 2017-03-31 Jakub Jelinek <jakub@redhat.com>
34630 PR middle-end/80173
34631 * expmed.c (store_bit_field_1): Don't attempt to create
34632 a word subreg out of hard registers wider than word if they
34633 have HARD_REGNO_NREGS of 1 for their mode.
34635 PR middle-end/80163
34636 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
34637 conversions to integer types wider than word and pointer.
34640 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
34641 (rtx_equal_for_cselib_p): Pass 0 to it.
34642 * cselib.c (cselib_hasher::equal): Likewise.
34643 (rtx_equal_for_cselib_1): Add depth argument. If depth
34644 is 128, don't look up VALUE locs and punt. Increment
34645 depth in recursive calls when walking VALUE locs.
34647 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
34649 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
34650 (make_gcov_file_name): Use the canonical path name for generating
34652 (read_line): Fix handling of files with ascii null bytes.
34654 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
34656 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
34657 to initialise a vector register instead
34658 of using a const_int.
34660 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34662 PR translation/80189
34663 * gimplify.c (omp_default_clause): Use %qs instead of %s in
34664 diagnostic messages.
34666 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
34669 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
34670 (dfp_diex_<mode>): Update mode of operand 1.
34671 * doc/extend.texi (dxex, dxexq): Document change to return type.
34672 (diex, diexq): Document change to argument type.
34674 2017-03-30 Martin Jambor <mjambor@suse.cz>
34677 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
34678 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
34679 it reflects the signature changes performed at the callee side.
34680 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
34681 to cgraph_build_function_type_skip_args.
34682 (build_function_decl_skip_args): Adjust call to the above function.
34684 2017-03-30 Jakub Jelinek <jakub@redhat.com>
34687 * config/i386/sse.md
34688 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
34689 register as dest whenever it is a MEM not rtx_equal_p to the
34690 corresponding dup operand, and when forcing into reg move the
34691 reg into the memory afterwards.
34692 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
34693 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
34694 for the force_reg mode.
34695 (avx512vl_vextractf128<mode>): Use register as dest either
34696 always when a MEM, or when it is a MEM not rtx_equal_p to the
34697 corresponding dup operand, or even not when it is a CONST_VECTOR
34698 depending on the mode and lo vs. hi.
34699 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
34701 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
34702 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
34703 Likewise. Require that operands[2] is even.
34704 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
34705 Remove extraneous parens. Require that operands[2] is a multiple
34707 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
34708 operands[0] is a MEM if <mask_applied>, the predicates/constraints
34709 disallow memory then.
34711 2017-03-30 Richard Biener <rguenther@suse.de>
34713 PR tree-optimization/77498
34714 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
34715 to non-constants over backedges.
34717 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
34719 PR rtl-optimization/80233
34720 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
34721 as last_combined_insn. Do not test for BARRIER_P separately.
34723 2017-03-29 Andreas Schwab <schwab@suse.de>
34726 * calls.c (prepare_call_address): Convert funexp to Pmode before
34727 copying to temp reg.
34729 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34731 PR tree-optimization/80158
34732 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
34733 Handle possible future case of more than one alternate
34735 (replace_rhs_if_not_dup): Likewise.
34736 (replace_one_candidate): Likewise.
34738 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
34740 PR rtl-optimization/80193
34741 * ira.c (ira): Do not check allocation for LRA.
34743 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
34745 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
34746 (nvptx_output_simt_exit): Declare.
34747 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
34748 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
34749 (init_softstack_frame): Move initialization of crtl->is_leaf to...
34750 (nvptx_declare_function_name): ...here. Emit declaration of local
34751 memory space buffer for omp_simt_enter insn.
34752 (nvptx_output_unisimt_switch): New.
34753 (nvptx_output_softstack_switch): New.
34754 (nvptx_output_simt_enter): New.
34755 (nvptx_output_simt_exit): New.
34756 * config/nvptx/nvptx.h (struct machine_function): New fields
34757 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
34758 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
34759 (UNSPECV_SIMT_EXIT): Ditto.
34760 (omp_simt_enter_insn): New insn.
34761 (omp_simt_enter): New expansion.
34762 (omp_simt_exit): New insn.
34763 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
34765 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
34766 (expand_GOMP_SIMT_ENTER_ALLOC): New.
34767 (expand_GOMP_SIMT_EXIT): New.
34768 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
34769 (GOMP_SIMT_ENTER_ALLOC): Ditto.
34770 (GOMP_SIMT_EXIT): Ditto.
34771 * target-insns.def (omp_simt_enter): New insn.
34772 (omp_simt_exit): Ditto.
34773 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
34775 (lower_rec_simd_input_clauses): Implement SIMT privatization.
34776 (lower_rec_input_clauses): Likewise.
34777 (lower_lastprivate_clauses): Handle SIMT privatization.
34779 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
34780 (ompdevlow_adjust_simt_enter): New.
34781 (find_simtpriv_var_op): New.
34782 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
34783 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
34785 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
34786 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
34787 (copy_decl_for_dup_finish): Ditto.
34789 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
34791 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
34794 * config/i386/i386.c (ix86_option_override_internal): Always
34795 allow -mpreferred-stack-boundary=3 for 64-bit targets.
34797 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34799 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
34801 2017-03-28 Bin Cheng <bin.cheng@arm.com>
34803 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
34804 mark new edge's irreducible flag accordign to it.
34805 (vect_do_peeling): Check loop preheader edge's irreducible flag
34806 and pass it to function slpeel_add_loop_guard.
34808 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
34810 PR tree-optimization/80218
34811 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34812 Update block frequencies and counts.
34814 2017-03-28 Richard Biener <rguenther@suse.de>
34816 PR tree-optimization/78644
34817 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
34818 of a simplification result we may not use it at all.
34820 2017-03-28 Richard Biener <rguenther@suse.de>
34823 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
34824 without arguments, generate default definition of a SSA name.
34826 2017-03-28 Richard Biener <rguenther@suse.de>
34828 PR middle-end/80222
34829 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
34830 TYPE_REF_CAN_ALIAS_ALL references.
34831 * fold-const.c (fold_indirect_ref_1): Likewise.
34833 2017-03-28 Martin Liska <mliska@suse.cz>
34836 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
34837 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
34839 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34840 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
34842 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
34843 (EXTRA_SPECS): Define.
34844 (SUBTARGET_EXTRA_SPECS): Likewise.
34845 (SUBTARGET_CPP_SPEC): Likewise.
34846 * config/arc/elf.h (EXTRA_SPECS): Renamed to
34847 SUBTARGET_EXTRA_SPECS.
34848 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
34850 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
34852 * config/arc/simdext.md (vst64_insn): Update pattern.
34853 (vld32wh_insn): Likewise.
34854 (vld32wl_insn): Likewise.
34855 (vld64_insn): Likewise.
34856 (vld32_insn): Likewise.
34858 2017-03-28 Marek Polacek <polacek@redhat.com>
34861 * fold-const.c (fold_comparison): Use protected_set_expr_location
34862 instead of SET_EXPR_LOCATION.
34864 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
34866 * tree.c (add_expr): Avoid name lookup warning.
34868 2017-03-27 Jeff Law <law@redhat.com>
34870 PR tree-optimization/80216
34871 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
34872 function name. Limit recursion depth.
34873 (record_temporary_equivalences): Corresponding changes.
34875 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
34877 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
34880 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34883 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
34885 * cfgcleanup.c (reg_note_cfa_p): New array.
34886 (insns_have_identical_cfa_notes): New function.
34887 (old_insns_match_p): Don't cross-jump in between /f
34888 and non-/f instructions. If both i1 and i2 are frame related,
34889 verify all CFA notes, their order and content.
34891 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
34894 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
34895 HImode and SImode with zero extend to DImode to one insn.
34896 (bswap<mode>2_extenddi): Likewise.
34897 (bswapsi2_extenddi): Likewise.
34898 (bswaphi2_extendsi): Likewise.
34899 (bswaphi2): Combine bswap HImode and SImode into one insn.
34900 Separate memory insns from swapping register.
34901 (bswapsi2): Likewise.
34902 (bswap<mode>2): Likewise.
34903 (bswaphi2_internal): Delete, no longer used.
34904 (bswapsi2_internal): Likewise.
34905 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
34906 store, and gpr<-gpr swap insns.
34907 (bswap<mode>2_store): Likewise.
34908 (bswaphi2_reg): Register only splitter, combine with the splitter.
34909 (bswaphi2 splitter): Likewise.
34910 (bswapsi2_reg): Likewise.
34911 (bswapsi2 splitter): Likewise.
34912 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
34913 the insns into load, store, and register/register insns.
34914 (bswapdi2_ldbrx): Likewise.
34915 (bswapdi2_load): Likewise.
34916 (bswapdi2_store): Likewise.
34917 (bswapdi2_reg): Likewise.
34919 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
34921 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
34922 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
34924 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
34927 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
34929 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34930 special handling for target option conflicts between dform
34931 options (-mpower9-dform, -mpower9-dform-vector,
34932 -mpower9-dform-scalar) and -mno-direct-move.
34934 2017-03-27 Richard Biener <rguenther@suse.de>
34936 PR tree-optimization/80181
34937 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
34939 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34941 * config/arc/predicates.md (move_double_src_operand): Replace the
34942 call to move_double_src_operand with a call to address_operand.
34944 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34946 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
34947 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
34948 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
34950 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34952 * config/arc/predicates.md (long_immediate_loadstore_operand):
34953 Consider scaled addresses cases.
34955 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
34957 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
34958 restored when in interrupt.
34959 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
34960 doesn't have delay slot.
34962 2017-03-27 Richard Biener <rguenther@suse.de>
34965 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
34966 inlined thunk clones.
34968 2017-03-27 Jakub Jelinek <jakub@redhat.com>
34971 * asan.c (instrument_derefs): Copy over last operand from
34972 original COMPONENT_REF to the new COMPONENT_REF with
34973 DECL_BIT_FIELD_REPRESENTATIVE.
34974 * ubsan.c (instrument_object_size): Likewise.
34976 2017-03-27 Richard Biener <rguenther@suse.de>
34978 PR tree-optimization/80170
34979 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
34980 sure DR/SCEV didnt fold in constants we do not see when looking
34981 at the reference base alignment.
34983 2017-03-27 Richard Biener <rguenther@suse.de>
34985 PR middle-end/80171
34986 * gimple-fold.c (fold_ctor_reference): Properly guard against
34987 NULL return value from canonicalize_constructor_val.
34989 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
34992 * config/i386/i386.c (ix86_expand_builtin)
34993 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
34994 flags reg setting and flags reg using instructions.
34995 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
34996 clobbering instructions to zero extend op2.
34998 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35000 * doc/install.texi (Configuration) <--with-aix-soname>:
35001 Update link to AIX ld.
35003 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35005 PR rtl-optimization/80160
35006 PR rtl-optimization/80159
35007 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35008 reg_alternate_class into account.
35010 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35013 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35014 to consider in curr_insn_transform.
35016 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35018 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35019 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35020 and emit_mode_inner.
35022 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35024 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35025 argument to the overloaded builtin variants. Use the new flag to
35026 deprecate certain builtin variants.
35027 * config/s390/s390-builtin-types.def: Add new builtin types.
35028 * config/s390/s390-builtins.h: Support new flags field for
35029 overloaded builtins.
35030 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35031 (s390_macro_to_expand): Enable vector float data type.
35032 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35033 builtins by incrementing the __VEC__ version number.
35034 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35036 (s390_resolve_overloaded_builtin): Emit error messages depending
35037 on the builtin flags.
35038 * config/s390/s390.c (s390_expand_builtin): Support additional
35039 flags argument. Change error message to match the messages
35040 emitted in s390-c.c.
35041 * config/s390/s390.md: New UNSPEC_* constants.
35042 (op_type): Add new instruction types.
35043 * config/s390/vecintrin.h: Add new builtins and test data class
35045 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35046 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35047 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35048 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35049 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35050 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35052 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35053 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35054 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35055 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35057 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35058 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35059 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35060 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35062 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35063 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35064 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35065 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35066 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35067 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35068 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35070 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35071 ("vec_scatter_element<V_HW_4:mode>_DI")
35072 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35073 ("vec_fpint<mode>", "vflls")
35074 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35075 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35076 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35077 ("*vec_cmphe<mode>_cc"): ... these.
35079 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35080 mode constant instead of magic value.
35082 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35084 * config/s390/s390.c (s390_expand_vec_compare): Support other
35085 vector floating point modes than just V2DF.
35086 (s390_expand_vcond): Likewise.
35087 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35088 (s390_cannot_change_mode_class): Prevent mode changes between TF
35089 and V1TF in vector registers.
35090 * config/s390/s390.md (DF, SF): New mode attributes.
35091 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35092 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35093 SFmode support for VRs.
35094 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35096 (VFT, VF_HW): New mode iterators.
35097 (vw, sdx): New mode attributes.
35098 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35099 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35100 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35101 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35102 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35103 also the new vector floating point modes. Renaming to ...
35105 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35106 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35107 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35108 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35109 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35110 ("vec_unordered<mode>"): ... these.
35112 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35113 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35114 ("*vec_extendv2df"): New insn definitions.
35116 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35118 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35119 ("mulditi3_2", "*muldi3_sign"): New patterns.
35120 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35121 rename the pattern definition.
35123 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35125 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35127 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35129 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35131 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35132 instruction if possible.
35133 * config/s390/vector.md (vec_halfnumelts): New mode
35135 ("*vec_vllezlf<mode>"): New pattern.
35137 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35139 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35140 ("popcountv4si2", "popcountv2di2"): Rename to ...
35141 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35142 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35144 ("popcount<mode>2_vxe"): New pattern.
35146 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35148 * common/config/s390/s390-common.c (processor_flags_table): Add
35150 * config.gcc: Add arch12.
35151 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35152 Default to arch12 for unknown CPU model numbers.
35153 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35154 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35155 PROCESSOR_max sanity check.
35156 * config/s390/s390-opts.h (enum processor_type): Add
35158 * config/s390/s390.c (processor_table): Add arch12.
35159 (s390_expand_builtin): Add check for B_VXE flag.
35160 (s390_issue_rate): Add PROCESSOR_ARCH12.
35161 (s390_get_sched_attrmask): Likewise.
35162 (s390_get_unit_mask): Likewise.
35163 (s390_sched_score): Enable z13 scheduling for arch12.
35164 (s390_sched_reorder): Likewise.
35165 (s390_sched_variable_issue): Likewise.
35166 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35168 (s390_tune_attr): Use z13 scheduling also for arch12.
35169 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35170 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35171 (TARGET_VXE_P): New macros.
35172 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35173 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35174 * config/s390/s390.opt: Add arch12 as processor_type.
35176 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35178 * config/s390/s390.md
35179 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35180 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35181 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35183 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35184 Rename expanders to ...
35186 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35187 ("fixuns_truncdddi2_emu"): ... these.
35189 ("fixuns_trunc<mode>si2_emu"): New expander.
35191 ("*fixuns_truncdfdi2_z13"): Rename to ...
35192 ("*fixuns_truncdfdi2_vx"): ... this.
35194 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35196 * config/s390/2964.md: Remove the single element vector compare
35197 instructions which are no longer used.
35198 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35200 (s390_canonicalize_comparison): Remove handling of DFmode
35202 (s390_expand_vec_compare_scalar): Remove function.
35203 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35204 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35206 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35208 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35210 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35212 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35213 will anyway by matched by mov<mode>_64dfp.
35215 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35217 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35218 vlef/vstef. Add missing operand to vleif.
35220 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35222 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35223 pair for all vector types with 64 bit elements.
35224 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35225 * config/s390/vector.md (V_HW_64): ... here.
35226 (V_128_NOSINGLE): New mode iterator.
35227 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35228 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35229 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35230 ("*vec_load_pairv2di"): Change to ...
35231 ("*vec_load_pair<mode>"): ... this one.
35233 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35235 * config/s390/constraints.md: Add comments.
35236 (jKK): Reject element sizes > 8 bytes.
35237 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35239 * config/s390/s390.md: Add the s_operand checks formerly in
35240 s390_split_ok_p to various splitters where they are still
35242 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35243 for 128 bit vectors. Plus two splitters.
35245 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35247 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35250 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35253 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35254 error if the boundary argument is not constant.
35256 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35258 PR rtl-optimization/80112
35259 * loop-doloop.c (doloop_condition_get): Don't check condition
35260 if cmp isn't SET with IF_THEN_ELSE src.
35262 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35264 PR tree-optimization/80158
35265 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35266 replacing a candidate statement, also replace it for the
35267 candidate's alternate interpretation.
35268 (replace_rhs_if_not_dup): Likewise.
35269 (replace_one_candidate): Likewise.
35271 2017-03-24 Richard Biener <rguenther@suse.de>
35273 PR tree-optimization/80167
35274 * graphite-isl-ast-to-gimple.c
35275 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35277 (translate_isl_ast_to_gimple::get_rename): Likewise.
35279 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35281 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35282 handling of certain combinations of target options, including the
35283 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35284 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35286 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35289 * config/arm/arm.md (*load_multiple): Add reload_completed to
35290 matching condition.
35292 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35293 Richard Biener <rguenth@suse.de>
35295 PR tree-optimization/79908
35296 PR tree-optimization/80136
35297 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35298 been cast away, gimplify_and_add suffices.
35300 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35302 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35304 2017-03-23 Richard Biener <rguenther@suse.de>
35306 PR tree-optimization/80032
35307 * gimplify.c (gimple_push_cleanup): Forced unconditional
35308 cleanups still have to go to the conditional_cleanups
35311 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35313 PR tree-optimization/80072
35314 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35316 (next_operand_entry_id): Change type to unsigned int.
35317 (sort_by_operand_rank): Make sure to return the right return value
35318 even if unsigned fields are bigger than INT_MAX.
35319 (struct oecount): Change cnt and id type to unsigned int.
35320 (oecount_hasher::equal): Formatting fix.
35321 (oecount_cmp): Make sure to return the right return value
35322 even if unsigned fields are bigger than INT_MAX.
35323 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35326 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35327 TREE_READONLY on result if writing it more than once.
35330 * doc/invoke.texi (-fsanitize=thread): Document that with
35331 -fnon-call-exceptions atomics are not able to throw
35335 * tsan.c: Include tree-eh.h.
35336 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35337 maybe_clean_or_replace_eh_stmt where needed.
35338 (instrument_memory_accesses): Add cfg_changed argument.
35339 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35340 if it returned true.
35341 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35343 PR rtl-optimization/63191
35344 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35345 wrapper function, moved the whole old content into ...
35346 (ix86_delegitimize_address_1): ... this. New inline function.
35347 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35348 true as last argument instead of ix86_delegitimize_address.
35350 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35352 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35353 cortexa57_branch_cost.
35355 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35357 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35359 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35362 * doc/md.texi (Constraints): Document wA constraint.
35363 * config/rs6000/constraints.md (wA): New.
35364 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35365 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35366 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35367 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35369 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35372 * gimplify.c (is_oacc_declared): New function.
35373 (oacc_default_clause): Use it to set default flags for acc declared
35374 variables inside parallel regions.
35375 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35376 declared variables.
35377 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35378 declare attribute to any decl as necessary.
35380 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35383 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35384 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35385 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35386 (arm_arch_lpae): This.
35387 * config/arm/arm.c (arm_arch7ve): Rename into ...
35388 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35389 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35392 2017-03-22 Martin Liska <mliska@suse.cz>
35395 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
35396 error message instead of an ICE.
35398 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35400 * doc/extend.texi (6.11 Additional Floating Types): Revise.
35402 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
35404 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35406 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35409 2017-03-21 Martin Sebor <msebor@redhat.com>
35411 * doc/extend.texi: Use "cannot" instead of "can't."
35412 * doc/hostconfig.texi: Same.
35413 * doc/install.texi: Same.
35414 * doc/invoke.texi: Same.
35415 * doc/loop.texi: Same.
35416 * doc/md.texi: Same.
35417 * doc/objc.texi: Same.
35418 * doc/rtl.texi: Same.
35419 * doc/tm.texi: Same.
35420 * doc/tm.texi.in: Same.
35421 * doc/trouble.texi: Same.
35423 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
35426 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
35427 (collect_checksum_attributes): Set it.
35428 (die_checksum_ordered): Use it.
35430 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35432 PR tree-optimization/79908
35433 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
35434 change: For a VA_ARG whose LHS has been cast away, use
35435 force_gimple_operand to construct the side effects.
35437 2017-03-21 David Malcolm <dmalcolm@redhat.com>
35439 PR translation/80001
35440 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
35441 more amenable to translation.
35442 (oacc_loop_auto_partitions): Likewise.
35444 2017-03-21 Marek Polacek <polacek@redhat.com>
35445 Martin Sebor <msebor@redhat.com>
35447 PR tree-optimization/80109
35448 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
35449 on INTEGRAL_TYPE_P.
35451 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35452 Segher Boessenkool <segher@kernel.crashing.org>
35455 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
35456 check reg_used_between_p between insn and one of succ or succ2
35457 depending on if succ is artificial insn not inserted into insn
35460 2017-03-21 Martin Liska <mliska@suse.cz>
35462 PR gcov-profile/80081
35463 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
35464 * doc/gcc.texi: Include gcov-dump stuff.
35465 * doc/gcov-dump.texi: New file.
35467 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
35469 PR rtl-optimization/79150
35470 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
35471 conditional jump, if the jump is the last insn of the loop.
35473 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35474 Richard Biener <rguenth@suse.de>
35476 PR tree-optimization/79908
35477 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35478 been cast away, use force_gimple_operand to construct the side
35481 2017-03-21 Martin Liska <mliska@suse.cz>
35483 PR libfortran/79956
35484 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
35487 2017-03-21 Brad Spengler <spender@grsecurity.net>
35490 * plugin.c (htab_hash_plugin): New function.
35491 (add_new_plugin): Use it and adjust.
35492 (parse_plugin_arg_opt): Adjust.
35493 (init_one_plugin): Likewise.
35495 2017-03-21 Richard Biener <rguenther@suse.de>
35497 PR tree-optimization/80032
35498 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
35499 if set force the cleanup to happen unconditionally.
35500 (gimplify_target_expr): Push inserted clobbers with force_uncond
35501 to avoid them being removed by control-dependent DCE.
35503 2017-03-21 Richard Biener <rguenther@suse.de>
35505 PR tree-optimization/80122
35506 * tree-inline.c (copy_bb): Do not expans va-arg packs or
35507 va_arg_pack_len when the inlined call stmt requires pack
35509 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
35511 2017-03-21 Jakub Jelinek <jakub@redhat.com>
35514 * tsan.c (instrument_builtin_call): If the memory model argument
35515 is not a constant, assume it is valid.
35518 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
35521 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35523 PR rtl-optimization/79910
35524 * combine.c (can_combine_p): Do not allow combining an I0 or I1
35525 if its dest is used by an insn before I2 (other than the combined
35526 insns themselves, which are properly handled already).
35528 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
35531 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35533 * combine.c (record_used_regs): New static function.
35534 (try_combine): Handle situations where there is an additional
35535 instruction between I2 and I3 which needs to have a LOG_LINK
35539 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35541 * combine.c (try_combine): Delete redundant i1 test. Call
35542 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35544 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35547 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
35548 alternatives 13/14.
35550 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35552 PR tree-optimization/80054
35553 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
35554 the optimization if a PHI or any of its arguments is not dominated
35555 by the candidate's basis. Use gphi* rather than gimple* as
35557 (replace_profitable_candidates): Clean up a gimple* variable that
35558 should be a gphi* variable.
35560 2017-03-20 Martin Sebor <msebor@redhat.com>
35563 * doc/extend.texi (attribute constructor): Document present limitation.
35565 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
35568 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
35569 __POWER9_VECTOR__ #ifdef control, change template definition to
35570 use Power9-specific built-in function.
35571 (vec_any_eq): Likewise.
35572 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
35573 to control outcomes from this test.
35574 (vector_ae_<mode>p): For VEC_F modes, likewise.
35576 2017-03-20 Ian Lance Taylor <iant@google.com>
35578 * config/i386/i386.c (ix86_function_regparm): Save an extra
35579 register for -fsplit-stack with DECL_STATIC_CHAIN.
35581 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35584 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
35585 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
35587 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
35589 * config/riscv/riscv.c (riscv_print_operand): Use "fence
35591 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
35594 2017-03-20 Marek Polacek <polacek@redhat.com>
35597 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
35599 2017-03-20 Richard Biener <rguenther@suse.de>
35601 PR tree-optimization/80113
35602 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
35603 allocate extra SSA name for PHI def.
35604 (add_close_phis_to_outer_loops): Likewise.
35605 (add_close_phis_to_merge_points): Likewise.
35606 (copy_loop_close_phi_args): Likewise.
35607 (copy_cond_phi_nodes): Likewise.
35609 2017-03-20 Martin Liska <mliska@suse.cz>
35611 PR middle-end/79753
35612 * tree-chkp.c (chkp_build_returned_bound): Do not build
35613 returned bounds for a LHS that's not a BOUNDED_P type.
35615 2017-03-20 Martin Liska <mliska@suse.cz>
35619 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
35620 COMPLEX_CST and VECTOR_CST.
35622 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35625 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
35626 target operand. A new splitter adds the clobber statement in case
35627 the target operand is dead anyway.
35629 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
35631 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
35632 to age-old versions of binutils and glibc.
35634 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
35636 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
35638 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35640 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
35642 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
35644 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
35645 requirement for binutils 2.13.
35647 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
35649 * combine.c (try_combine): Delete redundant i1 test. Call
35650 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
35652 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
35654 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
35655 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
35657 <riscv64-*-elf>: Re-arrange section
35658 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
35659 <riscv32-*-linux>: Likewise.
35660 <riscv64-*-elf>: Likewise
35661 <riscv64-*-linux>: Likewise.
35663 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
35666 * aarch64.opt(verbose-cost-dump): Fix typo.
35668 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
35671 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
35672 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
35674 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
35676 * reload.c (find_reloads): When reloading a nonoffsettable address,
35677 use RELOAD_OTHER for it and its address reloads.
35679 PR rtl-optimization/79910
35680 * combine.c (record_used_regs): New static function.
35681 (try_combine): Handle situations where there is an additional
35682 instruction between I2 and I3 which needs to have a LOG_LINK
35685 2017-03-17 Jeff Law <law@redhat.com>
35687 PR tree-optimization/71437
35688 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
35689 conditional in the hash table first.
35690 (vrp_dom_walker::before_dom_children): Extract condition from
35691 ASSERT_EXPR. Record condition, its inverion and any implied
35692 conditions as well.
35694 2017-03-17 Marek Polacek <polacek@redhat.com>
35695 Markus Trippelsdorf <markus@trippelsdorf.de>
35697 PR tree-optimization/80079
35698 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
35701 2017-03-17 Richard Biener <rguenther@suse.de>
35703 PR middle-end/80075
35704 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
35705 Properly verify the LHS before the RHS possibly claims to be
35707 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
35710 2017-03-17 Martin Jambor <mjambor@suse.cz>
35712 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
35713 (List of -O2 options): Likewise.
35714 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
35717 2017-03-17 Tom de Vries <tom@codesourcery.com>
35719 * gcov-dump.c (print_usage): Print bug_report_url.
35721 2017-03-17 Richard Biener <rguenther@suse.de>
35723 PR middle-end/80050
35724 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
35725 (parser::peek): Likewise.
35727 2017-03-17 Richard Biener <rguenther@suse.de>
35729 PR tree-optimization/80048
35730 * sese.c (free_sese_info): Properly release rename_map and
35731 copied_bb_map elements.
35733 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
35735 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
35736 Add linked-list forward and backlinks. Insert on
35737 construction, remove on destruction.
35738 (class pass_store_merging): Add m_stores_head field.
35739 (pass_store_merging::terminate_and_process_all_chains):
35740 Iterate over m_stores_head list.
35741 (pass_store_merging::terminate_all_aliasing_chains):
35743 (pass_store_merging::execute): Check for debug stmts first.
35744 Push new chains onto the m_stores_head stack.
35746 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
35749 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
35750 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
35751 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
35753 2017-03-16 Jeff Law <law@redhat.com>
35755 PR tree-optimization/71437
35756 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
35757 member function. Implementation moved into after_dom_children
35758 member function and into the threader's thread_outgoing_edges
35760 (dom_opt_dom_walker::after_dom_children): Simplify by moving
35761 some code into new thread_outgoing_edges.
35762 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
35763 definition. Simplify marker handling (do it here). Assume we always
35764 have the available expression and the const/copies tables.
35765 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
35767 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
35768 * tree-vrp.c (equiv_stack): No longer file scoped.
35769 (vrp_dom_walker): New class.
35770 (vrp_dom_walker::before_dom_children): New member function.
35771 (vrp_dom_walker::after_dom_children): Likewise.
35772 (identify_jump_threads): Setup domwalker. Use it rather than
35773 walking edges in a random order by hand. Simplify setup/finalization.
35774 (finalize_jump_threads): Remove.
35775 (vrp_finalize): Do not call identify_jump_threads here.
35776 (execute_vrp): Do it here instead and call thread_through_all_blocks
35779 PR tree-optimization/71437
35780 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
35782 (simplify_stmt_for_jump_threading): Add basic_block argument. All
35784 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
35785 (dom_opt_dom_walker::thread_across_edge): Remove
35786 handle_dominating_asserts argument. All callers changed.
35787 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
35788 changes. Remove calls to lhs_of_dominating_assert. Other
35789 uses of handle_dominating_asserts turn into unconditional code
35790 (simplify_control_stmt_condition_1): Likewise.
35791 (simplify_control_stmt_condition): Likewise.
35792 (thread_through_normal_block, thread_across_edge): Likewise.
35793 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
35794 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
35795 object if it is not an SSA_NAME.
35796 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
35797 before calling into the VRP specific simplifiers.
35798 (identify_jump_threads): Remove handle_dominating_asserts
35801 2017-03-16 Jakub Jelinek <jakub@redhat.com>
35804 * tree-diagnostic.c (default_tree_printer): No longer static.
35805 * tree-diagnostic.h (default_tree_printer): New prototype.
35807 2017-03-16 Tamar Christina <tamar.christina@arm.com>
35809 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
35810 Change ins into fmov.
35812 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35814 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
35815 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
35816 Use h_con constraint for operand 1.
35817 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
35818 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
35820 2017-03-15 Jeff Law <law@redhat.com>
35822 PR tree-optimization/71437
35823 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
35824 (record_temporary_equivalences): Use it.
35826 PR tree-optimization/71437
35827 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
35828 tree-ssa-scopedtables.
35829 (lookup_avail_expr, build_and_record_new_cond): Likewise.
35830 (record_conditions, record_cond, vuse_eq): Likewise.
35831 (record_edge_info): Adjust to API tweak of record_conditions.
35832 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
35833 (record_temporary_equivalences, optimize_stmt): Likewise.
35834 (eliminate_redundant_computations): Likewise.
35835 (record_equivalences_from_stmt): Likewise.
35836 * tree-ssa-scopedtables.c: Include options.h and params.h.
35837 (vuse_eq): New function, moved from tree-ssa-dom.c
35838 (build_and_record_new_cond): Likewise.
35839 (record_conditions): Likewise. Accept vector of conditions rather
35840 than edge_equivalence structure for first argument.
35841 for the first argument.
35842 (avail_exprs_stack::lookup_avail_expr): New member function, moved
35843 from tree-ssa-dom.c.
35844 (avail_exprs_stack::record_cond): Likewise.
35845 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
35846 from tree-ssa-dom.c.
35847 (avail_exprs_stack): Add new member functions lookup_avail_expr
35849 (record_conditions): Declare.
35851 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
35854 * lra-constraints.c (process_alt_operands): Increase reject for
35855 reloading an input/output operand.
35857 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
35860 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
35861 insns to convert from signed/unsigned char/short to IEEE 128-bit
35863 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
35865 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
35868 * config/i386/i386.c (ix86_vector_duplicate_value): Create
35869 subreg of inner mode for values already in registers.
35871 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
35873 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
35874 iteration reg is used after the loop.
35876 2017-03-14 Martin Sebor <msebor@redhat.com>
35878 PR tree-optimization/79800
35879 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
35880 precision in negative-positive range.
35881 (format_floating): Call non-const overload with adjusted precision.
35883 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
35886 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
35889 2017-03-14 Martin Sebor <msebor@redhat.com>
35891 PR middle-end/80020
35892 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
35893 * builtins.def (aligned_alloc): Use it.
35896 * Makefile.in (GTFILES): Add calls.c.
35897 * calls.c: Include "gt-calls.h".
35899 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
35901 PR rtl-optimization/79728
35902 * regs.h (struct target_regs): New field
35903 x_contains_allocatable_regs_of_mode.
35904 (contains_allocatable_regs_of_mode): New macro.
35905 * reginfo.c (init_reg_sets_1): Initialize it, and change
35906 contains_reg_of_mode so it includes global regs as well.
35907 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
35908 rather than contains_regs_of_mode.
35910 2017-03-14 Martin Liska <mliska@suse.cz>
35912 * doc/invoke.texi: Document options that can't be combined with
35913 -fcheck-pointer-bounds.
35915 2017-03-14 Martin Liska <mliska@suse.cz>
35917 PR middle-end/79831
35918 * doc/invoke.texi (-Wchkp): Document the option.
35920 2017-03-14 Martin Liska <mliska@suse.cz>
35922 * Makefile.in: Install gcov-dump.
35924 2017-03-14 Martin Liska <mliska@suse.cz>
35926 * multiple_target.c (expand_target_clones): Bail out for
35927 an invalid attribute.
35929 2017-03-14 Richard Biener <rguenther@suse.de>
35931 * alias.c (struct alias_set_entry): Pack properly.
35932 * cfgloop.h (struct loop): Likewise.
35933 * cse.c (struct set): Likewise.
35934 * ipa-utils.c (struct searchc_env): Likewise.
35935 * loop-invariant.c (struct invariant): Likewise.
35936 * lra-remat.c (struct cand): Likewise.
35937 * recog.c (struct change_t): Likewise.
35938 * rtl.h (struct address_info): Likewise.
35939 * symbol-summary.h (function_summary): Likewise.
35940 * tree-loop-distribution.c (struct partition): Likewise.
35941 * tree-object-size.c (struct object_size_info): Likewise.
35942 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
35943 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
35944 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
35945 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
35946 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
35947 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
35948 (struct _stmt_vec_info): Likewise.
35950 2017-03-14 Martin Liska <mliska@suse.cz>
35953 * multiple_target.c (create_dispatcher_calls): Check that
35954 a target can create a function dispatcher.
35956 2017-03-14 Martin Liska <mliska@suse.cz>
35959 * multiple_target.c (expand_target_clones): Drop local.local
35960 flag for default implementation.
35962 2017-03-14 Richard Biener <rguenther@suse.de>
35964 PR tree-optimization/80030
35965 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
35967 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
35969 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
35970 gcc_fallthrough() instead of __attribute__((fallthrough));
35972 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35974 * doc/gcc.texi: Remove "up" link to (DIR).
35975 * doc/gccint.texi: Ditto.
35977 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
35979 * doc/install.texi (Specific) <avr>: Remove reference to
35982 2017-03-13 Jeff Law <law@redhat.com>
35984 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
35985 attribute rather than comments.
35987 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
35988 match_scratch operand is highest.
35990 2017-03-13 Martin Liska <mliska@suse.cz>
35992 PR middle-end/78339
35993 * ipa-pure-const.c (warn_function_noreturn): If the declarations
35994 is a CHKP clone, use original declaration.
35996 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
35998 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
35999 (arc_conditional_register_usage): Use a different allocation order
36000 when optimizing for size.
36001 * common/config/arc/arc-common.c (arc_option_optimization_table):
36002 Section anchors default on when optimizing for size.
36004 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36006 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36008 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36010 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36011 * config/arc/arc.md (cpu_facility): Add cd variant.
36012 (*movqi_insn): Add code density variant.
36013 (*movhi_insn): Likewise.
36014 (*movqi_insn): Likewise.
36015 (*addsi3_mixed): Likewise.
36016 (subsi3_insn): Likewise.
36018 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36020 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36022 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36024 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36025 expressions with MINUS and UNARY ops.
36027 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36030 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36032 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36033 between vec_select and vector argument.
36034 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36035 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36036 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36037 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36038 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36039 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36041 2017-03-13 Richard Biener <rguenther@suse.de>
36044 * params.def (vect-max-peeling-for-alignment): Fix typo.
36046 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36048 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36049 issue that only occurred with binutils below 2.18.
36051 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36053 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36054 refer to binutils 2.11/2.12 minimum.
36056 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36058 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36059 ftp.kernel.org and simplify binutils requirement.
36061 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36063 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36065 (Optimize Options): Ditto. Also remove redundancy.
36067 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36069 PR translation/79848
36070 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36072 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36073 to G_ to avoid double translation.
36075 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36077 PR translation/79923
36078 * auto-profile.c (get_combined_location): Convert leading
36079 character of diagnostics to lower case and remove trailing period.
36080 (read_profile): Likewise for various diagnostics.
36081 * config/arm/arm.c (arm_option_override): Remove trailing period
36082 from various diagnostics.
36083 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36084 (msp430_expand_delay_cycles): Likewise.
36086 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36089 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36090 full command-line argument, rather than just "str".
36091 (aarch64_validate_march): Likewise.
36092 (aarch64_validate_mtune): Likewise.
36094 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36096 PR rtl-optimization/78911
36097 * lra-assigns.c (must_not_spill_p): New function.
36098 (spill_for): Use it.
36100 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36102 PR tree-optimization/79981
36103 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36104 ATOMIC_COMPARE_EXCHANGE ifn result.
36105 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36106 IFN_ATOMIC_COMPARE_EXCHANGE.
36108 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36111 * opts.c (parse_sanitizer_options): Add missing question mark to
36112 "did you mean" message.
36114 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36116 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36118 (VMULEUH_UNS): Likewise.
36119 (VMULOUB_UNS): Likewise.
36120 (VMULOUH_UNS): Likewise.
36121 * config/rs6000/rs6000.c (builtin_function_type): Remove
36122 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36124 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36127 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36128 x with result of extra_parsing_for_operand_code_0.
36129 (function_reader::extra_parsing_for_operand_code_0): Convert
36130 return type from void to rtx, returning x. When reading
36131 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36132 larger size containing struct block_symbol.
36134 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36136 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36137 -mfloat128-hardware without -m64.
36139 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36142 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36143 entries to the case statement that marks unsigned arguments to
36144 overloaded functions.
36146 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36148 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36149 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36151 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36154 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36155 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36157 2017-03-10 Martin Liska <mliska@suse.cz>
36161 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36162 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36165 2017-03-10 Olivier Hainque <hainque@adacore.com>
36167 * tree-switch-conversion (array_value_type): Start by resetting
36168 candidate type to it's main variant.
36170 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36172 PR rtl-optimization/79909
36173 * combine.c (try_combine): Use simplify_replace_rtx on individual
36174 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36175 of the whole CALL_INSN_FUNCTION_USAGE.
36177 PR tree-optimization/79972
36178 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36179 get_range_info on SSA_NAMEs. Formatting fixes.
36181 2017-03-10 Richard Biener <rguenther@suse.de>
36182 Jakub Jelinek <jakub@redhat.com>
36184 PR tree-optimization/77975
36185 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36186 edge to be constant.
36187 (get_val_for): For constant x return it. Formatting fix.
36188 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36189 would use the same bases as the current one.
36191 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36193 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36194 instead of vec_select for V1TImode.
36195 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36197 (VSX_LE_128): Add V1TI to this mode iterator.
36198 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36199 (*vsx_le_perm_store_<mode>): Likewise.
36200 (pre-reload splitter for VSX stores): Likewise.
36201 (post-reload splitter for VSX stores): Likewise.
36202 (*vsx_xxpermdi2_le_<mode>): Likewise.
36203 (*vsx_lxvd2x2_le_<mode>): Likewise.
36204 (*vsx_stxvd2x2_le_<mode>): Likewise.
36206 2017-03-09 Michael Eager <eager@eagercon.com>
36208 Correct failures with --enable-checking=yes,rtl.
36210 * config/microblaze/microblaze.c (microblaze_expand_shift):
36211 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36212 test for const0_rtx.
36213 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36214 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36216 2017-03-09 Richard Biener <rguenther@suse.de>
36218 PR tree-optimization/79977
36219 * graphite-scop-detection.c (scop_detection::merge_sese):
36220 Handle the case of extra exits to blocks dominating the entry.
36222 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36224 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36227 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36229 PR rtl-optimization/79949
36230 * lra-constraints.c (process_alt_operands): Check memory when
36231 trying to predict a cycle. Print about the overall increase.
36233 2017-03-09 Richard Biener <rguenther@suse.de>
36235 PR middle-end/79971
36236 * gimple-expr.c (useless_type_conversion_p): Preserve
36237 TYPE_SATURATING for fixed-point types.
36239 2017-03-09 Richard Biener <rguenther@suse.de>
36242 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36243 alignment of BLKmode params.
36245 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36248 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36249 (VALL_NO_V2Q): Likewise.
36251 * config/aarch64/aarch64-simd.md
36252 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36254 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36255 VALL_NO_V2Q mode iterator.
36256 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36258 2017-03-09 Martin Liska <mliska@suse.cz>
36260 PR tree-optimization/79631
36261 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36262 tree_int_cst_sign_bit just for INTEGER constants.
36264 2017-03-09 Martin Liska <mliska@suse.cz>
36268 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36271 2017-03-09 Marek Polacek <polacek@redhat.com>
36274 * tree.c (inchash::add_expr): Handle TREE_VEC.
36276 2017-03-09 Martin Liska <mliska@suse.cz>
36279 (chkp_narrow_size_and_offset): New function.
36280 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36281 (void chkp_parse_bit_field_ref): New function.
36282 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36283 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36285 2017-03-09 Martin Liska <mliska@suse.cz>
36288 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36289 (chkp_find_bounds_1): Remove gcc_unreachable.
36291 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36294 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36295 BUILT_IN_SYNC*, determine the access type from the size suffix and
36296 always build a MEM_REF with that type. Handle forgotten
36297 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36300 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36301 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36302 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36303 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36304 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36305 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36306 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36307 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36308 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36309 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36310 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36311 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36312 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36313 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36314 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36315 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36316 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36317 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36318 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36319 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36320 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36321 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36322 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36323 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36324 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36325 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36326 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36327 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36328 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36329 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36330 definitions outside of __OPTIMIZE__ guarded section.
36333 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36334 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36335 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36336 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36339 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36341 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36342 ("vfenez<mode>"): Add missing constraints.
36344 2017-03-08 Martin Sebor <msebor@redhat.com>
36347 * config/nds32/nds32.c (nds32_option_override):
36348 Fix misspelled diagnostic.
36350 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36353 * gimplify.c (gimplify_omp_for): Replace index var in outer
36354 taskloop statement with an artificial variable and add
36355 OMP_CLAUSE_PRIVATE clause for it.
36357 2017-03-08 Richard Biener <rguenther@suse.de>
36359 PR tree-optimization/79955
36360 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36361 for accesses that are completely outside of the variable.
36363 2017-03-08 Andrew Haley <aph@redhat.com>
36365 PR tree-optimization/79943
36366 * tree-ssa-loop-split.c (compute_new_first_bound): When
36367 calculating the new upper bound, (END-BEG) should be added, not
36370 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36372 * config/avr/avr.md (setmemhi): Make sure match_dup
36373 operand number comes before match_scratch.
36375 2017-03-08 Richard Biener <rguenther@suse.de>
36377 PR tree-optimization/79920
36378 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36379 with ncopies == 1 to ...
36380 (vect_transform_slp_perm_load): ... here. Properly compute
36381 all element loads by iterating VF times over the group. Do
36382 not handle ncopies (computed in a broken way) in
36383 vect_create_mask_and_perm.
36385 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36388 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36389 is a uniform vector, use uniform_vector_p return value instead of
36390 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36392 2017-03-07 Marek Polacek <polacek@redhat.com>
36394 PR middle-end/79809
36395 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
36396 (alloca_call_type): Likewise.
36398 2017-03-07 Martin Liska <mliska@suse.cz>
36400 * gcov.c (process_args): Put comment to correct location.
36402 2017-03-07 Martin Liska <mliska@suse.cz>
36404 PR middle-end/68270
36405 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
36406 Use array_at_struct_end_p instead of DECL_CHAIN (field).
36407 (chkp_narrow_bounds_for_field): Likewise.
36408 (chkp_parse_array_and_component_ref): Pass one more argument to
36411 2017-03-07 Richard Biener <rguenther@suse.de>
36413 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
36416 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
36418 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
36419 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
36421 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36424 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
36425 to end of description.
36426 (PARAM_MAX_STORES_TO_MERGE): Likewise.
36428 2017-03-07 Jakub Jelinek <jakub@redhat.com>
36430 PR rtl-optimization/79901
36431 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
36433 (*avx512f_<code><mode>3<mask_name>): ... this.
36434 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
36435 iterator instead of VI8_AVX2_AVX512BW.
36437 PR rtl-optimization/79901
36438 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
36439 min/max expander, expand it using expand_vec_cond_expr.
36442 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
36445 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36448 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
36449 to void * for PCH reasons.
36450 * dwarf2out.c (output_loc_operands, output_die): Cast
36451 v.val_vec.array to unsigned char *.
36453 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
36456 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
36459 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
36461 PR rtl-optimization/79571
36462 * lra-constraints.c (process_alt_operands): Calculate static
36463 reject and subtract it from overall when only addresses will be
36466 2017-03-06 Julia Koval <julia.koval@intel.com>
36469 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
36470 incoming stack boundary to 128 for 64-bit targets.
36472 2017-03-06 Richard Biener <rguenther@suse.de>
36474 PR tree-optimization/79894
36475 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
36476 to NULL after folding it.
36478 2017-03-06 Richard Biener <rguenther@suse.de>
36480 PR tree-optimization/79824
36481 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
36482 check disabling peeling for gaps.
36484 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
36486 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
36487 attributes): Document gettimeofday.
36489 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36491 * config/s390/s390.c (s390_option_override_internal): Set
36492 PARAM_MIN_VECT_LOOP_BOUND
36494 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
36496 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
36497 * config/s390/s390.md: Likewise.
36499 2017-03-06 Jakub Jelinek <jakub@redhat.com>
36502 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
36503 (<avx2_avx512>_perm<mode>): Rename to ...
36504 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
36506 (<avx512>_perm<mode>_mask): Rename to ...
36507 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
36509 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
36510 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
36511 instead of VI8F_256_512.
36512 (avx512f_perm<mode>): New define_expand.
36513 (avx512f_perm<mode>_mask): Likewise.
36514 (avx512f_perm<mode>_1<mask_name>): New define_insn.
36515 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
36517 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36519 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
36520 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
36522 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
36524 2017-03-06 Martin Liska <mliska@suse.cz>
36527 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
36528 when having a SSA NAME w/o VAR_DECL assigned to it.
36530 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36532 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
36533 msa_dpsub_<su>_d): Fix MODE for vec_select.
36535 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
36537 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
36539 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
36541 2017-03-06 Richard Biener <rguenther@suse.de>
36543 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
36544 * plugin.c (register_plugin_info): Likewise.
36545 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
36547 2017-03-05 Jakub Jelinek <jakub@redhat.com>
36549 * config/i386/sse.md (sse_storehps, sse_storelps,
36550 avx_<castmode><avxsizesuffix>_<castmode>,
36551 avx512f_<castmode><avxsizesuffix>_<castmode>,
36552 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
36553 in condition that at least one operand is not a MEM.
36555 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36557 PR middle-end/79805
36558 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
36559 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
36561 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
36562 gimple_call_nothrow_p flag based on whether original builtin can throw.
36563 If it can, emit following stmts on the fallthrough edge.
36564 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
36565 don't create new bb if inserting just debug stmts on the edge, try to
36566 insert them on the fallthru bb or just reset debug stmts.
36568 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
36571 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
36572 restore recog_data (including the operand rtxes inside it) around
36573 the call to get_insn_template.
36575 2017-03-03 Martin Sebor <msebor@redhat.com>
36577 PR tree-optimization/79699
36578 * context.c (context::~context): Free MPFR caches to avoid
36579 a memory leak on program exit.
36581 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36583 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
36584 Use wide_int::ulow () instead of .elt (0).
36586 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36588 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
36589 (*pushxf): Limit oF constraint to 32bit targets and add oC
36590 constraint for 64bit targets.
36591 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
36592 (*pushdf): Change rmF constraint to rmC.
36594 2017-03-03 Martin Liska <mliska@suse.cz>
36596 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
36597 Remove unused variable.
36599 2017-03-03 Jakub Jelinek <jakub@redhat.com>
36602 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
36603 is a memory operand, increase num_memory.
36604 (ix86_expand_args_builtin): Likewise.
36606 2017-03-03 Jan Hubicka <jh@suse.cz>
36609 * ipa-devirt.c (maybe_record_node): Properly handle
36610 __cxa_pure_virtual visibility.
36612 2017-03-03 Martin Liska <mliska@suse.cz>
36614 PR tree-optimization/79803
36615 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
36617 (pass_loop_prefetch::execute): Disabled optimization if an
36618 assumption about L1 cache size is not met.
36620 2017-03-03 Martin Liska <mliska@suse.cz>
36622 PR rtl-optimization/79574
36623 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
36624 (hash_scan_set): Likewise.
36625 (dump_hash_table): Likewise.
36626 (hoist_code): Likewise.
36628 2017-03-03 Richard Biener <rguenther@suse.de>
36630 * fixed-value.c (fixed_from_string): Restore use of elt (1)
36631 in place of uhigh ().
36632 (fixed_convert_from_real): Likewise.
36634 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
36637 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
36639 2017-03-03 Richard Biener <rguenther@suse.de>
36641 PR middle-end/79818
36642 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
36643 TYPE_OVERFLOW_UNDEFINED check.
36645 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36647 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
36649 (vector_ae_<mode>_p): Likewise.
36650 (vector_nez_<mode>_p): Likewise.
36651 (vector_ne_v2di_p): Likewise.
36652 (vector_ae_v2di_p): Likewise.
36653 (vector_ne_<mode>_p): Likewise.
36654 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
36656 (vsx_tsqrt<mode>2_fe): Likewise.
36658 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
36661 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
36663 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36665 PR rtl-optimization/79780
36666 * cprop.c (one_cprop_pass): When second and further conditional trap
36667 in a single basic block is turned into an unconditional trap, turn it
36668 into a deleted note to avoid RTL verification failures.
36670 2017-03-02 Richard Biener <rguenther@suse.de>
36672 * fold-const.c (const_binop): Use ulow () instead of elt (0).
36674 2017-03-02 Richard Biener <rguenther@suse.de>
36676 PR tree-optimization/79345
36678 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
36679 param and abort the walk, returning -1 if it is hit.
36680 (walk_aliased_vdefs): Take a limit param and pass it on.
36681 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
36682 defaulting to 0 and return a signed int.
36683 * tree-ssa-uninit.c (struct check_defs_data): New struct.
36684 (check_defs): New helper.
36685 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
36686 about uninitialized memory.
36687 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
36688 bogus uninitialized warning.
36689 (fixed_convert_from_real): Likewise.
36691 2017-03-02 Bin Cheng <bin.cheng@arm.com>
36693 PR tree-optimization/66768
36694 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
36695 iv_use if base object can't be determined.
36697 2017-03-02 Jakub Jelinek <jakub@redhat.com>
36699 PR tree-optimization/79345
36700 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
36701 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
36702 (get_pattern_stats): Initialize it.
36703 * genemit.c (gen_expand): Verify match_scratch numbers come after
36704 match_operand/match_dup numbers.
36705 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
36706 match_scratch numbers.
36707 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
36709 * config/s390/s390.md (trunctdsd2): Likewise.
36711 2017-03-02 Richard Biener <rguenther@suse.de>
36713 * wide-int.h (wide_int_storage::operator=): Implement in terms
36716 2017-03-02 Richard Biener <rguenther@suse.de>
36718 PR tree-optimization/79777
36719 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
36720 the to insert expression to sth existing.
36722 2017-03-01 Martin Sebor <msebor@redhat.com>
36724 PR middle-end/79692
36725 * gimple-ssa-sprintf.c
36726 (directive::known_width_and_precision): New function.
36727 (format_integer): Use it.
36728 (get_mpfr_format_length): Consider the full range of precision
36729 when computing %g output with the # flag. Set the likely byte
36730 count to 3 rather than 1 when precision is indeterminate.
36731 (format_floating): Correct the lower bound of precision.
36733 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36735 * doc/invoke.texi: Document default code model for 64-bit Linux.
36737 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36740 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
36741 udiv rather than div since input pattern is unsigned.
36743 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
36745 * config/i386/i386.c (print_reg): Warn for values of
36746 unsupported size in integer register.
36748 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
36751 * config/rs6000/predicates.md (current_file_function_operand): Do
36752 not allow self calls to be local if the function is replaceable.
36754 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
36757 * config/rs6000/altivec.h (vec_ctz and others): Change the
36758 preprocessor macro that controls conditional compilation from
36759 _ARCH_PWR9 to __POWER9_VECTOR__.
36760 (vec_all_ne): Change parameterization of __altivec_scalar_pred
36761 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
36762 control (instead of _ARCH_PWR9 control) so that template
36763 definition uses power9-specific function.
36764 (vec_any_eq): Likewise.
36765 (vec_all_ne): Change macro definition to use a power9-specific
36766 expansion under #ifdef __POWER9_VECTOR__ control (instead of
36767 _ARCH_PWR9 control).
36768 (vec_any_eq) Likewise.
36769 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
36770 expansion for CMPNEF to remove support for xvcmpnesp instruction.
36771 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
36772 support for xvcmpnedp instruction.
36773 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
36774 macro expansion so that Power9 implementation of vec_all_ne does
36775 not use the AltiVec predicate framework.
36776 (VCMPNEH_P): Likewise.
36777 (VCMPNEW_P): Likewise.
36778 (VCMPNED_P): Likewise.
36779 (VCMPNEFP_P): Likewise.
36780 (VCMPNEDP_P): Likewise.
36781 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
36782 implementation of vec_any_eq to not use AltiVec predicate
36784 (VCMPAEH_P): Likewise.
36785 (VCMPAEW_P): Likewise.
36786 (VCMPAED_P): Likewise.
36787 (VCMPAEFP_P): Likewise.
36788 (VCMPAEDP_P): Likewise.
36789 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
36790 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
36791 not use the AltiVec predicate framework.
36792 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
36793 of vec_any_eq to not use AltiVec predicate framework.
36794 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36795 support for predefined __POWER9_VECTOR__ macro to indicate that
36796 Power9 instruction selection is enabled.
36797 (altivec_overloaded_builtins): Remove extraneous
36798 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
36799 function argument types RS6000_BTI_bool_V16QI and
36800 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
36801 entry for overloaded function argument types RS6000_BTI_bool_V4SI
36802 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
36803 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
36804 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
36805 Power9 for implementations of vec_cmpne. Change the signature for
36806 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
36807 (representing vec_all_ne) to remove the previously described first
36808 argument of type RS6000_BTI_INTSI, as this was an artifact of
36809 reliance on the AltiVec predicate framework, which is no longer
36810 used in the implementation of these functions. Add
36811 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
36812 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
36813 since, unlike the AltiVec predicate framework implementation, we
36814 do not share function descriptors between vec_alle and vec_anyeq.
36815 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
36816 set of modes that receive special treatment even when
36817 TARGET_P9_VECTOR is true. The special treatment emits code that
36818 does not depend on Power9 instructions.
36819 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
36820 define_expand to not rely on AltiVec predicate framework.
36821 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36823 (vector_ne_v2di_p): Change this define_expand to not rely on
36824 AltiVec predicate framework.
36825 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
36827 (vector_ne_<mode>_p): Change this define_expand to not rely on
36828 AltiVec predicate framework.
36829 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
36831 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
36832 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
36833 define_insn pattern.
36834 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
36835 define_insn pattern because the xvcmpne<VSs>. instruction is not
36837 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
36838 instruction is not supported.
36840 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36842 * config/nvptx/nvptx.c: Include intl.h.
36844 2017-03-01 Martin Jambor <mjambor@suse.cz>
36847 * ipa-prop.h (ipa_bits): Removed field known.
36848 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
36849 to pointers. Adjusted their comments to warn about their sharing.
36850 (ipcp_transformation_summary): Change bits to a vector of pointers.
36851 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
36852 (ipa_get_ipa_bits_for_value): Declare.
36853 * tree-vrp.h (value_range): Mark as GTY((for_user)).
36854 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
36855 (ipa_bits_hash_table): Likewise.
36856 (ipa_vr_ggc_hash_traits): Likewise.
36857 (ipa_vr_hash_table): Likewise.
36858 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
36859 being pointers and vr_known being removed.
36860 (ipa_set_jf_unknown): Likewise.
36861 (ipa_get_ipa_bits_for_value): New function.
36862 (ipa_set_jfunc_bits): Likewise.
36863 (ipa_get_value_range): New overloaded functions.
36864 (ipa_set_jfunc_vr): Likewise.
36865 (ipa_compute_jump_functions_for_edge): Use the above functions to
36866 construct bits and vr parts of jump functions.
36867 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
36868 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36870 (ipcp_grow_transformations_if_necessary): Also allocate
36871 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
36873 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
36874 them. Fix too long lines.
36875 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
36876 vr_known being removed.
36877 (ipa_read_jump_function): Use new setter functions to construct bits
36878 and vr parts of jump functions or set them to NULL.
36879 (write_ipcp_transformation_info): Adjust for bits being pointers.
36880 (read_ipcp_transformation_info): Likewise.
36881 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
36883 Include gt-ipa-prop.h.
36884 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
36886 (ipcp_store_bits_results): Likewise.
36887 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
36888 Do not write to existing jump functions but use a temporary instead.
36890 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36893 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
36894 attempt to use its first operand as BIT_FIELD_REF base.
36896 2017-03-01 Richard Biener <rguenther@suse.de>
36898 PR middle-end/79721
36899 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
36900 interpolating formula in wrapping arithmetic.
36901 (chrec_apply): Convert chrec_evaluate return value to wanted type.
36903 2017-03-01 Jakub Jelinek <jakub@redhat.com>
36905 PR tree-optimization/79734
36906 * tree-vect-generic.c (expand_vector_condition): Optimize
36907 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
36908 Handle VEC_COND_EXPR where comparison has different inner width from
36909 type's inner width.
36911 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
36913 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
36914 markup, and similar issues. Remove @opindex entries for things
36915 that aren't options. Add missing -mmpy-option entries.
36917 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36919 PR tree-optimization/79737
36920 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
36921 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
36922 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
36923 instead of byte_size. Formatting fix.
36924 (shift_bytes_in_array_right): Formatting fix.
36926 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
36929 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
36930 condition on optimize for the leaf function test.
36932 2017-02-28 Martin Liska <mliska@suse.cz>
36935 * read-rtl-function.c (function_reader::handle_unknown_directive):
36936 Bail out when one uses -flto.
36938 2017-02-28 Martin Liska <mliska@suse.cz>
36940 * common.opt: Replace space with tabular for options of <number>
36942 * config/i386/i386.opt: Show <number> value for
36943 -mlarge-data-threshold.
36944 * opts.c (print_filtered_help): Do not display number in hexadecimal
36947 2017-02-28 Martin Liska <mliska@suse.cz>
36949 * common.opt: Fix --help=option -Q for options which are of
36952 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
36954 * config/i386/i386.c (print_reg): Error out for values
36955 of 8-bit size in invalid integer register.
36957 2017-02-28 Martin Sebor <msebor@redhat.com>
36959 PR tree-optimization/79691
36960 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
36962 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36965 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
36966 gcc_unreachable with output_operand_lossage.
36968 2017-02-28 Richard Biener <rguenther@suse.de>
36970 PR tree-optimization/79740
36971 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
36973 (visit_nary_op): Insert the nary into the hashtable if we
36974 pattern-matched sth.
36975 * tree-ssa-pre.c (eliminate_insert): Robustify.
36977 2017-02-28 Richard Biener <rguenther@suse.de>
36979 PR middle-end/79731
36980 * fold-const.c (decode_field_reference): Reject out-of-bound
36983 2017-02-28 Jakub Jelinek <jakub@redhat.com>
36985 * config/i386/i386.c: Include intl.h.
36986 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
36987 instead of just cond ? "..." : "...".
36988 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
36989 * coverage.c (read_counts_file): Likewise.
36990 * omp-offload.c: Include intl.h.
36991 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
36992 of just cond ? "..." : "...".
36993 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
36994 of just cond ? "..." : "...".
36996 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
36999 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37001 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37002 'tune for' CPU name.
37003 * config/arm/arm-cpu-data.h: Regenerated.
37005 2017-02-28 Richard Biener <rguenther@suse.de>
37007 PR tree-optimization/79732
37008 * tree-inline.c (expand_call_inline): Do not shadow var.
37010 2017-02-28 Richard Biener <rguenther@suse.de>
37012 PR tree-optimization/79723
37013 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37014 address-space properly.
37016 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37018 * doc/optinfo.texi (Optimization groups): Fix option used for
37020 * doc/invoke.texi (-fopt-info): Document "omp".
37021 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37022 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37023 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37024 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37025 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37027 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37029 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37030 OPTGROUP_OMP with "omp".
37032 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37035 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37036 for arithmetic shift of unsigned V2DI.
37038 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37040 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37041 arc/linux.h headers.
37042 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37043 (LINK_SPEC): Likewise.
37044 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37045 (EXTRA_SPECS): Likewise.
37046 (STARTFILE_SPEC): Likewise.
37047 (ENDFILE_SPEC): Likewise.
37048 (LIB_SPEC): Likewise.
37049 (TARGET_SDATA_DEFAULT): Likewise.
37050 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37051 (MULTILIB_DEFAULTS): Likewise.
37052 (DWARF2_UNWIND_INFO): Likewise.
37053 * config/arc/big.h: New file.
37054 * config/arc/elf.h: Likewise.
37055 * config/arc/linux.h: Likewise.
37056 * config/arc/t-uClibc: Remove.
37058 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37060 PR tree-optimization/77536
37061 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37062 (tree_transform_and_unroll_loop): Use above function to compute the
37063 estimated niter of unrolled loop and use it when scaling profile.
37064 Also use count info rather than frequency if it's non-zero.
37065 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37066 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37067 (vect_transform_loop): Call above function.
37069 2017-02-27 Richard Biener <rguenther@suse.de>
37071 PR tree-optimization/45397
37072 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37073 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37074 (visit_nary_op): Add pattern matching for CSEing sign-changed
37075 or truncated operations with wider ones.
37077 2017-02-27 Richard Biener <rguenther@suse.de>
37079 PR tree-optimization/79690
37080 * tree-vect-stmts.c (vectorizable_store): Use vector type
37081 built from the DR with address-space.
37083 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37085 * doc/invoke.texi (Optimize Options): Refine the description
37086 of asan-use-after-return.
37088 2017-02-25 Alan Modra <amodra@gmail.com>
37090 PR rtl-optimization/79584
37091 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37092 base, not ad->base_term, the reg within base. Remove assertion
37093 that ad->base == ad->base_term. Replace gen_int_mode using
37094 bogus mode with const0_rtx.
37096 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37098 PR middle-end/79396
37099 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37100 FMA_EXPR like tcc_binary or tcc_unary.
37102 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37105 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37107 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37108 (output_loc_operands): Handle DW_OP_call_ref and
37109 DW_OP_GNU_variable_value.
37110 (struct variable_value_struct): New type.
37111 (struct variable_value_hasher): Likewise.
37112 (variable_value_hash): New variable.
37113 (string_types): Remove.
37114 (copy_loc_descr): New function.
37115 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37116 (prepend_loc_descr_to_each): New function.
37117 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37118 instead of add_loc_descr_to_each if the first argument is single
37119 location list and the second has multiple.
37120 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37121 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37122 when looking for variable value which doesn't have other location info.
37123 (loc_list_from_tree): Formatting fix.
37124 (gen_array_type_die): Simplify DW_AT_string_length handling.
37125 (adjust_string_types): Remove.
37126 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37127 string_types. Call resolve_variable_values.
37128 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37129 (resolve_addr_in_expr): Likewise. Add A argument.
37130 (copy_deref_exprloc): Remove deref argument. Adjust for the
37131 original expression being DW_OP_GNU_variable_value with optionally
37132 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37133 optionally after it.
37134 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37135 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37136 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37137 (variable_value_hasher::hash, variable_value_hasher::equal): New
37139 (resolve_variable_value_in_expr, resolve_variable_value,
37140 resolve_variable_values, note_variable_value_in_expr,
37141 note_variable_value): New functions.
37142 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37145 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37148 * opts.h (handle_generated_option): Add GENERATED_P argument.
37149 * opts-common.c (handle_option): Adjust function comment.
37150 (handle_generated_option): Add GENERATED_P argument, pass it to
37152 (control_warning_option): Pass false to handle_generated_option
37154 * opts.c (maybe_default_option): Pass true to handle_generated_option
37156 * optc-gen.awk: Likewise.
37158 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37160 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37161 a REG, look at the REG it is a SUBREG of.
37162 (splitter for cmpeqsi_t): Ditto.
37164 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37166 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37167 the special USEs with the pattern of the insn, not the insn itself.
37169 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37172 * doc/invoke.texi: Document -mload-store-pairs.
37174 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37175 Sandra Loosemore <sandra@codesourcery.com>
37177 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37178 argument isn't a CONST_INT.
37179 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37180 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37181 (nios2_validate_compare): Bypass alternate compare logic if *op2
37182 is not a CONST_INT.
37183 (ldstwm_operation_p): Return false if first_base is not a REG or
37184 if first_offset is not a CONST_INT.
37186 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37188 * config/cris/cris.md: Use correct operand in a define_peephole2.
37190 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37192 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37194 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37196 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37197 this_insn if it is an INSN or JUMP_INSN.
37198 (force_offsettable): Look at base, not at addr.
37199 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37200 on things that aren't necessarily CONST_INTs.
37202 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37204 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37205 -mfpmath=sse is the default also for x86-32 targets with SSE2
37206 instruction set when @option{-ffast-math} is enabled
37208 2017-02-24 Jeff Law <law@redhat.com>
37210 PR rtl-optimizatoin/79286
37211 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37214 2017-02-24 Richard Biener <rguenther@suse.de>
37216 PR tree-optimization/79389
37217 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37220 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37222 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37223 function comment to reflect reality.
37224 (loop_exits_before_overflow): Fix typo in function description.
37226 2017-02-24 Richard Biener <rguenther@suse.de>
37228 PR tree-optimization/79389
37229 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37230 properly that a threading opportunity exists. Detect conditional
37231 copy/constant propagation opportunities.
37233 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37235 * config/visium/visium.md (type): Add trap.
37236 (b): New mode attribute.
37237 (*btst): Rename into...
37238 (*btst<mode>): ...this and adjust.
37239 (*cbranchsi4_btst_insn): Rename into...
37240 (*cbranch<mode>4_btst_insn): ...this and adjust.
37241 (trap): New define_insn.
37243 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37245 PR tree-optimization/79389
37246 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37247 (noce_reversed_cond_code): New function.
37248 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37249 reversed_comparison_code. Formatting fix.
37250 (noce_try_store_flag): Test rev_cond != NULL in addition to
37251 reversed_comparison_code.
37252 (noce_try_store_flag_constants): Likewise.
37253 (noce_try_store_flag_mask): Likewise.
37254 (noce_try_addcc): Use rev_cond if non-NULL instead of
37255 reversed_comparison_code.
37256 (noce_try_cmove_arith): Likewise. Formatting fixes.
37257 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37258 (noce_find_if_block): Initialize rev_cond.
37259 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37260 instead of false as last argument never attempt to reverse it
37263 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37265 PR tree-optimization/79663
37266 * tree-predcom.c (combine_chains): Process refs in reverse order
37267 only for ZERO length chains, and add explaining comment.
37269 2017-02-23 Jeff Law <law@redhat.com>
37271 PR tree-optimization/79578
37272 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37273 in call to operand_equal_p.
37275 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37278 * config/i386/cpuid.h: Fix another undefined behavior.
37280 2017-02-23 Richard Biener <rguenther@suse.de>
37282 PR tree-optimization/79683
37283 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37284 vector types for data-refs.
37286 2017-02-23 Martin Liska <mliska@suse.cz>
37288 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37290 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37292 PR middle-end/79665
37293 * internal-fn.c (get_range_pos_neg): Moved to ...
37294 * tree.c (get_range_pos_neg): ... here. No longer static.
37295 * tree.h (get_range_pos_neg): New prototype.
37296 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37297 are known to be in between 0 and signed maximum inclusive, try to
37298 expand both unsigned and signed divmod and use the cheaper one from
37301 2017-02-22 Jeff Law <law@redhat.com>
37303 PR tree-optimization/79578
37304 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37305 to compare base operands.
37307 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37310 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37311 gpc_reg_operand instead of fpr_reg_operand.
37313 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37315 * config/mips/mips.c (mips_return_in_memory): Force FP
37316 vector types to be returned in memory for o32 ABI.
37318 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37320 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37321 instead of DW_TAG_member for static data member declarations and don't
37322 set no_linkage_name for static inline data members.
37323 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37326 2017-02-22 Martin Liska <mliska@suse.cz>
37328 * doc/invoke.texi: Replace inequality signs with square brackets
37331 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37334 * lra-constraints.c (simplify_operand_subreg): Handle
37335 WORD_REGISTER_OPERATIONS targets.
37337 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37340 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37341 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37342 elimination by swapping fld*.
37344 2017-02-22 Richard Biener <rguenther@suse.de>
37346 PR tree-optimization/79673
37347 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37348 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37349 irrelevant address-space qualifiers and avoiding a
37350 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37352 2017-02-22 Richard Biener <rguenther@suse.de>
37354 PR tree-optimization/79666
37355 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37356 to not symbolically negate if that may introduce undefined
37359 2017-02-22 Martin Liska <mliska@suse.cz>
37362 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37363 * data-streamer-out.c (streamer_write_gcov_count_stream):
37365 * value-prof.c (stream_out_histogram_value): Make assert more
37366 precise based on type of counter.
37368 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37371 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37372 Use nonimmediate_operand instead of memory_operand for operand 1.
37373 (float-extend standard_x87sse_constant_load splitter): Ditto.
37375 2017-02-21 Jeff Law <law@redhat.com>
37377 PR tree-optimization/79621
37378 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37379 blocks with edges to themselves.
37381 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37384 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37385 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37386 Use gimple_call_builtin_p.
37389 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37390 on temporarily removed DEBUG_INSNs.
37392 PR tree-optimization/79649
37393 * tree-loop-distribution.c (classify_partition): Give up on
37394 non-generic address space loads/stores.
37396 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
37398 * doc/loop.texi (Loop manipulation): Remove nonexistent
37399 tree_ssa_loop_version from the documentation.
37400 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
37402 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37405 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
37406 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
37407 * config/rs6000/rs6000.c: Include except.h.
37408 (rs6000_expand_split_stack_prologue): Call
37409 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
37411 2017-02-21 Martin Jambor <mjambor@suse.cz>
37414 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
37415 have been analyzed.
37417 2017-02-21 Martin Jambor <mjambor@suse.cz>
37419 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
37420 for backward compatibility only.
37421 * doc/invoke.texi (Option Summary): Remove all references to
37422 -fipa-cp-alignment.
37424 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
37428 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37430 * lra-constraints.c (curr_insn_transform): Handle
37431 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37433 2017-02-21 Martin Liska <mliska@suse.cz>
37435 * config/i386/i386.opt: Replace -masm-dialect with -masm.
37437 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
37439 PR translation/79638
37440 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
37442 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
37445 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
37446 (arm_function_ok_for_sibcall): Return false for an indirect call by
37447 descriptor if all the argument registers are used.
37448 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
37449 alignment of the function.
37451 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37453 PR tree-optimization/61441
37454 * simplify-rtx.c (simplify_const_unary_operation): For
37455 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
37456 the sNaN unmodified.
37458 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
37460 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
37461 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
37462 instead of SYSTEM_HEADER_DIR.
37464 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
37465 Martin Liška <mliska@suse.cz>
37467 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
37468 Fix typos and grammar, use active voice, and clarify.
37470 2017-02-20 Marek Polacek <polacek@redhat.com>
37472 PR middle-end/79537
37473 * gimplify.c (gimplify_expr): Handle unused *&&L;.
37476 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
37478 2017-02-20 Jakub Jelinek <jakub@redhat.com>
37481 * config/i386/i386.c (ix86_expand_builtin): Handle
37482 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
37483 ix86_builtins_isa[fcode].isa as a requirement of those
37484 flags and any other flag in the bitmask.
37485 (ix86_init_mmx_sse_builtins): Use 0 instead of
37486 ~OPTION_MASK_ISA_64BIT as mask.
37487 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
37488 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
37489 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
37490 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
37492 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37495 * lra-constraints.c (split_reg): Check requested split mode
37496 is supported by the register.
37498 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37500 * lra-constraints.c (simplify_operand_subreg): Remove early
37503 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37506 * lra-constraints.c (curr_insn_transform): Tighten condition
37507 for converting SUBREG reloads from OP_OUT to OP_INOUT.
37509 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
37512 * lra-constraints.c (curr_insn_transform): Handle
37513 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
37515 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
37518 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
37520 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
37522 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
37525 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
37528 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37530 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
37533 2017-02-18 Jakub Jelinek <jakub@redhat.com>
37535 * final.c (last_columnnum, override_columnnum): New variables.
37536 (final_start_function): Set last_columnnum, pass it to begin_prologue
37537 hook and pass 0 to dwarf2out_begin_prologue.
37538 (final_scan_insn): Update override_columnnum. Pass last_columnnum
37539 to source_line debug hook.
37540 (notice_source_line): Compute last_columnnum and for debug_column_info
37541 return true on column changes.
37542 * debug.h (struct gcc_debug_hooks): Add column argument to
37543 source_line and begin_prologue hooks.
37544 (debug_nothing_int_charstar_int_bool): Remove prototype.
37545 (debug_nothing_int_int_charstar,
37546 debug_nothing_int_int_charstar_int_bool): New prototypes.
37547 (dwarf2out_begin_prologue): Add column argument.
37548 * debug.c (do_nothing_debug_hooks): Adjust source_line and
37549 begin_prologue hooks.
37550 (debug_nothing_int_charstar_int_bool): Remove.
37551 (debug_nothing_int_int_charstar,
37552 debug_nothing_int_int_charstar_int_bool): New functions.
37553 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
37554 through to dwarf2out_source_line.
37555 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
37556 (dwarf2out_source_line): Add column argument, emit it if requested.
37557 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
37559 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37560 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
37561 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
37562 through to dwarf2out_begin_prologue.
37563 (vmsdbgout_source_line): Add column argument, pass it through to
37564 dwarf2out_source_line.
37565 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
37566 dbxout_source_line caller.
37567 (dbxout_source_line): Add column argument.
37569 * common.opt (gno-column-info, gcolumn-info): New options.
37570 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
37571 (check_die): Also test for multiple DW_AT_decl_column attributes.
37572 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
37573 DW_AT_decl_column if requested.
37574 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
37576 (gen_variable_die): Likewise.
37577 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
37578 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
37581 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
37582 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
37583 (ix86_handle_option): Handle OPT_m3dnowa.
37584 * doc/invoke.texi (-m3dnowa): Document.
37585 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
37586 -m3dnowa instead of -m3dnow -march=athlon.
37589 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
37590 instead of gcc_assert for K, r and R code checks. Formatting fixes.
37592 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37595 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
37596 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
37597 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
37598 generator for vsx_xxpermdi_<mode>_be.
37599 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
37600 force big-endian semantics.
37601 (vsx_xxpermdi_<mode>_be): New define_expand with same
37602 implementation as previous version of vsx_xxpermdi_<mode>.
37604 2017-02-17 Jakub Jelinek <jakub@redhat.com>
37606 PR tree-optimization/79327
37607 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
37608 variable, its initialization and use.
37610 2017-02-17 Julia Koval <julia.koval@intel.com>
37612 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
37613 (OPTION_MASK_ISA_PKU_UNSET): New.
37614 (ix86_handle_option): Handle -mrdpid.
37615 * config/i386/cpuid.h (bit_RDPID): New.
37616 * config/i386/driver-i386.c (host_detect_local_cpu):
37617 Detect RDPID feature.
37618 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
37619 * config/i386/i386-c.c (ix86_target_macros_internal):
37621 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
37622 (ix86_valid_target_attribute_inner_p): Add "rdpid".
37623 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
37624 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
37625 * config/i386/i386.md (define_insn "rdpid"): New.
37626 * config/i386/i386.opt Add -mrdpid.
37627 * config/i386/immintrin.h (_rdpid_u32): New.
37629 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
37631 PR rtl-optimization/79541
37632 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
37633 instead of transforming it into USE.
37635 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
37637 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
37638 If HONOR_SNANS (SFmode) force the input to a register.
37639 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
37640 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
37641 an frsp or similar insn.
37643 2017-02-17 Martin Liska <mliska@suse.cz>
37645 PR rtl-optimization/79577
37646 * params.def (selsched-max-sched-times): Increase minimum to 1.
37648 2017-02-17 Martin Liska <mliska@suse.cz>
37650 PR rtl-optimization/79574
37651 * gcse.c (want_to_gcse_p): Prevent integer overflow.
37653 2017-02-17 Martin Liska <mliska@suse.cz>
37655 PR tree-optimization/79529
37656 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
37657 ssa_defined_default_def_p to handle cases which are implicitly
37659 * tree-ssa.c (ssa_defined_default_def_p): New function.
37660 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
37661 which are implicitly defined.
37662 * tree-ssa.h (ssa_defined_default_def_p): Declare.
37664 2017-02-17 Richard Biener <rguenther@suse.de>
37666 PR middle-end/79576
37667 * params.def (max-ssa-name-query-depth): Limit to 10.
37669 2017-02-17 Richard Biener <rguenther@suse.de>
37671 PR tree-optimization/79552
37672 * tree-ssa-structalias.c (visit_loadstore): Properly verify
37675 2017-02-17 Richard Biener <rguenther@suse.de>
37678 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
37680 2017-02-17 Marek Polacek <polacek@redhat.com>
37682 PR middle-end/79536
37683 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
37684 (fold_negate_expr): New wrapper.
37686 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
37688 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
37689 Correct terminology and de-emphasize pre-standard behavior.
37691 2017-02-16 Alan Modra <amodra@gmail.com>
37693 PR rtl-optimization/79286
37694 * ira.c (def_dominates_uses): New function.
37695 (update_equiv_regs): Don't create an equivalence for insns that
37696 may trap where the register def does not dominate the use.
37698 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
37700 PR rtl-optimization/78127
37701 * lra.c (lra): Call lra_eliminate before finish the loop after
37704 2017-02-16 Richard Biener <rguenther@suse.de>
37706 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
37708 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
37709 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
37710 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
37711 (isl_val_int_from_wi): New function.
37712 (extract_affine_gmp): Rename to ...
37713 (extract_affine_wi): ... this, take a widest_int.
37714 (extract_affine_int): Just wrap extract_affine_wi.
37715 (add_param_constraints): Use isl_val_int_from_wi.
37716 (add_loop_constraints): Likewise, and extract_affine_wi.
37718 2017-02-15 Jeff Law <law@redhat.com>
37720 PR middle-end/79521
37721 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
37722 ira_init_register_move_cost_if_necessary.
37724 2017-02-15 Martin Sebor <msebor@redhat.com>
37726 PR middle-end/32003
37727 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
37728 removed in a prior commit.
37730 2017-02-15 Bin Cheng <bin.cheng@arm.com>
37732 PR tree-optimization/79347
37733 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
37734 counters during peeling.
37736 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
37738 * Makefile.in (site.exp): Remove "set ISLVER".
37740 2017-02-15 Jakub Jelinek <jakub@redhat.com>
37743 * real.c (real_from_integer): Call real_convert even for decimal.
37745 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
37748 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
37750 2017-02-14 Andrew Pinski <apinski@cavium.com>
37752 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
37753 cores and change the partno/implementer to be correct.
37754 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
37755 the 'B" as the implementer.
37756 * config/aarch64/aarch64-tune.md: Regenerate.
37758 2017-02-14 Carl Love <cel@us.ibm.com>
37760 * config/rs6000/rs6000.c: Add case statement entry to make the
37761 xvcvuxdsp built-in argument unsigned.
37762 * config/rs6000/vsx.md: Fix the source and return operand types so they
37763 match the instruction definitions from the ISA document. Fix typo
37764 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
37767 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
37770 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
37771 member early_clobber_alts.
37772 * lra-lives.c (reg_early_clobber_p): New.
37773 (process_bb_lives): Use it.
37774 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
37775 (debug_operand_data): Initialize early_clobber_alts.
37776 (setup_operand_alternative): Set up early_clobber_alts.
37777 (collect_non_operand_hard_regs): Ditto. Pass early clobber
37778 alternatives to new_insn_reg.
37779 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
37781 (lra_update_insn_regno_info): Pass the new arg.
37783 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37785 PR middle-end/79505
37786 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
37787 (new_oacc_loop_raw): Don't clear already cleared fields.
37790 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
37791 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
37792 _mm512_prefetch_i64gather_ps): New inline functions and macros.
37794 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
37797 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
37799 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
37802 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
37803 the extra instruction to the right place to store 128-bit constant
37806 2017-02-14 Martin Sebor <msebor@redhat.com>
37808 PR middle-end/79448
37809 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
37810 warning for strings of unknown length.
37812 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
37814 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
37816 2017-02-14 Jeff Law <law@redhat.com>
37819 * ira-costs.c (scan_one_insn): Initialize register move costs
37820 for pseudos seen in USE/CLOBBER insns.
37822 PR tree-optimization/79095
37823 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
37824 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
37825 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
37826 if the operands are known to be not equal, then the resulting range
37828 (intersect_ranges): If the new range is ~[0,0] and the old range is
37829 wide, then prefer ~[0,0].
37830 * tree-vrp.c (overflow_comparison_p_1): New function.
37831 (overflow_comparison_p): New function.
37832 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
37833 if NAME is used in an overflow test.
37834 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
37835 overflow check that can be expressed as an equality test, then adjust
37836 ops to be that equality test.
37838 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37840 * config/s390/s390-builtin-types.def: Remove flags argument.
37841 * config/s390/s390.c (s390_init_builtins): Likewise.
37843 2017-02-14 Martin Liska <mliska@suse.cz>
37845 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
37846 vector. Fix trailing white spaces.
37848 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
37850 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
37853 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37855 PR rtl-optimization/68664
37856 * config/arm/arm.c (arm_sched_can_speculate_insn):
37857 New function. Declare prototype.
37858 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37860 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37862 PR rtl-optimization/68664
37863 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
37865 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
37867 2017-02-14 Amit Pawar <amit.pawar@amd.com>
37869 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
37870 max skip bytes for function, loop and jump.
37872 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
37874 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
37875 ABS_EXPR for gimple dump.
37877 2017-02-14 Jakub Jelinek <jakub@redhat.com>
37880 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
37882 PR tree-optimization/79408
37883 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
37884 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
37885 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
37886 also if rhs1 is INTEGER_CST.
37888 2017-02-14 Richard Biener <rguenther@suse.de>
37890 PR middle-end/79432
37891 * tree-into-ssa.c (insert_phi_nodes): When the function can
37892 have abnormal edges rewrite SSA names with broken use-def
37893 dominance out of SSA and register them for PHI insertion.
37895 2017-02-13 Martin Sebor <msebor@redhat.com>
37897 PR middle-end/79496
37898 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
37899 clearing info.nowrite flag when snprintf size argument is a range.
37901 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37903 * cprop.c (cprop_jump): Add missing space in string literal.
37904 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
37905 (get_constraint_for_component_ref): Likewise.
37906 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
37907 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
37908 * lra-constraints.c (process_alt_operands): Likewise.
37909 * ipa-inline.c (inline_small_functions): Likewise.
37910 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
37911 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
37912 * trans-mem.c (diagnose_tm_1_op): Likewise.
37913 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
37914 (grid_parallel_clauses_gridifiable): Likewise.
37916 * config/nvptx/mkoffload.c (process): Add space in between
37919 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
37920 "MOD4_SSE_REGS" and "ALL_REGS".
37922 * spellcheck.c (test_data): Add , in between "foo" and "food".
37924 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37927 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
37928 boundary crossing check and subsequent code generation agree.
37930 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37932 * config/aarch64/aarch64.c (has_memory_op): Delete.
37933 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
37936 2017-02-13 Jakub Jelinek <jakub@redhat.com>
37938 PR rtl-optimization/79388
37939 PR rtl-optimization/79450
37940 * combine.c (distribute_notes): When removing TEM_INSN for which
37941 corresponding dest has last value recorded, invalidate that last
37944 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37946 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
37947 of explicit '@'. Add missing assembly comment marker on branch costs
37950 2017-02-13 Nathan Sidwell <nathan@acm.org>
37952 * gengtype-lex.l (<in_struct>): Add '/'.
37954 2017-02-13 Martin Liska <mliska@suse.cz>
37957 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
37959 2017-02-13 Richard Biener <rguenther@suse.de>
37961 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
37963 * configure: Re-generate.
37964 * config.in: Likewise.
37965 * graphite-dependences.c: Simplify as if
37966 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
37967 * graphite-isl-ast-to-gimple.c: Likewise.
37968 * graphite-optimize-isl.c: Likewise.
37969 * graphite-poly.c: Likewise.
37970 * graphite-sese-to-poly.c: Likewise.
37971 * graphite.h: Likewise.
37972 * toplev.c: Include isl/version.h and use isl_version () for
37973 printing the ISL version.
37974 * doc/install.texi: Update ISL requirement.
37976 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37978 * doc/standards.texi (Standards): Update reference to
37981 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
37983 * doc/extend.texi (Named Address Spaces): sourceware.org now
37985 * doc/install.texi (Binaries): Ditto.
37988 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37990 * doc/cpp.texi: Replace "stringify"/"stringification" with C
37991 standard terminology "stringize"/"stringizing" throughout.
37992 * doc/cppinternals.texi: Likewise.
37994 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
37996 * doc/extend.texi: Fix some spelling mistakes and typos.
37997 * doc/invoke.texi: Likewise.
37999 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38002 * params.def (inline-min-speedup) Change from 10 to 8.
38004 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38006 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38009 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38012 * ipa-inline-analysis.c (get_minimal_bb): New function.
38013 (record_modified): Use it.
38014 (remap_edge_change_prob): Handle also ancestor functions.
38016 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38018 * doc/contrib.texi (Contributors): Remove broken link into
38019 the Mauve CVS repository.
38021 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38023 PR middle-end/79454
38024 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38025 result computation whenever lhs doesn't have vector mode, not
38026 just when it has BLKmode.
38028 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38030 * doc/makefile.texi (profiledbootstrap): Refer to the
38031 installation instructions only in textual form.
38033 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38036 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38038 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38040 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38041 (Specific): Update mingw-w64 reference.
38043 (Specific): Remove broken link to Renesas RX processor.
38045 2017-02-10 Richard Biener <rguenther@suse.de>
38047 * toplev.c (process_options): Do not mention obsolete graphite
38048 options when printing sorry message about missing graphite support.
38049 Mention -floop-nest-optimize.
38051 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38053 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38054 (vtst_p16): Likewise.
38055 (vtstq_p8): Likewise.
38056 (vtstq_p16): Likewise.
38058 (vtstq_p64): Likewise.
38059 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38060 (vset_lane_p64): New.
38061 (vsetq_lane_p64): New.
38063 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38065 PR tree-optimization/79411
38066 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38067 stmt operands are SSA_NAMEs used in abnormal phis.
38068 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38071 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38074 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38077 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38079 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38081 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38083 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38087 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38088 not arbitrary TREE_CONSTANT.
38091 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38092 "omp declare target link" attribute unless is_global_var.
38093 * omp-offload.c (find_link_var_op): Likewise.
38095 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38096 Chung-Lin Tang <cltang@codesourcery.com>
38098 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38100 (gimplify_adjust_omp_clauses): Don't delete TILE.
38101 (gimplify_omp_for): Deal with TILE.
38102 * internal-fn.c (expand_GOACC_TILE): New function.
38103 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38105 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38106 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38108 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38109 avoid DIV for outermost collapse var.
38110 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38111 Remove out of date comments, fix whitespace.
38112 * omp-general.c (omp_extract_for_data): Deal with tiling.
38113 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38114 adjust OLF_DIM_BASE value.
38115 (struct omp_for_data): Add tiling field.
38116 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38117 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38118 for auto loops. Remove default auto determining, moved to
38119 oacc_loop_fixed_partitions.
38120 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38121 stmts, add e_mask field.
38122 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38123 (oacc_thread_numbers): Use oacc_dim_call.
38124 (oacc_xform_tile): New.
38125 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38126 (finish_oacc_loop): Adjust for ifns vector.
38127 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38128 add case for GOACC_TILE fns.
38129 (oacc_loop_xform_loop): Delete.
38130 (oacc_loop_process): Iterate over call list directly, and add
38131 handling for GOACC_TILE fns.
38132 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38134 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38135 vector partitioning to outer loops. Assign 2 partitions to loops
38136 when available. Add TILE handling.
38137 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38138 (execite_oacc_device_lower): Process GOACC_TILE fns,
38139 ignore unknown specs.
38140 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38141 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38142 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38144 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38146 * configure.ac (ACX_BUGURL): Update.
38147 * configure: Regenerate.
38149 2017-02-09 Richard Biener <rguenther@suse.de>
38151 PR tree-optimization/69823
38152 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38153 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38155 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38157 * config/arc/arc-c.def: Add __NPS400__ definition.
38158 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38159 (TARGET_NPS400): Define.
38161 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38163 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38165 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38166 pointer, arch_info.
38167 (arc_cpu_types): Fill the arch_info field with a pointer into the
38168 arc_arch_types table.
38169 (arc_selected_cpu): Declare.
38170 * config/arc/arc.c (arc_selected_cpu): Make global.
38171 (arc_selected_arch): Delete.
38172 (arc_base_cpu): Delete.
38173 (arc_override_options): Remove references to deleted variables,
38174 update access to arch information.
38175 (ARC_OPT): Update access to arch information.
38176 (ARC_OPTX): Likewise.
38177 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38178 (TARGET_ARC600): Update access to arch information.
38179 (TARGET_ARC601): Likewise.
38180 (TARGET_ARC700): Likewise.
38181 (TARGET_EM): Likewise.
38182 (TARGET_HS): Likewise.
38183 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38186 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38189 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38190 condition/operands for integer GE/LE/GEU/LEU operations.
38192 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38194 PR translation/79397
38195 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38198 2017-02-08 Martin Jambor <mjambor@suse.cz>
38201 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38202 whether allocation happened.
38203 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38204 nothing was allocated.
38206 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38208 PR tree-optimization/79408
38209 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38210 constant, but SSA_NAME with a known integer range, use the minimum
38211 of that range instead of op1 to determine if modulo can be replaced
38212 with its first operand.
38214 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38216 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38218 2017-02-08 Richard Biener <rguenther@suse.de>
38220 PR tree-optimization/71824
38221 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38222 Check all loops contained in the merged region.
38224 2017-02-07 Andrew Pinski <apinski@cavium.com>
38226 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38228 2017-02-07 Andrew Pinski <apinski@cavium.com>
38230 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38231 (thunderxt88): Likewise.
38232 (thunderxt81): Disable LSE and change v8.1 to v8.
38233 (thunderxt83): Likewise.
38235 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38236 Richard Biener <rguenther@suse.de>
38238 PR middle-end/79399
38239 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38240 type from int to size_t.
38241 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38243 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38245 PR rtl-optimization/79386
38246 * cprop.c (bypass_conditional_jumps): Initialize
38247 bypass_last_basic_block already before splitting bbs after
38248 unconditional traps...
38249 (bypass_conditional_jumps): ... rather than here.
38252 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38253 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38254 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38255 fix -masm=intel patterns.
38257 2017-02-07 Richard Biener <rguenther@suse.de>
38259 PR tree-optimization/79256
38260 PR middle-end/79278
38261 * builtins.c (get_object_alignment_2): Use min_align_of_type
38262 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38263 and ADJUST_FIELD_ALIGN.
38265 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38267 * doc/tm.texi: Regenerate.
38268 * stor-layout.c (layout_decl): Adjust.
38269 (update_alignment_for_field): Likewise.
38270 (place_field): Likewise.
38271 (min_align_of_type): Likewise.
38272 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38273 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38274 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38275 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38276 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38277 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38278 * config/i386/i386.c (x86_field_alignment): Likewise.
38279 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38280 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38281 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38282 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38283 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38284 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38288 2017-01-30 Richard Biener <rguenther@suse.de>
38290 PR tree-optimization/79256
38291 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38292 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38295 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38297 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38298 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38299 builtins to SImode and emit a zero-extend, if necessary.
38301 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38303 * docs/invoke.texi (RISC-V Options): Alphabetize.
38305 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38307 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38310 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38312 * config/riscv/riscv.c: New file.
38313 * common/config/riscv/riscv-common.c: Likewise.
38314 * config.gcc: Likewise.
38315 * config/riscv/constraints.md: Likewise.
38316 * config/riscv/elf.h: Likewise.
38317 * config/riscv/generic.md: Likewise.
38318 * config/riscv/linux.h: Likewise.
38319 * config/riscv/multilib-generator: Likewise.
38320 * config/riscv/peephole.md: Likewise.
38321 * config/riscv/pic.md: Likewise.
38322 * config/riscv/predicates.md: Likewise.
38323 * config/riscv/riscv-builtins.c: Likewise.
38324 * config/riscv/riscv-c.c: Likewise.
38325 * config/riscv/riscv-ftypes.def: Likewise.
38326 * config/riscv/riscv-modes.def: Likewise.
38327 * config/riscv/riscv-opts.h: Likewise.
38328 * config/riscv/riscv-protos.h: Likewise.
38329 * config/riscv/riscv.h: Likewise.
38330 * config/riscv/riscv.md: Likewise.
38331 * config/riscv/riscv.opt: Likewise.
38332 * config/riscv/sync.md: Likewise.
38333 * config/riscv/t-elf-multilib: Likewise.
38334 * config/riscv/t-linux: Likewise.
38335 * config/riscv/t-linux-multilib: Likewise.
38336 * config/riscv/t-riscv: Likewise.
38337 * configure.ac: Likewise.
38338 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38339 Waterman as RISC-V maintainers.
38340 * doc/install.texi: Add RISC-V entries.
38341 * doc/invoke.texi: Add RISC-V options section.
38342 * doc/md.texi: Add RISC-V constraints section.
38343 * configure: Regenerated.
38345 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38348 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38349 false values to be constant vectors with all 0 or all 1 bits set.
38350 (vcondu<mode><mode>): Likewise.
38351 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38353 (fpmask_comparison_operator): Update comment.
38354 (vecint_comparison_operator): New predicate.
38355 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38356 vector conditionals when the true and false values are constant
38357 vectors with all 0 bits or all 1 bits set.
38359 2017-02-06 Martin Sebor <msebor@redhat.com>
38361 PR tree-optimization/79376
38362 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38364 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38366 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38367 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38368 to simplify split condition.
38370 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38372 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38373 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38376 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38378 PR rtl-optimization/68664
38379 * target.def (can_speculate_insn): New hook.
38380 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38381 * doc/tm.texi: Regenerate.
38382 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38383 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38384 (rs6000_sched_can_speculate_insn): New function.
38386 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38388 PR tree-optimization/79284
38389 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38390 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38391 vectorizable_mask_load_store, vectorizable_operation,
38392 vect_is_simple_cond, get_same_sized_vectype): Use it instead
38393 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
38394 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
38395 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
38396 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
38397 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
38398 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
38399 is_gimple_assign (stmt). Replace another such test with
38400 is_gimple_assign (stmt).
38402 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38405 * config/avr/avr.c (rtl-iter.h): Include it.
38406 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
38407 (avr_legitimate_combined_insn): ...and implementation.
38409 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
38411 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
38412 * config/s390/s390.c (s390_const_operand_ok)
38413 (s390_canonicalize_comparison, s390_extract_part)
38414 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
38415 (s390_contiguous_bitmask_p, s390_rtx_costs)
38416 (legitimize_pic_address): Likewise.
38417 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
38418 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
38419 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
38420 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
38421 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
38423 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
38425 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
38426 REGNO($0) == REGNO($1).
38428 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38430 * config/s390/linux.h(SIZE_TYPE): Add comment.
38432 2017-02-06 Julian Brown <julian@codesourcery.com>
38433 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38434 Virendra Pathak <virendra.pathak@broadcom.com>
38436 * config/aarch64/aarch64-cores.def: Change the scheduler
38438 * config/aarch64/aarch64.md: Include thunderx2t99.md.
38439 * config/aarch64/thunderx2t99.md: New file.
38441 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38443 * doc/standards.texi (Go Language): Update link to language
38446 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38448 * tree-eh.c (lower_resx): Sanitize profile.
38449 (cleanup_empty_eh_move_lp): Likewise.
38451 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
38454 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
38456 * cfgloopmanip.h (loop_version): Update prototype.
38457 * modulo-sched.c (sms_schedule): Update call of loop_version.
38458 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
38459 * tree-parloops.c (gen_parallel_loop): Likewise.
38460 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
38461 * tree-ssa-loop-split.c (split_loop): Likewise.
38462 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
38463 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
38465 2017-02-05 Martin Liska <mliska@suse.cz>
38468 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
38470 (print_operand_address): Initialize a struct to zero.
38472 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38474 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
38475 garbage collector only in textual form.
38477 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
38479 * doc/extend.texi (x86 specific memory model extensions for
38480 transactional memory): Simplify a phrase.
38482 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
38485 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
38486 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
38487 (atomic_storedi_1): Likewise.
38489 2017-02-04 Jakub Jelinek <jakub@redhat.com>
38491 PR tree-optimization/79338
38492 * tree-parloops.c (gather_scalar_reductions): Don't call
38493 vect_analyze_loop_form for loop->inner before destroying loop's
38496 2017-02-03 Martin Sebor <msebor@redhat.com>
38498 PR tree-optimization/79327
38499 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
38500 when precision has resulted in leading zeros.
38501 (format_integer): Adjust the likely counter to assume an unknown
38502 argument that may be zero is non-zero.
38504 2017-02-03 Jason Merrill <jason@redhat.com>
38507 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
38508 avoid copying non-taken branch.
38510 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38512 PR tree-optimization/79340
38513 * tree-vect-loop.c (vectorizable_reduction): Release
38514 vec_defs elements after safe_splicing them into other vectors.
38517 PR tree-optimization/79327
38518 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
38519 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
38521 (format_integer): Use wide_int_to_tree instead of build_int_cst
38522 + to_?hwi. If argmin is NULL, just set argmin and argmax to
38523 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
38524 of shortest and longest sequence.
38526 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
38528 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
38529 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
38531 2017-02-03 Walter Lee <walt@tilera.com>
38534 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
38535 after initial stackframe link reg save.
38536 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
38538 2017-02-03 Jakub Jelinek <jakub@redhat.com>
38541 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
38542 wu for stxssp alternative.
38544 2017-02-03 Martin Sebor <msebor@redhat.com>
38546 PR tree-optimization/79352
38547 * gimple-fold.c (get_range_strlen): Add argument.
38548 (get_range_strlen): Change return type to bool.
38549 (get_maxval_strlen): Pass in a dummy argument.
38550 * gimple-fold.h (get_range_strlen): Change return type to bool.
38551 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
38552 * tree.h (array_at_struct_end_p): Add argument.
38553 * tree.c (array_at_struct_end_p): Handle it.
38555 2017-02-03 Martin Liska <mliska@suse.cz>
38558 * multiple_target.c (create_dispatcher_calls): Redirect edge
38559 from a caller of a dispatcher.
38560 (expand_target_clones): Make the clones local.
38561 (ipa_target_clone): Do both target clones and resolvers.
38562 (ipa_dispatcher_calls): Remove the pass.
38563 (pass_dispatcher_calls::gate): Likewise.
38564 (make_pass_dispatcher_calls): Likewise.
38565 * passes.def (pass_target_clone): Put as very first IPA early
38568 2017-02-03 Martin Liska <mliska@suse.cz>
38570 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
38571 in case of a function with ifunc attribute.
38573 2017-02-03 Martin Liska <mliska@suse.cz>
38575 * cgraph.c (cgraph_node::dump): Dump function version info.
38576 * symtab.c (symtab_node::dump_base): Add missing new line.
38578 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38580 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
38581 (ifcombine_ifandif): Use it.
38583 2017-02-03 Martin Liska <mliska@suse.cz>
38585 * doc/invoke.texi: Document default value for
38586 use-after-scope-direct-emission-threshold.
38588 2017-02-03 Martin Liska <mliska@suse.cz>
38590 PR tree-optimization/79339
38591 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
38592 (format_floating): Likewise.
38594 2017-02-03 Martin Liska <mliska@suse.cz>
38597 * ipa-prop.c (ipa_node_params_t::insert): Remove current
38599 (ipa_node_params_t::remove): Likewise.
38600 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
38601 initialization from removed ipa_node_params_t::insert.
38602 (ipa_node_params::~ipa_node_params): Move from removed
38603 ipa_node_params_t::release.
38604 * symbol-summary.h (symbol_summary::m_released): New member.
38605 Do not release a summary twice. Do not allow to call finalizer
38606 for types of a summary that live in GGC memory.
38608 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
38610 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
38613 2017-02-02 Martin Sebor <msebor@redhat.com>
38615 PR middle-end/79275
38616 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
38617 (format_string): Tighten up the range of output for non-constant
38618 strings and correct the expected range for wide non-constant strings.
38620 2017-02-02 Martin Sebor <msebor@redhat.com>
38622 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
38624 PR middle-end/32003
38625 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
38627 (-fdump-tree-@var): Add to index and document how to come up
38628 with pass-specific option and dump file names.
38629 (-fdump-passes): Clarify where to look for output.
38631 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
38633 PR middle-end/77445
38634 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
38635 statistics of the analyzed path; allow threading for speed when
38636 any of BBs along the path are optimized for speed.
38638 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
38640 PR middle-end/78468
38641 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
38642 settings of the virtual registers.
38645 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
38647 * explow.c (get_dynamic_stack_size): Take known alignment of stack
38648 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
38651 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38653 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
38654 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
38656 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38658 * config/s390/s390.md: Add missing comments with the expanded
38660 * config/s390/vector.md: Likewise.
38661 * config/s390/vx-builtins.md: Likewise.
38663 2017-02-02 Jakub Jelinek <jakub@redhat.com>
38666 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
38667 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
38668 conditions on a single line.
38670 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38672 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38673 __S390_VX__ to __VX__.
38675 2017-02-01 Andrew Pinski <apinski@cavium.com>
38677 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
38678 stmt_info to record_stmt_cost.
38679 (vect_get_known_peeling_cost): Pass stmt_info if known to
38681 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
38682 cpu_vector_cost field into
38683 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
38684 field into vec_int_stmt_cost and vec_fp_stmt_cost.
38685 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
38686 splitting of scalar_stmt_cost and vec_stmt_cost.
38687 (thunderx_vector_cost): Likewise.
38688 (cortexa57_vector_cost): LIkewise.
38689 (exynosm1_vector_cost): Likewise.
38690 (xgene1_vector_cost): Likewise.
38691 (thunderx2t99_vector_cost): Improve after the splitting of the two
38693 (aarch64_builtin_vectorization_cost): Update for the splitting of
38694 scalar_stmt_cost and vec_stmt_cost.
38696 2017-02-01 Torvald Riegel <triegel@redhat.com>
38697 Richard Henderson <rth@redhat.com>
38699 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
38700 conditional on existance of a fast atomic load.
38701 * optabs-query.c (can_atomic_load_p): New function.
38702 * optabs-query.h (can_atomic_load_p): Declare it.
38703 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
38704 no fast atomic load is available for the particular size of access.
38705 (expand_atomic_compare_and_swap): Likewise.
38706 (expand_atomic_load): Likewise.
38707 (expand_atomic_store): Likewise.
38708 (expand_atomic_fetch_op): Likewise.
38709 * testsuite/lib/target-supports.exp
38710 (check_effective_target_sync_int_128): Remove x86 because it provides
38711 no fast atomic load.
38712 (check_effective_target_sync_int_128_runtime): Likewise.
38714 2017-02-01 Richard Biener <rguenther@suse.de>
38716 * graphite.c: Include tree-vectorizer.h for find_loop_location.
38717 (graphite_transform_loops): Provide opt-info for optimized nests.
38718 * tree-parloop.c (parallelize_loops): Provide opt-info for
38719 parallelized loops.
38721 2017-02-01 Richard Biener <rguenther@suse.de>
38723 PR middle-end/79315
38724 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
38725 was not set before.
38727 2017-02-01 Richard Biener <rguenther@suse.de>
38729 PR tree-optimization/71824
38730 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38731 Verify the loops are valid in the merged SESE region.
38732 (scop_detection::can_represent_loop_1): Check analyzing the
38733 evolution of the number of iterations in the region succeeds.
38735 2017-01-31 Ian Lance Taylor <iant@golang.org>
38737 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
38738 REG_ARGS_SIZE note to 32-bit push insns and call insn.
38740 2017-01-31 David Malcolm <dmalcolm@redhat.com>
38742 PR preprocessor/79210
38743 * input.c (get_substring_ranges_for_loc): Replace line_width
38744 assertion with error-handling.
38746 2017-01-31 Richard Biener <rguenther@suse.de>
38748 PR tree-optimization/77318
38749 * graphite-sese-to-poly.c (extract_affine): Fix assert.
38750 (create_pw_aff_from_tree): Take loop parameter.
38751 (add_condition_to_pbb): Pass loop of the condition to
38752 create_pw_aff_from_tree.
38754 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38756 * config/s390/s390.c (s390_asan_shadow_offset): New function.
38757 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
38759 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
38763 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
38765 (convert_int_to_float128): Likewise.
38766 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
38767 (convert_int_to_float128): Likewise.
38768 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
38769 (UNSPEC_IEEE128_CONVERT): Likewise.
38770 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
38771 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
38772 Use local variables for IBM extended format.
38773 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
38774 (fix_trunc<mode>si2_fprs): Likewise.
38775 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
38776 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
38777 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
38778 to know that we can now have integers of all sizes in vector
38780 (fix<uns>_<mode>di2_hw): Likewise.
38781 (float<uns>_<mode>si2_hw): Likewise.
38782 (fix_<mode>si2_hw): Likewise.
38783 (fixuns_<mode>si2_hw): Likewise.
38784 (float<uns>_<mode>di2_hw): Likewise.
38785 (float_<mode>di2_hw): Likewise.
38786 (float_<mode>si2_hw): Likewise.
38787 (floatuns_<mode>di2_hw): Likewise.
38788 (floatuns_<mode>si2_hw): Likewise.
38789 (xscvqp<su>wz_<mode>): Delete, no longer used.
38790 (xscvqp<su>dz_<mode>): Likewise.
38791 (xscv<su>dqp_<mode>): Likewise.
38792 (ieee128_mfvsrd_64bit): Likewise.
38793 (ieee128_mfvsrd_32bit): Likewise.
38794 (ieee128_mfvsrwz): Likewise.
38795 (ieee128_mtvsrw): Likewise.
38796 (ieee128_mtvsrd_64bit): Likewise.
38797 (ieee128_mtvsrd_32bit): Likewise.
38799 2017-01-31 Martin Liska <mliska@suse.cz>
38802 * ipa-prop.c (ipa_free_all_node_params): Call release method
38803 instead of ~sumbol_summary to not to trigger double times
38806 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
38808 PR tree-optimization/71691
38809 * bitmap.h (class auto_bitmap): New.
38810 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
38811 is_maybe_undefined instead of ssa_undefined_value_p.
38813 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38815 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
38816 __S390_ARCH_LEVEL__ to __ARCH__.
38818 2017-01-31 Jakub Jelinek <jakub@redhat.com>
38820 PR tree-optimization/79267
38821 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
38822 if should_remove_lhs_p is true.
38824 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
38827 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
38828 (add_alignment_attribute): New.
38829 (base_type_die): Add alignment attribute.
38830 (subrange_type_die): Likewise.
38831 (modified_type_die): Likewise.
38832 (gen_array_type_die): Likewise.
38833 (gen_descr_array_type_die: Likewise.
38834 (gen_enumeration_type_die): Likewise.
38835 (gen_subprogram_die): Likewise.
38836 (gen_variable_die): Likewise.
38837 (gen_field_die): Likewise.
38838 (gen_ptr_to_mbr_type_die): Likewise.
38839 (gen_struct_or_union_type_die): Likewise.
38840 (gen_subroutine_type_die): Likewise.
38841 (gen_typedef_die): Likewise.
38842 (base_type_cmp): Compare alignment attribute.
38844 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38847 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
38848 (setb_unsigned) New pattern for setb with CCUNS.
38849 * config/rs6000/rs6000.c (expand_block_compare): Use a different
38850 subfc./subfe sequence to avoid overflow problems. Generate a
38851 shorter sequence with cmpld/setb for power9.
38852 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
38853 for generating subfc. instruction.
38854 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
38855 now uses this instruction.
38857 2017-01-30 Ian Lance Taylor <iant@google.com>
38860 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
38861 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
38863 2017-01-30 Martin Sebor <msebor@redhat.com>
38865 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
38866 Move constant to the right of a relational operator.
38867 (get_mpfr_format_length, format_character, format_string): Ditto.
38868 (should_warn_p, maybe_warn): Same.
38870 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
38872 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
38875 * asan.c (get_translation_unit_decl): Remove function.
38876 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
38878 2017-01-30 Martin Liska <mliska@suse.cz>
38880 PR gcov-profile/79259
38881 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
38882 -fprofile-generate.
38884 2017-01-30 Martin Liska <mliska@suse.cz>
38887 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
38888 Initialize variables with NULL value.
38890 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
38893 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
38895 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
38897 2017-01-30 Richard Biener <rguenther@suse.de>
38899 PR tree-optimization/79276
38900 * tree-vrp.c (process_assert_insertions): Properly adjust common
38901 when removing a duplicate.
38903 2017-01-30 Richard Biener <rguenther@suse.de>
38905 PR tree-optimization/79256
38906 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38907 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38909 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
38911 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
38914 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
38915 ("*r<noxa>sbg_<mode>_sll_bitmask")
38916 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
38917 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
38918 Use contiguous_bitmask_nowrap_operand.
38920 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38923 * config/rs6000/altivec.h (vec_xl): Revise #define.
38924 (vec_xst): Likewise.
38926 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
38928 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
38930 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
38932 PR rtl-optimization/79194
38933 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
38934 traps before call to bypass_conditional_jumps.
38936 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38938 PR tree-optimization/71374
38939 * lra-constraints.c (check_conflict_input_operands): New.
38940 (match_reload): Use it.
38942 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
38945 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
38946 account to calculate conflict_set.
38948 2017-01-27 Bin Cheng <bin.cheng@arm.com>
38950 PR rtl-optimization/78559
38951 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
38952 other_insn in combine.
38954 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
38956 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
38957 uint16_type_node for BT_UINT16.
38959 2017-01-27 David Malcolm <dmalcolm@redhat.com>
38961 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
38962 "RTL Tests" to menu.
38963 (GIMPLE Tests): New node.
38964 (RTL Tests): New node.
38966 2017-01-27 Richard Biener <rguenther@suse.de>
38968 PR tree-optimization/79245
38969 * tree-loop-distribution.c (distribute_loop): Apply cost
38970 modeling also to detected patterns.
38972 2017-01-27 Richard Biener <rguenther@suse.de>
38974 PR tree-optimization/71433
38975 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
38976 (compare_assert_loc): New function.
38977 (process_assert_insertions): Sort and optimize assert locations
38978 to remove duplicates and push down identical assertions on
38979 edges to their destination block.
38981 2017-01-27 Richard Biener <rguenther@suse.de>
38983 PR tree-optimization/79244
38984 * tree-vrp.c (remove_range_assertions): Forcefully propagate
38985 out SSA names even if abnormal.
38987 2017-01-27 Jakub Jelinek <jakub@redhat.com>
38989 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
38990 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
38991 instead of MPFR_RNDN.
38993 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
38996 * arm.c (arm_option_override): Don't call build_target_option_node
38997 until after doing all option overrides.
38998 (arm_valid_target_attribute_tree): Likewise.
39000 2017-01-27 Martin Liska <mliska@suse.cz>
39002 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39003 for {cd}tors and C++ {cd}tors.
39005 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39007 * config/s390/s390.md ("*setmem_long_and")
39008 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39010 2017-01-26 Martin Sebor <msebor@redhat.com>
39012 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39015 2017-01-26 Martin Sebor <msebor@redhat.com>
39017 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39018 HAVE_DFmode before using XFmode or DFmode.
39019 (parse_directive): Avoid using the z length modifier to avoid
39020 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39022 PR middle-end/78703
39023 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39024 to accept adjustment as an array.
39025 (get_int_range): New function.
39026 (struct directive): Make width and prec arrays.
39027 (directive::set_width, directive::set_precision): Call get_int_range.
39028 (format_integer, format_floating): Handle width and precision ranges.
39029 (format_string, parse_directive): Same.
39031 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39034 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39035 comdat_type_p set, just clone them, but keep the children in the
39039 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39040 which have direct callers with -fvar-tracking-assignments enabled
39042 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39043 inside of type units.
39045 2017-01-26 Martin Sebor <msebor@redhat.com>
39047 PR middle-end/78703
39048 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39050 (struct format_result): Replace number_chars, number_chars_min,
39051 and number_chars_max with a single member of struct result_range.
39053 (format_result::operator+=): Adjust.
39054 (struct fmtresult): Remove bounded. Handle likely and unlikely
39056 (fmtresult::adjust_for_width_or_precision): New function.
39057 (fmtresult:type_max_digits): New function.
39058 (bytes_remaining): Handle likely and unlikely counters.
39059 (min_bytes_remaining): Remove.
39060 (format_percent): Simplify.
39061 (format_integer, format_floating): Set likely and unlikely counters.
39062 (get_string_length, format_character, format_string): Same.
39063 (format_plain, should_warn_p): New function.
39064 (maybe_warn): Call should_warn_p. Update diagnostic messages
39065 and handle those for all directives, including plain strings.
39066 (format_directive): Handle likely and unlikely counters.
39067 Remove unnecessary quoting from diagnostics. Add an informational
39069 (add_bytes): Remove.
39070 (pass_sprintf_length::compute_format_length): Simplify.
39071 (try_substitute_return_value): Handle likely and unlikely counters.
39073 2017-01-26 Carl Love <cel@us.ibm.com>
39075 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39076 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39078 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39081 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39082 endianess for subregs into account.
39083 * lra-constraints.c (lra_constraints): Do risky transformations
39084 always on the first iteration.
39085 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39086 last_call_used_reg_set.
39087 (process_bb_lives): Define and use last_call_used_reg_set.
39088 * lra.c (lra): Always continue after lra_constraints on the first
39091 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39093 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39095 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39097 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39099 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39100 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39101 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39102 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39103 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39104 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39105 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39106 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39107 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39109 2017-01-26 Marek Polacek <polacek@redhat.com>
39112 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39113 for the third operand.
39115 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39117 PR middle-end/79236
39118 * omp-low.c (struct omp_context): Add simt_stmt field.
39119 (scan_omp_for): Return omp_context *.
39120 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39121 context to the _simt_ SIMD stmt.
39122 (lower_omp_for): For combined SIMD with sibling _simt_
39123 SIMD, make sure to use the same decls in _looptemp_
39124 clauses as in the sibling.
39126 2017-01-26 David Sherwood <david.sherwood@arm.com>
39128 PR middle-end/79212
39129 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39132 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39135 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39136 emit fld b; fld a; if possible.
39138 * brig-builtins.def: Update copyright years.
39139 * config/arm/arm_acle_builtins.def: Update copyright years.
39141 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39144 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39145 constraint instead of o for the stxsd instruction.
39147 2017-01-25 Carl Love <cel@us.ibm.com>
39149 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39150 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39152 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39154 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39156 2017-01-25 Richard Biener <rguenther@suse.de>
39158 PR tree-optimization/69264
39159 * target.def (vector_alignment_reachable): Improve documentation.
39160 * doc/tm.texi: Regenerate.
39161 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39163 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39164 earlier changes with respect to TYPE_USER_ALIGN.
39165 (vector_alignment_reachable_p): Likewise. Improve dumping.
39167 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39170 * config/arm/arm.md (xordi3): Force constant operand into a register
39173 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39175 * doc/invoke.texi (-fstore-merging): Correct default optimization
39176 levels at which it is enabled.
39177 (-O): Move -fstore-merging from list to...
39180 2017-01-25 Richard Biener <rguenther@suse.de>
39183 * omp-expand.c: Include debug.h.
39184 (expand_omp_taskreg): Make sure to generate early debug before
39185 outlining anything from a function.
39186 (expand_omp_target): Likewise.
39187 (grid_expand_target_grid_body): Likewise.
39189 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39192 * asan.c (get_translation_unit_decl): New function.
39193 (asan_add_global): Extract modules file name from globals
39194 TRANSLATION_UNIT_DECL name.
39196 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39199 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39200 for long calls with APCS frame and VFP.
39202 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39204 * cfg.c (original_copy_tables_initialized_p): New function.
39205 * cfg.h (original_copy_tables_initialized_p): New decl.
39206 * cfgrtl.c (relink_block_chain): Guard the call to
39207 free_original_copy_tables with a call to
39208 original_copy_tables_initialized_p.
39209 * cgraph.h (symtab_node::native_rtl_p): New decl.
39210 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39211 (symtab_node::needed_p): Don't assert for early assembly output
39212 for __RTL functions.
39213 (cgraph_node::finalize_function): Set "force_output" for __RTL
39215 (cgraph_node::analyze): Bail out early for __RTL functions.
39216 (analyze_functions): Update assertion to support __RTL functions.
39217 (cgraph_node::expand): Bail out early for __RTL functions.
39218 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39220 * function.h (struct function): Update comment for field
39222 * gimple-expr.c: Include "tree-pass.h".
39223 (gimple_has_body_p): Return false for __RTL functions.
39224 * Makefile.in (OBJS): Add run-rtl-passes.o.
39225 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39227 (gcc::pass_manager::get_clean_slate): New accessor.
39228 * passes.c: Include "insn-addr.h".
39229 (should_skip_pass_p): Add logging. Update logic for running
39230 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39231 property-provider override so it is only done for gimple passes.
39233 (skip_pass): New function.
39234 (execute_one_pass): Call skip_pass when skipping passes.
39235 * read-md.c (md_reader::read_char): Support filtering
39236 the input to a subset of line numbers.
39237 (md_reader::md_reader): Initialize fields
39238 m_first_line and m_last_line.
39239 (md_reader::read_file_fragment): New function.
39240 * read-md.h (md_reader::read_file_fragment): New decl.
39241 (md_reader::m_first_line): New field.
39242 (md_reader::m_last_line): New field.
39243 * read-rtl-function.c (function_reader::create_function): Only
39244 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39245 curr_properties. Set DECL_INITIAL to a dummy block.
39246 (read_rtl_function_body_from_file_range): New function.
39247 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39249 * run-rtl-passes.c: New file.
39250 * run-rtl-passes.h: New file.
39252 2017-01-24 Jeff Law <law@redhat.com>
39254 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39257 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39259 PR tree-optimization/79159
39260 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39261 (record_nonwrapping_iv): Improve boundary using above function if no
39262 value range information.
39264 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39265 Martin Jambor <mjambor@suse.cz>
39267 * brig-builtins.def: New file.
39268 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39269 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39270 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39271 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39272 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39273 * builtin-types.def (BT_INT8): New.
39274 (BT_INT16): Likewise.
39275 (BT_UINT8): Likewise.
39276 (BT_UINT16): Likewise.
39277 (BT_FN_ULONG): Likewise.
39278 (BT_FN_UINT_INT): Likewise.
39279 (BT_FN_UINT_ULONG): Likewise.
39280 (BT_FN_UINT_LONG): Likewise.
39281 (BT_FN_UINT_PTR): Likewise.
39282 (BT_FN_ULONG_PTR): Likewise.
39283 (BT_FN_INT8_FLOAT): Likewise.
39284 (BT_FN_INT16_FLOAT): Likewise.
39285 (BT_FN_UINT32_FLOAT): Likewise.
39286 (BT_FN_UINT16_FLOAT): Likewise.
39287 (BT_FN_UINT8_FLOAT): Likewise.
39288 (BT_FN_UINT64_FLOAT): Likewise.
39289 (BT_FN_UINT16_UINT32): Likewise.
39290 (BT_FN_UINT32_UINT16): Likewise.
39291 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39292 (BT_FN_INT_PTR_INT): Likewise.
39293 (BT_FN_UINT_PTR_UINT): Likewise.
39294 (BT_FN_LONG_PTR_LONG): Likewise.
39295 (BT_FN_ULONG_PTR_ULONG): Likewise.
39296 (BT_FN_VOID_UINT64_UINT64): Likewise.
39297 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39298 (BT_FN_INT8_INT8_INT8): Likewise.
39299 (BT_FN_INT16_INT16_INT16): Likewise.
39300 (BT_FN_INT_INT_INT): Likewise.
39301 (BT_FN_UINT_FLOAT_UINT): Likewise.
39302 (BT_FN_FLOAT_UINT_UINT): Likewise.
39303 (BT_FN_ULONG_UINT_UINT): Likewise.
39304 (BT_FN_ULONG_UINT_PTR): Likewise.
39305 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39306 (BT_FN_UINT_UINT_UINT): Likewise.
39307 (BT_FN_VOID_UINT_PTR): Likewise.
39308 (BT_FN_UINT_UINT_PTR: Likewise.
39309 (BT_FN_UINT32_UINT64_PTR): Likewise.
39310 (BT_FN_INT_INT_UINT_UINT): Likewise.
39311 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39312 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39313 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39314 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39315 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39316 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39317 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39318 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39319 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39320 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39321 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39322 * doc/frontends.texi: List BRIG FE.
39323 * doc/install.texi (Testing): Add BRIG tesring requirements.
39324 * doc/invoke.texi (Overall Options): Mention BRIG.
39325 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39327 2017-01-24 Richard Biener <rguenther@suse.de>
39329 PR translation/79208
39330 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39332 2017-01-24 Martin Jambor <mjambor@suse.cz>
39335 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39336 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39337 and known_contexts.
39339 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39341 PR middle-end/79123
39342 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39343 casts from signed to unsigned really don't have a range.
39345 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39347 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39348 GMP_RNDx for compatiblity.
39350 2017-01-24 Martin Liska <mliska@suse.cz>
39353 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39354 that would prevent us to call alloca with -1 as argument.
39356 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39358 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39359 Avoid -Wformat-security warning.
39361 2017-01-23 Andrew Pinski <apinski@cavium.com>
39363 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39366 2017-01-23 Martin Sebor <msebor@redhat.com>
39368 PR middle-end/78703
39369 * gimple-ssa-sprintf.c (warn_level): New global.
39370 (format_integer): Use it here and throughout the rest of the file.
39371 Use the same switch to compute sign as base.
39372 (maybe_warn): New function.
39373 (format_directive): Factor out warnings into maybe_warn.
39374 Add debugging output. Use warn_level.
39375 (add_bytes): Use warn_level.
39376 (pass_sprintf_length::compute_format_length): Add debugging output.
39377 (try_substitute_return_value): Same.
39378 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39380 PR middle-end/78703
39381 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39382 (struct fmtresult, format_integer, format_floating): Adjust.
39383 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39384 (get_string_length, format_string,format_directive): Same.
39385 (pass_sprintf_length::compute_format_length): Same.
39386 (try_substitute_return_value): Simplify slightly.
39388 PR middle-end/78703
39389 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39390 (fmtresult::operator+=): Outlined.
39391 (struct fmtresult): Add ctors.
39392 (struct conversion_spec): Rename...
39393 (struct directive): ...to this. Add and remove data members.
39394 (directive::set_width, directive::set_precision): New functions.
39395 (format_percent): Use fmtresult ctor.
39396 (get_width_and_precision): Remove.
39397 (format_integer): Make naming changes. Avoid computing width and
39399 (format_floating): Same. Adjust indentation.
39400 (format_character, format_none): New functions.
39401 (format_string): Moved character handling to format_character.
39402 (format_directive): Remove arguments, change return type.
39403 (parse_directive): New function.
39404 (pass_sprintf_length::compute_format_length): Move directive
39405 parsing to parse_directive.
39407 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39409 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
39410 (assign_assembler_name_if_needed): ... this.
39411 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
39412 (assign_assembler_name_if_needed): ... this.
39413 (free_lang_data_in_cgraph): Adjust callers.
39414 * cgraphunit.c (cgraph_node::analyze): Likewise.
39415 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
39418 2017-01-23 Richard Biener <rguenther@suse.de>
39420 PR tree-optimization/79088
39421 PR tree-optimization/79188
39422 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
39423 resetting loop bounds after last path deletion. Reset loop
39424 bounds of the target loop, make code match the comments.
39425 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
39426 Make sure loops need no fixups.
39428 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
39430 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
39431 exponent support with double type for first argument.
39432 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
39433 type returned by __builtin_vec_extract_sig,
39434 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
39435 functions from "vector int" to "vector unsigned int" or from
39436 "vector long long int" to "vector unsigned long long int".
39437 Changed type returned by __builtin_vec_extract_exp,
39438 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
39439 functions from "vector int" to "vector unsigned int" or from
39440 "vector long long int" to "vector unsigned long long int".
39441 Changed return type of __builtin_vec_test_data_class,
39442 __builtin_vec_test_data_class_sp, and
39443 __builtin_vec_test_data_class_dp from "vector int" to
39444 "vector bool int" or from "vector long long int" to "vector bool
39445 long long int" and changed second argument type from "unsigned
39446 int" to "int". Added new overloaded function forms "vector float
39447 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
39448 "vector float __builtin_vec_insert_exp_sp (vector float, vector
39449 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
39450 double, vector unsigned long long int)" and "vector double
39451 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
39452 long int)". Changed return type of
39453 __builtin_scalar_test_data_class and
39454 __builtin_scalar_test_data_class_sp and
39455 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
39456 int" and changed second argument from "unsigned int" to "int".
39457 Changed type returned by __builtin_scalar_test_neg,
39458 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
39459 from "int" to "bool int". Added new overloaded function form
39460 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
39461 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
39462 exponent double-precision with floating point first argument.
39463 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
39464 documentation of scalar_test_data_class, scalar_test_neg,
39465 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
39466 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
39467 vec_test_data_class built-in functions to reflect refinements in
39468 their type signatures.
39470 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
39472 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
39474 (aarch64_elf_asm_destructor): Likewise.
39476 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
39478 PR rtl-optimization/78634
39479 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
39480 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
39481 * ifcvt.c (noce_try_cmove): Add missing cost check.
39483 PR rtl-optimization/71724
39484 * combine.c (if_then_else_cond): Look for situations where it is
39485 beneficial to undo the work of one of the recursive calls.
39487 2017-01-23 Bin Cheng <bin.cheng@arm.com>
39489 PR tree-optimization/70754
39490 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
39491 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
39492 combined stmt before it if not NULL.
39493 (combine_chains): Process refs reversely and compute dominance point
39496 2017-01-23 Martin Liska <mliska@suse.cz>
39498 PR tree-optimization/79196
39499 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
39500 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
39502 (strlen_optimize_stmt): Call the renamed function.
39504 2017-01-23 Michael Matz <matz@suse.de>
39506 PR tree-optimization/78384
39507 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
39509 2017-01-23 Richard Biener <rguenther@suse.de>
39511 PR tree-optimization/79186
39512 * tree-vrp.c (register_new_assert_for): Make sure we've seen
39513 both incoming edges before moving an assert.
39515 2017-01-23 Martin Jambor <mjambor@suse.cz>
39517 * ipa-prop.c (load_from_param_1): Removed.
39518 (load_from_unmodified_param): Bits from load_from_param_1 put back
39520 (load_from_param): Removed.
39521 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
39522 with stmt. Reverted back to use of load_from_unmodified_param.
39524 2017-01-23 Martin Jambor <mjambor@suse.cz>
39527 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
39528 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
39529 field a pointer to garbage collected vector, mark lattices and
39530 ipcp_orig_node with GTY((skip)).
39531 (ipa_get_param_count): Adjust to descriptors being a pointer.
39532 (ipa_get_param): Likewise.
39533 (ipa_get_type): Likewise.
39534 (ipa_get_param_move_cost): Likewise.
39535 (ipa_set_param_used): Likewise.
39536 (ipa_get_controlled_uses): Likewise.
39537 (ipa_set_controlled_uses): Likewise.
39538 (ipa_is_param_used): Likewise.
39539 (ipa_node_params_t): Move into garbage collector. New methods insert
39541 (ipa_node_params_sum): Annotate wth GTY(()).
39542 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
39544 (ipa_load_from_parm_agg): Adjust declaration.
39545 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
39546 * ipa-profile.c (ipa_profile): Likewise.
39547 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
39548 (ipa_populate_param_decls): Make descriptors parameter garbage
39550 (ipa_dump_param): Adjust to descriptors being a pointer.
39551 (ipa_alloc_node_params): Likewise.
39552 (ipa_initialize_node_params): Likewise.
39553 (load_from_param_1): Make descriptors parameter garbage collected.
39554 (load_from_unmodified_param): Likewise.
39555 (load_from_param): Likewise.
39556 (ipa_load_from_parm_agg): Likewise.
39557 (ipa_node_params::~ipa_node_params): Removed.
39558 (ipa_free_all_node_params): Remove call to delete operator.
39559 (ipa_node_params_t::insert): New.
39560 (ipa_node_params_t::remove): Likewise.
39561 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
39562 copy known_csts and known_contexts vectors.
39563 (ipa_read_node_info): Adjust to descriptors being a pointer.
39564 (ipcp_modif_dom_walker): Make m_descriptors field garbage
39566 (ipcp_transform_function): Make descriptors variable garbage
39569 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
39571 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
39572 * config/i386/avx512dqintrin.h: Ditto.
39573 * config/i386/avx512fintrin.h: Ditto.
39574 * config/i386/i386.c: Handle new builtins.
39575 * config/i386/i386-builtin.def: Add new builtins.
39576 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
39577 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
39579 2017-01-23 Jakub Jelinek <jakub@redhat.com>
39580 Martin Liska <mliska@suse.cz>
39582 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
39583 * asan.c (asan_expand_poison_ifn): Support stores and use
39584 appropriate ASAN report function.
39585 * internal-fn.c (expand_ASAN_POISON_USE): New function.
39586 * internal-fn.def (ASAN_POISON_USE): Declare.
39587 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
39588 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
39589 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
39590 ASAN_POISON calls w/o LHS.
39591 * tree-ssa.c (execute_update_addresses_taken): Create clobber
39592 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
39593 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
39594 * gimplify.c (asan_poison_variables): Add attribute
39595 use_after_scope_memory to variables that really needs to live
39597 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
39598 having the attribute.
39600 2017-01-23 Martin Liska <mliska@suse.cz>
39602 * asan.c (create_asan_shadow_var): New function.
39603 (asan_expand_poison_ifn): Likewise.
39604 * asan.h (asan_expand_poison_ifn): New declaration.
39605 * internal-fn.c (expand_ASAN_POISON): Likewise.
39606 * internal-fn.def (ASAN_POISON): New builtin.
39607 * sanopt.c (pass_sanopt::execute): Expand
39608 asan_expand_poison_ifn.
39609 * tree-inline.c (copy_decl_for_dup_finish): Make function
39611 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
39612 * tree-ssa.c (is_asan_mark_p): New function.
39613 (execute_update_addresses_taken): Rewrite local variables
39614 (identified just by use-after-scope as addressable) into SSA.
39616 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39618 * doc/install.texi (Specific): opensource.apple.com uses https
39619 now. Remove trailing slash.
39621 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
39623 * README.Portability: Remove note on an Irix compatibility issue.
39625 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
39627 * gcov.c (INCLUDE_ALGORITHM): Define.
39628 (INCLUDE_VECTOR): Define.
39629 No longer include <vector> and <algorithm> directly.
39631 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
39633 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
39635 * doc/invoke.texi (Code Gen Options): Ditto.
39637 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
39640 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
39642 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
39644 rtl-optimization/79125
39645 * cprop.c (local_cprop_pass): Handle cases where we make an
39646 unconditional trap.
39648 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
39652 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
39653 read from, for big endian.
39655 2017-01-20 Jiong Wang <jiong.wang@arm.com>
39657 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
39658 register pauth builtins for LP64 only.
39660 2017-01-20 Marek Polacek <polacek@redhat.com>
39663 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
39666 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39668 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
39670 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
39671 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
39672 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
39674 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39677 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
39678 in big-endian mode when they are not a single duplicated value.
39680 2017-01-20 Richard Biener <rguenther@suse.de>
39682 * BASE-VER: Bump to 7.0.1.
39684 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
39686 * omp-low.c (omplow_simd_context): New struct. Use it...
39687 (lower_rec_simd_input_clauses): ...here and...
39688 (lower_rec_input_clauses): ...here to hold common data. Adjust all
39689 references to idx, lane, max_vf, is_simt.
39691 2017-01-20 Graham Markall <graham.markall@embecosm.com>
39693 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
39696 2017-01-20 Martin Jambor <mjambor@suse.cz>
39698 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
39699 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
39701 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
39702 (GTFILES): Rename hsa.c to hsa-common.c.
39703 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
39704 * hsa-dump.c: Likewise.
39705 * hsa-gen.c: Likewise.
39706 * hsa-regalloc.c: Likewise.
39707 * ipa-hsa.c: Likewise.
39708 * omp-expand.c: Likewise.
39709 * omp-low.c: Likewise.
39710 * toplev.c: Likewise.
39712 2017-01-20 Marek Polacek <polacek@redhat.com>
39715 * doc/invoke.texi: Document -Wduplicated-branches.
39716 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
39717 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
39718 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
39719 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
39720 return 0 only when not OEP_LEXICOGRAPHIC.
39721 (fold_build_cleanup_point_expr): Use the expression
39722 location when building CLEANUP_POINT_EXPR.
39723 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
39724 * tree.c (add_expr): Handle error_mark_node.
39726 2017-01-20 Martin Liska <mliska@suse.cz>
39729 * tree-profile.c (init_ic_make_global_vars): Do not call
39731 (gimple_init_gcov_profiler): Likewise.
39733 2017-01-20 Martin Liska <mliska@suse.cz>
39736 * cgraph.h (maybe_create_reference): Remove argument and
39738 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
39740 * ipa-cp.c (create_specialized_node): Likewise.
39741 * symtab.c (symtab_node::maybe_create_reference): Handle
39742 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
39744 2017-01-20 Martin Liska <mliska@suse.cz>
39746 * read-rtl-function.c (function_reader::create_function): Use
39747 build_decl instread of build_decl_stat.
39749 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
39751 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
39752 * config/i386/avx512dqintrin.h: Ditto.
39753 * config/i386/avx512fintrin.h: Ditto.
39754 * config/i386/i386-builtin-types.def: Add new types.
39755 * config/i386/i386.c: Handle new types.
39756 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
39757 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
39758 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
39759 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
39760 (__builtin_ia32_kshiftridi): New.
39761 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
39763 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
39767 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
39768 define to rs6000_init_stack_protect_guard.
39769 (rs6000_init_stack_protect_guard): New function.
39771 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39772 Yunqiang Su <yunqiang.su@imgtec.com>
39774 * config.gcc (supported_defaults): Add madd4.
39775 (with_madd4): Add validation.
39776 (all_defaults): Add madd4.
39777 * config/mips/mips.opt (mmadd4): New option.
39778 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39780 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39782 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
39783 (ISA_HAS_FUSED_MADD4): Likewise.
39784 * doc/invoke.texi (-mmadd4): Document the new option.
39785 * doc/install.texi (--with-madd4): Document the new option.
39787 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39789 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
39790 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
39791 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
39792 (aarch64_init_pauth_hint_builtins): New.
39793 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
39794 (aarch64_expand_builtin): Expand new builtins.
39796 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39798 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
39799 * combine-stack-adj.c (no_unhandled_cfa): Handle
39800 REG_CFA_TOGGLE_RA_MANGLE.
39801 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
39802 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
39803 info for return address signing.
39804 (aarch64_expand_epilogue): Likewise.
39806 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39808 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
39809 * config/aarch64/aarch64-protos.h
39810 (aarch64_return_address_signing_enabled): New declaration.
39811 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
39813 (aarch64_expand_prologue): Sign return address before it's pushed onto
39815 (aarch64_expand_epilogue): Authenticate return address fetched from
39817 (aarch64_override_options): Sanity check for ILP32 and ISA level.
39818 (aarch64_attributes): New function attributes for "sign-return-address".
39819 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
39820 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
39821 ("*do_return"): Generate combined instructions according to key index.
39822 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
39823 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
39825 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
39826 * config/aarch64/aarch64.opt (msign-return-address=): New.
39827 * doc/extend.texi (AArch64 Function Attributes): Documents
39828 "sign-return-address=".
39829 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
39831 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
39833 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
39834 overall option summary.
39836 2017-01-19 Jiong Wang <jiong.wang@arm.com>
39838 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
39839 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
39840 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
39841 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
39843 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
39845 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
39846 -mpower9-minmax by default for -mcpu=power9.
39847 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
39848 128-bit floating point.
39850 2017-01-20 Alan Modra <amodra@gmail.com>
39852 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
39853 optimizing for size.
39855 2017-01-20 Alan Modra <amodra@gmail.com>
39858 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
39859 for strcmp and strncmp from corresponding builtin decl.
39861 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39863 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
39864 instead of i386/rtems-64.h.
39865 * config/i386/rtems-64.h: Remove.
39867 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
39871 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
39873 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
39875 2017-01-19 Tamar Christina <tamar.christina@arm.com>
39877 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
39878 Change int to HOST_WIDE_INT.
39879 * config/aarch64/aarch64-protos.h
39880 (aarch64_simd_gen_const_vector_dup): Likewise.
39881 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
39883 2017-01-19 David Malcolm <dmalcolm@redhat.com>
39885 * langhooks-def.h (lhd_type_for_size): New decl.
39886 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
39887 * langhooks.c (lhd_type_for_size): New function, taken from
39890 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
39892 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
39893 define_bypass for CR latency.
39894 (power9-cracked-alu): Update bypass latency and remove power9-branch.
39895 (power9-alu2): Add define_bypass for CR latency.
39897 (power9-mul): Update insn latency.
39898 (power9-mul-compare): Update insn latency, bypass latency and remove
39901 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39903 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
39905 * config/aarch64/aarch64.md
39906 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
39907 aarch64_nopcrelative_literal_loads.
39908 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
39910 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
39912 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
39913 TARGET_LOONGSON_3A.
39914 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
39916 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
39919 * config.gcc (supported_defaults): Add lxc1-sxc1.
39920 (with_lxc1_sxc1): Add validation.
39921 (all_defaults): Add lxc1-sxc1.
39922 * config/mips/mips.opt (mlxc1-sxc1): New option.
39923 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
39925 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
39926 __mips_no_lxc1_sxc1.
39927 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
39928 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
39929 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
39931 2017-01-19 Richard Biener <rguenther@suse.de>
39933 PR tree-optimization/72488
39934 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
39935 sure to restore SSA info.
39936 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
39938 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
39940 PR rtl-optimization/79121
39941 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
39942 of the inner type when shifting an extended value.
39944 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39947 * symtab.c (symtab_node::equal_address_to): Fix comparing of
39948 interposable aliases.
39950 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
39953 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
39954 Use the evmergelohi instruction.
39955 (mov_si<mode>_e500_subreg4_2_le): Likewise.
39956 (mov_sitf_e500_subreg8_2_be): Likewise.
39957 (mov_sitf_e500_subreg12_2_le): Likewise.
39958 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
39959 (mov_si<mode>_e500_subreg4_2_be): Likewise.
39960 (mov_sitf_e500_subreg8_2_le): Likewise.
39961 (mov_sitf_e500_subreg12_2_be): Likewise.
39963 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39965 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
39966 attribute from vecsimple to vecperm.
39967 (altivec_vbpermq2): Likewise.
39969 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39972 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
39974 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39975 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
39976 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
39977 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
39978 case where N arg is SIZE_MAX.
39979 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
39980 (cmpstrsi): Add pattern.
39982 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
39984 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
39985 __builtin_vec_revb builtins.
39986 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
39987 built-in functions to support generation of the ISA 3.0 XXBR<x>
39988 vector byte reverse instructions.
39989 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
39990 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
39991 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
39992 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
39993 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
39994 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
39995 (P9V_BUILTIN_VEC_REVB): Likewise.
39996 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
39997 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
39998 (p9_xxbrq_v16qi): Likewise.
39999 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40000 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40001 (p9_xxbrh_v8hi): Likewise.
40002 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40003 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40004 vec_revb built-in functions.
40006 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40008 PR rtl-optimization/78952
40009 * config/i386/i386.md (any_extract): New code iterator.
40010 (*insvqi_2): Use any_extract for source operand.
40011 (*insvqi_3): Use any_shiftrt for source operand.
40013 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40015 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40017 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40019 2017-01-18 Matthias Klose <doko@ubuntu.com>
40021 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40023 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40025 * config/rs6000/altivec.h (vec_bperm): Change #define.
40026 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40027 (altivec_vbpermq2): New define_insn.
40028 (altivec_vbpermd): Likewise.
40029 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40030 function interface.
40031 (VBPERMD): Likewise.
40032 (VBPERM): New polymorphic function interface.
40033 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40034 Add entries for P9V_BUILTIN_VEC_VBPERM.
40035 * doc/extend.texi: Add interfaces for vec_bperm.
40037 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40039 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40040 first letter of error messages.
40041 (s390_resolve_overloaded_builtin): Likewise.
40042 * config/s390/s390.c (s390_expand_builtin): Likewise.
40043 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40044 (s390_valid_target_attribute_inner_p): Likewise.
40045 * config/s390/s390.md ("tabort"): Likewise.
40047 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40049 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40050 (ISA_AVOID_DIV_HILO): New macro.
40051 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40052 (ISA_HAS_DDIV): Likewise.
40054 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40056 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40058 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40060 * doc/invoke.texi (fabi-version): Spelling fix.
40062 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40065 * doc/invoke.texi (fabi-version): Mention mangling fix for
40068 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40071 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40073 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40076 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40077 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40079 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40081 (stack_protect_test): Ditto.
40082 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40083 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40085 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40086 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40087 -mstack-protector-guard-offset=.
40088 (RS/6000 and PowerPC Options): Ditto.
40090 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40092 * config/i386/i386.h (MASK_CLASS_P): New define.
40093 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40094 there are no registers from different register sets also when
40095 mask registers are used. Update function comment.
40096 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40097 to (*k/*r) and (*k/*km) alternatives.
40099 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40101 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40102 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40103 (EH_RETURN_HANDLER_RTX): New define.
40104 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40105 Force frame pointer in EH return functions.
40106 (aarch64_expand_epilogue): Add barrier for eh_return.
40107 (aarch64_final_eh_return_addr): Remove.
40108 (aarch64_eh_return_handler_rtx): New function.
40109 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40111 (aarch64_eh_return_handler_rtx): New prototype.
40113 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40115 * config/rs6000/altivec.h (vec_rlmi): New #define.
40116 (vec_vrlnm): Likewise.
40117 (vec_rlnm): Likewise.
40118 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40119 (UNSPEC_VRLNM): Likewise.
40120 (VIlong): New mode iterator.
40121 (altivec_vrl<VI_char>mi): New define_insn.
40122 (altivec_vrl<VI_char>nm): Likewise.
40123 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40125 (VRLDNM): Likewise.
40126 (RLNM): New polymorphic function entry.
40127 (VRLWMI): New monomorphic function entry.
40128 (VRLDMI): Likewise.
40129 (RLMI): New polymorphic function entry.
40130 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40131 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40132 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40135 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40138 * dwarf2out.c (field_byte_offset): Restore the
40139 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40140 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40141 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40144 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40147 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40149 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40152 * dwarf2out.c (add_data_member_location_attribute): For constant
40153 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40154 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40155 DW_AT_byte_size attributes.
40157 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40159 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40160 after forcing to constant memory when the code model is medium.
40162 2017-01-17 Julia Koval <julia.koval@intel.com>
40165 * config/i386/avx512fintrin.h
40166 (_mm512_i32gather_ps): Change __addr type to void const*.
40167 (_mm512_mask_i32gather_ps): Ditto.
40168 (_mm512_i32gather_pd): Ditto.
40169 (_mm512_mask_i32gather_pd): Ditto.
40170 (_mm512_i64gather_ps): Ditto.
40171 (_mm512_mask_i64gather_ps): Ditto.
40172 (_mm512_i64gather_pd): Ditto.
40173 (_mm512_mask_i64gather_pd): Ditto.
40174 (_mm512_i32gather_epi32): Ditto.
40175 (_mm512_mask_i32gather_epi32): Ditto.
40176 (_mm512_i32gather_epi64): Ditto.
40177 (_mm512_mask_i32gather_epi64): Ditto.
40178 (_mm512_i64gather_epi32): Ditto.
40179 (_mm512_mask_i64gather_epi32): Ditto.
40180 (_mm512_i64gather_epi64): Ditto.
40181 (_mm512_mask_i64gather_epi64): Ditto.
40182 (_mm512_i32scatter_ps): Change __addr type to void*.
40183 (_mm512_mask_i32scatter_ps): Ditto.
40184 (_mm512_i32scatter_pd): Ditto.
40185 (_mm512_mask_i32scatter_pd): Ditto.
40186 (_mm512_i64scatter_ps): Ditto.
40187 (_mm512_mask_i64scatter_ps): Ditto.
40188 (_mm512_i64scatter_pd): Ditto.
40189 (_mm512_mask_i64scatter_pd): Ditto.
40190 (_mm512_i32scatter_epi32): Ditto.
40191 (_mm512_mask_i32scatter_epi32): Ditto.
40192 (_mm512_i32scatter_epi64): Ditto.
40193 (_mm512_mask_i32scatter_epi64): Ditto.
40194 (_mm512_i64scatter_epi32): Ditto.
40195 (_mm512_mask_i64scatter_epi32): Ditto.
40196 (_mm512_i64scatter_epi64): Ditto.
40197 (_mm512_mask_i64scatter_epi64): Ditto.
40198 * config/i386/avx512pfintrin.h
40199 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40200 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40201 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40202 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40203 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40204 (_mm512_prefetch_i32scatter_ps): Ditto.
40205 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40206 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40207 (_mm512_prefetch_i64scatter_pd): Ditto.
40208 (_mm512_prefetch_i64scatter_ps): Ditto.
40209 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40210 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40211 * config/i386/avx512vlintrin.h
40212 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40213 (_mm_mmask_i32gather_ps): Ditto.
40214 (_mm256_mmask_i32gather_pd): Ditto.
40215 (_mm_mmask_i32gather_pd): Ditto.
40216 (_mm256_mmask_i64gather_ps): Ditto.
40217 (_mm_mmask_i64gather_ps): Ditto.
40218 (_mm256_mmask_i64gather_pd): Ditto.
40219 (_mm_mmask_i64gather_pd): Ditto.
40220 (_mm256_mmask_i32gather_epi32): Ditto.
40221 (_mm_mmask_i32gather_epi32): Ditto.
40222 (_mm256_mmask_i32gather_epi64): Ditto.
40223 (_mm_mmask_i32gather_epi64): Ditto.
40224 (_mm256_mmask_i64gather_epi32): Ditto.
40225 (_mm_mmask_i64gather_epi32): Ditto.
40226 (_mm256_mmask_i64gather_epi64): Ditto.
40227 (_mm_mmask_i64gather_epi64): Ditto.
40228 (_mm256_i32scatter_ps): Change __addr type to void*.
40229 (_mm256_mask_i32scatter_ps): Ditto.
40230 (_mm_i32scatter_ps): Ditto.
40231 (_mm_mask_i32scatter_ps): Ditto.
40232 (_mm256_i32scatter_pd): Ditto.
40233 (_mm256_mask_i32scatter_pd): Ditto.
40234 (_mm_i32scatter_pd): Ditto.
40235 (_mm_mask_i32scatter_pd): Ditto.
40236 (_mm256_i64scatter_ps): Ditto.
40237 (_mm256_mask_i64scatter_ps): Ditto.
40238 (_mm_i64scatter_ps): Ditto.
40239 (_mm_mask_i64scatter_ps): Ditto.
40240 (_mm256_i64scatter_pd): Ditto.
40241 (_mm256_mask_i64scatter_pd): Ditto.
40242 (_mm_i64scatter_pd): Ditto.
40243 (_mm_mask_i64scatter_pd): Ditto.
40244 (_mm256_i32scatter_epi32): Ditto.
40245 (_mm256_mask_i32scatter_epi32): Ditto.
40246 (_mm_i32scatter_epi32): Ditto.
40247 (_mm_mask_i32scatter_epi32): Ditto.
40248 (_mm256_i32scatter_epi64): Ditto.
40249 (_mm256_mask_i32scatter_epi64): Ditto.
40250 (_mm_i32scatter_epi64): Ditto.
40251 (_mm_mask_i32scatter_epi64): Ditto.
40252 (_mm256_i64scatter_epi32): Ditto.
40253 (_mm256_mask_i64scatter_epi32): Ditto.
40254 (_mm_i64scatter_epi32): Ditto.
40255 (_mm_mask_i64scatter_epi32): Ditto.
40256 (_mm256_i64scatter_epi64): Ditto.
40257 (_mm256_mask_i64scatter_epi64): Ditto.
40258 (_mm_i64scatter_epi64): Ditto.
40259 (_mm_mask_i64scatter_epi64): Ditto.
40260 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40261 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40262 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40263 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40264 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40265 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40266 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40267 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40268 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40269 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40270 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40271 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40272 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40273 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40274 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40275 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40276 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40277 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40278 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40279 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40280 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40281 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40282 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40283 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40284 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40285 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40286 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40287 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40288 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40289 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40290 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40291 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40292 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40293 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40294 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40295 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40296 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40297 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40298 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40299 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40300 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40301 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40302 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40303 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40304 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40305 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40306 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40307 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40308 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40309 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40310 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40311 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40312 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40313 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40314 definitions accordingly.
40316 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40317 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40320 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40323 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40326 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40327 hard regs for inner regno.
40329 2017-01-17 Martin Liska <mliska@suse.cz>
40332 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40333 assumption and add comment.
40335 2017-01-17 Nathan Sidwell <nathan@acm.org>
40337 * ipa-visibility.c (localize_node): New function, broken out of ...
40338 (function_and_variable_visibility): ... here. Call it.
40340 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40342 PR middle-end/77445
40343 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40344 correctly set frequency of oudgoing edge.
40345 (duplicate_thread_path): Fix profile updating.
40347 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40350 * configure.ac: Add GCC_BASE_VER.
40351 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40352 version from BASE-VER file.
40353 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40354 (gcc.o): Depend on $(BASEVER).
40355 * common.opt (dumpfullversion): New option.
40356 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40357 * doc/invoke.texi: Document -dumpfullversion.
40358 * doc/install.texi: Document --with-gcc-major-version-only.
40359 * configure: Regenerated.
40361 2017-01-17 Richard Biener <rguenther@suse.de>
40363 PR tree-optimization/71433
40364 * tree-vrp.c (register_new_assert_for): Merge same asserts
40365 on all incoming edges.
40366 (process_assert_insertions_for): Handle insertions at the
40369 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40371 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40372 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40374 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40377 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40380 2017-01-17 Alan Modra <amodra@gmail.com>
40383 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40384 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40385 symbolic stack limit when pic.
40387 2017-01-16 Martin Sebor <msebor@redhat.com>
40389 PR tree-optimization/78608
40390 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40392 2017-01-16 Jeff Law <law@redhat.com>
40395 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
40396 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
40397 for several include directories that may be relative to sysroot.
40398 * config/i386/x-mingw32 (gplus_includedir): Define.
40399 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
40400 (native_system_includedir): Likewise.
40401 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
40402 override if TARGET_SYSTEM_ROOT is defined.
40403 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
40405 PR tree-optimization/79090
40406 PR tree-optimization/33562
40407 PR tree-optimization/61912
40408 PR tree-optimization/77485
40409 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
40410 and computed trims into the dump file.
40412 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40414 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
40416 2017-01-16 Jakub Jelinek <jakub@redhat.com>
40419 * gimplify.c (gimplify_init_constructor): If want_value and
40420 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
40424 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
40425 sequence. Formatting fixes.
40426 (doloop_optimize): Formatting fixes.
40429 * gcc.c (debug_level_greater_than_spec_func): New function.
40430 (static_spec_functions): Add debug-level-gt spec function.
40431 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
40433 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40434 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
40435 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
40436 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
40437 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
40438 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
40440 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
40442 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
40443 QImode fixups to general and mask registers only.
40445 2017-01-16 Carl Love <cel@us.ibm.com>
40447 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
40448 for built-in functions
40449 vector signed char vec_nabs (vector signed char)
40450 vector signed short vec_nabs (vector signed short)
40451 vector signed int vec_nabs (vector signed int)
40452 vector signed long long vec_nabs (vector signed long long)
40453 vector float vec_nabs (vector float)
40454 vector double vec_nabs (vector double)
40455 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
40457 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
40458 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
40459 * doc/extend.texi: Update the documentation file for the new built-in
40462 2017-01-16 Martin Sebor <msebor@redhat.com>
40464 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
40467 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40469 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
40470 UNSPEC_VSX__XXSPLTD to require special splat handling.
40472 2017-01-16 David Malcolm <dmalcolm@redhat.com>
40475 * system.h: Poison strndup.
40477 2017-01-16 Alan Modra <amodra@gmail.com>
40480 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
40483 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
40485 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
40487 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
40489 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
40490 call recog here. Assert that INSN_CODE (insn) is non-negative.
40492 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40495 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
40497 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
40498 in the currently scheduled RTL fragment.
40500 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
40502 PR rtl-optimization/78751
40503 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
40506 2017-01-14 Jeff Law <law@redhat.com>
40508 PR tree-optimization/79090
40509 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
40510 variable length stores.
40511 (compute_trims): Delete dead assignment to *trim_tail.
40512 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
40515 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
40517 PR rtl-optimization/78626
40518 PR rtl-optimization/78727
40519 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
40520 of a block, and split such blocks after everything else is finished.
40522 2017-01-14 Alan Modra <amodra@gmail.com>
40525 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
40526 target legitimate_combined_insn.
40527 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
40528 (rs6000_legitimate_combined_insn): New function.
40529 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
40531 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
40532 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
40533 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
40535 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
40537 * doc/frontends.texi (G++ and GCC): Remove references to Java.
40539 2017-01-13 Jeff Law <law@redhat.com>
40541 PR tree-optimization/33562
40542 PR tree-optimization/61912
40543 PR tree-optimization/77485
40544 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
40546 (delete_dead_assignment): Likewise.
40547 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
40548 statement to delete_dead_call and delete_dead_assignment.
40550 2017-01-13 David Malcolm <dmalcolm@redhat.com>
40553 * substring-locations.c (format_warning_va): Strengthen case 1 so
40554 that both endpoints of the substring must be within the format
40555 range for just the substring to be printed.
40557 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
40559 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
40560 * config/i386/i386.c (ix86_target_string): Add missing options
40561 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
40562 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
40563 flags_other and ix86_target_other to flags2_other. Display unknown
40565 (ix86_valid_target_attribute_inner_p): Add missing options and
40566 reorder options by implied ISAs, as in ix86_target_string.
40568 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40570 * hash-table.h (hash_table::too_empty_p): New function.
40571 (hash_table::expand): Use it.
40572 (hash_table::traverse): Likewise.
40573 (hash_table::empty_slot): Use sizeof (value_type) instead of
40574 sizeof (PTR) to convert bytes to elements. Shrink the table
40575 if the current size is excessive for the current number of
40578 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
40580 * ira-costs.c (record_reg_classes): Break from the inner loop
40581 early once alt_fail is known to be true. Update outer loop
40582 handling accordingly.
40584 2017-01-13 Jeff Law <law@redhat.com>
40586 * tree-ssa-dse.c (decrement_count): New function.
40587 (increment_start_addr, maybe_trim_memstar_call): Likewise.
40588 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
40589 when we know the partially dead statement is a mem* function.
40591 PR tree-optimization/61912
40592 PR tree-optimization/77485
40593 * tree-ssa-dse.c: Include expr.h.
40594 (maybe_trim_constructor_store): New function.
40595 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
40597 PR tree-optimization/33562
40598 PR tree-optimization/61912
40599 PR tree-optimization/77485
40600 * doc/invoke.texi: Document new dse-max-object-size param.
40601 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
40602 * tree-ssa-dse.c: Include params.h.
40603 (dse_store_status): New enum.
40604 (initialize_ao_ref_for_dse): New, partially extracted from
40606 (valid_ao_ref_for_dse, normalize_ref): New.
40607 (setup_live_bytes_from_ref, compute_trims): Likewise.
40608 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
40609 (maybe_trim_partially_dead_store): Likewise.
40610 (maybe_trim_complex_store): Likewise.
40611 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
40612 Track what bytes live from the original store. Return tri-state
40613 for dead, partially dead or live.
40614 (dse_dom_walker): Add constructor, destructor and new private members.
40615 (delete_dead_call, delete_dead_assignment): New extracted from
40617 (dse_optimize_stmt): Make a member of dse_dom_walker.
40618 Use initialize_ao_ref_for_dse.
40620 PR tree-optimization/33562
40621 PR tree-optimization/61912
40622 PR tree-optimization/77485
40623 * sbitmap.h (bitmap_count_bits): Prototype.
40624 (bitmap_clear_range, bitmap_set_range): Likewise.
40625 * sbitmap.c (bitmap_clear_range): New function.
40626 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
40628 2017-01-13 Martin Liska <mliska@suse.cz>
40631 * function.c (set_cfun): Add new argument force.
40632 * function.h (set_cfun): Likewise.
40633 * ipa-inline-transform.c (inline_call): Use the function when
40634 strict alising from is dropped for function we inline to.
40636 2017-01-13 Richard Biener <rguenther@suse.de>
40638 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
40639 for dumping GIMPLE INTEGER_CSTs.
40641 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
40643 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
40644 to 201112L since C++17.
40646 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
40649 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
40650 if -fsanitize=kernel-address is present.
40652 2017-01-13 Richard Biener <rguenther@suse.de>
40654 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
40655 as _Literal ( type ) number in case usual suffixes do not
40656 preserve all information.
40658 2017-01-13 Richard Biener <rguenther@suse.de>
40660 PR tree-optimization/77283
40661 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
40662 and ssa-iterators.h.
40663 (is_feasible_trace): Implement a cost model based on joiner
40666 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
40669 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
40670 char or short to __float128/_Float128 directly.
40672 2017-01-12 Martin Sebor <msebor@redhat.com>
40674 to -Wformat-overflow.
40675 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
40676 (min_bytes_remaining): Same.
40677 (get_string_length): Same.
40678 (format_string): Same.
40679 (format_directive): Same.
40681 (pass_sprintf_length::handle_gimple_call): Same.
40683 2017-01-12 Jakub Jelinek <jakub@redhat.com>
40685 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
40686 info.nowrite calls with no lhs that can't throw. Return bool
40687 whether gsi_remove has been called or not.
40688 (pass_sprintf_length::handle_gimple_call): Return bool whether
40689 try_substitute_return_value called gsi_remove. Formatting fix.
40690 (pass_sprintf_length::execute): Don't use gsi_remove if
40691 handle_gimple_call returned true.
40694 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
40695 be removed due to side-effects, don't remove following barrier nor
40696 turn the successor edge into fallthru edge.
40698 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40701 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
40702 element-reversing loads and stores as not swappable.
40704 2017-01-12 Nathan Sidwell <nathan@acm.org>
40705 Nicolai Stange <nicstange@gmail.com>
40707 * combine.c (try_combine): Don't ignore result of overlap checking
40708 loop. Combine overlap & asm check into single loop.
40710 2017-01-12 Richard Biener <rguenther@suse.de>
40712 * tree-pretty-print.c (dump_generic_node): Provide -gimple
40713 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
40715 2017-01-12 Richard Biener <rguenther@suse.de>
40717 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
40718 and TS_TARGET_OPTION directly derive from TS_BASE.
40719 * tree-core.h (tree_optimization_option): Derive from tree_base.
40720 (tree_target_option): Likewise.
40722 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
40724 * config/i386/i386.c (memory_address_length): Increase len
40725 only when rip_relative_addr_p returns false.
40727 2017-01-11 Julia Koval <julia.koval@intel.com>
40729 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
40730 (OPTION_MASK_ISA_SGX_SET): New.
40731 (ix86_handle_option): Handle OPT_msgx.
40732 * config.gcc: Added sgxintrin.h.
40733 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
40734 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
40735 * config/i386/i386.c (ix86_target_string): Add -msgx.
40737 (ix86_option_override_internal): Handle new options.
40738 (ix86_valid_target_attribute_inner_p): Add sgx.
40739 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
40740 * config/i386/i386.opt: Add msgx.
40741 * config/i386/sgxintrin.h: New file.
40742 * config/i386/x86intrin.h: Add sgxintrin.h.
40744 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40747 * fold-const.c (maybe_nonzero_address): Return 1 for function
40749 (tree_single_nonzero_warnv_p): Don't handle function local objects
40753 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
40756 2017-01-11 David Malcolm <dmalcolm@redhat.com>
40759 * opts.c: Include "spellcheck.h"
40760 (struct string_fragment): New struct.
40761 (struct edit_distance_traits<const string_fragment &>): New
40763 (get_closest_sanitizer_option): New function.
40764 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
40766 2017-01-11 Jakub Jelinek <jakub@redhat.com>
40768 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
40770 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
40771 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
40772 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
40773 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
40774 for initial die_offset if dwarf_split_debug_info.
40775 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
40776 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
40778 (output_skeleton_debug_sections): Formatting fix. Use
40779 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
40780 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
40782 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
40784 * config/arm/cortex-a53.md: Add bypasses for
40785 cortex_a53_r2f_cvt.
40786 (cortex_a53_r2f): Only use for transfers.
40787 (cortex_a53_f2r): Likewise.
40788 (cortex_a53_r2f_cvt): Add reservation for conversions.
40789 (cortex_a53_f2r_cvt): Likewise.
40791 2017-01-11 Tamar Christina <tamar.christina@arm.com>
40793 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
40794 to all inlined functions, change static to extern.
40796 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
40799 * config/arm/arm.c (legitimize_pic_address): Handle reference to
40801 (arm_assemble_integer): Likewise.
40803 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
40805 * config.gcc: Use new awk script to check CPU, FPU and architecture
40806 parameters for --with-... options.
40807 * config/arm/parsecpu.awk: New file
40808 * config/arm/arm-cpus.in: New file.
40809 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
40811 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
40813 * config/arm/t-arm: Update dependency rules.
40814 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
40815 of processing .def files.
40816 * config/arm/genopt.sh: Deleted.
40817 * config/arm/gentune.sh: Deleted.
40818 * config/arm/arm-cores.def: Deleted.
40819 * config/arm/arm-arches.def: Deleted.
40820 * config/arm/arm-fpus.def: Deleted.
40821 * config/arm/arm-tune.md: Regenerated.
40822 * config/arm/arm-tables.opt: Regenerated.
40823 * config/arm/arm-cpu.h: New generated file.
40824 * config/arm/arm-cpu-data.h: New generated file.
40825 * config/arm/arm-cpu-cdata.h: New generated file.
40827 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
40830 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
40832 (input_varpool_node): Unpack dynamically_initialized bit.
40834 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
40836 PR rtl-optimization/79032
40837 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
40838 the alignment of the adjusted memory reference against that of MODE,
40839 instead of the alignment of the original memory reference.
40841 2017-01-11 Martin Jambor <mjambor@suse.cz>
40843 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
40845 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
40846 decorated functions.
40848 2017-01-11 Richard Biener <rguenther@suse.de>
40850 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
40851 set range/nonnull info for PHI results. Do not set it on
40852 stmts marked for removal.
40854 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
40856 * expr.c (store_field): In the bitfield case, fetch the return value
40857 from the registers before applying a single big-endian adjustment.
40858 Always do a final load for a BLKmode value not larger than a word.
40860 2017-01-10 David Malcolm <dmalcolm@redhat.com>
40863 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
40864 that we correctly handle column numbers greater than
40865 LINE_MAP_MAX_COLUMN_NUMBER.
40867 2017-01-10 Martin Sebor <msebor@redhat.com>
40869 PR middle-end/78245
40870 * gimple-ssa-sprintf.c (get_destination_size): Call
40871 {init,fini}object_sizes.
40872 * tree-object-size.c (addr_object_size): Adjust.
40873 (pass_through_call): Adjust.
40874 (pass_object_sizes::execute): Adjust.
40875 * tree-object-size.h (fini_object_sizes): Declare.
40877 2017-01-10 Martin Sebor <msebor@redhat.com>
40879 PR tree-optimization/78775
40880 * builtins.c (get_size_range): Move...
40881 * calls.c: ...to here.
40882 (alloc_max_size): Accept zero argument.
40883 (operand_signed_p): Remove.
40884 (maybe_warn_alloc_args_overflow): Call get_size_range.
40885 * calls.h (get_size_range): Declare.
40887 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
40889 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
40890 from TI's devices.csv file as of September 2016.
40891 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
40893 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
40895 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
40896 * doc/invoke.texi: Likewise.
40897 * doc/md.texi: Likewise.
40898 * doc/objc.texi: Likewise.
40900 2017-01-10 Joshua Conner <joshconner@google.com>
40902 * config/arm/fuchsia-elf.h: New file.
40903 * config/fuchsia.h: New file.
40904 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
40905 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
40907 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
40909 2016-01-10 Richard Biener <rguenther@suse.de>
40911 PR tree-optimization/79034
40912 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
40913 Propagate out degenerate PHIs in the joiner.
40915 2017-01-10 Martin Liska <mliska@suse.cz>
40917 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
40918 (sort_congruence_classes_by_decl_uid): Likewise.
40919 (sort_congruence_class_groups_by_decl_uid): Likewise.
40920 (sem_item_optimizer::merge_classes): Sort class, groups in these
40921 classes and members in the groups by DECL_UID of declarations.
40922 This would make merge operations stable.
40924 2017-01-10 Martin Liska <mliska@suse.cz>
40926 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
40927 usage of m_classes_vec.
40928 (sem_item_optimizer::~sem_item_optimizer): Likewise.
40929 (sem_item_optimizer::get_group_by_hash): Likewise.
40930 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
40931 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
40932 (sem_item_optimizer::verify_classes): Likewise.
40933 (sem_item_optimizer::process_cong_reduction): Likewise.
40934 (sem_item_optimizer::dump_cong_classes): Likewise.
40935 (sem_item_optimizer::merge_classes): Likewise.
40936 * ipa-icf.h (congruence_class_hash): Rename from
40937 congruence_class_group_hash. Remove declaration of m_classes_vec.
40939 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
40941 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
40942 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
40943 * config.gcc: Add avx512vpopcntdqintrin.h.
40944 * config/i386/avx512vpopcntdqintrin.h: New.
40945 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
40946 * config/i386/i386-builtin-types.def: Add new types.
40947 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
40948 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
40949 __builtin_ia32_vpopcountq_v8di_mask): New.
40950 * config/i386/i386-c.c (ix86_target_macros_internal): Define
40951 __AVX512VPOPCNTDQ__.
40952 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
40953 (PTA_AVX512VPOPCNTDQ): Define.
40954 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
40955 TARGET_AVX512VPOPCNTDQ_P): Define.
40956 * config/i386/i386.opt: Add mavx512vpopcntdq.
40957 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
40958 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
40960 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40962 PR middle-end/77484
40963 * predict.def (PRED_CALL): Set to 67.
40965 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
40967 * expr.c (store_field): In the bitfield case, if the value comes from
40968 a function call and is of an aggregate type returned in registers, do
40969 not modify the field mode; extract the value in all cases if the mode
40970 is BLKmode and the size is not larger than a word.
40972 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
40975 * config/i386/cpuid.h: Fix undefined behavior.
40977 2017-01-04 Jeff Law <law@redhat.com>
40979 PR tree-optimization/79007
40980 PR tree-optimization/67955
40981 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
40982 conservative for pt.null when flag_non_call_exceptions is on.
40984 2017-01-09 Jakub Jelinek <jakub@redhat.com>
40986 PR translation/79019
40987 PR translation/79020
40988 * params.def (PARAM_INLINE_MIN_SPEEDUP,
40989 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
40990 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
40992 * config/avr/avr.opt (maccumulate-args): Likewise.
40993 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
40994 * common.opt (freport-bug): Likewise.
40995 * cif-code.def (CIF_FINAL_ERROR): Likewise.
40996 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
40997 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
40998 translatable string.
40999 * config/i386/i386.c (function_value_32): Likewise.
41000 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41001 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41003 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41004 * common/config/msp430/msp430-common.c (msp430_handle_option):
41006 * symtab.c (symtab_node::verify_base): Likewise.
41007 * opts.c (set_debug_level): Likewise.
41008 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41009 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41010 missing whitespace to translatable strings.
41011 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41012 * config/sh/superh.h: Likewise.
41013 * config/i386/xopintrin.h: Likewise.
41014 * config/i386/znver1.md: Likewise.
41015 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41016 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41017 * double-int.h (struct double_int): Likewise.
41018 * double-int.c (div_and_round_double): Likewise.
41019 * wide-int.cc: Likewise.
41020 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41021 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41022 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41023 (crossjumps_occurred): ... this.
41024 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41027 PR tree-optimization/78899
41028 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41029 returning bool return struct loop *, NULL for failure and the new
41031 (versionable_outer_loop_p): Don't version outer loop if it has
41032 dont_vectorized bit set.
41033 (tree_if_conversion): When versioning outer loop, ensure
41034 tree_if_conversion is performed also on the inner loop of the
41035 non-vectorizable outer loop copy.
41036 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41037 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41038 prevent vectorization of it.
41039 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41040 the outer loop vectorization of the non-scalar version is attempted
41041 before vectorization of the inner loop in scalar version. If
41042 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41043 vectorization of its inner loop.
41044 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41045 has 2 inner loops, rename also on edges from bb whose single pred
41046 is outer_loop->header. Fix typo in function comment.
41048 2017-01-09 Martin Sebor <msebor@redhat.com>
41051 * asan.c (asan_emit_stack_protection): Increase local buffer size
41052 to avoid snprintf truncation warning.
41054 2017-01-09 Andrew Pinski <apinski@cavium.com>
41056 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41057 to reference thunderx2t99 for the tuning structure
41058 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41060 (thunderx2t99_extra_costs): This.
41061 * config/aarch64/aarch64-tune.md: Regenerate.
41062 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41063 (vulcan_addrcost_table): This.
41064 (vulcan_regmove_cost): Rename to ...
41065 (thunderx2t99_regmove_cost): This.
41066 (vulcan_vector_cost): Rename to ...
41067 (thunderx2t99_vector_cost): this.
41068 (vulcan_branch_cost): Rename to ...
41069 (thunderx2t99_branch_cost): This.
41070 (vulcan_tunings): Rename to ...
41071 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41072 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41074 2017-01-09 Martin Jambor <mjambor@suse.cz>
41078 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41079 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41080 (propagate_vr_accross_jump_function): Use the above function for all
41081 value range computations for pass-through jump functions and type
41082 converasion from explicit value range values.
41083 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41084 parameters from TYPE_ARG_TYPES.
41085 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41086 (ipa_write_node_info): Stream type of the actual argument.
41087 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41089 2017-01-09 Martin Liska <mliska@suse.cz>
41092 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41093 (lookup_compiler): Do not show error message with have_E.
41095 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41097 PR tree-optimization/78938
41098 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41099 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41100 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41101 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41104 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41106 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41109 2017-01-09 Richard Biener <rguenther@suse.de>
41111 PR tree-optimization/78997
41112 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41113 name condition properly.
41115 2017-01-09 Richard Biener <rguenther@suse.de>
41118 * dwarf2out.c (is_cxx): New overload with context.
41119 (is_naming_typedef_decl): Use it.
41121 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41123 * invoke.texi (Option Summary): Correct spacing in option lists
41124 and add line breaks to fix over-long lines.
41126 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41128 PR middle-end/17660
41130 * extend.texi (Common Variable Attributes): Add xref to GCC
41131 Internals manual to explain mode attribute keywords.
41133 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41136 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41137 and Preprocessor Options.
41138 (Options for Linking): Document -pthread here....
41139 (RS/6000 and PowerPC Options): ...not here.
41140 (Solaris 2 Options): ...or here.
41141 * doc/cppopts.texi: Document -pthread.
41143 2017-01-08 Martin Sebor <msebor@redhat.com>
41145 PR middle-end/77708
41146 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41147 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41148 New member functions.
41149 (format_directive): Used them.
41151 (pass_sprintf_length::handle_gimple_call): Same.
41152 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41153 to avoid truncation for any argument.
41154 (extract_affine_mul): Same.
41155 * tree.c (get_file_function_name): Same.
41157 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41159 PR middle-end/77484
41160 * predict.def (PRED_INDIR_CALL): Set to 86.
41162 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41164 PR preprocessor/54124
41165 * doc/cppopts.texi: Reformat -d subtable to list the full name
41166 of the options. Add cross-reference to the docs for the general
41167 compiler -d options.
41168 * doc/invoke.texi (Developer Options): Add cross-reference to the
41169 preprocessor-specific -d option documentation.
41171 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41173 PR preprocessor/13498
41174 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41175 redudant material, and reflect new command-line options.
41176 (System Headers): Likewise.
41178 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41180 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41181 -isystem, and -idirafter. Copy-edit.
41182 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41183 default for -ftrack-macro-expansion. Delete obsolete and
41184 badly-formatted implementation details about -fdebug-cpp output.
41185 * doc/cppwarnopts.texi: Copy-edit.
41187 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41190 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41191 that the transition from a max line width >= 1<<10 to narrower
41192 lines works correctly.
41194 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41196 * doc/options.texi (PerFunction): New.
41197 * opt-functions.awk (switch_flags): Map both Optimization and
41198 PerFunction to CL_OPTIMIZATION.
41199 * opth-gen.awk: Test for PerFunction flag along with
41201 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41202 it only when the latter is present. Skip those that don't in
41203 the hash function generator.
41204 * common.opt (fvar-tracking): Mark as PerFunction instead of
41206 (fvar-tracking-assignments): Likewise.
41207 (fvar-tracking-assignments-toggle): Likewise.
41208 (fvar-tracking-uninit): Likewise.
41210 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41212 PR translation/79018
41213 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41216 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41219 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41220 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41221 TARGET_LONG_JUMP_TABLE_OFFSETS.
41222 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41223 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41224 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41225 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41226 * config/m68k/m68k.md (tablejump expander): Likewise.
41227 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41228 TARGET_LONG_JUMP_TABLE_OFFSETS.
41229 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41230 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41232 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41233 David Holsgrove <david.holsgrove@xilinx.com>
41235 * common/config/microblaze/microblaze-common.c
41236 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41237 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41239 * config/microblaze/microblaze.c (microblaze_must_save_register)
41240 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41242 (microblaze_eh_return): New function.
41243 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41244 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41245 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41246 * config/microblaze/microblaze.md (eh_return): New pattern.
41248 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41250 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41251 GCC_DIAGNOSTIC_STRINGIFY): Define.
41253 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41255 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41257 * config/arm/arm.md (<mcrr>): New.
41259 * config/arm/arm.c (arm_arch5te): New.
41260 (arm_option_override): Set arm_arch5te.
41261 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41263 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41264 (arm_mcrr_qualifiers): ... this. New.
41265 (MRRC_QUALIFIERS): Define to...
41266 (arm_mrrc_qualifiers): ... this. New.
41267 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41269 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41270 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41271 (MRRCI, mrrc, MRRC): New.
41272 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41273 VUNSPEC_MRRC2): New.
41275 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41277 * config/arm/arm.md (<mcr>): New.
41279 * config/arm/arm.c (arm_coproc_builtin_available): Add
41280 support for mcr, mrc, mcr2 and mrc2.
41281 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41282 (arm_mcr_qualifiers): ... this. New.
41283 (MRC_QUALIFIERS): Define to ...
41284 (arm_mrc_qualifiers): ... this. New.
41285 (MCR_QUALIFIERS): Define to ...
41286 (arm_mcr_qualifiers): ... this. New.
41287 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41289 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41290 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41291 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41292 VUNSPEC_MRC2): New.
41294 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41296 * config/arm/arm.md (*ldc): New.
41300 * config/arm/arm.c (arm_coproc_builtin_available): Add
41301 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41302 (arm_coproc_ldc_stc_legitimate_address): New.
41303 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41304 'qualifier_const_pointer'.
41305 (LDC_QUALIFIERS): Define to...
41306 (arm_ldc_qualifiers): ... this. New.
41307 (STC_QUALIFIERS): Define to...
41308 (arm_stc_qualifiers): ... this. New.
41309 * config/arm/arm-protos.h
41310 (arm_coproc_ldc_stc_legitimate_address): New.
41311 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41312 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41313 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41314 stc2, stcl, stc2l): New.
41315 * config/arm/constraints.md (Uz): New.
41316 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41317 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41318 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41319 VUNSPEC_STC2L): New.
41321 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41323 * config/arm/arm.md (<cdp>): New.
41324 * config/arm/arm.c (neon_const_bounds): Rename this ...
41325 (arm_const_bounds): ... this.
41326 (arm_coproc_builtin_available): New.
41327 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41328 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41329 (CDP_QUALIFIERS): Define to...
41330 (arm_cdp_qualifiers): ... this. New.
41332 (arm_expand_builtin_args): Add case for 6 arguments.
41333 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41334 (arm_const_bounds): ... this.
41335 (arm_coproc_builtin_available): New.
41336 * config/arm/arm_acle.h (__arm_cdp): New.
41338 * config/arm/arm_acle_builtins.def (cdp): New.
41340 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41341 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41342 'arm_const_bounds'.
41343 * config/arm/types.md (coproc): New.
41344 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41345 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41346 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41347 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41349 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41351 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41352 (UBINOP_QUALIFIERS): New.
41354 (acle_builtin_data): New. Change comment.
41355 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41356 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41357 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41358 arm_acle_builtins.def.
41359 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41360 (arm_init_acle_builtins): New.
41361 (CRC32_BUILTIN): Remove.
41362 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41363 crc32cb, crc32ch and crc32cw.
41364 (arm_init_crc32_builtins): Remove.
41365 (arm_init_builtins): Use arm_init_acle_builtins rather
41366 than arm_init_crc32_builtins.
41367 (arm_expand_acle_builtin): New.
41368 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41369 * config/arm/arm_acle_builtins.def: New.
41371 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41373 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41374 (arm_builtin_datum): ... this.
41375 (arm_init_neon_builtin): Rename to ...
41376 (arm_init_builtin): ... this. Add a new parameters PREFIX
41377 and USE_SIG_IN_NAME.
41378 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41379 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41380 'arm_builtin_datum'.
41381 (arm_init_vfp_builtins): Likewise.
41382 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41383 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41384 (arm_expand_neon_args): Rename to ...
41385 (arm_expand_builtin_args): ... this. Rename builtin_arg
41386 enum values and differentiate between ARG_BUILTIN_MEMORY
41387 and ARG_BUILTIN_NEON_MEMORY.
41388 (arm_expand_neon_builtin_1): Rename to ...
41389 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41390 values, arm_expand_builtin_args and add bool parameter NEON.
41391 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41392 (arm_expand_vfp_builtin): Likewise.
41393 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
41395 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41397 PR middle-end/77484
41398 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
41399 * predict.c (tree_estimate_probability_bb): Reverse direction of
41400 polymorphic call predictor.
41402 2017-01-06 David Malcolm <dmalcolm@redhat.com>
41404 * passes.c (execute_one_pass): Split out pass-skipping logic into...
41405 (determine_pass_name_match): ...this new function and...
41406 (should_skip_pass_p): ...this new function.
41408 2017-01-06 Nathan Sidwell <nathan@acm.org>
41410 * ipa-visibility.c (function_and_variable_visibility): Reformat
41411 comments and long lines. Remove extrneous if.
41412 * symtab.c (symtab_node::make_decl_local): Fix code format.
41413 (symtab_node::set_section_for_node): Fix comment typo.
41415 2017-01-06 Martin Liska <mliska@suse.cz>
41418 * lra-constraints.c: Rename invariant to lra_invariant.
41419 * predict.c (set_even_probabilities): Initialize e to NULL.
41421 2017-01-05 Martin Sebor <msebor@redhat.com>
41423 PR tree-optimization/78910
41424 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
41425 (format_integer): Correct off-by-one error in the handling
41426 of precision with negative numbers in signed conversions..
41428 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
41430 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
41432 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41434 PR tree-optimization/71016
41435 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
41436 factor_out_conditional_conversion. Formatting fix.
41437 (factor_out_conditional_conversion): Add cond_stmt argument.
41438 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
41439 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
41442 2017-01-05 David Malcolm <dmalcolm@redhat.com>
41444 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
41445 read-rtl-function.o, and selftest-rtl.o.
41446 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
41447 (selftest::aarch64_test_loading_full_dump): New function.
41448 (selftest::aarch64_run_selftests): New function.
41449 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
41450 selftest::aarch64_run_selftests.
41451 * config/i386/i386.c
41452 (selftest::ix86_test_loading_dump_fragment_1): New function.
41453 (selftest::ix86_test_loading_call_insn): New function.
41454 (selftest::ix86_test_loading_full_dump): New function.
41455 (selftest::ix86_test_loading_unspec): New function.
41456 (selftest::ix86_run_selftests): Call the new functions.
41457 * emit-rtl.c (maybe_set_max_label_num): New function.
41458 * emit-rtl.h (maybe_set_max_label_num): New decl.
41459 * function.c (instantiate_decls): Guard call to
41460 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
41461 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
41463 * gensupport.c (gen_reader::gen_reader): Pass "false"
41464 for new "compact" param of rtx_reader.
41465 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
41466 rather than an empty string for NULL strings.
41467 * read-md.c: Potentially include config.h rather than bconfig.h.
41468 Wrap include of errors.h with #ifdef GENERATOR_FILE.
41469 (have_error): New global, copied from errors.c.
41470 (md_reader::read_name): Rename to...
41471 (md_reader::read_name_1): ...this, adding "out_loc" param,
41472 and converting "missing name or number" to returning false, rather
41474 (md_reader::read_name): Reimplement in terms of read_name_1.
41475 (md_reader::read_name_or_nil): New function.
41476 (md_reader::read_string): Handle "(nil)" by returning NULL.
41477 (md_reader::md_reader): Add new param "compact".
41478 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
41479 (md_reader::read_file): New method.
41480 * read-md.h (md_reader::md_reader): Add new param "compact".
41481 (md_reader::read_file): New method.
41482 (md_reader::is_compact): New accessor.
41483 (md_reader::read_name): Convert return type from void to file_location.
41484 (md_reader::read_name_or_nil): New decl.
41485 (md_reader::read_name_1): New decl.
41486 (md_reader::m_compact): New field.
41487 (noop_reader::noop_reader): Pass "false" for new "compact" param
41489 (rtx_reader::rtx_reader): Add new "compact" param.
41490 (rtx_reader::read_rtx_operand): Make virtual and convert return
41491 type from void to rtx.
41492 (rtx_reader::read_until): New decl.
41493 (rtx_reader::handle_any_trailing_information): New virtual function.
41494 (rtx_reader::postprocess): New virtual function.
41495 (rtx_reader::finalize_string): New virtual function.
41496 (rtx_reader::m_in_call_function_usage): New field.
41497 (rtx_reader::m_reuse_rtx_by_id): New field.
41498 * read-rtl-function.c: New file.
41499 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
41500 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
41501 (selftest::verify_three_block_rtl_cfg): New decl.
41502 * read-rtl-function.h: New file.
41503 * read-rtl.c: Potentially include config.h rather than bconfig.h.
41504 For host, include function.h, memmodel.h, and emit-rtl.h.
41505 (one_time_initialization): New function.
41506 (struct compact_insn_name): New struct.
41507 (compact_insn_names): New array.
41508 (find_code): Handle insn codes in compact dumps.
41509 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
41510 (bind_subst_iter_and_attr): Likewise.
41511 (add_condition_to_string): Likewise.
41512 (add_condition_to_rtx): Likewise.
41513 (apply_attribute_uses): Likewise.
41514 (add_current_iterators): Likewise.
41515 (apply_iterators): Likewise.
41516 (initialize_iterators): Guard usage of apply_subst_iterator with
41517 #ifdef GENERATOR_FILE.
41518 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
41519 (md_reader::read_mapping): Likewise.
41520 (add_define_attr_for_define_subst): Likewise.
41521 (add_define_subst_attr): Likewise.
41522 (read_subst_mapping): Likewise.
41523 (check_code_iterator): Likewise.
41524 (rtx_reader::read_rtx): Likewise. Move one-time initialization
41526 (one_time_initialization): New function.
41527 (rtx_reader::read_until): New method.
41528 (read_flags): New function.
41529 (parse_reg_note_name): New function.
41530 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
41531 Handle reuse_rtx ids.
41532 Wrap iterator lookup within #ifdef GENERATOR_FILE.
41533 Add parsing support for RTL dumps, mirroring the special-cases in
41534 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
41535 values, and calling handle_any_trailing_information.
41536 (rtx_reader::read_rtx_operand): Convert return type from void
41537 to rtx, returning return_rtx. Handle case 'e'. Call
41538 finalize_string on XSTR and XTMPL fields.
41539 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
41540 "(nil)" values were omitted. Call the postprocess vfunc on the
41542 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
41543 class ctor. Initialize m_in_call_function_usage. Call
41544 one_time_initialization.
41545 * rtl-tests.c (selftest::test_uncond_jump): Call
41546 set_new_first_and_last_insn.
41547 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
41548 * selftest-rtl.c: New file.
41549 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
41550 (selftest::get_insn_by_uid): New decl.
41551 * selftest-run-tests.c (selftest::run_tests): Call
41552 read_rtl_function_c_tests.
41553 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
41554 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
41557 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
41559 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
41560 operands in a special way. Assert that pos+len <= mode precision.
41562 2017-01-05 Jakub Jelinek <jakub@redhat.com>
41564 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
41565 3 argument Alias with unlimited for the negative form.
41566 (fno-vect-cost-model): Removed.
41568 2017-01-05 Martin Liska <mliska@suse.cz>
41570 * hsa-gen.c (gen_hsa_divmod): New function.
41571 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
41573 2017-01-05 Martin Liska <mliska@suse.cz>
41576 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
41579 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41581 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
41582 small constant length operands.
41584 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
41586 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
41587 between loop iterations.
41589 2017-01-05 Martin Liska <mliska@suse.cz>
41592 * gimplify.c (gimplify_decl_expr): Compare to
41593 asan_poisoned_variables instread of checking flags.
41594 (gimplify_target_expr): Likewise.
41595 (gimplify_expr): Likewise.
41596 (gimplify_function_tree): Conditionally initialize
41597 asan_poisoned_variables.
41599 2017-01-04 Jeff Law <law@redhat.com>
41601 PR tree-optimizatin/78812
41602 * rtl.h (contains_mem_rtx_p): Prototype.
41603 * ifcvt.c (containts_mem_rtx_p): Move from here to...
41604 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
41605 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
41606 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
41607 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
41609 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41611 * input.c (assert_char_at_range): Default-initialize actual_range.
41613 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41615 * df-scan.c (df_ref_create_structure): Make regno unsigned,
41616 to match the caller.
41618 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41620 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
41621 insns after final jump in test to emit dummy move.
41623 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41625 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
41626 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
41628 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
41630 * multiple_target.c (create_dispatcher_calls): Init e_next.
41631 * tree-ssa-loop-split.c (split_loop): Init border.
41632 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
41635 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
41640 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
41641 (altivec_register_operand): Do not return true if the operand
41642 contains a SUBREG mixing SImode and SFmode.
41643 (vsx_register_operand): Likewise.
41644 (vsx_reg_sfsubreg_ok): New predicate.
41645 (vfloat_operand): Do not return true if the operand contains a
41646 SUBREG mixing SImode and SFmode.
41647 (vint_operand): Likewise.
41648 (vlogical_operand): Likewise.
41649 (gpc_reg_operand): Likewise.
41650 (int_reg_operand): Likewise.
41651 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
41652 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
41653 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
41655 (rs6000_emit_move_si_sf_subreg): New helper function.
41656 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
41657 fixup SUBREGs involving SImode and SFmode.
41658 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
41659 numbers for the new peephole2 optimization.
41660 (peephole2 for SFmode unions): New peephole2 to optimize cases in
41661 the GLIBC math library that do AND/IOR/XOR operations on single
41662 precision floating point.
41663 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
41664 target macros to say whether we need to avoid SUBREGs mixing
41666 (TARGET_ALLOW_SF_SUBREG): Likewise.
41667 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
41668 (UNSPEC_SI_FROM_SF): Likewise.
41669 (iorxor): Change spacing.
41670 (and_ior_xor): New iterator for AND, IOR, and XOR.
41671 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
41672 (movdi_from_sf_zero_ext): Likewise.
41673 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
41674 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
41675 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
41676 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
41677 (fms<mode>4): Likewise.
41678 (fnma<mode>4): Likewise.
41679 (fnms<mode>4): Likewise.
41680 (nfma<mode>4): Likewise.
41681 (nfms<mode>4): Likewise.
41683 2017-01-04 Marek Polacek <polacek@redhat.com>
41686 * doc/invoke.texi: Document -Wpointer-compare.
41688 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41690 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
41693 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
41694 descriptions for -gdwarf-5 and emit them as uleb128 instead of
41697 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
41700 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
41701 documentation of the powerpc_popcntb_ok attribute.
41702 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
41703 code to issue warning messages if a requested CPU configuration is
41704 not supported by the binary (assembler and loader) toolchain.
41705 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
41706 made to define a built-in function that has been disabled.
41707 (paired_init_builtins): Add assertion to prevent ICE if attempt is
41708 made to define a built-in function that has been disabled.
41709 (altivec_init_builtins): Add comment explaining why definition
41710 of the DST built-in functions is not preceded by an assertion
41711 check. Add assertions to prevent ICE if attempts are made to
41712 define an altivec predicate or an abs* built-in function that has
41714 (htm_init_builtins): Add comment explaining why definition of the
41715 htm built-in functions is not preceded by an assertion check.
41717 2017-01-04 Jeff Law <law@redhat.com>
41719 PR tree-optimizatin/67955
41720 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
41721 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
41722 the points-to solution does not include pt_null. Use DECL_PT_UID
41725 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
41727 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
41728 Use gen_int_mode instead of gen_lopwart for const_int operands.
41730 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41732 PR tree-optimization/71563
41733 * match.pd: Simplify X << Y into X if Y is known to be 0 or
41734 out of range value - has low bits known to be zero.
41736 2017-01-04 Alan Modra <amodra@gmail.com>
41738 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
41739 * configure: Regenerate.
41740 * config.in: Regenerate.
41742 2017-01-04 Jakub Jelinek <jakub@redhat.com>
41745 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
41746 a substring of the message, but strcmp with the whole message. Ifdef
41747 ENABLE_NLS, translate the message first using dgettext.
41749 2017-01-03 Jeff Law <law@redhat.com>
41751 PR tree-optimizatin/78856
41752 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
41753 (mark_threaded_blocks): Remove code to truncate thread paths that
41754 cross multiple loop headers. Instead invalidate the cached loop
41755 iteration information and handle case of a thread path walking
41756 into an irreducible region.
41758 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
41761 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
41762 assertions. Add support for doing the signbit if the IEEE 128-bit
41763 floating point value is in a GPR.
41764 * config/rs6000/rs6000.md (Fsignbit): Delete.
41765 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
41766 Update the length attribute if the value is in a GPR.
41767 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
41768 the sign or zero extension instruction, since the value is always 0/1.
41769 (signbit<mode>2_dm2): Delete using <Fsignbit>.
41772 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
41773 extracting SImode to a GPR register so that we can generate a
41774 store, limit the vector to be in a traditional Altivec register
41775 for the vextuwrx instruction.
41777 2017-01-03 Ian Lance Taylor <iant@google.com>
41779 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
41781 2017-01-03 Martin Sebor <msebor@redhat.com>
41783 PR tree-optimization/78696
41784 * gimple-ssa-sprintf.c (format_floating): Correct handling of
41785 precision. Use MPFR for %f for greater fidelity. Correct handling
41787 (pass_sprintf_length::compute_format_length): Set width and precision
41788 specified by asrerisk to void_node for vararg functions.
41789 (try_substitute_return_value): Adjust dump output.
41791 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
41793 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
41795 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
41797 * doc/invoke.texi (SPARC options): Document -mlra as the default.
41798 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
41799 -mlra/-mno-lra was passed to the compiler.
41801 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
41803 PR rtl-optimization/65618
41804 * emit-rtl.c (try_split): Move initialization of "before" and
41805 "after" to just before the call to emit_insn_after_setloc.
41807 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
41809 * doc/md.texi (Standard Names): Remove reference to Java frontend.
41811 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
41813 * dwarf2out.c (gen_enumeration_type_die): When
41814 -gno-strict-dwarf, add a DW_AT_encoding attribute.
41816 2017-01-03 Jakub Jelinek <jakub@redhat.com>
41818 PR tree-optimization/78965
41819 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
41820 Change first argument from const call_info & to call_info &. For %n
41821 set info.nowrite to false.
41823 PR middle-end/78901
41824 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
41825 possibly throwing calls.
41827 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
41828 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
41829 and fns handling, rather than in a separate case SSA_NAME.
41831 2017-01-02 Jeff Law <law@redhat.com>
41833 * config/darwin-driver.c (darwin_driver_init): Const-correctness
41834 fixes for first_period and second_period variables.
41836 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
41839 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
41840 (*insvqi_1): New insn pattern.
41841 (*insvqi_1_mem_rex64): Ditto.
41842 (*insvqi_2): Ditto.
41843 (*insvqi_3): Rename from *insvqi.
41845 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
41847 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
41849 * doc/cfg.texi (Edges): Remove reference to Java.
41850 (Maintaining the CFG): Ditto.
41852 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41854 PR middle-end/77674
41855 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
41856 transparent aliases.
41858 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41860 PR middle-end/77484
41861 * predict.def (PRED_CALL): Update hitrate.
41862 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
41863 * predict.c (tree_estimate_probability_bb): Split CALL predictor
41864 into direct/indirect/polymorphic variants.
41866 2017-01-01 Jakub Jelinek <jakub@redhat.com>
41868 Update copyright years.
41870 * gcc.c (process_command): Update copyright notice dates.
41871 * gcov-dump.c (print_version): Ditto.
41872 * gcov.c (print_version): Ditto.
41873 * gcov-tool.c (print_version): Ditto.
41874 * gengtype.c (create_file): Ditto.
41875 * doc/cpp.texi: Bump @copying's copyright year.
41876 * doc/cppinternals.texi: Ditto.
41877 * doc/gcc.texi: Ditto.
41878 * doc/gccint.texi: Ditto.
41879 * doc/gcov.texi: Ditto.
41880 * doc/install.texi: Ditto.
41881 * doc/invoke.texi: Ditto.
41883 Copyright (C) 2017 Free Software Foundation, Inc.
41885 Copying and distribution of this file, with or without modification,
41886 are permitted in any medium without royalty provided the copyright
41887 notice and this notice are preserved.