1 2017-03-09 Martin Liska <mliska@suse.cz>
3 PR tree-optimization/79631
4 * tree-chkp-opt.c (chkp_is_constant_addr): Call
5 tree_int_cst_sign_bit just for INTEGER constants.
7 2017-03-09 Martin Liska <mliska@suse.cz>
11 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
14 2017-03-09 Marek Polacek <polacek@redhat.com>
17 * tree.c (inchash::add_expr): Handle TREE_VEC.
19 2017-03-09 Martin Liska <mliska@suse.cz>
22 (chkp_narrow_size_and_offset): New function.
23 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24 (void chkp_parse_bit_field_ref): New function.
25 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
26 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28 2017-03-09 Martin Liska <mliska@suse.cz>
31 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
32 (chkp_find_bounds_1): Remove gcc_unreachable.
34 2017-03-09 Jakub Jelinek <jakub@redhat.com>
37 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
38 BUILT_IN_SYNC*, determine the access type from the size suffix and
39 always build a MEM_REF with that type. Handle forgotten
40 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
43 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
44 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
45 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
46 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
47 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
48 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
49 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
50 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
51 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
52 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
53 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
54 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
55 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
56 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
57 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
58 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
59 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
60 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
61 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
62 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
63 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
64 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
65 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
66 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
67 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
68 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
69 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
70 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
71 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
72 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
73 definitions outside of __OPTIMIZE__ guarded section.
76 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
77 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
78 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
79 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
82 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
84 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
85 ("vfenez<mode>"): Add missing constraints.
87 2017-03-08 Martin Sebor <msebor@redhat.com>
90 * config/nds32/nds32.c (nds32_option_override):
92 2017-03-08 Jakub Jelinek <jakub@redhat.com>
95 * gimplify.c (gimplify_omp_for): Replace index var in outer
96 taskloop statement with an artificial variable and add
97 OMP_CLAUSE_PRIVATE clause for it.
99 2017-03-08 Richard Biener <rguenther@suse.de>
101 PR tree-optimization/79955
102 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
103 for accesses that are completely outside of the variable.
105 2017-03-08 Andrew Haley <aph@redhat.com>
107 PR tree-optimization/79943
108 * tree-ssa-loop-split.c (compute_new_first_bound): When
109 calculating the new upper bound, (END-BEG) should be added, not
112 2017-03-08 Jakub Jelinek <jakub@redhat.com>
114 * config/avr/avr.md (setmemhi): Make sure match_dup
115 operand number comes before match_scratch.
117 2017-03-08 Richard Biener <rguenther@suse.de>
119 PR tree-optimization/79920
120 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
121 with ncopies == 1 to ...
122 (vect_transform_slp_perm_load): ... here. Properly compute
123 all element loads by iterating VF times over the group. Do
124 not handle ncopies (computed in a broken way) in
125 vect_create_mask_and_perm.
127 2017-03-08 Jakub Jelinek <jakub@redhat.com>
130 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
131 is a uniform vector, use uniform_vector_p return value instead of
132 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
134 2017-03-07 Marek Polacek <polacek@redhat.com>
137 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
138 (alloca_call_type): Likewise.
140 2017-03-07 Martin Liska <mliska@suse.cz>
142 * gcov.c (process_args): Put comment to correct location.
144 2017-03-07 Martin Liska <mliska@suse.cz>
147 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
148 Use array_at_struct_end_p instead of DECL_CHAIN (field).
149 (chkp_narrow_bounds_for_field): Likewise.
150 (chkp_parse_array_and_component_ref): Pass one more argument to
153 2017-03-07 Richard Biener <rguenther@suse.de>
155 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
158 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
160 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
161 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
163 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
166 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
167 to end of description.
168 (PARAM_MAX_STORES_TO_MERGE): Likewise.
170 2017-03-07 Jakub Jelinek <jakub@redhat.com>
172 PR rtl-optimization/79901
173 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
175 (*avx512f_<code><mode>3<mask_name>): ... this.
176 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
177 iterator instead of VI8_AVX2_AVX512BW.
179 PR rtl-optimization/79901
180 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
181 min/max expander, expand it using expand_vec_cond_expr.
184 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
187 2017-03-06 Jakub Jelinek <jakub@redhat.com>
190 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
191 to void * for PCH reasons.
192 * dwarf2out.c (output_loc_operands, output_die): Cast
193 v.val_vec.array to unsigned char *.
195 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
198 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
201 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
203 PR rtl-optimization/79571
204 * lra-constraints.c (process_alt_operands): Calculate static
205 reject and subtract it from overall when only addresses will be
208 2017-03-06 Julia Koval <julia.koval@intel.com>
211 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
212 incoming stack boundary to 128 for 64-bit targets.
214 2017-03-06 Richard Biener <rguenther@suse.de>
216 PR tree-optimization/79894
217 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
218 to NULL after folding it.
220 2017-03-06 Richard Biener <rguenther@suse.de>
222 PR tree-optimization/79824
223 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
224 check disabling peeling for gaps.
226 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
228 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
229 attributes): Document gettimeofday.
231 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
233 * config/s390/s390.c (s390_option_override_internal): Set
234 PARAM_MIN_VECT_LOOP_BOUND
236 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
238 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
239 * config/s390/s390.md: Likewise.
241 2017-03-06 Jakub Jelinek <jakub@redhat.com>
244 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
245 (<avx2_avx512>_perm<mode>): Rename to ...
246 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
248 (<avx512>_perm<mode>_mask): Rename to ...
249 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
251 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
252 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
253 instead of VI8F_256_512.
254 (avx512f_perm<mode>): New define_expand.
255 (avx512f_perm<mode>_mask): Likewise.
256 (avx512f_perm<mode>_1<mask_name>): New define_insn.
257 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
259 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
261 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
262 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
264 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
266 2017-03-06 Martin Liska <mliska@suse.cz>
269 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
270 when having a SSA NAME w/o VAR_DECL assigned to it.
272 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
274 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
275 msa_dpsub_<su>_d): Fix MODE for vec_select.
277 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
279 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
281 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
283 2017-03-06 Richard Biener <rguenther@suse.de>
285 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
286 * plugin.c (register_plugin_info): Likewise.
287 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
289 2017-03-05 Jakub Jelinek <jakub@redhat.com>
291 * config/i386/sse.md (sse_storehps, sse_storelps,
292 avx_<castmode><avxsizesuffix>_<castmode>,
293 avx512f_<castmode><avxsizesuffix>_<castmode>,
294 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
295 in condition that at least one operand is not a MEM.
297 2017-03-03 Jakub Jelinek <jakub@redhat.com>
300 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
301 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
303 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
304 gimple_call_nothrow_p flag based on whether original builtin can throw.
305 If it can, emit following stmts on the fallthrough edge.
306 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
307 don't create new bb if inserting just debug stmts on the edge, try to
308 insert them on the fallthru bb or just reset debug stmts.
310 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
313 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
314 restore recog_data (including the operand rtxes inside it) around
315 the call to get_insn_template.
317 2017-03-03 Martin Sebor <msebor@redhat.com>
319 PR tree-optimization/79699
320 * context.c (context::~context): Free MPFR caches to avoid
321 a memory leak on program exit.
323 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
325 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
326 Use wide_int::ulow () instead of .elt (0).
328 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
330 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
331 (*pushxf): Limit oF constraint to 32bit targets and add oC
332 constraint for 64bit targets.
333 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
334 (*pushdf): Change rmF constraint to rmC.
336 2017-03-03 Martin Liska <mliska@suse.cz>
338 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
339 Remove unused variable.
341 2017-03-03 Jakub Jelinek <jakub@redhat.com>
344 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
345 is a memory operand, increase num_memory.
346 (ix86_expand_args_builtin): Likewise.
348 2017-03-03 Jan Hubicka <jh@suse.cz>
351 * ipa-devirt.c (maybe_record_node): Properly handle
352 __cxa_pure_virtual visibility.
354 2017-03-03 Martin Liska <mliska@suse.cz>
356 PR tree-optimization/79803
357 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
359 (pass_loop_prefetch::execute): Disabled optimization if an
360 assumption about L1 cache size is not met.
362 2017-03-03 Martin Liska <mliska@suse.cz>
364 PR rtl-optimization/79574
365 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
366 (hash_scan_set): Likewise.
367 (dump_hash_table): Likewise.
368 (hoist_code): Likewise.
370 2017-03-03 Richard Biener <rguenther@suse.de>
372 * fixed-value.c (fixed_from_string): Restore use of elt (1)
373 in place of uhigh ().
374 (fixed_convert_from_real): Likewise.
376 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
379 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
381 2017-03-03 Richard Biener <rguenther@suse.de>
384 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
385 TYPE_OVERFLOW_UNDEFINED check.
387 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
389 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
391 (vector_ae_<mode>_p): Likewise.
392 (vector_nez_<mode>_p): Likewise.
393 (vector_ne_v2di_p): Likewise.
394 (vector_ae_v2di_p): Likewise.
395 (vector_ne_<mode>_p): Likewise.
396 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
398 (vsx_tsqrt<mode>2_fe): Likewise.
400 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
403 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
405 2017-03-02 Jakub Jelinek <jakub@redhat.com>
407 PR rtl-optimization/79780
408 * cprop.c (one_cprop_pass): When second and further conditional trap
409 in a single basic block is turned into an unconditional trap, turn it
410 into a deleted note to avoid RTL verification failures.
412 2017-03-02 Richard Biener <rguenther@suse.de>
414 * fold-const.c (const_binop): Use ulow () instead of elt (0).
416 2017-03-02 Richard Biener <rguenther@suse.de>
418 PR tree-optimization/79345
420 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
421 param and abort the walk, returning -1 if it is hit.
422 (walk_aliased_vdefs): Take a limit param and pass it on.
423 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
424 defaulting to 0 and return a signed int.
425 * tree-ssa-uninit.c (struct check_defs_data): New struct.
426 (check_defs): New helper.
427 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
428 about uninitialized memory.
429 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
430 bogus uninitialized warning.
431 (fixed_convert_from_real): Likewise.
433 2017-03-02 Bin Cheng <bin.cheng@arm.com>
435 PR tree-optimization/66768
436 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
437 iv_use if base object can't be determined.
439 2017-03-02 Jakub Jelinek <jakub@redhat.com>
441 PR tree-optimization/79345
442 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
443 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
444 (get_pattern_stats): Initialize it.
445 * genemit.c (gen_expand): Verify match_scratch numbers come after
446 match_operand/match_dup numbers.
447 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
448 match_scratch numbers.
449 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
451 * config/s390/s390.md (trunctdsd2): Likewise.
453 2017-03-02 Richard Biener <rguenther@suse.de>
455 * wide-int.h (wide_int_storage::operator=): Implement in terms
458 2017-03-02 Richard Biener <rguenther@suse.de>
460 PR tree-optimization/79777
461 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
462 the to insert expression to sth existing.
464 2017-03-01 Martin Sebor <msebor@redhat.com>
467 * gimple-ssa-sprintf.c
468 (directive::known_width_and_precision): New function.
469 (format_integer): Use it.
470 (get_mpfr_format_length): Consider the full range of precision
471 when computing %g output with the # flag. Set the likely byte
472 count to 3 rather than 1 when precision is indeterminate.
473 (format_floating): Correct the lower bound of precision.
475 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
477 * doc/invoke.texi: Document default code model for 64-bit Linux.
479 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
482 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
483 udiv rather than div since input pattern is unsigned.
485 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
487 * config/i386/i386.c (print_reg): Warn for values of
488 unsupported size in integer register.
490 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
493 * config/rs6000/predicates.md (current_file_function_operand): Do
494 not allow self calls to be local if the function is replaceable.
496 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
499 * config/rs6000/altivec.h (vec_ctz and others): Change the
500 preprocessor macro that controls conditional compilation from
501 _ARCH_PWR9 to __POWER9_VECTOR__.
502 (vec_all_ne): Change parameterization of __altivec_scalar_pred
503 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
504 control (instead of _ARCH_PWR9 control) so that template
505 definition uses power9-specific function.
506 (vec_any_eq): Likewise.
507 (vec_all_ne): Change macro definition to use a power9-specific
508 expansion under #ifdef __POWER9_VECTOR__ control (instead of
510 (vec_any_eq) Likewise.
511 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
512 expansion for CMPNEF to remove support for xvcmpnesp instruction.
513 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
514 support for xvcmpnedp instruction.
515 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
516 macro expansion so that Power9 implementation of vec_all_ne does
517 not use the AltiVec predicate framework.
518 (VCMPNEH_P): Likewise.
519 (VCMPNEW_P): Likewise.
520 (VCMPNED_P): Likewise.
521 (VCMPNEFP_P): Likewise.
522 (VCMPNEDP_P): Likewise.
523 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
524 implementation of vec_any_eq to not use AltiVec predicate
526 (VCMPAEH_P): Likewise.
527 (VCMPAEW_P): Likewise.
528 (VCMPAED_P): Likewise.
529 (VCMPAEFP_P): Likewise.
530 (VCMPAEDP_P): Likewise.
531 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
532 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
533 not use the AltiVec predicate framework.
534 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
535 of vec_any_eq to not use AltiVec predicate framework.
536 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
537 support for predefined __POWER9_VECTOR__ macro to indicate that
538 Power9 instruction selection is enabled.
539 (altivec_overloaded_builtins): Remove extraneous
540 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
541 function argument types RS6000_BTI_bool_V16QI and
542 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
543 entry for overloaded function argument types RS6000_BTI_bool_V4SI
544 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
545 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
546 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
547 Power9 for implementations of vec_cmpne. Change the signature for
548 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
549 (representing vec_all_ne) to remove the previously described first
550 argument of type RS6000_BTI_INTSI, as this was an artifact of
551 reliance on the AltiVec predicate framework, which is no longer
552 used in the implementation of these functions. Add
553 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
554 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
555 since, unlike the AltiVec predicate framework implementation, we
556 do not share function descriptors between vec_alle and vec_anyeq.
557 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
558 set of modes that receive special treatment even when
559 TARGET_P9_VECTOR is true. The special treatment emits code that
560 does not depend on Power9 instructions.
561 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
562 define_expand to not rely on AltiVec predicate framework.
563 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
565 (vector_ne_v2di_p): Change this define_expand to not rely on
566 AltiVec predicate framework.
567 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
569 (vector_ne_<mode>_p): Change this define_expand to not rely on
570 AltiVec predicate framework.
571 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
573 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
574 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
576 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
577 define_insn pattern because the xvcmpne<VSs>. instruction is not
579 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
580 instruction is not supported.
582 2017-03-01 Jakub Jelinek <jakub@redhat.com>
584 * config/nvptx/nvptx.c: Include intl.h.
586 2017-03-01 Martin Jambor <mjambor@suse.cz>
589 * ipa-prop.h (ipa_bits): Removed field known.
590 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
591 to pointers. Adjusted their comments to warn about their sharing.
592 (ipcp_transformation_summary): Change bits to a vector of pointers.
593 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
594 (ipa_get_ipa_bits_for_value): Declare.
595 * tree-vrp.h (value_range): Mark as GTY((for_user)).
596 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
597 (ipa_bits_hash_table): Likewise.
598 (ipa_vr_ggc_hash_traits): Likewise.
599 (ipa_vr_hash_table): Likewise.
600 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
601 being pointers and vr_known being removed.
602 (ipa_set_jf_unknown): Likewise.
603 (ipa_get_ipa_bits_for_value): New function.
604 (ipa_set_jfunc_bits): Likewise.
605 (ipa_get_value_range): New overloaded functions.
606 (ipa_set_jfunc_vr): Likewise.
607 (ipa_compute_jump_functions_for_edge): Use the above functions to
608 construct bits and vr parts of jump functions.
609 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
610 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
612 (ipcp_grow_transformations_if_necessary): Also allocate
613 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
615 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
616 them. Fix too long lines.
617 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
618 vr_known being removed.
619 (ipa_read_jump_function): Use new setter functions to construct bits
620 and vr parts of jump functions or set them to NULL.
621 (write_ipcp_transformation_info): Adjust for bits being pointers.
622 (read_ipcp_transformation_info): Likewise.
623 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
625 Include gt-ipa-prop.h.
626 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
628 (ipcp_store_bits_results): Likewise.
629 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
630 Do not write to existing jump functions but use a temporary instead.
632 2017-03-01 Jakub Jelinek <jakub@redhat.com>
635 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
636 attempt to use its first operand as BIT_FIELD_REF base.
638 2017-03-01 Richard Biener <rguenther@suse.de>
641 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
642 interpolating formula in wrapping arithmetic.
643 (chrec_apply): Convert chrec_evaluate return value to wanted type.
645 2017-03-01 Jakub Jelinek <jakub@redhat.com>
647 PR tree-optimization/79734
648 * tree-vect-generic.c (expand_vector_condition): Optimize
649 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
650 Handle VEC_COND_EXPR where comparison has different inner width from
653 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
655 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
656 markup, and similar issues. Remove @opindex entries for things
657 that aren't options. Add missing -mmpy-option entries.
659 2017-02-28 Jakub Jelinek <jakub@redhat.com>
661 PR tree-optimization/79737
662 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
663 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
664 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
665 instead of byte_size. Formatting fix.
666 (shift_bytes_in_array_right): Formatting fix.
668 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
671 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
672 condition on optimize for the leaf function test.
674 2017-02-28 Martin Liska <mliska@suse.cz>
677 * read-rtl-function.c (function_reader::handle_unknown_directive):
678 Bail out when one uses -flto.
680 2017-02-28 Martin Liska <mliska@suse.cz>
682 * common.opt: Replace space with tabular for options of <number>
684 * config/i386/i386.opt: Show <number> value for
685 -mlarge-data-threshold.
686 * opts.c (print_filtered_help): Do not display number in hexadecimal
689 2017-02-28 Martin Liska <mliska@suse.cz>
691 * common.opt: Fix --help=option -Q for options which are of
694 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
696 * config/i386/i386.c (print_reg): Error out for values
697 of 8-bit size in invalid integer register.
699 2017-02-28 Martin Sebor <msebor@redhat.com>
701 PR tree-optimization/79691
702 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
704 2017-02-28 Jakub Jelinek <jakub@redhat.com>
707 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
708 gcc_unreachable with output_operand_lossage.
710 2017-02-28 Richard Biener <rguenther@suse.de>
712 PR tree-optimization/79740
713 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
715 (visit_nary_op): Insert the nary into the hashtable if we
717 * tree-ssa-pre.c (eliminate_insert): Robustify.
719 2017-02-28 Richard Biener <rguenther@suse.de>
722 * fold-const.c (decode_field_reference): Reject out-of-bound
725 2017-02-28 Jakub Jelinek <jakub@redhat.com>
727 * config/i386/i386.c: Include intl.h.
728 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
729 instead of just cond ? "..." : "...".
730 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
731 * coverage.c (read_counts_file): Likewise.
732 * omp-offload.c: Include intl.h.
733 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
734 of just cond ? "..." : "...".
735 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
736 of just cond ? "..." : "...".
738 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
741 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
743 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
745 * config/arm/arm-cpu-data.h: Regenerated.
747 2017-02-28 Richard Biener <rguenther@suse.de>
749 PR tree-optimization/79732
750 * tree-inline.c (expand_call_inline): Do not shadow var.
752 2017-02-28 Richard Biener <rguenther@suse.de>
754 PR tree-optimization/79723
755 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
756 address-space properly.
758 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
760 * doc/optinfo.texi (Optimization groups): Fix option used for
762 * doc/invoke.texi (-fopt-info): Document "omp".
763 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
764 (OPTGROUP_ALL): Add OPTGROUP_OMP.
765 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
766 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
767 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
769 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
771 * dumpfile.c (optgroup_options): Instead of "openmp", associate
772 OPTGROUP_OMP with "omp".
774 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
777 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
778 for arithmetic shift of unsigned V2DI.
780 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
782 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
784 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
785 (LINK_SPEC): Likewise.
786 (ARC_TLS_EXTRA_START_SPEC): Likewise.
787 (EXTRA_SPECS): Likewise.
788 (STARTFILE_SPEC): Likewise.
789 (ENDFILE_SPEC): Likewise.
790 (LIB_SPEC): Likewise.
791 (TARGET_SDATA_DEFAULT): Likewise.
792 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
793 (MULTILIB_DEFAULTS): Likewise.
794 (DWARF2_UNWIND_INFO): Likewise.
795 * config/arc/big.h: New file.
796 * config/arc/elf.h: Likewise.
797 * config/arc/linux.h: Likewise.
798 * config/arc/t-uClibc: Remove.
800 2017-02-27 Bin Cheng <bin.cheng@arm.com>
802 PR tree-optimization/77536
803 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
804 (tree_transform_and_unroll_loop): Use above function to compute the
805 estimated niter of unrolled loop and use it when scaling profile.
806 Also use count info rather than frequency if it's non-zero.
807 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
808 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
809 (vect_transform_loop): Call above function.
811 2017-02-27 Richard Biener <rguenther@suse.de>
813 PR tree-optimization/45397
814 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
815 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
816 (visit_nary_op): Add pattern matching for CSEing sign-changed
817 or truncated operations with wider ones.
819 2017-02-27 Richard Biener <rguenther@suse.de>
821 PR tree-optimization/79690
822 * tree-vect-stmts.c (vectorizable_store): Use vector type
823 built from the DR with address-space.
825 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
827 * doc/invoke.texi (Optimize Options): Refine the description
828 of asan-use-after-return.
830 2017-02-25 Alan Modra <amodra@gmail.com>
832 PR rtl-optimization/79584
833 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
834 base, not ad->base_term, the reg within base. Remove assertion
835 that ad->base == ad->base_term. Replace gen_int_mode using
836 bogus mode with const0_rtx.
838 2017-02-25 Jakub Jelinek <jakub@redhat.com>
841 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
842 FMA_EXPR like tcc_binary or tcc_unary.
844 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
847 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
849 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
850 (output_loc_operands): Handle DW_OP_call_ref and
851 DW_OP_GNU_variable_value.
852 (struct variable_value_struct): New type.
853 (struct variable_value_hasher): Likewise.
854 (variable_value_hash): New variable.
855 (string_types): Remove.
856 (copy_loc_descr): New function.
857 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
858 (prepend_loc_descr_to_each): New function.
859 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
860 instead of add_loc_descr_to_each if the first argument is single
861 location list and the second has multiple.
862 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
863 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
864 when looking for variable value which doesn't have other location info.
865 (loc_list_from_tree): Formatting fix.
866 (gen_array_type_die): Simplify DW_AT_string_length handling.
867 (adjust_string_types): Remove.
868 (gen_subprogram_die): Don't call adjust_string_types nor test/set
869 string_types. Call resolve_variable_values.
870 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
871 (resolve_addr_in_expr): Likewise. Add A argument.
872 (copy_deref_exprloc): Remove deref argument. Adjust for the
873 original expression being DW_OP_GNU_variable_value with optionally
874 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
876 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
877 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
878 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
879 (variable_value_hasher::hash, variable_value_hasher::equal): New
881 (resolve_variable_value_in_expr, resolve_variable_value,
882 resolve_variable_values, note_variable_value_in_expr,
883 note_variable_value): New functions.
884 (dwarf2out_early_finish): Call note_variable_value on all toplevel
887 2017-02-24 Jakub Jelinek <jakub@redhat.com>
890 * opts.h (handle_generated_option): Add GENERATED_P argument.
891 * opts-common.c (handle_option): Adjust function comment.
892 (handle_generated_option): Add GENERATED_P argument, pass it to
894 (control_warning_option): Pass false to handle_generated_option
896 * opts.c (maybe_default_option): Pass true to handle_generated_option
898 * optc-gen.awk: Likewise.
900 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
902 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
903 a REG, look at the REG it is a SUBREG of.
904 (splitter for cmpeqsi_t): Ditto.
906 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
908 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
909 the special USEs with the pattern of the insn, not the insn itself.
911 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
914 * doc/invoke.texi: Document -mload-store-pairs.
916 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
917 Sandra Loosemore <sandra@codesourcery.com>
919 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
920 argument isn't a CONST_INT.
921 (nios2_alternate_compare_const): Assert op is a CONST_INT.
922 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
923 (nios2_validate_compare): Bypass alternate compare logic if *op2
925 (ldstwm_operation_p): Return false if first_base is not a REG or
926 if first_offset is not a CONST_INT.
928 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
930 * config/cris/cris.md: Use correct operand in a define_peephole2.
932 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
934 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
936 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
938 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
939 this_insn if it is an INSN or JUMP_INSN.
940 (force_offsettable): Look at base, not at addr.
941 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
942 on things that aren't necessarily CONST_INTs.
944 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
946 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
947 -mfpmath=sse is the default also for x86-32 targets with SSE2
948 instruction set when @option{-ffast-math} is enabled
950 2017-02-24 Jeff Law <law@redhat.com>
952 PR rtl-optimizatoin/79286
953 * ira.c (update_equiv_regs): Drop may_trap_p exception to
956 2017-02-24 Richard Biener <rguenther@suse.de>
958 PR tree-optimization/79389
959 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
962 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
964 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
965 function comment to reflect reality.
966 (loop_exits_before_overflow): Fix typo in function description.
968 2017-02-24 Richard Biener <rguenther@suse.de>
970 PR tree-optimization/79389
971 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
972 properly that a threading opportunity exists. Detect conditional
973 copy/constant propagation opportunities.
975 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
977 * config/visium/visium.md (type): Add trap.
978 (b): New mode attribute.
979 (*btst): Rename into...
980 (*btst<mode>): ...this and adjust.
981 (*cbranchsi4_btst_insn): Rename into...
982 (*cbranch<mode>4_btst_insn): ...this and adjust.
983 (trap): New define_insn.
985 2017-02-23 Jakub Jelinek <jakub@redhat.com>
987 PR tree-optimization/79389
988 * ifcvt.c (struct noce_if_info): Add rev_cond field.
989 (noce_reversed_cond_code): New function.
990 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
991 reversed_comparison_code. Formatting fix.
992 (noce_try_store_flag): Test rev_cond != NULL in addition to
993 reversed_comparison_code.
994 (noce_try_store_flag_constants): Likewise.
995 (noce_try_store_flag_mask): Likewise.
996 (noce_try_addcc): Use rev_cond if non-NULL instead of
997 reversed_comparison_code.
998 (noce_try_cmove_arith): Likewise. Formatting fixes.
999 (noce_try_minmax, noce_try_abs): Clear rev_cond.
1000 (noce_find_if_block): Initialize rev_cond.
1001 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
1002 instead of false as last argument never attempt to reverse it
1005 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1007 PR tree-optimization/79663
1008 * tree-predcom.c (combine_chains): Process refs in reverse order
1009 only for ZERO length chains, and add explaining comment.
1011 2017-02-23 Jeff Law <law@redhat.com>
1013 PR tree-optimization/79578
1014 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
1015 in call to operand_equal_p.
1017 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
1020 * config/i386/cpuid.h: Fix another undefined behavior.
1022 2017-02-23 Richard Biener <rguenther@suse.de>
1024 PR tree-optimization/79683
1025 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
1026 vector types for data-refs.
1028 2017-02-23 Martin Liska <mliska@suse.cz>
1030 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
1032 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1035 * internal-fn.c (get_range_pos_neg): Moved to ...
1036 * tree.c (get_range_pos_neg): ... here. No longer static.
1037 * tree.h (get_range_pos_neg): New prototype.
1038 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
1039 are known to be in between 0 and signed maximum inclusive, try to
1040 expand both unsigned and signed divmod and use the cheaper one from
1043 2017-02-22 Jeff Law <law@redhat.com>
1045 PR tree-optimization/79578
1046 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
1047 to compare base operands.
1049 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
1052 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
1053 gpc_reg_operand instead of fpr_reg_operand.
1055 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
1057 * config/mips/mips.c (mips_return_in_memory): Force FP
1058 vector types to be returned in memory for o32 ABI.
1060 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1062 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
1063 instead of DW_TAG_member for static data member declarations and don't
1064 set no_linkage_name for static inline data members.
1065 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
1068 2017-02-22 Martin Liska <mliska@suse.cz>
1070 * doc/invoke.texi: Replace inequality signs with square brackets
1073 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1075 PR tree-optimization/68644
1076 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
1078 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
1081 * lra-constraints.c (simplify_operand_subreg): Handle
1082 WORD_REGISTER_OPERATIONS targets.
1084 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1087 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
1088 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
1089 elimination by swapping fld*.
1091 2017-02-22 Richard Biener <rguenther@suse.de>
1093 PR tree-optimization/79673
1094 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
1095 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
1096 irrelevant address-space qualifiers and avoiding a
1097 ADDR_SPACE_CONVERT_EXPR from fold_convert.
1099 2017-02-22 Richard Biener <rguenther@suse.de>
1101 PR tree-optimization/79666
1102 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
1103 to not symbolically negate if that may introduce undefined
1106 2017-02-22 Martin Liska <mliska@suse.cz>
1109 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
1110 * data-streamer-out.c (streamer_write_gcov_count_stream):
1112 * value-prof.c (stream_out_histogram_value): Make assert more
1113 precise based on type of counter.
1115 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
1118 * config/i386/i386.md (standard_x87sse_constant_load splitter):
1119 Use nonimmediate_operand instead of memory_operand for operand 1.
1120 (float-extend standard_x87sse_constant_load splitter): Ditto.
1122 2017-02-21 Jeff Law <law@redhat.com>
1124 PR tree-optimization/79621
1125 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
1126 blocks with edges to themselves.
1128 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1131 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
1132 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
1133 Use gimple_call_builtin_p.
1136 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
1137 on temporarily removed DEBUG_INSNs.
1139 PR tree-optimization/79649
1140 * tree-loop-distribution.c (classify_partition): Give up on
1141 non-generic address space loads/stores.
1143 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
1145 * doc/loop.texi (Loop manipulation): Remove nonexistent
1146 tree_ssa_loop_version from the documentation.
1147 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
1149 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1152 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
1153 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
1154 * config/rs6000/rs6000.c: Include except.h.
1155 (rs6000_expand_split_stack_prologue): Call
1156 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
1158 2017-02-21 Martin Jambor <mjambor@suse.cz>
1161 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
1164 2017-02-21 Martin Jambor <mjambor@suse.cz>
1166 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
1167 for backward compatibility only.
1168 * doc/invoke.texi (Option Summary): Remove all references to
1171 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
1175 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1177 * lra-constraints.c (curr_insn_transform): Handle
1178 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1180 2017-02-21 Martin Liska <mliska@suse.cz>
1182 * config/i386/i386.opt: Replace -masm-dialect with -masm.
1184 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
1186 PR translation/79638
1187 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
1189 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
1192 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
1193 (arm_function_ok_for_sibcall): Return false for an indirect call by
1194 descriptor if all the argument registers are used.
1195 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
1196 alignment of the function.
1198 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1200 PR tree-optimization/61441
1201 * simplify-rtx.c (simplify_const_unary_operation): For
1202 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
1203 the sNaN unmodified.
1205 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1207 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1208 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1209 instead of SYSTEM_HEADER_DIR.
1211 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
1212 Martin Liška <mliska@suse.cz>
1214 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
1215 Fix typos and grammar, use active voice, and clarify.
1217 2017-02-20 Marek Polacek <polacek@redhat.com>
1220 * gimplify.c (gimplify_expr): Handle unused *&&L;.
1223 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
1225 2017-02-20 Jakub Jelinek <jakub@redhat.com>
1228 * config/i386/i386.c (ix86_expand_builtin): Handle
1229 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
1230 ix86_builtins_isa[fcode].isa as a requirement of those
1231 flags and any other flag in the bitmask.
1232 (ix86_init_mmx_sse_builtins): Use 0 instead of
1233 ~OPTION_MASK_ISA_64BIT as mask.
1234 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
1235 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
1236 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
1237 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
1239 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1242 * lra-constraints.c (split_reg): Check requested split mode
1243 is supported by the register.
1245 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1247 * lra-constraints.c (simplify_operand_subreg): Remove early
1250 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1253 * lra-constraints.c (curr_insn_transform): Tighten condition
1254 for converting SUBREG reloads from OP_OUT to OP_INOUT.
1256 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1259 * lra-constraints.c (curr_insn_transform): Handle
1260 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1262 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
1265 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
1267 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
1269 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
1272 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
1275 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1277 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
1280 2017-02-18 Jakub Jelinek <jakub@redhat.com>
1282 * final.c (last_columnnum, override_columnnum): New variables.
1283 (final_start_function): Set last_columnnum, pass it to begin_prologue
1284 hook and pass 0 to dwarf2out_begin_prologue.
1285 (final_scan_insn): Update override_columnnum. Pass last_columnnum
1286 to source_line debug hook.
1287 (notice_source_line): Compute last_columnnum and for debug_column_info
1288 return true on column changes.
1289 * debug.h (struct gcc_debug_hooks): Add column argument to
1290 source_line and begin_prologue hooks.
1291 (debug_nothing_int_charstar_int_bool): Remove prototype.
1292 (debug_nothing_int_int_charstar,
1293 debug_nothing_int_int_charstar_int_bool): New prototypes.
1294 (dwarf2out_begin_prologue): Add column argument.
1295 * debug.c (do_nothing_debug_hooks): Adjust source_line and
1296 begin_prologue hooks.
1297 (debug_nothing_int_charstar_int_bool): Remove.
1298 (debug_nothing_int_int_charstar,
1299 debug_nothing_int_int_charstar_int_bool): New functions.
1300 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
1301 through to dwarf2out_source_line.
1302 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
1303 (dwarf2out_source_line): Add column argument, emit it if requested.
1304 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
1306 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1307 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1308 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
1309 through to dwarf2out_begin_prologue.
1310 (vmsdbgout_source_line): Add column argument, pass it through to
1311 dwarf2out_source_line.
1312 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
1313 dbxout_source_line caller.
1314 (dbxout_source_line): Add column argument.
1316 * common.opt (gno-column-info, gcolumn-info): New options.
1317 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
1318 (check_die): Also test for multiple DW_AT_decl_column attributes.
1319 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
1320 DW_AT_decl_column if requested.
1321 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
1323 (gen_variable_die): Likewise.
1324 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
1325 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
1328 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
1329 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
1330 (ix86_handle_option): Handle OPT_m3dnowa.
1331 * doc/invoke.texi (-m3dnowa): Document.
1332 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
1333 -m3dnowa instead of -m3dnow -march=athlon.
1336 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
1337 instead of gcc_assert for K, r and R code checks. Formatting fixes.
1339 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1342 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
1343 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
1344 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
1345 generator for vsx_xxpermdi_<mode>_be.
1346 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
1347 force big-endian semantics.
1348 (vsx_xxpermdi_<mode>_be): New define_expand with same
1349 implementation as previous version of vsx_xxpermdi_<mode>.
1351 2017-02-17 Jakub Jelinek <jakub@redhat.com>
1353 PR tree-optimization/79327
1354 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
1355 variable, its initialization and use.
1357 2017-02-17 Julia Koval <julia.koval@intel.com>
1359 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
1360 (OPTION_MASK_ISA_PKU_UNSET): New.
1361 (ix86_handle_option): Handle -mrdpid.
1362 * config/i386/cpuid.h (bit_RDPID): New.
1363 * config/i386/driver-i386.c (host_detect_local_cpu):
1364 Detect RDPID feature.
1365 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
1366 * config/i386/i386-c.c (ix86_target_macros_internal):
1368 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
1369 (ix86_valid_target_attribute_inner_p): Add "rdpid".
1370 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
1371 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
1372 * config/i386/i386.md (define_insn "rdpid"): New.
1373 * config/i386/i386.opt Add -mrdpid.
1374 * config/i386/immintrin.h (_rdpid_u32): New.
1376 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
1378 PR rtl-optimization/79541
1379 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
1380 instead of transforming it into USE.
1382 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
1384 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
1385 If HONOR_SNANS (SFmode) force the input to a register.
1386 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
1387 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
1388 an frsp or similar insn.
1390 2017-02-17 Martin Liska <mliska@suse.cz>
1392 PR rtl-optimization/79577
1393 * params.def (selsched-max-sched-times): Increase minimum to 1.
1395 2017-02-17 Martin Liska <mliska@suse.cz>
1397 PR rtl-optimization/79574
1398 * gcse.c (want_to_gcse_p): Prevent integer overflow.
1400 2017-02-17 Martin Liska <mliska@suse.cz>
1402 PR tree-optimization/79529
1403 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
1404 ssa_defined_default_def_p to handle cases which are implicitly
1406 * tree-ssa.c (ssa_defined_default_def_p): New function.
1407 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
1408 which are implicitly defined.
1409 * tree-ssa.h (ssa_defined_default_def_p): Declare.
1411 2017-02-17 Richard Biener <rguenther@suse.de>
1414 * params.def (max-ssa-name-query-depth): Limit to 10.
1416 2017-02-17 Richard Biener <rguenther@suse.de>
1418 PR tree-optimization/79552
1419 * tree-ssa-structalias.c (visit_loadstore): Properly verify
1422 2017-02-17 Richard Biener <rguenther@suse.de>
1425 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
1427 2017-02-17 Marek Polacek <polacek@redhat.com>
1430 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
1431 (fold_negate_expr): New wrapper.
1433 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
1435 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
1436 Correct terminology and de-emphasize pre-standard behavior.
1438 2017-02-16 Alan Modra <amodra@gmail.com>
1440 PR rtl-optimization/79286
1441 * ira.c (def_dominates_uses): New function.
1442 (update_equiv_regs): Don't create an equivalence for insns that
1443 may trap where the register def does not dominate the use.
1445 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
1447 PR rtl-optimization/78127
1448 * lra.c (lra): Call lra_eliminate before finish the loop after
1451 2017-02-16 Richard Biener <rguenther@suse.de>
1453 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
1455 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
1456 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
1457 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
1458 (isl_val_int_from_wi): New function.
1459 (extract_affine_gmp): Rename to ...
1460 (extract_affine_wi): ... this, take a widest_int.
1461 (extract_affine_int): Just wrap extract_affine_wi.
1462 (add_param_constraints): Use isl_val_int_from_wi.
1463 (add_loop_constraints): Likewise, and extract_affine_wi.
1465 2017-02-15 Jeff Law <law@redhat.com>
1468 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
1469 ira_init_register_move_cost_if_necessary.
1471 2017-02-15 Martin Sebor <msebor@redhat.com>
1474 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
1475 removed in a prior commit.
1477 2017-02-15 Bin Cheng <bin.cheng@arm.com>
1479 PR tree-optimization/79347
1480 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
1481 counters during peeling.
1483 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
1485 * Makefile.in (site.exp): Remove "set ISLVER".
1487 2017-02-15 Jakub Jelinek <jakub@redhat.com>
1490 * real.c (real_from_integer): Call real_convert even for decimal.
1492 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
1495 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
1497 2017-02-14 Andrew Pinski <apinski@cavium.com>
1499 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
1500 cores and change the partno/implementer to be correct.
1501 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
1502 the 'B" as the implementer.
1503 * config/aarch64/aarch64-tune.md: Regenerate.
1505 2017-02-14 Carl Love <cel@us.ibm.com>
1507 * config/rs6000/rs6000.c: Add case statement entry to make the
1508 xvcvuxdsp built-in argument unsigned.
1509 * config/rs6000/vsx.md: Fix the source and return operand types so they
1510 match the instruction definitions from the ISA document. Fix typo
1511 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
1514 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
1517 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
1518 member early_clobber_alts.
1519 * lra-lives.c (reg_early_clobber_p): New.
1520 (process_bb_lives): Use it.
1521 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
1522 (debug_operand_data): Initialize early_clobber_alts.
1523 (setup_operand_alternative): Set up early_clobber_alts.
1524 (collect_non_operand_hard_regs): Ditto. Pass early clobber
1525 alternatives to new_insn_reg.
1526 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
1528 (lra_update_insn_regno_info): Pass the new arg.
1530 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1533 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
1534 (new_oacc_loop_raw): Don't clear already cleared fields.
1537 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
1538 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
1539 _mm512_prefetch_i64gather_ps): New inline functions and macros.
1541 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
1544 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
1546 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
1549 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
1550 the extra instruction to the right place to store 128-bit constant
1553 2017-02-14 Martin Sebor <msebor@redhat.com>
1556 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
1557 warning for strings of unknown length.
1559 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
1561 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
1563 2017-02-14 Jeff Law <law@redhat.com>
1566 * ira-costs.c (scan_one_insn): Initialize register move costs
1567 for pseudos seen in USE/CLOBBER insns.
1569 PR tree-optimization/79095
1570 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
1571 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
1572 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
1573 if the operands are known to be not equal, then the resulting range
1575 (intersect_ranges): If the new range is ~[0,0] and the old range is
1576 wide, then prefer ~[0,0].
1577 * tree-vrp.c (overflow_comparison_p_1): New function.
1578 (overflow_comparison_p): New function.
1579 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
1580 if NAME is used in an overflow test.
1581 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
1582 overflow check that can be expressed as an equality test, then adjust
1583 ops to be that equality test.
1585 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1587 * config/s390/s390-builtin-types.def: Remove flags argument.
1588 * config/s390/s390.c (s390_init_builtins): Likewise.
1590 2017-02-14 Martin Liska <mliska@suse.cz>
1592 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
1593 vector. Fix trailing white spaces.
1595 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
1597 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
1600 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1602 PR rtl-optimization/68664
1603 * config/arm/arm.c (arm_sched_can_speculate_insn):
1604 New function. Declare prototype.
1605 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1607 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1609 PR rtl-optimization/68664
1610 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
1612 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
1614 2017-02-14 Amit Pawar <amit.pawar@amd.com>
1616 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
1617 max skip bytes for function, loop and jump.
1619 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1621 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
1622 ABS_EXPR for gimple dump.
1624 2017-02-14 Jakub Jelinek <jakub@redhat.com>
1627 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
1629 PR tree-optimization/79408
1630 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
1631 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
1632 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
1633 also if rhs1 is INTEGER_CST.
1635 2017-02-14 Richard Biener <rguenther@suse.de>
1638 * tree-into-ssa.c (insert_phi_nodes): When the function can
1639 have abnormal edges rewrite SSA names with broken use-def
1640 dominance out of SSA and register them for PHI insertion.
1642 2017-02-13 Martin Sebor <msebor@redhat.com>
1645 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
1646 clearing info.nowrite flag when snprintf size argument is a range.
1648 2017-02-13 Jakub Jelinek <jakub@redhat.com>
1650 * cprop.c (cprop_jump): Add missing space in string literal.
1651 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
1652 (get_constraint_for_component_ref): Likewise.
1653 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
1654 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
1655 * lra-constraints.c (process_alt_operands): Likewise.
1656 * ipa-inline.c (inline_small_functions): Likewise.
1657 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
1658 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
1659 * trans-mem.c (diagnose_tm_1_op): Likewise.
1660 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
1661 (grid_parallel_clauses_gridifiable): Likewise.
1663 * config/nvptx/mkoffload.c (process): Add space in between
1666 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
1667 "MOD4_SSE_REGS" and "ALL_REGS".
1669 * spellcheck.c (test_data): Add , in between "foo" and "food".
1671 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1674 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
1675 boundary crossing check and subsequent code generation agree.
1677 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1679 * config/aarch64/aarch64.c (has_memory_op): Delete.
1680 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
1683 2017-02-13 Jakub Jelinek <jakub@redhat.com>
1685 PR rtl-optimization/79388
1686 PR rtl-optimization/79450
1687 * combine.c (distribute_notes): When removing TEM_INSN for which
1688 corresponding dest has last value recorded, invalidate that last
1691 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1693 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
1694 of explicit '@'. Add missing assembly comment marker on branch costs
1697 2017-02-13 Nathan Sidwell <nathan@acm.org>
1699 * gengtype-lex.l (<in_struct>): Add '/'.
1701 2017-02-13 Martin Liska <mliska@suse.cz>
1704 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
1706 2017-02-13 Richard Biener <rguenther@suse.de>
1708 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
1710 * configure: Re-generate.
1711 * config.in: Likewise.
1712 * graphite-dependences.c: Simplify as if
1713 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
1714 * graphite-isl-ast-to-gimple.c: Likewise.
1715 * graphite-optimize-isl.c: Likewise.
1716 * graphite-poly.c: Likewise.
1717 * graphite-sese-to-poly.c: Likewise.
1718 * graphite.h: Likewise.
1719 * toplev.c: Include isl/version.h and use isl_version () for
1720 printing the ISL version.
1721 * doc/install.texi: Update ISL requirement.
1723 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
1725 * doc/standards.texi (Standards): Update reference to
1728 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
1730 * doc/extend.texi (Named Address Spaces): sourceware.org now
1732 * doc/install.texi (Binaries): Ditto.
1735 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
1737 * doc/cpp.texi: Replace "stringify"/"stringification" with C
1738 standard terminology "stringize"/"stringizing" throughout.
1739 * doc/cppinternals.texi: Likewise.
1741 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
1743 * doc/extend.texi: Fix some spelling mistakes and typos.
1744 * doc/invoke.texi: Likewise.
1746 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
1749 * params.def (inline-min-speedup) Change from 10 to 8.
1751 2017-02-11 Jakub Jelinek <jakub@redhat.com>
1753 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
1756 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
1759 * ipa-inline-analysis.c (get_minimal_bb): New function.
1760 (record_modified): Use it.
1761 (remap_edge_change_prob): Handle also ancestor functions.
1763 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
1765 * doc/contrib.texi (Contributors): Remove broken link into
1766 the Mauve CVS repository.
1768 2017-02-11 Jakub Jelinek <jakub@redhat.com>
1771 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
1772 result computation whenever lhs doesn't have vector mode, not
1773 just when it has BLKmode.
1775 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
1777 * doc/makefile.texi (profiledbootstrap): Refer to the
1778 installation instructions only in textual form.
1780 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1783 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
1785 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
1787 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
1788 (Specific): Update mingw-w64 reference.
1790 (Specific): Remove broken link to Renesas RX processor.
1792 2017-02-10 Richard Biener <rguenther@suse.de>
1794 * toplev.c (process_options): Do not mention obsolete graphite
1795 options when printing sorry message about missing graphite support.
1796 Mention -floop-nest-optimize.
1798 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
1800 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
1801 (vtst_p16): Likewise.
1802 (vtstq_p8): Likewise.
1803 (vtstq_p16): Likewise.
1805 (vtstq_p64): Likewise.
1806 * config/arm/arm_neon.h (vgetq_lane_p64): New.
1807 (vset_lane_p64): New.
1808 (vsetq_lane_p64): New.
1810 2017-02-10 Jakub Jelinek <jakub@redhat.com>
1812 PR tree-optimization/79411
1813 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
1814 stmt operands are SSA_NAMEs used in abnormal phis.
1815 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
1818 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
1821 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
1824 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
1826 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
1828 2017-02-09 Jakub Jelinek <jakub@redhat.com>
1830 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
1834 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
1835 not arbitrary TREE_CONSTANT.
1838 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
1839 "omp declare target link" attribute unless is_global_var.
1840 * omp-offload.c (find_link_var_op): Likewise.
1842 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
1843 Chung-Lin Tang <cltang@codesourcery.com>
1845 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
1847 (gimplify_adjust_omp_clauses): Don't delete TILE.
1848 (gimplify_omp_for): Deal with TILE.
1849 * internal-fn.c (expand_GOACC_TILE): New function.
1850 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
1852 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
1853 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
1855 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
1856 avoid DIV for outermost collapse var.
1857 (expand_oacc_for): Insert tile element loop as needed. Adjust.
1858 Remove out of date comments, fix whitespace.
1859 * omp-general.c (omp_extract_for_data): Deal with tiling.
1860 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
1861 adjust OLF_DIM_BASE value.
1862 (struct omp_for_data): Add tiling field.
1863 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
1864 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
1865 for auto loops. Remove default auto determining, moved to
1866 oacc_loop_fixed_partitions.
1867 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
1868 stmts, add e_mask field.
1869 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
1870 (oacc_thread_numbers): Use oacc_dim_call.
1871 (oacc_xform_tile): New.
1872 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
1873 (finish_oacc_loop): Adjust for ifns vector.
1874 (oacc_loop_discover_walk): Append loop abstraction sites to list,
1875 add case for GOACC_TILE fns.
1876 (oacc_loop_xform_loop): Delete.
1877 (oacc_loop_process): Iterate over call list directly, and add
1878 handling for GOACC_TILE fns.
1879 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
1881 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
1882 vector partitioning to outer loops. Assign 2 partitions to loops
1883 when available. Add TILE handling.
1884 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
1885 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
1886 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
1887 * tree.c (omp_clause_num_ops): Adjust TILE ops.
1888 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
1890 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
1892 * configure.ac (ACX_BUGURL): Update.
1893 * configure: Regenerate.
1895 2017-02-09 Richard Biener <rguenther@suse.de>
1897 PR tree-optimization/69823
1898 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
1899 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
1901 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
1903 * config/arc/arc-c.def: Add __NPS400__ definition.
1904 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
1905 (TARGET_NPS400): Define.
1907 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
1909 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
1911 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
1913 (arc_cpu_types): Fill the arch_info field with a pointer into the
1914 arc_arch_types table.
1915 (arc_selected_cpu): Declare.
1916 * config/arc/arc.c (arc_selected_cpu): Make global.
1917 (arc_selected_arch): Delete.
1918 (arc_base_cpu): Delete.
1919 (arc_override_options): Remove references to deleted variables,
1920 update access to arch information.
1921 (ARC_OPT): Update access to arch information.
1922 (ARC_OPTX): Likewise.
1923 * config/arc/arc.h (arc_base_cpu): Remove declaration.
1924 (TARGET_ARC600): Update access to arch information.
1925 (TARGET_ARC601): Likewise.
1926 (TARGET_ARC700): Likewise.
1927 (TARGET_EM): Likewise.
1928 (TARGET_HS): Likewise.
1929 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
1932 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
1935 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
1936 condition/operands for integer GE/LE/GEU/LEU operations.
1938 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
1940 PR translation/79397
1941 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
1944 2017-02-08 Martin Jambor <mjambor@suse.cz>
1947 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
1948 whether allocation happened.
1949 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
1950 nothing was allocated.
1952 2017-02-08 Jakub Jelinek <jakub@redhat.com>
1954 PR tree-optimization/79408
1955 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
1956 constant, but SSA_NAME with a known integer range, use the minimum
1957 of that range instead of op1 to determine if modulo can be replaced
1958 with its first operand.
1960 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1962 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
1964 2017-02-08 Richard Biener <rguenther@suse.de>
1966 PR tree-optimization/71824
1967 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
1968 Check all loops contained in the merged region.
1970 2017-02-07 Andrew Pinski <apinski@cavium.com>
1972 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
1974 2017-02-07 Andrew Pinski <apinski@cavium.com>
1976 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
1977 (thunderxt88): Likewise.
1978 (thunderxt81): Disable LSE and change v8.1 to v8.
1979 (thunderxt83): Likewise.
1981 2017-02-07 Jakub Jelinek <jakub@redhat.com>
1982 Richard Biener <rguenther@suse.de>
1985 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
1986 type from int to size_t.
1987 * ira-costs.c (struct_costs_size): Change type from int to size_t.
1989 2017-02-07 Jakub Jelinek <jakub@redhat.com>
1991 PR rtl-optimization/79386
1992 * cprop.c (bypass_conditional_jumps): Initialize
1993 bypass_last_basic_block already before splitting bbs after
1994 unconditional traps...
1995 (bypass_conditional_jumps): ... rather than here.
1998 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
1999 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
2000 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
2001 fix -masm=intel patterns.
2003 2017-02-07 Richard Biener <rguenther@suse.de>
2005 PR tree-optimization/79256
2007 * builtins.c (get_object_alignment_2): Use min_align_of_type
2008 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
2009 and ADJUST_FIELD_ALIGN.
2011 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
2013 * doc/tm.texi: Regenerate.
2014 * stor-layout.c (layout_decl): Adjust.
2015 (update_alignment_for_field): Likewise.
2016 (place_field): Likewise.
2017 (min_align_of_type): Likewise.
2018 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
2019 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
2020 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
2021 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
2022 * config/frv/frv.c (frv_adjust_field_align): Likewise.
2023 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
2024 * config/i386/i386.c (x86_field_alignment): Likewise.
2025 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
2026 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
2027 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
2028 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
2029 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
2030 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
2034 2017-01-30 Richard Biener <rguenther@suse.de>
2036 PR tree-optimization/79256
2037 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2038 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2041 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
2043 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
2044 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
2045 builtins to SImode and emit a zero-extend, if necessary.
2047 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2049 * docs/invoke.texi (RISC-V Options): Alphabetize.
2051 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2053 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
2056 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2058 * config/riscv/riscv.c: New file.
2059 * gcc/common/config/riscv/riscv-common.c: Likewise.
2060 * config.gcc: Likewise.
2061 * config/riscv/constraints.md: Likewise.
2062 * config/riscv/elf.h: Likewise.
2063 * config/riscv/generic.md: Likewise.
2064 * config/riscv/linux.h: Likewise.
2065 * config/riscv/multilib-generator: Likewise.
2066 * config/riscv/peephole.md: Likewise.
2067 * config/riscv/pic.md: Likewise.
2068 * config/riscv/predicates.md: Likewise.
2069 * config/riscv/riscv-builtins.c: Likewise.
2070 * config/riscv/riscv-c.c: Likewise.
2071 * config/riscv/riscv-ftypes.def: Likewise.
2072 * config/riscv/riscv-modes.def: Likewise.
2073 * config/riscv/riscv-opts.h: Likewise.
2074 * config/riscv/riscv-protos.h: Likewise.
2075 * config/riscv/riscv.h: Likewise.
2076 * config/riscv/riscv.md: Likewise.
2077 * config/riscv/riscv.opt: Likewise.
2078 * config/riscv/sync.md: Likewise.
2079 * config/riscv/t-elf-multilib: Likewise.
2080 * config/riscv/t-linux: Likewise.
2081 * config/riscv/t-linux-multilib: Likewise.
2082 * config/riscv/t-riscv: Likewise.
2083 * configure.ac: Likewise.
2084 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
2085 Waterman as RISC-V maintainers.
2086 * doc/install.texi: Add RISC-V entries.
2087 * doc/invoke.texi: Add RISC-V options section.
2088 * doc/md.texi: Add RISC-V constraints section.
2089 * configure: Regenerated.
2091 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2094 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
2095 false values to be constant vectors with all 0 or all 1 bits set.
2096 (vcondu<mode><mode>): Likewise.
2097 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
2099 (fpmask_comparison_operator): Update comment.
2100 (vecint_comparison_operator): New predicate.
2101 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
2102 vector conditionals when the true and false values are constant
2103 vectors with all 0 bits or all 1 bits set.
2105 2017-02-06 Martin Sebor <msebor@redhat.com>
2107 PR tree-optimization/79376
2108 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
2110 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
2112 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
2113 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
2114 to simplify split condition.
2116 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2118 * omp-expand.c (oxpand_omp_atomic_fetch_op,
2119 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
2122 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
2124 PR rtl-optimization/68664
2125 * target.def (can_speculate_insn): New hook.
2126 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
2127 * doc/tm.texi: Regenerate.
2128 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
2129 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
2130 (rs6000_sched_can_speculate_insn): New function.
2132 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2134 PR tree-optimization/79284
2135 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
2136 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
2137 vectorizable_mask_load_store, vectorizable_operation,
2138 vect_is_simple_cond, get_same_sized_vectype): Use it instead
2139 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
2140 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
2141 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
2142 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2143 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2144 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
2145 is_gimple_assign (stmt). Replace another such test with
2146 is_gimple_assign (stmt).
2148 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2151 * config/avr/avr.c (rtl-iter.h): Include it.
2152 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
2153 (avr_legitimate_combined_insn): ...and implementation.
2155 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
2157 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
2158 * config/s390/s390.c (s390_const_operand_ok)
2159 (s390_canonicalize_comparison, s390_extract_part)
2160 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
2161 (s390_contiguous_bitmask_p, s390_rtx_costs)
2162 (legitimize_pic_address): Likewise.
2163 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
2164 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
2165 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
2166 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
2167 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
2169 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2171 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
2172 REGNO($0) == REGNO($1).
2174 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2176 * config/s390/linux.h(SIZE_TYPE): Add comment.
2178 2017-02-06 Julian Brown <julian@codesourcery.com>
2179 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2180 Virendra Pathak <virendra.pathak@broadcom.com>
2182 * config/aarch64/aarch64-cores.def: Change the scheduler
2184 * config/aarch64/aarch64.md: Include thunderx2t99.md.
2185 * config/aarch64/thunderx2t99.md: New file.
2187 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2189 * doc/standards.texi (Go Language): Update link to language
2192 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2194 * tree-eh.c (lower_resx): Sanitize profile.
2195 (cleanup_empty_eh_move_lp): Likewise.
2197 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2200 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
2202 * cfgloopmanip.h (loop_version): Update prototype.
2203 * modulo-sched.c (sms_schedule): Update call of loop_version.
2204 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
2205 * tree-parloops.c (gen_parallel_loop): Likewise.
2206 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
2207 * tree-ssa-loop-split.c (split_loop): Likewise.
2208 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
2209 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2211 2017-02-05 Martin Liska <mliska@suse.cz>
2214 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
2216 (print_operand_address): Initialize a struct to zero.
2218 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2220 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
2221 garbage collector only in textual form.
2223 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2225 * doc/extend.texi (x86 specific memory model extensions for
2226 transactional memory): Simplify a phrase.
2228 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
2231 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
2232 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
2233 (atomic_storedi_1): Likewise.
2235 2017-02-04 Jakub Jelinek <jakub@redhat.com>
2237 PR tree-optimization/79338
2238 * tree-parloops.c (gather_scalar_reductions): Don't call
2239 vect_analyze_loop_form for loop->inner before destroying loop's
2242 2017-02-03 Martin Sebor <msebor@redhat.com>
2244 PR tree-optimization/79327
2245 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
2246 when precision has resulted in leading zeros.
2247 (format_integer): Adjust the likely counter to assume an unknown
2248 argument that may be zero is non-zero.
2250 2017-02-03 Jason Merrill <jason@redhat.com>
2253 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
2254 avoid copying non-taken branch.
2256 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2258 PR tree-optimization/79340
2259 * tree-vect-loop.c (vectorizable_reduction): Release
2260 vec_defs elements after safe_splicing them into other vectors.
2263 PR tree-optimization/79327
2264 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
2265 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
2267 (format_integer): Use wide_int_to_tree instead of build_int_cst
2268 + to_?hwi. If argmin is NULL, just set argmin and argmax to
2269 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
2270 of shortest and longest sequence.
2272 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
2274 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
2275 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
2277 2017-02-03 Walter Lee <walt@tilera.com>
2280 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
2281 after initial stackframe link reg save.
2282 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
2284 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2287 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
2288 wu for stxssp alternative.
2290 2017-02-03 Martin Sebor <msebor@redhat.com>
2292 PR tree-optimization/79352
2293 * gimple-fold.c (get_range_strlen): Add argument.
2294 (get_range_strlen): Change return type to bool.
2295 (get_maxval_strlen): Pass in a dummy argument.
2296 * gimple-fold.h (get_range_strlen): Change return type to bool.
2297 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
2298 * tree.h (array_at_struct_end_p): Add argument.
2299 * tree.c (array_at_struct_end_p): Handle it.
2301 2017-02-03 Martin Liska <mliska@suse.cz>
2304 * multiple_target.c (create_dispatcher_calls): Redirect edge
2305 from a caller of a dispatcher.
2306 (expand_target_clones): Make the clones local.
2307 (ipa_target_clone): Do both target clones and resolvers.
2308 (ipa_dispatcher_calls): Remove the pass.
2309 (pass_dispatcher_calls::gate): Likewise.
2310 (make_pass_dispatcher_calls): Likewise.
2311 * passes.def (pass_target_clone): Put as very first IPA early
2314 2017-02-03 Martin Liska <mliska@suse.cz>
2316 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
2317 in case of a function with ifunc attribute.
2319 2017-02-03 Martin Liska <mliska@suse.cz>
2321 * cgraph.c (cgraph_node::dump): Dump function version info.
2322 * symtab.c (symtab_node::dump_base): Add missing new line.
2324 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2326 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
2327 (ifcombine_ifandif): Use it.
2329 2017-02-03 Martin Liska <mliska@suse.cz>
2331 * doc/invoke.texi: Document default value for
2332 use-after-scope-direct-emission-threshold.
2334 2017-02-03 Martin Liska <mliska@suse.cz>
2336 PR tree-optimization/79339
2337 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
2338 (format_floating): Likewise.
2340 2017-02-03 Martin Liska <mliska@suse.cz>
2343 * ipa-prop.c (ipa_node_params_t::insert): Remove current
2345 (ipa_node_params_t::remove): Likewise.
2346 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
2347 initialization from removed ipa_node_params_t::insert.
2348 (ipa_node_params::~ipa_node_params): Move from removed
2349 ipa_node_params_t::release.
2350 * symbol-summary.h (symbol_summary::m_released): New member.
2351 Do not release a summary twice. Do not allow to call finalizer
2352 for types of a summary that live in GGC memory.
2354 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2356 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
2359 2017-02-02 Martin Sebor <msebor@redhat.com>
2362 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
2363 (format_string): Tighten up the range of output for non-constant
2364 strings and correct the expected range for wide non-constant strings.
2366 2017-02-02 Martin Sebor <msebor@redhat.com>
2368 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
2371 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
2373 (-fdump-tree-@var): Add to index and document how to come up
2374 with pass-specific option and dump file names.
2375 (-fdump-passes): Clarify where to look for output.
2377 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2380 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
2381 statistics of the analyzed path; allow threading for speed when
2382 any of BBs along the path are optimized for speed.
2384 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
2387 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
2388 settings of the virtual registers.
2391 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
2393 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2394 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
2397 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2399 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
2400 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
2402 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2404 * config/s390/s390.md: Add missing comments with the expanded
2406 * config/s390/vector.md: Likewise.
2407 * config/s390/vx-builtins.md: Likewise.
2409 2017-02-02 Jakub Jelinek <jakub@redhat.com>
2412 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
2413 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
2414 conditions on a single line.
2416 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2418 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2419 __S390_VX__ to __VX__.
2421 2017-02-01 Andrew Pinski <apinski@cavium.com>
2423 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
2424 stmt_info to record_stmt_cost.
2425 (vect_get_known_peeling_cost): Pass stmt_info if known to
2427 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
2428 cpu_vector_cost field into
2429 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
2430 field into vec_int_stmt_cost and vec_fp_stmt_cost.
2431 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
2432 splitting of scalar_stmt_cost and vec_stmt_cost.
2433 (thunderx_vector_cost): Likewise.
2434 (cortexa57_vector_cost): LIkewise.
2435 (exynosm1_vector_cost): Likewise.
2436 (xgene1_vector_cost): Likewise.
2437 (thunderx2t99_vector_cost): Improve after the splitting of the two
2439 (aarch64_builtin_vectorization_cost): Update for the splitting of
2440 scalar_stmt_cost and vec_stmt_cost.
2442 2017-02-01 Torvald Riegel <triegel@redhat.com>
2443 Richard Henderson <rth@redhat.com>
2445 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
2446 conditional on existance of a fast atomic load.
2447 * optabs-query.c (can_atomic_load_p): New function.
2448 * optabs-query.h (can_atomic_load_p): Declare it.
2449 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
2450 no fast atomic load is available for the particular size of access.
2451 (expand_atomic_compare_and_swap): Likewise.
2452 (expand_atomic_load): Likewise.
2453 (expand_atomic_store): Likewise.
2454 (expand_atomic_fetch_op): Likewise.
2455 * testsuite/lib/target-supports.exp
2456 (check_effective_target_sync_int_128): Remove x86 because it provides
2457 no fast atomic load.
2458 (check_effective_target_sync_int_128_runtime): Likewise.
2460 2017-02-01 Richard Biener <rguenther@suse.de>
2462 * graphite.c: Include tree-vectorizer.h for find_loop_location.
2463 (graphite_transform_loops): Provide opt-info for optimized nests.
2464 * tree-parloop.c (parallelize_loops): Provide opt-info for
2467 2017-02-01 Richard Biener <rguenther@suse.de>
2470 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
2473 2017-02-01 Richard Biener <rguenther@suse.de>
2475 PR tree-optimization/71824
2476 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2477 Verify the loops are valid in the merged SESE region.
2478 (scop_detection::can_represent_loop_1): Check analyzing the
2479 evolution of the number of iterations in the region succeeds.
2481 2017-01-31 Ian Lance Taylor <iant@golang.org>
2483 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
2484 REG_ARGS_SIZE note to 32-bit push insns and call insn.
2486 2017-01-31 David Malcolm <dmalcolm@redhat.com>
2488 PR preprocessor/79210
2489 * input.c (get_substring_ranges_for_loc): Replace line_width
2490 assertion with error-handling.
2492 2017-01-31 Richard Biener <rguenther@suse.de>
2494 PR tree-optimization/77318
2495 * graphite-sese-to-poly.c (extract_affine): Fix assert.
2496 (create_pw_aff_from_tree): Take loop parameter.
2497 (add_condition_to_pbb): Pass loop of the condition to
2498 create_pw_aff_from_tree.
2500 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2502 * config/s390/s390.c (s390_asan_shadow_offset): New function.
2503 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
2505 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2509 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
2511 (convert_int_to_float128): Likewise.
2512 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
2513 (convert_int_to_float128): Likewise.
2514 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
2515 (UNSPEC_IEEE128_CONVERT): Likewise.
2516 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
2517 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
2518 Use local variables for IBM extended format.
2519 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
2520 (fix_trunc<mode>si2_fprs): Likewise.
2521 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
2522 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
2523 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
2524 to know that we can now have integers of all sizes in vector
2526 (fix<uns>_<mode>di2_hw): Likewise.
2527 (float<uns>_<mode>si2_hw): Likewise.
2528 (fix_<mode>si2_hw): Likewise.
2529 (fixuns_<mode>si2_hw): Likewise.
2530 (float<uns>_<mode>di2_hw): Likewise.
2531 (float_<mode>di2_hw): Likewise.
2532 (float_<mode>si2_hw): Likewise.
2533 (floatuns_<mode>di2_hw): Likewise.
2534 (floatuns_<mode>si2_hw): Likewise.
2535 (xscvqp<su>wz_<mode>): Delete, no longer used.
2536 (xscvqp<su>dz_<mode>): Likewise.
2537 (xscv<su>dqp_<mode>): Likewise.
2538 (ieee128_mfvsrd_64bit): Likewise.
2539 (ieee128_mfvsrd_32bit): Likewise.
2540 (ieee128_mfvsrwz): Likewise.
2541 (ieee128_mtvsrw): Likewise.
2542 (ieee128_mtvsrd_64bit): Likewise.
2543 (ieee128_mtvsrd_32bit): Likewise.
2545 2017-01-31 Martin Liska <mliska@suse.cz>
2548 * ipa-prop.c (ipa_free_all_node_params): Call release method
2549 instead of ~sumbol_summary to not to trigger double times
2552 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
2554 PR tree-optimization/71691
2555 * bitmap.h (class auto_bitmap): New.
2556 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
2557 is_maybe_undefined instead of ssa_undefined_value_p.
2559 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2561 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2562 __S390_ARCH_LEVEL__ to __ARCH__.
2564 2017-01-31 Jakub Jelinek <jakub@redhat.com>
2566 PR tree-optimization/79267
2567 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
2568 if should_remove_lhs_p is true.
2570 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
2573 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
2574 (add_alignment_attribute): New.
2575 (base_type_die): Add alignment attribute.
2576 (subrange_type_die): Likewise.
2577 (modified_type_die): Likewise.
2578 (gen_array_type_die): Likewise.
2579 (gen_descr_array_type_die: Likewise.
2580 (gen_enumeration_type_die): Likewise.
2581 (gen_subprogram_die): Likewise.
2582 (gen_variable_die): Likewise.
2583 (gen_field_die): Likewise.
2584 (gen_ptr_to_mbr_type_die): Likewise.
2585 (gen_struct_or_union_type_die): Likewise.
2586 (gen_subroutine_type_die): Likewise.
2587 (gen_typedef_die): Likewise.
2588 (base_type_cmp): Compare alignment attribute.
2590 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2593 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
2594 (setb_unsigned) New pattern for setb with CCUNS.
2595 * config/rs6000/rs6000.c (expand_block_compare): Use a different
2596 subfc./subfe sequence to avoid overflow problems. Generate a
2597 shorter sequence with cmpld/setb for power9.
2598 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
2599 for generating subfc. instruction.
2600 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
2601 now uses this instruction.
2603 2017-01-30 Ian Lance Taylor <iant@google.com>
2606 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
2607 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
2609 2017-01-30 Martin Sebor <msebor@redhat.com>
2611 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
2612 Move constant to the right of a relational operator.
2613 (get_mpfr_format_length, format_character, format_string): Ditto.
2614 (should_warn_p, maybe_warn): Same.
2616 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
2618 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2621 * asan.c (get_translation_unit_decl): Remove function.
2622 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
2624 2017-01-30 Martin Liska <mliska@suse.cz>
2626 PR gcov-profile/79259
2627 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
2630 2017-01-30 Martin Liska <mliska@suse.cz>
2633 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
2634 Initialize variables with NULL value.
2636 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
2639 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
2641 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
2643 2017-01-30 Richard Biener <rguenther@suse.de>
2645 PR tree-optimization/79276
2646 * tree-vrp.c (process_assert_insertions): Properly adjust common
2647 when removing a duplicate.
2649 * gcc.dg/torture/pr79276.c: New testcase.
2651 2017-01-30 Richard Biener <rguenther@suse.de>
2653 PR tree-optimization/79256
2654 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2655 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2657 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
2659 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
2662 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
2663 ("*r<noxa>sbg_<mode>_sll_bitmask")
2664 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
2665 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
2666 Use contiguous_bitmask_nowrap_operand.
2668 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2671 * config/rs6000/altivec.h (vec_xl): Revise #define.
2672 (vec_xst): Likewise.
2674 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
2676 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
2678 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
2680 PR rtl-optimization/79194
2681 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
2682 traps before call to bypass_conditional_jumps.
2684 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
2686 PR tree-optimization/71374
2687 * lra-constraints.c (check_conflict_input_operands): New.
2688 (match_reload): Use it.
2690 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
2693 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
2694 account to calculate conflict_set.
2696 2017-01-27 Bin Cheng <bin.cheng@arm.com>
2698 PR rtl-optimization/78559
2699 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
2700 other_insn in combine.
2702 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
2704 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
2705 uint16_type_node for BT_UINT16.
2707 2017-01-27 David Malcolm <dmalcolm@redhat.com>
2709 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
2710 "RTL Tests" to menu.
2711 (GIMPLE Tests): New node.
2712 (RTL Tests): New node.
2714 2017-01-27 Richard Biener <rguenther@suse.de>
2716 PR tree-optimization/79245
2717 * tree-loop-distribution.c (distribute_loop): Apply cost
2718 modeling also to detected patterns.
2720 2017-01-27 Richard Biener <rguenther@suse.de>
2722 PR tree-optimization/71433
2723 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
2724 (compare_assert_loc): New function.
2725 (process_assert_insertions): Sort and optimize assert locations
2726 to remove duplicates and push down identical assertions on
2727 edges to their destination block.
2729 2017-01-27 Richard Biener <rguenther@suse.de>
2731 PR tree-optimization/79244
2732 * tree-vrp.c (remove_range_assertions): Forcefully propagate
2733 out SSA names even if abnormal.
2735 2017-01-27 Jakub Jelinek <jakub@redhat.com>
2737 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
2738 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
2739 instead of MPFR_RNDN.
2741 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
2744 * arm.c (arm_option_override): Don't call build_target_option_node
2745 until after doing all option overrides.
2746 (arm_valid_target_attribute_tree): Likewise.
2748 2017-01-27 Martin Liska <mliska@suse.cz>
2750 * doc/invoke.texi (-fprofile-arcs): Document profiling support
2751 for {cd}tors and C++ {cd}tors.
2753 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
2755 * config/s390/s390.md ("*setmem_long_and")
2756 ("*setmem_long_and_31z"): Use zero_extend instead of and.
2758 2017-01-26 Martin Sebor <msebor@redhat.com>
2760 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
2763 2017-01-26 Martin Sebor <msebor@redhat.com>
2765 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
2766 HAVE_DFmode before using XFmode or DFmode.
2767 (parse_directive): Avoid using the z length modifier to avoid
2768 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
2771 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
2772 to accept adjustment as an array.
2773 (get_int_range): New function.
2774 (struct directive): Make width and prec arrays.
2775 (directive::set_width, directive::set_precision): Call get_int_range.
2776 (format_integer, format_floating): Handle width and precision ranges.
2777 (format_string, parse_directive): Same.
2779 2017-01-26 Jakub Jelinek <jakub@redhat.com>
2782 * dwarf2out.c (generate_skeleton_bottom_up): For children with
2783 comdat_type_p set, just clone them, but keep the children in the
2787 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
2788 which have direct callers with -fvar-tracking-assignments enabled
2790 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
2791 inside of type units.
2793 2017-01-26 Martin Sebor <msebor@redhat.com>
2796 * gimple-ssa-sprintf.c (struct result_range): Add likely and
2798 (struct format_result): Replace number_chars, number_chars_min,
2799 and number_chars_max with a single member of struct result_range.
2801 (format_result::operator+=): Adjust.
2802 (struct fmtresult): Remove bounded. Handle likely and unlikely
2804 (fmtresult::adjust_for_width_or_precision): New function.
2805 (fmtresult:type_max_digits): New function.
2806 (bytes_remaining): Handle likely and unlikely counters.
2807 (min_bytes_remaining): Remove.
2808 (format_percent): Simplify.
2809 (format_integer, format_floating): Set likely and unlikely counters.
2810 (get_string_length, format_character, format_string): Same.
2811 (format_plain, should_warn_p): New function.
2812 (maybe_warn): Call should_warn_p. Update diagnostic messages
2813 and handle those for all directives, including plain strings.
2814 (format_directive): Handle likely and unlikely counters.
2815 Remove unnecessary quoting from diagnostics. Add an informational
2817 (add_bytes): Remove.
2818 (pass_sprintf_length::compute_format_length): Simplify.
2819 (try_substitute_return_value): Handle likely and unlikely counters.
2821 2017-01-26 Carl Love <cel@us.ibm.com>
2823 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
2824 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
2826 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
2829 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
2830 endianess for subregs into account.
2831 * lra-constraints.c (lra_constraints): Do risky transformations
2832 always on the first iteration.
2833 * lra-lives.c (check_pseudos_live_through_calls): Add arg
2834 last_call_used_reg_set.
2835 (process_bb_lives): Define and use last_call_used_reg_set.
2836 * lra.c (lra): Always continue after lra_constraints on the first
2839 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
2841 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
2843 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
2845 2017-01-26 Jakub Jelinek <jakub@redhat.com>
2847 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
2848 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
2849 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
2850 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
2851 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
2852 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
2853 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
2854 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
2855 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
2857 2017-01-26 Marek Polacek <polacek@redhat.com>
2860 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
2861 for the third operand.
2863 2017-01-26 Jakub Jelinek <jakub@redhat.com>
2866 * omp-low.c (struct omp_context): Add simt_stmt field.
2867 (scan_omp_for): Return omp_context *.
2868 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
2869 context to the _simt_ SIMD stmt.
2870 (lower_omp_for): For combined SIMD with sibling _simt_
2871 SIMD, make sure to use the same decls in _looptemp_
2872 clauses as in the sibling.
2874 2017-01-26 David Sherwood <david.sherwood@arm.com>
2877 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
2880 2017-01-26 Jakub Jelinek <jakub@redhat.com>
2883 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
2884 emit fld b; fld a; if possible.
2886 * brig-builtins.def: Update copyright years.
2887 * config/arm/arm_acle_builtins.def: Update copyright years.
2889 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
2892 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
2893 constraint instead of o for the stxsd instruction.
2895 2017-01-25 Carl Love <cel@us.ibm.com>
2897 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
2898 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
2900 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
2902 * doc/invoke.texi (C++ Dialect Options): Fix typo.
2904 2017-01-25 Richard Biener <rguenther@suse.de>
2906 PR tree-optimization/69264
2907 * target.def (vector_alignment_reachable): Improve documentation.
2908 * doc/tm.texi: Regenerate.
2909 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
2911 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
2912 earlier changes with respect to TYPE_USER_ALIGN.
2913 (vector_alignment_reachable_p): Likewise. Improve dumping.
2915 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2918 * config/arm/arm.md (xordi3): Force constant operand into a register
2921 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2923 * doc/invoke.texi (-fstore-merging): Correct default optimization
2924 levels at which it is enabled.
2925 (-O): Move -fstore-merging from list to...
2928 2017-01-25 Richard Biener <rguenther@suse.de>
2931 * omp-expand.c: Include debug.h.
2932 (expand_omp_taskreg): Make sure to generate early debug before
2933 outlining anything from a function.
2934 (expand_omp_target): Likewise.
2935 (grid_expand_target_grid_body): Likewise.
2937 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
2940 * asan.c (get_translation_unit_decl): New function.
2941 (asan_add_global): Extract modules file name from globals
2942 TRANSLATION_UNIT_DECL name.
2944 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
2947 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
2948 for long calls with APCS frame and VFP.
2950 2017-01-24 David Malcolm <dmalcolm@redhat.com>
2952 * cfg.c (original_copy_tables_initialized_p): New function.
2953 * cfg.h (original_copy_tables_initialized_p): New decl.
2954 * cfgrtl.c (relink_block_chain): Guard the call to
2955 free_original_copy_tables with a call to
2956 original_copy_tables_initialized_p.
2957 * cgraph.h (symtab_node::native_rtl_p): New decl.
2958 * cgraphunit.c (symtab_node::native_rtl_p): New function.
2959 (symtab_node::needed_p): Don't assert for early assembly output
2960 for __RTL functions.
2961 (cgraph_node::finalize_function): Set "force_output" for __RTL
2963 (cgraph_node::analyze): Bail out early for __RTL functions.
2964 (analyze_functions): Update assertion to support __RTL functions.
2965 (cgraph_node::expand): Bail out early for __RTL functions.
2966 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
2968 * function.h (struct function): Update comment for field
2970 * gimple-expr.c: Include "tree-pass.h".
2971 (gimple_has_body_p): Return false for __RTL functions.
2972 * Makefile.in (OBJS): Add run-rtl-passes.o.
2973 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
2975 (gcc::pass_manager::get_clean_slate): New accessor.
2976 * passes.c: Include "insn-addr.h".
2977 (should_skip_pass_p): Add logging. Update logic for running
2978 "expand" to be compatible with both __GIMPLE and __RTL. Guard
2979 property-provider override so it is only done for gimple passes.
2981 (skip_pass): New function.
2982 (execute_one_pass): Call skip_pass when skipping passes.
2983 * read-md.c (md_reader::read_char): Support filtering
2984 the input to a subset of line numbers.
2985 (md_reader::md_reader): Initialize fields
2986 m_first_line and m_last_line.
2987 (md_reader::read_file_fragment): New function.
2988 * read-md.h (md_reader::read_file_fragment): New decl.
2989 (md_reader::m_first_line): New field.
2990 (md_reader::m_last_line): New field.
2991 * read-rtl-function.c (function_reader::create_function): Only
2992 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
2993 curr_properties. Set DECL_INITIAL to a dummy block.
2994 (read_rtl_function_body_from_file_range): New function.
2995 * read-rtl-function.h (read_rtl_function_body_from_file_range):
2997 * run-rtl-passes.c: New file.
2998 * run-rtl-passes.h: New file.
3000 2017-01-24 Jeff Law <law@redhat.com>
3002 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
3005 2017-01-24 Bin Cheng <bin.cheng@arm.com>
3007 PR tree-optimization/79159
3008 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
3009 (record_nonwrapping_iv): Improve boundary using above function if no
3010 value range information.
3012 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
3013 Martin Jambor <mjambor@suse.cz>
3015 * brig-builtins.def: New file.
3016 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
3017 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
3018 (DEF_HSAIL_SAT_BUILTIN): Likewise.
3019 (DEF_HSAIL_INTR_BUILTIN): Likewise.
3020 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
3021 * builtin-types.def (BT_INT8): New.
3022 (BT_INT16): Likewise.
3023 (BT_UINT8): Likewise.
3024 (BT_UINT16): Likewise.
3025 (BT_FN_ULONG): Likewise.
3026 (BT_FN_UINT_INT): Likewise.
3027 (BT_FN_UINT_ULONG): Likewise.
3028 (BT_FN_UINT_LONG): Likewise.
3029 (BT_FN_UINT_PTR): Likewise.
3030 (BT_FN_ULONG_PTR): Likewise.
3031 (BT_FN_INT8_FLOAT): Likewise.
3032 (BT_FN_INT16_FLOAT): Likewise.
3033 (BT_FN_UINT32_FLOAT): Likewise.
3034 (BT_FN_UINT16_FLOAT): Likewise.
3035 (BT_FN_UINT8_FLOAT): Likewise.
3036 (BT_FN_UINT64_FLOAT): Likewise.
3037 (BT_FN_UINT16_UINT32): Likewise.
3038 (BT_FN_UINT32_UINT16): Likewise.
3039 (BT_FN_UINT16_UINT16_UINT16): Likewise.
3040 (BT_FN_INT_PTR_INT): Likewise.
3041 (BT_FN_UINT_PTR_UINT): Likewise.
3042 (BT_FN_LONG_PTR_LONG): Likewise.
3043 (BT_FN_ULONG_PTR_ULONG): Likewise.
3044 (BT_FN_VOID_UINT64_UINT64): Likewise.
3045 (BT_FN_UINT8_UINT8_UINT8): Likewise.
3046 (BT_FN_INT8_INT8_INT8): Likewise.
3047 (BT_FN_INT16_INT16_INT16): Likewise.
3048 (BT_FN_INT_INT_INT): Likewise.
3049 (BT_FN_UINT_FLOAT_UINT): Likewise.
3050 (BT_FN_FLOAT_UINT_UINT): Likewise.
3051 (BT_FN_ULONG_UINT_UINT): Likewise.
3052 (BT_FN_ULONG_UINT_PTR): Likewise.
3053 (BT_FN_ULONG_ULONG_ULONG): Likewise.
3054 (BT_FN_UINT_UINT_UINT): Likewise.
3055 (BT_FN_VOID_UINT_PTR): Likewise.
3056 (BT_FN_UINT_UINT_PTR: Likewise.
3057 (BT_FN_UINT32_UINT64_PTR): Likewise.
3058 (BT_FN_INT_INT_UINT_UINT): Likewise.
3059 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
3060 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
3061 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
3062 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
3063 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
3064 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
3065 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
3066 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
3067 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
3068 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
3069 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
3070 * doc/frontends.texi: List BRIG FE.
3071 * doc/install.texi (Testing): Add BRIG tesring requirements.
3072 * doc/invoke.texi (Overall Options): Mention BRIG.
3073 * doc/standards.texi (Standards): Doucment BRIG HSA version.
3075 2017-01-24 Richard Biener <rguenther@suse.de>
3077 PR translation/79208
3078 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
3080 2017-01-24 Martin Jambor <mjambor@suse.cz>
3083 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
3084 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
3087 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
3090 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
3091 casts from signed to unsigned really don't have a range.
3093 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
3095 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
3096 GMP_RNDx for compatiblity.
3098 2017-01-24 Martin Liska <mliska@suse.cz>
3101 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
3102 that would prevent us to call alloca with -1 as argument.
3104 2017-01-24 Jakub Jelinek <jakub@redhat.com>
3106 * dwarf2out.c (output_compilation_unit_header, output_file_names):
3107 Avoid -Wformat-security warning.
3109 2017-01-23 Andrew Pinski <apinski@cavium.com>
3111 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
3114 2017-01-23 Martin Sebor <msebor@redhat.com>
3117 * gimple-ssa-sprintf.c (warn_level): New global.
3118 (format_integer): Use it here and throughout the rest of the file.
3119 Use the same switch to compute sign as base.
3120 (maybe_warn): New function.
3121 (format_directive): Factor out warnings into maybe_warn.
3122 Add debugging output. Use warn_level.
3123 (add_bytes): Use warn_level.
3124 (pass_sprintf_length::compute_format_length): Add debugging output.
3125 (try_substitute_return_value): Same.
3126 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
3129 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
3130 (struct fmtresult, format_integer, format_floating): Adjust.
3131 (fmtresult::fmtresult): Set max correctly in two argument ctor.
3132 (get_string_length, format_string,format_directive): Same.
3133 (pass_sprintf_length::compute_format_length): Same.
3134 (try_substitute_return_value): Simplify slightly.
3137 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
3138 (fmtresult::operator+=): Outlined.
3139 (struct fmtresult): Add ctors.
3140 (struct conversion_spec): Rename...
3141 (struct directive): ...to this. Add and remove data members.
3142 (directive::set_width, directive::set_precision): New functions.
3143 (format_percent): Use fmtresult ctor.
3144 (get_width_and_precision): Remove.
3145 (format_integer): Make naming changes. Avoid computing width and
3147 (format_floating): Same. Adjust indentation.
3148 (format_character, format_none): New functions.
3149 (format_string): Moved character handling to format_character.
3150 (format_directive): Remove arguments, change return type.
3151 (parse_directive): New function.
3152 (pass_sprintf_length::compute_format_length): Move directive
3153 parsing to parse_directive.
3155 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3157 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
3158 (assign_assembler_name_if_needed): ... this.
3159 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
3160 (assign_assembler_name_if_needed): ... this.
3161 (free_lang_data_in_cgraph): Adjust callers.
3162 * cgraphunit.c (cgraph_node::analyze): Likewise.
3163 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
3166 2017-01-23 Richard Biener <rguenther@suse.de>
3168 PR tree-optimization/79088
3169 PR tree-optimization/79188
3170 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
3171 resetting loop bounds after last path deletion. Reset loop
3172 bounds of the target loop, make code match the comments.
3173 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
3174 Make sure loops need no fixups.
3176 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3178 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
3179 exponent support with double type for first argument.
3180 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
3181 type returned by __builtin_vec_extract_sig,
3182 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
3183 functions from "vector int" to "vector unsigned int" or from
3184 "vector long long int" to "vector unsigned long long int".
3185 Changed type returned by __builtin_vec_extract_exp,
3186 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
3187 functions from "vector int" to "vector unsigned int" or from
3188 "vector long long int" to "vector unsigned long long int".
3189 Changed return type of __builtin_vec_test_data_class,
3190 __builtin_vec_test_data_class_sp, and
3191 __builtin_vec_test_data_class_dp from "vector int" to
3192 "vector bool int" or from "vector long long int" to "vector bool
3193 long long int" and changed second argument type from "unsigned
3194 int" to "int". Added new overloaded function forms "vector float
3195 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
3196 "vector float __builtin_vec_insert_exp_sp (vector float, vector
3197 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
3198 double, vector unsigned long long int)" and "vector double
3199 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
3200 long int)". Changed return type of
3201 __builtin_scalar_test_data_class and
3202 __builtin_scalar_test_data_class_sp and
3203 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
3204 int" and changed second argument from "unsigned int" to "int".
3205 Changed type returned by __builtin_scalar_test_neg,
3206 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
3207 from "int" to "bool int". Added new overloaded function form
3208 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
3209 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
3210 exponent double-precision with floating point first argument.
3211 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3212 documentation of scalar_test_data_class, scalar_test_neg,
3213 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
3214 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
3215 vec_test_data_class built-in functions to reflect refinements in
3216 their type signatures.
3218 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
3220 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
3222 (aarch64_elf_asm_destructor): Likewise.
3224 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
3226 PR rtl-optimization/78634
3227 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
3228 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
3229 * ifcvt.c (noce_try_cmove): Add missing cost check.
3231 PR rtl-optimization/71724
3232 * combine.c (if_then_else_cond): Look for situations where it is
3233 beneficial to undo the work of one of the recursive calls.
3235 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3237 PR tree-optimization/70754
3238 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3239 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3240 combined stmt before it if not NULL.
3241 (combine_chains): Process refs reversely and compute dominance point
3244 2017-01-23 Martin Liska <mliska@suse.cz>
3246 PR tree-optimization/79196
3247 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
3248 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
3250 (strlen_optimize_stmt): Call the renamed function.
3252 2017-01-23 Michael Matz <matz@suse.de>
3254 PR tree-optimization/78384
3255 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
3257 2017-01-23 Richard Biener <rguenther@suse.de>
3259 PR tree-optimization/79186
3260 * tree-vrp.c (register_new_assert_for): Make sure we've seen
3261 both incoming edges before moving an assert.
3263 2017-01-23 Martin Jambor <mjambor@suse.cz>
3265 * ipa-prop.c (load_from_param_1): Removed.
3266 (load_from_unmodified_param): Bits from load_from_param_1 put back
3268 (load_from_param): Removed.
3269 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
3270 with stmt. Reverted back to use of load_from_unmodified_param.
3272 2017-01-23 Martin Jambor <mjambor@suse.cz>
3275 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
3276 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
3277 field a pointer to garbage collected vector, mark lattices and
3278 ipcp_orig_node with GTY((skip)).
3279 (ipa_get_param_count): Adjust to descriptors being a pointer.
3280 (ipa_get_param): Likewise.
3281 (ipa_get_type): Likewise.
3282 (ipa_get_param_move_cost): Likewise.
3283 (ipa_set_param_used): Likewise.
3284 (ipa_get_controlled_uses): Likewise.
3285 (ipa_set_controlled_uses): Likewise.
3286 (ipa_is_param_used): Likewise.
3287 (ipa_node_params_t): Move into garbage collector. New methods insert
3289 (ipa_node_params_sum): Annotate wth GTY(()).
3290 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
3292 (ipa_load_from_parm_agg): Adjust declaration.
3293 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
3294 * ipa-profile.c (ipa_profile): Likewise.
3295 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
3296 (ipa_populate_param_decls): Make descriptors parameter garbage
3298 (ipa_dump_param): Adjust to descriptors being a pointer.
3299 (ipa_alloc_node_params): Likewise.
3300 (ipa_initialize_node_params): Likewise.
3301 (load_from_param_1): Make descriptors parameter garbage collected.
3302 (load_from_unmodified_param): Likewise.
3303 (load_from_param): Likewise.
3304 (ipa_load_from_parm_agg): Likewise.
3305 (ipa_node_params::~ipa_node_params): Removed.
3306 (ipa_free_all_node_params): Remove call to delete operator.
3307 (ipa_node_params_t::insert): New.
3308 (ipa_node_params_t::remove): Likewise.
3309 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
3310 copy known_csts and known_contexts vectors.
3311 (ipa_read_node_info): Adjust to descriptors being a pointer.
3312 (ipcp_modif_dom_walker): Make m_descriptors field garbage
3314 (ipcp_transform_function): Make descriptors variable garbage
3317 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
3319 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
3320 * config/i386/avx512dqintrin.h: Ditto.
3321 * config/i386/avx512fintrin.h: Ditto.
3322 * gcc/config/i386/i386.c: Handle new builtins.
3323 * config/i386/i386-builtin.def: Add new builtins.
3324 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
3325 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
3327 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3328 Martin Liska <mliska@suse.cz>
3330 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
3331 * asan.c (asan_expand_poison_ifn): Support stores and use
3332 appropriate ASAN report function.
3333 * internal-fn.c (expand_ASAN_POISON_USE): New function.
3334 * internal-fn.def (ASAN_POISON_USE): Declare.
3335 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
3336 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
3337 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
3338 ASAN_POISON calls w/o LHS.
3339 * tree-ssa.c (execute_update_addresses_taken): Create clobber
3340 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
3341 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
3342 * gimplify.c (asan_poison_variables): Add attribute
3343 use_after_scope_memory to variables that really needs to live
3345 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
3346 having the attribute.
3348 2017-01-23 Martin Liska <mliska@suse.cz>
3350 * asan.c (create_asan_shadow_var): New function.
3351 (asan_expand_poison_ifn): Likewise.
3352 * asan.h (asan_expand_poison_ifn): New declaration.
3353 * internal-fn.c (expand_ASAN_POISON): Likewise.
3354 * internal-fn.def (ASAN_POISON): New builtin.
3355 * sanopt.c (pass_sanopt::execute): Expand
3356 asan_expand_poison_ifn.
3357 * tree-inline.c (copy_decl_for_dup_finish): Make function
3359 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
3360 * tree-ssa.c (is_asan_mark_p): New function.
3361 (execute_update_addresses_taken): Rewrite local variables
3362 (identified just by use-after-scope as addressable) into SSA.
3364 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3366 * doc/install.texi (Specific): opensource.apple.com uses https
3367 now. Remove trailing slash.
3369 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3371 * README.Portability: Remove note on an Irix compatibility issue.
3373 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
3375 * gcov.c (INCLUDE_ALGORITHM): Define.
3376 (INCLUDE_VECTOR): Define.
3377 No longer include <vector> and <algorithm> directly.
3379 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
3381 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
3383 * doc/invoke.texi (Code Gen Options): Ditto.
3385 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
3388 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
3390 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
3392 rtl-optimization/79125
3393 * cprop.c (local_cprop_pass): Handle cases where we make an
3396 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
3400 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
3401 read from, for big endian.
3403 2017-01-20 Jiong Wang <jiong.wang@arm.com>
3405 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
3406 register pauth builtins for LP64 only.
3408 2017-01-20 Marek Polacek <polacek@redhat.com>
3411 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
3414 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3416 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
3418 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
3419 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
3420 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
3422 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3425 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
3426 in big-endian mode when they are not a single duplicated value.
3428 2017-01-20 Richard Biener <rguenther@suse.de>
3430 * BASE-VER: Bump to 7.0.1.
3432 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3434 * omp-low.c (omplow_simd_context): New struct. Use it...
3435 (lower_rec_simd_input_clauses): ...here and...
3436 (lower_rec_input_clauses): ...here to hold common data. Adjust all
3437 references to idx, lane, max_vf, is_simt.
3439 2017-01-20 Graham Markall <graham.markall@embecosm.com>
3441 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
3444 2017-01-20 Martin Jambor <mjambor@suse.cz>
3446 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
3447 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
3449 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
3450 (GTFILES): Rename hsa.c to hsa-common.c.
3451 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
3452 * hsa-dump.c: Likewise.
3453 * hsa-gen.c: Likewise.
3454 * hsa-regalloc.c: Likewise.
3455 * ipa-hsa.c: Likewise.
3456 * omp-expand.c: Likewise.
3457 * omp-low.c: Likewise.
3458 * toplev.c: Likewise.
3460 2017-01-20 Marek Polacek <polacek@redhat.com>
3463 * doc/invoke.texi: Document -Wduplicated-branches.
3464 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
3465 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
3466 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
3467 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
3468 return 0 only when not OEP_LEXICOGRAPHIC.
3469 (fold_build_cleanup_point_expr): Use the expression
3470 location when building CLEANUP_POINT_EXPR.
3471 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
3472 * tree.c (add_expr): Handle error_mark_node.
3474 2017-01-20 Martin Liska <mliska@suse.cz>
3477 * tree-profile.c (init_ic_make_global_vars): Do not call
3479 (gimple_init_gcov_profiler): Likewise.
3481 2017-01-20 Martin Liska <mliska@suse.cz>
3484 * cgraph.h (maybe_create_reference): Remove argument and
3486 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
3488 * ipa-cp.c (create_specialized_node): Likewise.
3489 * symtab.c (symtab_node::maybe_create_reference): Handle
3490 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
3492 2017-01-20 Martin Liska <mliska@suse.cz>
3494 * read-rtl-function.c (function_reader::create_function): Use
3495 build_decl instread of build_decl_stat.
3497 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
3499 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
3500 * config/i386/avx512dqintrin.h: Ditto.
3501 * config/i386/avx512fintrin.h: Ditto.
3502 * config/i386/i386-builtin-types.def: Add new types.
3503 * gcc/config/i386/i386.c: Handle new types.
3504 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
3505 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
3506 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
3507 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
3508 (__builtin_ia32_kshiftridi): New.
3509 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
3511 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
3515 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
3516 define to rs6000_init_stack_protect_guard.
3517 (rs6000_init_stack_protect_guard): New function.
3519 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3520 Yunqiang Su <yunqiang.su@imgtec.com>
3522 * config.gcc (supported_defaults): Add madd4.
3523 (with_madd4): Add validation.
3524 (all_defaults): Add madd4.
3525 * config/mips/mips.opt (mmadd4): New option.
3526 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3528 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3530 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
3531 (ISA_HAS_FUSED_MADD4): Likewise.
3532 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
3533 * gcc/doc/install.texi (--with-madd4): Document the new option.
3535 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3537 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
3538 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
3539 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
3540 (aarch64_init_pauth_hint_builtins): New.
3541 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
3542 (aarch64_expand_builtin): Expand new builtins.
3544 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3546 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
3547 * combine-stack-adj.c (no_unhandled_cfa): Handle
3548 REG_CFA_TOGGLE_RA_MANGLE.
3549 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
3550 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
3551 info for return address signing.
3552 (aarch64_expand_epilogue): Likewise.
3554 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3556 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
3557 * config/aarch64/aarch64-protos.h
3558 (aarch64_return_address_signing_enabled): New declaration.
3559 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
3561 (aarch64_expand_prologue): Sign return address before it's pushed onto
3563 (aarch64_expand_epilogue): Authenticate return address fetched from
3565 (aarch64_override_options): Sanity check for ILP32 and ISA level.
3566 (aarch64_attributes): New function attributes for "sign-return-address".
3567 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
3568 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
3569 ("*do_return"): Generate combined instructions according to key index.
3570 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
3571 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
3573 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
3574 * config/aarch64/aarch64.opt (msign-return-address=): New.
3575 * doc/extend.texi (AArch64 Function Attributes): Documents
3576 "sign-return-address=".
3577 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
3579 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
3581 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
3582 overall option summary.
3584 2017-01-19 Jiong Wang <jiong.wang@arm.com>
3586 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
3587 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
3588 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
3589 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
3591 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
3593 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
3594 -mpower9-minmax by default for -mcpu=power9.
3595 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
3596 128-bit floating point.
3598 2017-01-20 Alan Modra <amodra@gmail.com>
3600 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
3601 optimizing for size.
3603 2017-01-20 Alan Modra <amodra@gmail.com>
3606 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
3607 for strcmp and strncmp from corresponding builtin decl.
3609 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3611 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
3612 instead of i386/rtems-64.h.
3613 * config/i386/rtems-64.h: Remove.
3615 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
3619 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
3621 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
3623 2017-01-19 Tamar Christina <tamar.christina@arm.com>
3625 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3626 Change int to HOST_WIDE_INT.
3627 * config/aarch64/aarch64-protos.h
3628 (aarch64_simd_gen_const_vector_dup): Likewise.
3629 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
3631 2017-01-19 David Malcolm <dmalcolm@redhat.com>
3633 * langhooks-def.h (lhd_type_for_size): New decl.
3634 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
3635 * langhooks.c (lhd_type_for_size): New function, taken from
3638 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
3640 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
3641 define_bypass for CR latency.
3642 (power9-cracked-alu): Update bypass latency and remove power9-branch.
3643 (power9-alu2): Add define_bypass for CR latency.
3645 (power9-mul): Update insn latency.
3646 (power9-mul-compare): Update insn latency, bypass latency and remove
3649 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3651 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
3653 * config/aarch64/aarch64.md
3654 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
3655 aarch64_nopcrelative_literal_loads.
3656 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
3658 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
3660 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
3662 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
3664 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
3667 * config.gcc (supported_defaults): Add lxc1-sxc1.
3668 (with_lxc1_sxc1): Add validation.
3669 (all_defaults): Add lxc1-sxc1.
3670 * config/mips/mips.opt (mlxc1-sxc1): New option.
3671 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
3673 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
3674 __mips_no_lxc1_sxc1.
3675 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
3676 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
3677 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
3679 2017-01-19 Richard Biener <rguenther@suse.de>
3681 PR tree-optimization/72488
3682 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
3683 sure to restore SSA info.
3684 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
3686 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
3688 PR rtl-optimization/79121
3689 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
3690 of the inner type when shifting an extended value.
3692 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
3695 * symtab.c (symtab_node::equal_address_to): Fix comparing of
3696 interposable aliases.
3698 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
3701 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
3702 Use the evmergelohi instruction.
3703 (mov_si<mode>_e500_subreg4_2_le): Likewise.
3704 (mov_sitf_e500_subreg8_2_be): Likewise.
3705 (mov_sitf_e500_subreg12_2_le): Likewise.
3706 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
3707 (mov_si<mode>_e500_subreg4_2_be): Likewise.
3708 (mov_sitf_e500_subreg8_2_le): Likewise.
3709 (mov_sitf_e500_subreg12_2_be): Likewise.
3711 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3713 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
3714 attribute from vecsimple to vecperm.
3715 (altivec_vbpermq2): Likewise.
3717 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3720 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
3722 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3723 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
3724 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
3725 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
3726 case where N arg is SIZE_MAX.
3727 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
3728 (cmpstrsi): Add pattern.
3730 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
3732 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3733 __builtin_vec_revb builtins.
3734 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
3735 built-in functions to support generation of the ISA 3.0 XXBR<x>
3736 vector byte reverse instructions.
3737 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
3738 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
3739 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
3740 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
3741 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
3742 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
3743 (P9V_BUILTIN_VEC_REVB): Likewise.
3744 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
3745 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
3746 (p9_xxbrq_v16qi): Likewise.
3747 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
3748 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
3749 (p9_xxbrh_v8hi): Likewise.
3750 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
3751 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
3752 vec_revb built-in functions.
3754 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
3756 PR rtl-optimization/78952
3757 * config/i386/i386.md (any_extract): New code iterator.
3758 (*insvqi_2): Use any_extract for source operand.
3759 (*insvqi_3): Use any_shiftrt for source operand.
3761 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
3763 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
3765 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
3767 2017-01-18 Matthias Klose <doko@ubuntu.com>
3769 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
3771 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3773 * config/rs6000/altivec.h (vec_bperm): Change #define.
3774 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
3775 (altivec_vbpermq2): New define_insn.
3776 (altivec_vbpermd): Likewise.
3777 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
3779 (VBPERMD): Likewise.
3780 (VBPERM): New polymorphic function interface.
3781 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
3782 Add entries for P9V_BUILTIN_VEC_VBPERM.
3783 * doc/extend.texi: Add interfaces for vec_bperm.
3785 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3787 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
3788 first letter of error messages.
3789 (s390_resolve_overloaded_builtin): Likewise.
3790 * config/s390/s390.c (s390_expand_builtin): Likewise.
3791 (s390_invalid_arg_for_unprototyped_fn): Likewise.
3792 (s390_valid_target_attribute_inner_p): Likewise.
3793 * config/s390/s390.md ("tabort"): Likewise.
3795 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
3797 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
3798 (ISA_AVOID_DIV_HILO): New macro.
3799 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
3800 (ISA_HAS_DDIV): Likewise.
3802 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
3804 * doc/invoke.texi (fabi-version): Correct number of occurrences.
3806 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
3808 * doc/invoke.texi (fabi-version): Spelling fix.
3810 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
3813 * doc/invoke.texi (fabi-version): Mention mangling fix for
3816 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
3819 * doc/invoke.texi (fabi-version): Document discriminator mangling.
3821 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
3824 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
3825 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
3827 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
3829 (stack_protect_test): Ditto.
3830 * config/rs6000/rs6000.opt (mstack-protector-guard=,
3831 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
3833 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
3834 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
3835 -mstack-protector-guard-offset=.
3836 (RS/6000 and PowerPC Options): Ditto.
3838 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
3840 * config/i386/i386.h (MASK_CLASS_P): New define.
3841 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
3842 there are no registers from different register sets also when
3843 mask registers are used. Update function comment.
3844 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
3845 to (*k/*r) and (*k/*km) alternatives.
3847 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
3849 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
3850 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
3851 (EH_RETURN_HANDLER_RTX): New define.
3852 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
3853 Force frame pointer in EH return functions.
3854 (aarch64_expand_epilogue): Add barrier for eh_return.
3855 (aarch64_final_eh_return_addr): Remove.
3856 (aarch64_eh_return_handler_rtx): New function.
3857 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
3859 (aarch64_eh_return_handler_rtx): New prototype.
3861 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3863 * config/rs6000/altivec.h (vec_rlmi): New #define.
3864 (vec_vrlnm): Likewise.
3865 (vec_rlnm): Likewise.
3866 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
3867 (UNSPEC_VRLNM): Likewise.
3868 (VIlong): New mode iterator.
3869 (altivec_vrl<VI_char>mi): New define_insn.
3870 (altivec_vrl<VI_char>nm): Likewise.
3871 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
3874 (RLNM): New polymorphic function entry.
3875 (VRLWMI): New monomorphic function entry.
3877 (RLMI): New polymorphic function entry.
3878 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
3879 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
3880 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
3883 2017-01-17 Jakub Jelinek <jakub@redhat.com>
3886 * dwarf2out.c (field_byte_offset): Restore the
3887 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
3888 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
3889 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
3892 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
3895 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
3897 2017-01-17 Jakub Jelinek <jakub@redhat.com>
3900 * dwarf2out.c (add_data_member_location_attribute): For constant
3901 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
3902 instead of DW_AT_data_member_location, DW_AT_bit_offset and
3903 DW_AT_byte_size attributes.
3905 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
3907 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
3908 after forcing to constant memory when the code model is medium.
3910 2017-01-17 Julia Koval <julia.koval@intel.com>
3913 * config/i386/avx512fintrin.h
3914 (_mm512_i32gather_ps): Change __addr type to void const*.
3915 (_mm512_mask_i32gather_ps): Ditto.
3916 (_mm512_i32gather_pd): Ditto.
3917 (_mm512_mask_i32gather_pd): Ditto.
3918 (_mm512_i64gather_ps): Ditto.
3919 (_mm512_mask_i64gather_ps): Ditto.
3920 (_mm512_i64gather_pd): Ditto.
3921 (_mm512_mask_i64gather_pd): Ditto.
3922 (_mm512_i32gather_epi32): Ditto.
3923 (_mm512_mask_i32gather_epi32): Ditto.
3924 (_mm512_i32gather_epi64): Ditto.
3925 (_mm512_mask_i32gather_epi64): Ditto.
3926 (_mm512_i64gather_epi32): Ditto.
3927 (_mm512_mask_i64gather_epi32): Ditto.
3928 (_mm512_i64gather_epi64): Ditto.
3929 (_mm512_mask_i64gather_epi64): Ditto.
3930 (_mm512_i32scatter_ps): Change __addr type to void*.
3931 (_mm512_mask_i32scatter_ps): Ditto.
3932 (_mm512_i32scatter_pd): Ditto.
3933 (_mm512_mask_i32scatter_pd): Ditto.
3934 (_mm512_i64scatter_ps): Ditto.
3935 (_mm512_mask_i64scatter_ps): Ditto.
3936 (_mm512_i64scatter_pd): Ditto.
3937 (_mm512_mask_i64scatter_pd): Ditto.
3938 (_mm512_i32scatter_epi32): Ditto.
3939 (_mm512_mask_i32scatter_epi32): Ditto.
3940 (_mm512_i32scatter_epi64): Ditto.
3941 (_mm512_mask_i32scatter_epi64): Ditto.
3942 (_mm512_i64scatter_epi32): Ditto.
3943 (_mm512_mask_i64scatter_epi32): Ditto.
3944 (_mm512_i64scatter_epi64): Ditto.
3945 (_mm512_mask_i64scatter_epi64): Ditto.
3946 * config/i386/avx512pfintrin.h
3947 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
3948 (_mm512_mask_prefetch_i32gather_ps): Ditto.
3949 (_mm512_mask_prefetch_i64gather_pd): Ditto.
3950 (_mm512_mask_prefetch_i64gather_ps): Ditto.
3951 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
3952 (_mm512_prefetch_i32scatter_ps): Ditto.
3953 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
3954 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
3955 (_mm512_prefetch_i64scatter_pd): Ditto.
3956 (_mm512_prefetch_i64scatter_ps): Ditto.
3957 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
3958 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
3959 * config/i386/avx512vlintrin.h
3960 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
3961 (_mm_mmask_i32gather_ps): Ditto.
3962 (_mm256_mmask_i32gather_pd): Ditto.
3963 (_mm_mmask_i32gather_pd): Ditto.
3964 (_mm256_mmask_i64gather_ps): Ditto.
3965 (_mm_mmask_i64gather_ps): Ditto.
3966 (_mm256_mmask_i64gather_pd): Ditto.
3967 (_mm_mmask_i64gather_pd): Ditto.
3968 (_mm256_mmask_i32gather_epi32): Ditto.
3969 (_mm_mmask_i32gather_epi32): Ditto.
3970 (_mm256_mmask_i32gather_epi64): Ditto.
3971 (_mm_mmask_i32gather_epi64): Ditto.
3972 (_mm256_mmask_i64gather_epi32): Ditto.
3973 (_mm_mmask_i64gather_epi32): Ditto.
3974 (_mm256_mmask_i64gather_epi64): Ditto.
3975 (_mm_mmask_i64gather_epi64): Ditto.
3976 (_mm256_i32scatter_ps): Change __addr type to void*.
3977 (_mm256_mask_i32scatter_ps): Ditto.
3978 (_mm_i32scatter_ps): Ditto.
3979 (_mm_mask_i32scatter_ps): Ditto.
3980 (_mm256_i32scatter_pd): Ditto.
3981 (_mm256_mask_i32scatter_pd): Ditto.
3982 (_mm_i32scatter_pd): Ditto.
3983 (_mm_mask_i32scatter_pd): Ditto.
3984 (_mm256_i64scatter_ps): Ditto.
3985 (_mm256_mask_i64scatter_ps): Ditto.
3986 (_mm_i64scatter_ps): Ditto.
3987 (_mm_mask_i64scatter_ps): Ditto.
3988 (_mm256_i64scatter_pd): Ditto.
3989 (_mm256_mask_i64scatter_pd): Ditto.
3990 (_mm_i64scatter_pd): Ditto.
3991 (_mm_mask_i64scatter_pd): Ditto.
3992 (_mm256_i32scatter_epi32): Ditto.
3993 (_mm256_mask_i32scatter_epi32): Ditto.
3994 (_mm_i32scatter_epi32): Ditto.
3995 (_mm_mask_i32scatter_epi32): Ditto.
3996 (_mm256_i32scatter_epi64): Ditto.
3997 (_mm256_mask_i32scatter_epi64): Ditto.
3998 (_mm_i32scatter_epi64): Ditto.
3999 (_mm_mask_i32scatter_epi64): Ditto.
4000 (_mm256_i64scatter_epi32): Ditto.
4001 (_mm256_mask_i64scatter_epi32): Ditto.
4002 (_mm_i64scatter_epi32): Ditto.
4003 (_mm_mask_i64scatter_epi32): Ditto.
4004 (_mm256_i64scatter_epi64): Ditto.
4005 (_mm256_mask_i64scatter_epi64): Ditto.
4006 (_mm_i64scatter_epi64): Ditto.
4007 (_mm_mask_i64scatter_epi64): Ditto.
4008 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
4009 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
4010 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
4011 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
4012 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
4013 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
4014 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
4015 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
4016 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
4017 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
4018 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
4019 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
4020 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
4021 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
4022 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
4023 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
4024 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
4025 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
4026 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
4027 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
4028 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
4029 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
4030 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
4031 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
4032 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
4033 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
4034 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
4035 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
4036 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
4037 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
4038 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
4039 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
4040 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
4041 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
4042 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
4043 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
4044 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
4045 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
4046 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
4047 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
4048 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
4049 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
4050 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
4051 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
4052 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
4053 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
4054 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
4055 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
4056 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
4057 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
4058 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
4059 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
4060 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
4061 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
4062 definitions accordingly.
4064 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
4065 Kuan-Lin Chen <kuanlinchentw@gmail.com>
4068 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
4071 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
4074 * ira-conflicts.c (ira_build_conflicts): Update total conflict
4075 hard regs for inner regno.
4077 2017-01-17 Martin Liska <mliska@suse.cz>
4080 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
4081 assumption and add comment.
4083 2017-01-17 Nathan Sidwell <nathan@acm.org>
4085 * ipa-visibility.c (localize_node): New function, broken out of ...
4086 (function_and_variable_visibility): ... here. Call it.
4088 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4091 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
4092 correctly set frequency of oudgoing edge.
4093 (duplicate_thread_path): Fix profile updating.
4095 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4098 * configure.ac: Add GCC_BASE_VER.
4099 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
4100 version from BASE-VER file.
4101 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
4102 (gcc.o): Depend on $(BASEVER).
4103 * common.opt (dumpfullversion): New option.
4104 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
4105 * doc/invoke.texi: Document -dumpfullversion.
4106 * doc/install.texi: Document --with-gcc-major-version-only.
4107 * configure: Regenerated.
4109 2017-01-17 Richard Biener <rguenther@suse.de>
4111 PR tree-optimization/71433
4112 * tree-vrp.c (register_new_assert_for): Merge same asserts
4113 on all incoming edges.
4114 (process_assert_insertions_for): Handle insertions at the
4117 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
4119 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
4120 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
4122 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
4125 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
4128 2017-01-17 Alan Modra <amodra@gmail.com>
4131 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
4132 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
4133 symbolic stack limit when pic.
4135 2017-01-16 Martin Sebor <msebor@redhat.com>
4137 PR tree-optimization/78608
4138 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
4140 2017-01-16 Jeff Law <law@redhat.com>
4143 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
4144 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
4145 for several include directories that may be relative to sysroot.
4146 * config/i386/x-mingw32 (gplus_includedir): Define.
4147 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
4148 (native_system_includedir): Likewise.
4149 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
4150 override if TARGET_SYSTEM_ROOT is defined.
4151 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
4153 PR tree-optimization/79090
4154 PR tree-optimization/33562
4155 PR tree-optimization/61912
4156 PR tree-optimization/77485
4157 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
4158 and computed trims into the dump file.
4160 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4162 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
4164 2017-01-16 Jakub Jelinek <jakub@redhat.com>
4167 * gimplify.c (gimplify_init_constructor): If want_value and
4168 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
4172 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
4173 sequence. Formatting fixes.
4174 (doloop_optimize): Formatting fixes.
4177 * gcc.c (debug_level_greater_than_spec_func): New function.
4178 (static_spec_functions): Add debug-level-gt spec function.
4179 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
4181 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4182 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4183 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
4184 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
4185 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
4186 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
4188 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
4190 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
4191 QImode fixups to general and mask registers only.
4193 2017-01-16 Carl Love <cel@us.ibm.com>
4195 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
4196 for built-in functions
4197 vector signed char vec_nabs (vector signed char)
4198 vector signed short vec_nabs (vector signed short)
4199 vector signed int vec_nabs (vector signed int)
4200 vector signed long long vec_nabs (vector signed long long)
4201 vector float vec_nabs (vector float)
4202 vector double vec_nabs (vector double)
4203 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
4205 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
4206 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
4207 * doc/extend.texi: Update the documentation file for the new built-in
4210 2017-01-16 Martin Sebor <msebor@redhat.com>
4212 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
4215 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4217 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
4218 UNSPEC_VSX__XXSPLTD to require special splat handling.
4220 2017-01-16 David Malcolm <dmalcolm@redhat.com>
4223 * system.h: Poison strndup.
4225 2017-01-16 Alan Modra <amodra@gmail.com>
4228 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
4231 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
4233 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
4235 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
4237 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
4238 call recog here. Assert that INSN_CODE (insn) is non-negative.
4240 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4243 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
4245 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
4246 in the currently scheduled RTL fragment.
4248 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4250 PR rtl-optimization/78751
4251 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
4254 2017-01-14 Jeff Law <law@redhat.com>
4256 PR tree-optimization/79090
4257 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
4258 variable length stores.
4259 (compute_trims): Delete dead assignment to *trim_tail.
4260 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
4263 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
4265 PR rtl-optimization/78626
4266 PR rtl-optimization/78727
4267 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
4268 of a block, and split such blocks after everything else is finished.
4270 2017-01-14 Alan Modra <amodra@gmail.com>
4273 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
4274 target legitimate_combined_insn.
4275 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
4276 (rs6000_legitimate_combined_insn): New function.
4277 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
4279 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
4280 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
4281 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
4283 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
4285 * doc/frontends.texi (G++ and GCC): Remove references to Java.
4287 2017-01-13 Jeff Law <law@redhat.com>
4289 PR tree-optimization/33562
4290 PR tree-optimization/61912
4291 PR tree-optimization/77485
4292 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
4294 (delete_dead_assignment): Likewise.
4295 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
4296 statement to delete_dead_call and delete_dead_assignment.
4298 2017-01-13 David Malcolm <dmalcolm@redhat.com>
4301 * substring-locations.c (format_warning_va): Strengthen case 1 so
4302 that both endpoints of the substring must be within the format
4303 range for just the substring to be printed.
4305 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
4307 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
4308 * config/i386/i386.c (ix86_target_string): Add missing options
4309 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
4310 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
4311 flags_other and ix86_target_other to flags2_other. Display unknown
4313 (ix86_valid_target_attribute_inner_p): Add missing options and
4314 reorder options by implied ISAs, as in ix86_target_string.
4316 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4318 * hash-table.h (hash_table::too_empty_p): New function.
4319 (hash_table::expand): Use it.
4320 (hash_table::traverse): Likewise.
4321 (hash_table::empty_slot): Use sizeof (value_type) instead of
4322 sizeof (PTR) to convert bytes to elements. Shrink the table
4323 if the current size is excessive for the current number of
4326 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4328 * ira-costs.c (record_reg_classes): Break from the inner loop
4329 early once alt_fail is known to be true. Update outer loop
4330 handling accordingly.
4332 2017-01-13 Jeff Law <law@redhat.com>
4334 * tree-ssa-dse.c (decrement_count): New function.
4335 (increment_start_addr, maybe_trim_memstar_call): Likewise.
4336 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
4337 when we know the partially dead statement is a mem* function.
4339 PR tree-optimization/61912
4340 PR tree-optimization/77485
4341 * tree-ssa-dse.c: Include expr.h.
4342 (maybe_trim_constructor_store): New function.
4343 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
4345 PR tree-optimization/33562
4346 PR tree-optimization/61912
4347 PR tree-optimization/77485
4348 * doc/invoke.texi: Document new dse-max-object-size param.
4349 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
4350 * tree-ssa-dse.c: Include params.h.
4351 (dse_store_status): New enum.
4352 (initialize_ao_ref_for_dse): New, partially extracted from
4354 (valid_ao_ref_for_dse, normalize_ref): New.
4355 (setup_live_bytes_from_ref, compute_trims): Likewise.
4356 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
4357 (maybe_trim_partially_dead_store): Likewise.
4358 (maybe_trim_complex_store): Likewise.
4359 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
4360 Track what bytes live from the original store. Return tri-state
4361 for dead, partially dead or live.
4362 (dse_dom_walker): Add constructor, destructor and new private members.
4363 (delete_dead_call, delete_dead_assignment): New extracted from
4365 (dse_optimize_stmt): Make a member of dse_dom_walker.
4366 Use initialize_ao_ref_for_dse.
4368 PR tree-optimization/33562
4369 PR tree-optimization/61912
4370 PR tree-optimization/77485
4371 * sbitmap.h (bitmap_count_bits): Prototype.
4372 (bitmap_clear_range, bitmap_set_range): Likewise.
4373 * sbitmap.c (bitmap_clear_range): New function.
4374 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
4376 2017-01-13 Martin Liska <mliska@suse.cz>
4379 * function.c (set_cfun): Add new argument force.
4380 * function.h (set_cfun): Likewise.
4381 * ipa-inline-transform.c (inline_call): Use the function when
4382 strict alising from is dropped for function we inline to.
4384 2017-01-13 Richard Biener <rguenther@suse.de>
4386 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
4387 for dumping GIMPLE INTEGER_CSTs.
4389 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4391 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
4392 to 201112L since C++17.
4394 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
4397 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
4398 if -fsanitize=kernel-address is present.
4400 2017-01-13 Richard Biener <rguenther@suse.de>
4402 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
4403 as _Literal ( type ) number in case usual suffixes do not
4404 preserve all information.
4406 2017-01-13 Richard Biener <rguenther@suse.de>
4408 PR tree-optimization/77283
4409 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
4410 and ssa-iterators.h.
4411 (is_feasible_trace): Implement a cost model based on joiner
4414 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4417 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
4418 char or short to __float128/_Float128 directly.
4420 2017-01-12 Martin Sebor <msebor@redhat.com>
4422 to -Wformat-overflow.
4423 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
4424 (min_bytes_remaining): Same.
4425 (get_string_length): Same.
4426 (format_string): Same.
4427 (format_directive): Same.
4429 (pass_sprintf_length::handle_gimple_call): Same.
4431 2017-01-12 Jakub Jelinek <jakub@redhat.com>
4433 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
4434 info.nowrite calls with no lhs that can't throw. Return bool
4435 whether gsi_remove has been called or not.
4436 (pass_sprintf_length::handle_gimple_call): Return bool whether
4437 try_substitute_return_value called gsi_remove. Formatting fix.
4438 (pass_sprintf_length::execute): Don't use gsi_remove if
4439 handle_gimple_call returned true.
4442 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
4443 be removed due to side-effects, don't remove following barrier nor
4444 turn the successor edge into fallthru edge.
4446 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4449 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
4450 element-reversing loads and stores as not swappable.
4452 2017-01-12 Nathan Sidwell <nathan@acm.org>
4453 Nicolai Stange <nicstange@gmail.com>
4455 * combine.c (try_combine): Don't ignore result of overlap checking
4456 loop. Combine overlap & asm check into single loop.
4458 2017-01-12 Richard Biener <rguenther@suse.de>
4460 * tree-pretty-print.c (dump_generic_node): Provide -gimple
4461 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
4463 2017-01-12 Richard Biener <rguenther@suse.de>
4465 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
4466 and TS_TARGET_OPTION directly derive from TS_BASE.
4467 * tree-core.h (tree_optimization_option): Derive from tree_base.
4468 (tree_target_option): Likewise.
4470 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
4472 * config/i386/i386.c (memory_address_length): Increase len
4473 only when rip_relative_addr_p returns false.
4475 2017-01-11 Julia Koval <julia.koval@intel.com>
4477 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
4478 (OPTION_MASK_ISA_SGX_SET): New.
4479 (ix86_handle_option): Handle OPT_msgx.
4480 * config.gcc: Added sgxintrin.h.
4481 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
4482 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
4483 * config/i386/i386.c (ix86_target_string): Add -msgx.
4485 (ix86_option_override_internal): Handle new options.
4486 (ix86_valid_target_attribute_inner_p): Add sgx.
4487 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
4488 * config/i386/i386.opt: Add msgx.
4489 * config/i386/sgxintrin.h: New file.
4490 * config/i386/x86intrin.h: Add sgxintrin.h.
4492 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4495 * fold-const.c (maybe_nonzero_address): Return 1 for function
4497 (tree_single_nonzero_warnv_p): Don't handle function local objects
4501 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
4504 2017-01-11 David Malcolm <dmalcolm@redhat.com>
4507 * opts.c: Include "spellcheck.h"
4508 (struct string_fragment): New struct.
4509 (struct edit_distance_traits<const string_fragment &>): New
4511 (get_closest_sanitizer_option): New function.
4512 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
4514 2017-01-11 Jakub Jelinek <jakub@redhat.com>
4516 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
4518 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
4519 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
4520 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
4521 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
4522 for initial die_offset if dwarf_split_debug_info.
4523 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
4524 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
4526 (output_skeleton_debug_sections): Formatting fix. Use
4527 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
4528 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
4530 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
4532 * config/arm/cortex-a53.md: Add bypasses for
4534 (cortex_a53_r2f): Only use for transfers.
4535 (cortex_a53_f2r): Likewise.
4536 (cortex_a53_r2f_cvt): Add reservation for conversions.
4537 (cortex_a53_f2r_cvt): Likewise.
4539 2017-01-11 Tamar Christina <tamar.christina@arm.com>
4541 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
4542 to all inlined functions, change static to extern.
4544 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
4547 * config/arm/arm.c (legitimize_pic_address): Handle reference to
4549 (arm_assemble_integer): Likewise.
4551 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
4553 * config.gcc: Use new awk script to check CPU, FPU and architecture
4554 parameters for --with-... options.
4555 * config/arm/parsecpu.awk: New file
4556 * config/arm/arm-cpus.in: New file.
4557 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
4559 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
4561 * config/arm/t-arm: Update dependency rules.
4562 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
4563 of processing .def files.
4564 * config/arm/genopt.sh: Deleted.
4565 * config/arm/gentune.sh: Deleted.
4566 * config/arm/arm-cores.def: Deleted.
4567 * config/arm/arm-arches.def: Deleted.
4568 * config/arm/arm-fpus.def: Deleted.
4569 * config/arm/arm-tune.md: Regenerated.
4570 * config/arm/arm-tables.opt: Regenerated.
4571 * config/arm/arm-cpu.h: New generated file.
4572 * config/arm/arm-cpu-data.h: New generated file.
4573 * config/arm/arm-cpu-cdata.h: New generated file.
4575 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
4578 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
4580 (input_varpool_node): Unpack dynamically_initialized bit.
4582 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
4584 PR rtl-optimization/79032
4585 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
4586 the alignment of the adjusted memory reference against that of MODE,
4587 instead of the alignment of the original memory reference.
4589 2017-01-11 Martin Jambor <mjambor@suse.cz>
4591 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
4593 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
4594 decorated functions.
4596 2017-01-11 Richard Biener <rguenther@suse.de>
4598 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
4599 set range/nonnull info for PHI results. Do not set it on
4600 stmts marked for removal.
4602 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
4604 * expr.c (store_field): In the bitfield case, fetch the return value
4605 from the registers before applying a single big-endian adjustment.
4606 Always do a final load for a BLKmode value not larger than a word.
4608 2017-01-10 David Malcolm <dmalcolm@redhat.com>
4611 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
4612 that we correctly handle column numbers greater than
4613 LINE_MAP_MAX_COLUMN_NUMBER.
4615 2017-01-10 Martin Sebor <msebor@redhat.com>
4618 * gimple-ssa-sprintf.c (get_destination_size): Call
4619 {init,fini}object_sizes.
4620 * tree-object-size.c (addr_object_size): Adjust.
4621 (pass_through_call): Adjust.
4622 (pass_object_sizes::execute): Adjust.
4623 * tree-object-size.h (fini_object_sizes): Declare.
4625 2017-01-10 Martin Sebor <msebor@redhat.com>
4627 PR tree-optimization/78775
4628 * builtins.c (get_size_range): Move...
4629 * calls.c: ...to here.
4630 (alloc_max_size): Accept zero argument.
4631 (operand_signed_p): Remove.
4632 (maybe_warn_alloc_args_overflow): Call get_size_range.
4633 * calls.h (get_size_range): Declare.
4635 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
4637 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
4638 from TI's devices.csv file as of September 2016.
4639 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
4641 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
4643 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
4644 * doc/invoke.texi: Likewise.
4645 * doc/md.texi: Likewise.
4646 * doc/objc.texi: Likewise.
4648 2017-01-10 Joshua Conner <joshconner@google.com>
4650 * config/arm/fuchsia-elf.h: New file.
4651 * config/fuchsia.h: New file.
4652 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
4653 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
4655 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
4657 2016-01-10 Richard Biener <rguenther@suse.de>
4659 PR tree-optimization/79034
4660 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
4661 Propagate out degenerate PHIs in the joiner.
4663 2017-01-10 Martin Liska <mliska@suse.cz>
4665 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
4666 (sort_congruence_classes_by_decl_uid): Likewise.
4667 (sort_congruence_class_groups_by_decl_uid): Likewise.
4668 (sem_item_optimizer::merge_classes): Sort class, groups in these
4669 classes and members in the groups by DECL_UID of declarations.
4670 This would make merge operations stable.
4672 2017-01-10 Martin Liska <mliska@suse.cz>
4674 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
4675 usage of m_classes_vec.
4676 (sem_item_optimizer::~sem_item_optimizer): Likewise.
4677 (sem_item_optimizer::get_group_by_hash): Likewise.
4678 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
4679 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
4680 (sem_item_optimizer::verify_classes): Likewise.
4681 (sem_item_optimizer::process_cong_reduction): Likewise.
4682 (sem_item_optimizer::dump_cong_classes): Likewise.
4683 (sem_item_optimizer::merge_classes): Likewise.
4684 * ipa-icf.h (congruence_class_hash): Rename from
4685 congruence_class_group_hash. Remove declaration of m_classes_vec.
4687 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
4689 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
4690 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
4691 * config.gcc: Add avx512vpopcntdqintrin.h.
4692 * config/i386/avx512vpopcntdqintrin.h: New.
4693 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
4694 * config/i386/i386-builtin-types.def: Add new types.
4695 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
4696 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
4697 __builtin_ia32_vpopcountq_v8di_mask): New.
4698 * config/i386/i386-c.c (ix86_target_macros_internal): Define
4699 __AVX512VPOPCNTDQ__.
4700 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
4701 (PTA_AVX512VPOPCNTDQ): Define.
4702 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
4703 TARGET_AVX512VPOPCNTDQ_P): Define.
4704 * config/i386/i386.opt: Add mavx512vpopcntdq.
4705 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
4706 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
4708 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4711 * predict.def (PRED_CALL): Set to 67.
4713 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
4715 * expr.c (store_field): In the bitfield case, if the value comes from
4716 a function call and is of an aggregate type returned in registers, do
4717 not modify the field mode; extract the value in all cases if the mode
4718 is BLKmode and the size is not larger than a word.
4720 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
4723 * config/i386/cpuid.h: Fix undefined behavior.
4725 2017-01-04 Jeff Law <law@redhat.com>
4727 PR tree-optimization/79007
4728 PR tree-optimization/67955
4729 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
4730 conservative for pt.null when flag_non_call_exceptions is on.
4732 2017-01-09 Jakub Jelinek <jakub@redhat.com>
4734 PR translation/79019
4735 PR translation/79020
4736 * params.def (PARAM_INLINE_MIN_SPEEDUP,
4737 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
4738 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
4740 * config/avr/avr.opt (maccumulate-args): Likewise.
4741 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
4742 * common.opt (freport-bug): Likewise.
4743 * cif-code.def (CIF_FINAL_ERROR): Likewise.
4744 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
4745 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
4746 translatable string.
4747 * config/i386/i386.c (function_value_32): Likewise.
4748 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
4749 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
4751 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
4752 * common/config/msp430/msp430-common.c (msp430_handle_option):
4754 * symtab.c (symtab_node::verify_base): Likewise.
4755 * opts.c (set_debug_level): Likewise.
4756 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
4757 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
4758 missing whitespace to translatable strings.
4759 * config/avr/avr.md (bswapsi2): Fix typo in comment.
4760 * config/sh/superh.h: Likewise.
4761 * config/i386/xopintrin.h: Likewise.
4762 * config/i386/znver1.md: Likewise.
4763 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
4764 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
4765 * double-int.h (struct double_int): Likewise.
4766 * double-int.c (div_and_round_double): Likewise.
4767 * wide-int.cc: Likewise.
4768 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
4769 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
4770 * cfgcleanup.c (crossjumps_occured): Renamed to ...
4771 (crossjumps_occurred): ... this.
4772 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
4775 PR tree-optimization/78899
4776 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
4777 returning bool return struct loop *, NULL for failure and the new
4779 (versionable_outer_loop_p): Don't version outer loop if it has
4780 dont_vectorized bit set.
4781 (tree_if_conversion): When versioning outer loop, ensure
4782 tree_if_conversion is performed also on the inner loop of the
4783 non-vectorizable outer loop copy.
4784 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
4785 LOOP_VECTORIZED in inner loop of the scalar outer loop and
4786 prevent vectorization of it.
4787 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
4788 the outer loop vectorization of the non-scalar version is attempted
4789 before vectorization of the inner loop in scalar version. If
4790 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
4791 vectorization of its inner loop.
4792 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
4793 has 2 inner loops, rename also on edges from bb whose single pred
4794 is outer_loop->header. Fix typo in function comment.
4796 2017-01-09 Martin Sebor <msebor@redhat.com>
4799 * asan.c (asan_emit_stack_protection): Increase local buffer size
4800 to avoid snprintf truncation warning.
4802 2017-01-09 Andrew Pinski <apinski@cavium.com>
4804 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
4805 to reference thunderx2t99 for the tuning structure
4806 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
4808 (thunderx2t99_extra_costs): This.
4809 * config/aarch64/aarch64-tune.md: Regenerate.
4810 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
4811 (vulcan_addrcost_table): This.
4812 (vulcan_regmove_cost): Rename to ...
4813 (thunderx2t99_regmove_cost): This.
4814 (vulcan_vector_cost): Rename to ...
4815 (thunderx2t99_vector_cost): this.
4816 (vulcan_branch_cost): Rename to ...
4817 (thunderx2t99_branch_cost): This.
4818 (vulcan_tunings): Rename to ...
4819 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
4820 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
4822 2017-01-09 Martin Jambor <mjambor@suse.cz>
4826 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
4827 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
4828 (propagate_vr_accross_jump_function): Use the above function for all
4829 value range computations for pass-through jump functions and type
4830 converasion from explicit value range values.
4831 (ipcp_propagate_stage): Do not attempt to deduce types of formal
4832 parameters from TYPE_ARG_TYPES.
4833 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
4834 (ipa_write_node_info): Stream type of the actual argument.
4835 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
4837 2017-01-09 Martin Liska <mliska@suse.cz>
4840 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
4841 (lookup_compiler): Do not show error message with have_E.
4843 2017-01-09 Jakub Jelinek <jakub@redhat.com>
4845 PR tree-optimization/78938
4846 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
4847 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
4848 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
4849 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
4852 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4854 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
4857 2017-01-09 Richard Biener <rguenther@suse.de>
4859 PR tree-optimization/78997
4860 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
4861 name condition properly.
4863 2017-01-09 Richard Biener <rguenther@suse.de>
4866 * dwarf2out.c (is_cxx): New overload with context.
4867 (is_naming_typedef_decl): Use it.
4869 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
4871 * invoke.texi (Option Summary): Correct spacing in option lists
4872 and add line breaks to fix over-long lines.
4874 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
4878 * extend.texi (Common Variable Attributes): Add xref to GCC
4879 Internals manual to explain mode attribute keywords.
4881 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
4884 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
4885 and Preprocessor Options.
4886 (Options for Linking): Document -pthread here....
4887 (RS/6000 and PowerPC Options): ...not here.
4888 (Solaris 2 Options): ...or here.
4889 * doc/cppopts.texi: Document -pthread.
4891 2017-01-08 Martin Sebor <msebor@redhat.com>
4894 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
4895 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
4896 New member functions.
4897 (format_directive): Used them.
4899 (pass_sprintf_length::handle_gimple_call): Same.
4900 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
4901 to avoid truncation for any argument.
4902 (extract_affine_mul): Same.
4903 * tree.c (get_file_function_name): Same.
4905 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
4908 * predict.def (PRED_INDIR_CALL): Set to 86.
4910 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
4912 PR preprocessor/54124
4913 * doc/cppopts.texi: Reformat -d subtable to list the full name
4914 of the options. Add cross-reference to the docs for the general
4915 compiler -d options.
4916 * doc/invoke.texi (Developer Options): Add cross-reference to the
4917 preprocessor-specific -d option documentation.
4919 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
4921 PR preprocessor/13498
4922 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
4923 redudant material, and reflect new command-line options.
4924 (System Headers): Likewise.
4926 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
4928 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
4929 -isystem, and -idirafter. Copy-edit.
4930 * doc/cppopts.texi: Copy-edit. Remove contradiction about
4931 default for -ftrack-macro-expansion. Delete obsolete and
4932 badly-formatted implementation details about -fdebug-cpp output.
4933 * doc/cppwarnopts.texi: Copy-edit.
4935 2017-01-07 David Malcolm <dmalcolm@redhat.com>
4938 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
4939 that the transition from a max line width >= 1<<10 to narrower
4940 lines works correctly.
4942 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
4944 * doc/options.texi (PerFunction): New.
4945 * opt-functions.awk (switch_flags): Map both Optimization and
4946 PerFunction to CL_OPTIMIZATION.
4947 * opth-gen.awk: Test for PerFunction flag along with
4949 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
4950 it only when the latter is present. Skip those that don't in
4951 the hash function generator.
4952 * common.opt (fvar-tracking): Mark as PerFunction instead of
4954 (fvar-tracking-assignments): Likewise.
4955 (fvar-tracking-assignments-toggle): Likewise.
4956 (fvar-tracking-uninit): Likewise.
4958 2017-01-07 Jakub Jelinek <jakub@redhat.com>
4960 PR translation/79018
4961 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
4964 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
4967 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
4968 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
4969 TARGET_LONG_JUMP_TABLE_OFFSETS.
4970 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
4971 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
4972 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
4973 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
4974 * config/m68k/m68k.md (tablejump expander): Likewise.
4975 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
4976 TARGET_LONG_JUMP_TABLE_OFFSETS.
4977 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
4978 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
4980 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
4981 David Holsgrove <david.holsgrove@xilinx.com>
4983 * common/config/microblaze/microblaze-common.c
4984 (TARGET_EXCEPT_UNWIND_INFO): Remove.
4985 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
4987 * config/microblaze/microblaze.c (microblaze_must_save_register)
4988 (microblaze_expand_epilogue, microblaze_return_addr): Handle
4990 (microblaze_eh_return): New function.
4991 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
4992 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
4993 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
4994 * config/microblaze/microblaze.md (eh_return): New pattern.
4996 2017-01-06 Jakub Jelinek <jakub@redhat.com>
4998 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
4999 GCC_DIAGNOSTIC_STRINGIFY): Define.
5001 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
5003 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5005 * config/arm/arm.md (<mcrr>): New.
5007 * config/arm/arm.c (arm_arch5te): New.
5008 (arm_option_override): Set arm_arch5te.
5009 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
5011 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
5012 (arm_mcrr_qualifiers): ... this. New.
5013 (MRRC_QUALIFIERS): Define to...
5014 (arm_mrrc_qualifiers): ... this. New.
5015 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
5017 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
5018 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
5019 (MRRCI, mrrc, MRRC): New.
5020 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
5021 VUNSPEC_MRRC2): New.
5023 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5025 * config/arm/arm.md (<mcr>): New.
5027 * config/arm/arm.c (arm_coproc_builtin_available): Add
5028 support for mcr, mrc, mcr2 and mrc2.
5029 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
5030 (arm_mcr_qualifiers): ... this. New.
5031 (MRC_QUALIFIERS): Define to ...
5032 (arm_mrc_qualifiers): ... this. New.
5033 (MCR_QUALIFIERS): Define to ...
5034 (arm_mcr_qualifiers): ... this. New.
5035 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
5037 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
5038 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
5039 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
5042 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5044 * config/arm/arm.md (*ldc): New.
5048 * config/arm/arm.c (arm_coproc_builtin_available): Add
5049 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
5050 (arm_coproc_ldc_stc_legitimate_address): New.
5051 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5052 'qualifier_const_pointer'.
5053 (LDC_QUALIFIERS): Define to...
5054 (arm_ldc_qualifiers): ... this. New.
5055 (STC_QUALIFIERS): Define to...
5056 (arm_stc_qualifiers): ... this. New.
5057 * config/arm/arm-protos.h
5058 (arm_coproc_ldc_stc_legitimate_address): New.
5059 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
5060 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
5061 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
5062 stc2, stcl, stc2l): New.
5063 * config/arm/constraints.md (Uz): New.
5064 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
5065 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
5066 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
5067 VUNSPEC_STC2L): New.
5069 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5071 * config/arm/arm.md (<cdp>): New.
5072 * config/arm/arm.c (neon_const_bounds): Rename this ...
5073 (arm_const_bounds): ... this.
5074 (arm_coproc_builtin_available): New.
5075 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
5076 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
5077 (CDP_QUALIFIERS): Define to...
5078 (arm_cdp_qualifiers): ... this. New.
5080 (arm_expand_builtin_args): Add case for 6 arguments.
5081 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
5082 (arm_const_bounds): ... this.
5083 (arm_coproc_builtin_available): New.
5084 * config/arm/arm_acle.h (__arm_cdp): New.
5086 * config/arm/arm_acle_builtins.def (cdp): New.
5088 * config/arm/iterators.md (CDPI,CDP,cdp): New.
5089 * config/arm/neon.md: Rename all 'neon_const_bounds' to
5091 * config/arm/types.md (coproc): New.
5092 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
5093 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
5094 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
5095 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
5097 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5099 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
5100 (UBINOP_QUALIFIERS): New.
5102 (acle_builtin_data): New. Change comment.
5103 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
5104 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
5105 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
5106 arm_acle_builtins.def.
5107 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
5108 (arm_init_acle_builtins): New.
5109 (CRC32_BUILTIN): Remove.
5110 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
5111 crc32cb, crc32ch and crc32cw.
5112 (arm_init_crc32_builtins): Remove.
5113 (arm_init_builtins): Use arm_init_acle_builtins rather
5114 than arm_init_crc32_builtins.
5115 (arm_expand_acle_builtin): New.
5116 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
5117 * config/arm/arm_acle_builtins.def: New.
5119 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5121 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
5122 (arm_builtin_datum): ... this.
5123 (arm_init_neon_builtin): Rename to ...
5124 (arm_init_builtin): ... this. Add a new parameters PREFIX
5125 and USE_SIG_IN_NAME.
5126 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
5127 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
5128 'arm_builtin_datum'.
5129 (arm_init_vfp_builtins): Likewise.
5130 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
5131 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
5132 (arm_expand_neon_args): Rename to ...
5133 (arm_expand_builtin_args): ... this. Rename builtin_arg
5134 enum values and differentiate between ARG_BUILTIN_MEMORY
5135 and ARG_BUILTIN_NEON_MEMORY.
5136 (arm_expand_neon_builtin_1): Rename to ...
5137 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
5138 values, arm_expand_builtin_args and add bool parameter NEON.
5139 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
5140 (arm_expand_vfp_builtin): Likewise.
5141 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
5143 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5146 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
5147 * predict.c (tree_estimate_probability_bb): Reverse direction of
5148 polymorphic call predictor.
5150 2017-01-06 David Malcolm <dmalcolm@redhat.com>
5152 * passes.c (execute_one_pass): Split out pass-skipping logic into...
5153 (determine_pass_name_match): ...this new function and...
5154 (should_skip_pass_p): ...this new function.
5156 2017-01-06 Nathan Sidwell <nathan@acm.org>
5158 * ipa-visibility.c (function_and_variable_visibility): Reformat
5159 comments and long lines. Remove extrneous if.
5160 * symtab.c (symtab_node::make_decl_local): Fix code format.
5161 (symtab_node::set_section_for_node): Fix comment typo.
5163 2017-01-06 Martin Liska <mliska@suse.cz>
5166 * lra-constraints.c: Rename invariant to lra_invariant.
5167 * predict.c (set_even_probabilities): Initialize e to NULL.
5169 2017-01-05 Martin Sebor <msebor@redhat.com>
5171 PR tree-optimization/78910
5172 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
5173 (format_integer): Correct off-by-one error in the handling
5174 of precision with negative numbers in signed conversions..
5176 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
5178 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
5180 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5182 PR tree-optimization/71016
5183 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
5184 factor_out_conditional_conversion. Formatting fix.
5185 (factor_out_conditional_conversion): Add cond_stmt argument.
5186 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
5187 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
5190 2017-01-05 David Malcolm <dmalcolm@redhat.com>
5192 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
5193 read-rtl-function.o, and selftest-rtl.o.
5194 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
5195 (selftest::aarch64_test_loading_full_dump): New function.
5196 (selftest::aarch64_run_selftests): New function.
5197 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
5198 selftest::aarch64_run_selftests.
5199 * config/i386/i386.c
5200 (selftest::ix86_test_loading_dump_fragment_1): New function.
5201 (selftest::ix86_test_loading_call_insn): New function.
5202 (selftest::ix86_test_loading_full_dump): New function.
5203 (selftest::ix86_test_loading_unspec): New function.
5204 (selftest::ix86_run_selftests): Call the new functions.
5205 * emit-rtl.c (maybe_set_max_label_num): New function.
5206 * emit-rtl.h (maybe_set_max_label_num): New decl.
5207 * function.c (instantiate_decls): Guard call to
5208 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
5209 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
5211 * gensupport.c (gen_reader::gen_reader): Pass "false"
5212 for new "compact" param of rtx_reader.
5213 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
5214 rather than an empty string for NULL strings.
5215 * read-md.c: Potentially include config.h rather than bconfig.h.
5216 Wrap include of errors.h with #ifdef GENERATOR_FILE.
5217 (have_error): New global, copied from errors.c.
5218 (md_reader::read_name): Rename to...
5219 (md_reader::read_name_1): ...this, adding "out_loc" param,
5220 and converting "missing name or number" to returning false, rather
5222 (md_reader::read_name): Reimplement in terms of read_name_1.
5223 (md_reader::read_name_or_nil): New function.
5224 (md_reader::read_string): Handle "(nil)" by returning NULL.
5225 (md_reader::md_reader): Add new param "compact".
5226 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
5227 (md_reader::read_file): New method.
5228 * read-md.h (md_reader::md_reader): Add new param "compact".
5229 (md_reader::read_file): New method.
5230 (md_reader::is_compact): New accessor.
5231 (md_reader::read_name): Convert return type from void to file_location.
5232 (md_reader::read_name_or_nil): New decl.
5233 (md_reader::read_name_1): New decl.
5234 (md_reader::m_compact): New field.
5235 (noop_reader::noop_reader): Pass "false" for new "compact" param
5237 (rtx_reader::rtx_reader): Add new "compact" param.
5238 (rtx_reader::read_rtx_operand): Make virtual and convert return
5239 type from void to rtx.
5240 (rtx_reader::read_until): New decl.
5241 (rtx_reader::handle_any_trailing_information): New virtual function.
5242 (rtx_reader::postprocess): New virtual function.
5243 (rtx_reader::finalize_string): New virtual function.
5244 (rtx_reader::m_in_call_function_usage): New field.
5245 (rtx_reader::m_reuse_rtx_by_id): New field.
5246 * read-rtl-function.c: New file.
5247 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
5248 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
5249 (selftest::verify_three_block_rtl_cfg): New decl.
5250 * read-rtl-function.h: New file.
5251 * read-rtl.c: Potentially include config.h rather than bconfig.h.
5252 For host, include function.h, memmodel.h, and emit-rtl.h.
5253 (one_time_initialization): New function.
5254 (struct compact_insn_name): New struct.
5255 (compact_insn_names): New array.
5256 (find_code): Handle insn codes in compact dumps.
5257 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
5258 (bind_subst_iter_and_attr): Likewise.
5259 (add_condition_to_string): Likewise.
5260 (add_condition_to_rtx): Likewise.
5261 (apply_attribute_uses): Likewise.
5262 (add_current_iterators): Likewise.
5263 (apply_iterators): Likewise.
5264 (initialize_iterators): Guard usage of apply_subst_iterator with
5265 #ifdef GENERATOR_FILE.
5266 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
5267 (md_reader::read_mapping): Likewise.
5268 (add_define_attr_for_define_subst): Likewise.
5269 (add_define_subst_attr): Likewise.
5270 (read_subst_mapping): Likewise.
5271 (check_code_iterator): Likewise.
5272 (rtx_reader::read_rtx): Likewise. Move one-time initialization
5274 (one_time_initialization): New function.
5275 (rtx_reader::read_until): New method.
5276 (read_flags): New function.
5277 (parse_reg_note_name): New function.
5278 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
5279 Handle reuse_rtx ids.
5280 Wrap iterator lookup within #ifdef GENERATOR_FILE.
5281 Add parsing support for RTL dumps, mirroring the special-cases in
5282 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
5283 values, and calling handle_any_trailing_information.
5284 (rtx_reader::read_rtx_operand): Convert return type from void
5285 to rtx, returning return_rtx. Handle case 'e'. Call
5286 finalize_string on XSTR and XTMPL fields.
5287 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
5288 "(nil)" values were omitted. Call the postprocess vfunc on the
5290 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
5291 class ctor. Initialize m_in_call_function_usage. Call
5292 one_time_initialization.
5293 * rtl-tests.c (selftest::test_uncond_jump): Call
5294 set_new_first_and_last_insn.
5295 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
5296 * selftest-rtl.c: New file.
5297 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
5298 (selftest::get_insn_by_uid): New decl.
5299 * selftest-run-tests.c (selftest::run_tests): Call
5300 read_rtl_function_c_tests.
5301 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
5302 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
5305 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
5307 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
5308 operands in a special way. Assert that pos+len <= mode precision.
5310 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5312 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
5313 3 argument Alias with unlimited for the negative form.
5314 (fno-vect-cost-model): Removed.
5316 2017-01-05 Martin Liska <mliska@suse.cz>
5318 * hsa-gen.c (gen_hsa_divmod): New function.
5319 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
5321 2017-01-05 Martin Liska <mliska@suse.cz>
5324 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
5327 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5329 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
5330 small constant length operands.
5332 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5334 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
5335 between loop iterations.
5337 2017-01-05 Martin Liska <mliska@suse.cz>
5340 * gimplify.c (gimplify_decl_expr): Compare to
5341 asan_poisoned_variables instread of checking flags.
5342 (gimplify_target_expr): Likewise.
5343 (gimplify_expr): Likewise.
5344 (gimplify_function_tree): Conditionally initialize
5345 asan_poisoned_variables.
5347 2017-01-04 Jeff Law <law@redhat.com>
5349 PR tree-optimizatin/78812
5350 * rtl.h (contains_mem_rtx_p): Prototype.
5351 * ifcvt.c (containts_mem_rtx_p): Move from here to...
5352 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
5353 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
5354 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
5355 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
5357 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5359 * input.c (assert_char_at_range): Default-initialize actual_range.
5361 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5363 * df-scan.c (df_ref_create_structure): Make regno unsigned,
5364 to match the caller.
5366 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5368 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
5369 insns after final jump in test to emit dummy move.
5371 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5373 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
5374 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
5376 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5378 * multiple_target.c (create_dispatcher_calls): Init e_next.
5379 * tree-ssa-loop-split.c (split_loop): Init border.
5380 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
5383 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
5388 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
5389 (altivec_register_operand): Do not return true if the operand
5390 contains a SUBREG mixing SImode and SFmode.
5391 (vsx_register_operand): Likewise.
5392 (vsx_reg_sfsubreg_ok): New predicate.
5393 (vfloat_operand): Do not return true if the operand contains a
5394 SUBREG mixing SImode and SFmode.
5395 (vint_operand): Likewise.
5396 (vlogical_operand): Likewise.
5397 (gpc_reg_operand): Likewise.
5398 (int_reg_operand): Likewise.
5399 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
5400 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
5401 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
5403 (rs6000_emit_move_si_sf_subreg): New helper function.
5404 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
5405 fixup SUBREGs involving SImode and SFmode.
5406 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
5407 numbers for the new peephole2 optimization.
5408 (peephole2 for SFmode unions): New peephole2 to optimize cases in
5409 the GLIBC math library that do AND/IOR/XOR operations on single
5410 precision floating point.
5411 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
5412 target macros to say whether we need to avoid SUBREGs mixing
5414 (TARGET_ALLOW_SF_SUBREG): Likewise.
5415 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
5416 (UNSPEC_SI_FROM_SF): Likewise.
5417 (iorxor): Change spacing.
5418 (and_ior_xor): New iterator for AND, IOR, and XOR.
5419 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
5420 (movdi_from_sf_zero_ext): Likewise.
5421 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
5422 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
5423 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
5424 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
5425 (fms<mode>4): Likewise.
5426 (fnma<mode>4): Likewise.
5427 (fnms<mode>4): Likewise.
5428 (nfma<mode>4): Likewise.
5429 (nfms<mode>4): Likewise.
5431 2017-01-04 Marek Polacek <polacek@redhat.com>
5434 * doc/invoke.texi: Document -Wpointer-compare.
5436 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5438 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
5441 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
5442 descriptions for -gdwarf-5 and emit them as uleb128 instead of
5445 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
5448 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
5449 documentation of the powerpc_popcntb_ok attribute.
5450 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5451 code to issue warning messages if a requested CPU configuration is
5452 not supported by the binary (assembler and loader) toolchain.
5453 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
5454 made to define a built-in function that has been disabled.
5455 (paired_init_builtins): Add assertion to prevent ICE if attempt is
5456 made to define a built-in function that has been disabled.
5457 (altivec_init_builtins): Add comment explaining why definition
5458 of the DST built-in functions is not preceded by an assertion
5459 check. Add assertions to prevent ICE if attempts are made to
5460 define an altivec predicate or an abs* built-in function that has
5462 (htm_init_builtins): Add comment explaining why definition of the
5463 htm built-in functions is not preceded by an assertion check.
5465 2017-01-04 Jeff Law <law@redhat.com>
5467 PR tree-optimizatin/67955
5468 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
5469 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
5470 the points-to solution does not include pt_null. Use DECL_PT_UID
5473 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
5475 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
5476 Use gen_int_mode instead of gen_lopwart for const_int operands.
5478 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5480 PR tree-optimization/71563
5481 * match.pd: Simplify X << Y into X if Y is known to be 0 or
5482 out of range value - has low bits known to be zero.
5484 2017-01-04 Alan Modra <amodra@gmail.com>
5486 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
5487 * configure: Regenerate.
5488 * config.in: Regenerate.
5490 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5493 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
5494 a substring of the message, but strcmp with the whole message. Ifdef
5495 ENABLE_NLS, translate the message first using dgettext.
5497 2017-01-03 Jeff Law <law@redhat.com>
5499 PR tree-optimizatin/78856
5500 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
5501 (mark_threaded_blocks): Remove code to truncate thread paths that
5502 cross multiple loop headers. Instead invalidate the cached loop
5503 iteration information and handle case of a thread path walking
5504 into an irreducible region.
5506 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5509 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
5510 assertions. Add support for doing the signbit if the IEEE 128-bit
5511 floating point value is in a GPR.
5512 * config/rs6000/rs6000.md (Fsignbit): Delete.
5513 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
5514 Update the length attribute if the value is in a GPR.
5515 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
5516 the sign or zero extension instruction, since the value is always 0/1.
5517 (signbit<mode>2_dm2): Delete using <Fsignbit>.
5520 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
5521 extracting SImode to a GPR register so that we can generate a
5522 store, limit the vector to be in a traditional Altivec register
5523 for the vextuwrx instruction.
5525 2017-01-03 Ian Lance Taylor <iant@google.com>
5527 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
5529 2017-01-03 Martin Sebor <msebor@redhat.com>
5531 PR tree-optimization/78696
5532 * gimple-ssa-sprintf.c (format_floating): Correct handling of
5533 precision. Use MPFR for %f for greater fidelity. Correct handling
5535 (pass_sprintf_length::compute_format_length): Set width and precision
5536 specified by asrerisk to void_node for vararg functions.
5537 (try_substitute_return_value): Adjust dump output.
5539 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
5541 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
5543 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
5545 * doc/invoke.texi (SPARC options): Document -mlra as the default.
5546 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
5547 -mlra/-mno-lra was passed to the compiler.
5549 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
5551 PR rtl-optimization/65618
5552 * emit-rtl.c (try_split): Move initialization of "before" and
5553 "after" to just before the call to emit_insn_after_setloc.
5555 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
5557 * doc/md.texi (Standard Names): Remove reference to Java frontend.
5559 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
5561 * dwarf2out.c (gen_enumeration_type_die): When
5562 -gno-strict-dwarf, add a DW_AT_encoding attribute.
5564 2017-01-03 Jakub Jelinek <jakub@redhat.com>
5566 PR tree-optimization/78965
5567 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
5568 Change first argument from const call_info & to call_info &. For %n
5569 set info.nowrite to false.
5572 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
5573 possibly throwing calls.
5575 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
5576 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
5577 and fns handling, rather than in a separate case SSA_NAME.
5579 2017-01-02 Jeff Law <law@redhat.com>
5581 * config/darwin-driver.c (darwin_driver_init): Const-correctness
5582 fixes for first_period and second_period variables.
5584 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
5587 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
5588 (*insvqi_1): New insn pattern.
5589 (*insvqi_1_mem_rex64): Ditto.
5591 (*insvqi_3): Rename from *insvqi.
5593 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
5595 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
5597 * doc/cfg.texi (Edges): Remove reference to Java.
5598 (Maintaining the CFG): Ditto.
5600 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5603 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
5604 transparent aliases.
5606 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5609 * predict.def (PRED_CALL): Update hitrate.
5610 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
5611 * predict.c (tree_estimate_probability_bb): Split CALL predictor
5612 into direct/indirect/polymorphic variants.
5614 2017-01-01 Jakub Jelinek <jakub@redhat.com>
5616 Update copyright years.
5618 * gcc.c (process_command): Update copyright notice dates.
5619 * gcov-dump.c (print_version): Ditto.
5620 * gcov.c (print_version): Ditto.
5621 * gcov-tool.c (print_version): Ditto.
5622 * gengtype.c (create_file): Ditto.
5623 * doc/cpp.texi: Bump @copying's copyright year.
5624 * doc/cppinternals.texi: Ditto.
5625 * doc/gcc.texi: Ditto.
5626 * doc/gccint.texi: Ditto.
5627 * doc/gcov.texi: Ditto.
5628 * doc/install.texi: Ditto.
5629 * doc/invoke.texi: Ditto.
5631 Copyright (C) 2017 Free Software Foundation, Inc.
5633 Copying and distribution of this file, with or without modification,
5634 are permitted in any medium without royalty provided the copyright
5635 notice and this notice are preserved.