1 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4 * profile-count.c (profile_count::to_sreal_scale): Fix return value
5 for uninitialied counts.
7 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
10 * tree-inline.c (estimate_num_insns): Set to 1 at least.
12 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
15 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane
18 2017-11-19 Tom de Vries <tom@codesourcery.com>
20 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
23 2017-11-19 Tom de Vries <tom@codesourcery.com>
25 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
28 2017-11-19 Tom de Vries <tom@codesourcery.com>
30 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
33 2017-11-19 Tom de Vries <tom@codesourcery.com>
35 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
37 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
39 2017-11-19 Tom de Vries <tom@codesourcery.com>
41 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
43 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
44 semicolon after MCORE_EXPORT_NAME call.
46 2017-11-19 Tom de Vries <tom@codesourcery.com>
49 * vmsdbgout.c (vmsdbgout_early_finish): New function.
50 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
52 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
54 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
55 (cgraph_edge::clone): Cleanup updating of profile.
56 * ipa-cp.c (update_profiling_info): Likewise.
57 * ipa-inline-transform.c (inline_transform): Likewise.
58 * ipa-inline.c (inline_small_functions): Add missing space to dump.
59 * ipa-split.c (execute_split_functions): Do not split when function
61 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
62 * profile-count.c (profile_count::dump): Add global0.
63 (profile_count::to_cgraph_frequency): Do not ICE when entry is
65 (profile_count::to_sreal_scale): Likewise.
66 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
67 (profile_count::combine_with_ipa_count): New function.
68 * profile-count.h (profile_guessed_global0adjusted): New.
69 (profile_count::adjusted_zero): New.
70 (profile_count::global0adjusted): New.
71 (profile_count::combine_with_ipa_count): New.
72 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
73 correct profile of return block of split functions.
74 (copy_cfg_body): Remove unused profile_count.
75 (copy_body): Likewise.
76 (expand_call_inline): Update.
77 (tree_function_versioning): Update.
79 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
81 * hash-set.h (hash_set::empty): New.
82 * tree-ssa-threadbackward.h: Delete.
83 * tree-ssa-threadbackward.c (class thread_jumps): New.
84 Move max_threaded_paths into class.
85 (fsm_find_thread_path): Remove arguments that are now in class.
86 (profitable_jump_thread_path): Rename to...
87 (thread_jumps::profitable_jump_thread_path): ...this.
88 (convert_and_register_jump_thread_path): Rename to...
89 (thread_jumps::convert_and_register_current_path): ...this.
90 (check_subpath_and_update_thread_path): Rename to...
91 (thread_jumps::check_subpath_and_update_thread_path): ...this.
92 (register_jump_thread_path_if_profitable): Rename to...
93 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
94 (handle_phi): Rename to...
95 (thread_jumps::handle_phi): ...this.
96 (handle_assignment): Rename to...
97 (thread_jumps::handle_assignment): ...this.
98 (fsm_find_control_statement_thread_paths): Rename to...
99 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
100 (find_jump_threads_backwards): Rename to...
101 (thread_jumps::find_jump_threads_backwards): ...this.
102 Initialize path local data.
103 (pass_thread_jumps::execute): Call find_jump_threads_backwards
104 from within thread_jumps class.
105 (pass_early_thread_jumps::execute): Same.
107 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
109 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
111 2017-11-17 Steve Ellcey <sellcey@cavium.com>
113 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
114 to canonical location.
116 2017-11-17 Steve Ellcey <sellcey@cavium.com>
119 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
121 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
123 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
125 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
126 rcrt1.o%s/grcrt1.o%s for -static-pie.
128 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
130 * i386.c (ix86_multiplication_cost, ix86_division_cost,
131 ix86_shift_rotate_cost): Break out from ...
132 (ix86_rtx_costs): ... here.
133 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
136 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
138 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
139 when entry block was promoted unlikely.
140 (estimate_bb_frequencies): Increase frequency scale.
141 * profile-count.h (profile_count): Export precision info.
143 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
145 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
146 disturbin profile of entry block.
148 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
150 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
153 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
155 * ipa-cp.c (update_profiling_info): Handle conversion to local
157 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
159 2017-11-17 Jeff Law <law@redhat.com>
161 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
162 from evrp_dom_walker class. Various methods moved into new class.
163 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
164 (evrp_range_analyzer::enter): New method.
165 (evrp_range_analyzer::leave): New method.
166 (evrp_dom_walker): Remove delegators no longer needed by this class.
167 Replace vr_values data member with evrp_range_analyzer
169 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
170 method extracted from evrp_dom_walker::before_dom_children.
171 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
172 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
174 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
175 Add private copy constructor and move assignment operators.
176 Privatize methods and class data where trivially possible.
177 (evrp_dom_walker::cleanup): New function, extracted from
178 execute_early_vrp. Simplify access to class data.
180 * vr-values.h (get_output_for_vrp): Prototype.
181 * vr-values.c (get_output_for_vrp): New function extracted from
182 vrp_visit_assignment_or_call and extract_range_from_stmt.
183 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
185 2017-11-17 Luis Machado <luis.machado@linaro.org>
187 * config/aarch64/aarch64.c
188 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
189 (qdf24xx_tunings) <autoprefetcher_model>: Set to
190 tune_params::AUTOPREFETCHER_WEAK.
192 2017-11-17 Tamar Christina <tamar.christina@arm.com>
195 * config/arm/arm.c (arm_valid_target_attribute_rec):
196 Parse "arch=" and "+<ext>".
197 (arm_valid_target_attribute_tree): Re-init global options.
198 (arm_option_override): Make non-static.
199 (arm_options_perform_arch_sanity_checks): Make errors fatal.
200 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
201 (__ARM_FEATURE_CRC32): Support undef.
202 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
203 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
205 2017-11-17 David Malcolm <dmalcolm@redhat.com>
207 * gdbinit.in (break-on-diagnostic): New command.
209 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
211 * config/i386/i386.c (ix86_expand_epilogue): Change simple
212 return to indirect jump for EH return if control-flow
213 protection is enabled. Change explicit 'false' argument in
214 pro_epilogue_adjust_stack with a value of
216 * config/i386/i386.md (simple_return_indirect_internal):
217 Remove SImode restriction to support 64-bit.
219 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
221 * combine.c (added_notes_insn): New.
222 (try_combine): Handle added_notes_insn like added_links_insn.
223 Rewrite return value code.
224 (distribute_notes): Set added_notes_insn to the earliest insn we added
227 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
229 PR rtl-optimization/82621
230 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
231 dest of one of those SETs is unused.
233 2017-11-17 Richard Biener <rguenther@suse.de>
236 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
237 * tree-pretty-print.c (dump_generic_node): Handle
238 annot_expr_parallel_kind.
239 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
240 * gimplify.c (gimple_boolify): Likewise.
242 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
244 * config.gcc (extra_headers): Add cet.h for x86 targets.
245 * config/i386/cet.h: New file.
246 * doc/install.texi: Add --enable-cet/--disable-cet.
248 2017-11-17 Richard Biener <rguenther@suse.de>
250 PR tree-optimization/83017
251 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
252 (gen_parallel_loop): Properly count iterations.
253 (parallelize_loops): Handle loop->can_be_parallel independent
254 of flag_loop_parallelize_all. Make static profitability test match
256 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
257 * invoke.texi (parloops-min-per-thread): Document.
259 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
261 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
262 upstreaming review comments.
264 2017-11-17 Tamar Christina <tamar.christina@arm.com>
266 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
267 with fast unaligned access.
268 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
270 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
272 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
273 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
275 2017-11-17 Richard Biener <rguenther@suse.de>
277 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
278 folding of references.
280 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
283 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
284 of replacing call to strncmp with corresponding call to strcmp when
287 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
289 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
290 option prefer-avx256 for skylake-avx512 configuration.
291 * config/i386/i386.c (ix86_option_override_internal): Ditto.
292 (get_builtin_code_for_version): Ditto.
294 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
295 Monk Chiang <sh.chiang04@gmail.com>
297 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
298 (FIXED_REGISTERS): Reserve more register numbers.
299 (CALL_USED_REGISTERS): Likewise.
300 (REG_ALLOC_ORDER): Likewise.
301 (REG_CLASS_CONTENTS): Likewise.
302 (REGISTER_NAMES): Likewise.
304 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
305 Kito Cheng <kito.cheng@gmail.com>
307 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
309 * config/nds32/iterators.md: Add vector mode iterators and attributes.
311 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
313 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
314 parameter list for vec_splats.
316 2017-11-16 Joseph Myers <joseph@codesourcery.com>
318 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
320 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
323 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
326 * doc/install.texi: Document incompatibility of Perl >=5.6.26
327 with the required version of automake 1.11.6.
329 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
331 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
333 (DU_C2_3_power9): Correct reservation combinations.
334 (FP_DIV_power9, VEC_DIV_power9): New.
335 (power9-alu): Split out rotate/shift...
336 (power9-rot): ...to here, correct dispatch resource.
337 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
339 (power9-fp): Correct latency.
340 (power9-sdiv): Add div/sqrt resource.
341 (power9-ddiv): Correct latency, add div/sqrt resource.
342 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
343 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
345 (power9-qpdiv, power9-qpmul): Adjust resource usage.
347 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
349 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
350 switch statement mapping KF built-ins to TF built-ins if we don't
351 have the proper ISA 3.0 assembler support.
353 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
355 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
356 (lower_emutls_function_body): Do not compute it.
358 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
360 * ipa-split.c (split_bb_info): Turn time to sreal.
361 (split_point): Likewise.
362 (dump_split_point): Likewise.
363 (fine_split_points): Likewise.
364 (execute_split_functions): Only zero split_bbs; turn time to sreals.
366 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
368 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
371 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
373 * predict.c (combine_predictions_for_bb): Preserve zero predicted
375 (expensive_function_p): Remove useless assert.
376 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
378 2017-11-16 Martin Sebor <msebor@redhat.com>
380 PR tree-optimization/82588
381 PR tree-optimization/82583
382 * tree-vrp.c (check_array_ref): Handle flexible array members,
383 string literals, and inner indices.
384 (search_for_addr_array): Add detail to diagnostics.
386 2017-11-16 Nathan Sidwell <nathan@acm.org>
390 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
391 (SET_DECL_ASSEMBLER_NAME): Forward to
392 overwrite_decl_assembler_name.
393 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
394 (overwrite_decl_assembler_name): Declare.
395 * tree.c (overwrite_decl_assembler_name): New.
396 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
397 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
398 (LANG_HOOKS_INITIALIZER): Add it.
399 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
400 * langhooks.c (lhd_set_decl_assembler_name): Use
401 SET_DECL_ASSEMBLER_NAME.
402 (lhd_overwrite_decl_assembler_name): Default implementation.
404 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
405 Jackson Woodruff <jackson.woodruff@arm.com>
407 PR tree-optimization/71026
408 * match.pd: Canonicalize constant multiplies in division.
410 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
412 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
413 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
414 Remove OPT_fomit_frame_pointer entry.
415 * common/config/arc/arc-common.c: Likewise.
416 * common/config/arm/arm-common.c: Likewise.
417 * common/config/avr/avr-common.c: Likewise.
418 * common/config/c6x/c6x-common.c: Likewise.
419 * common/config/cr16/cr16-common.c: Likewise.
420 * common/config/cris/cris-common.c: Likewise.
421 * common/config/epiphany/epiphany-common.c: Likewise.
422 * common/config/fr30/fr30-common.c: Likewise.
423 * common/config/frv/frv-common.c: Likewise.
424 * common/config/ia64/ia64-common.c: Likewise.
425 * common/config/iq2000/iq2000-common.c: Likewise.
426 * common/config/lm32/lm32-common.c: Likewise.
427 * common/config/m32r/m32r-common.c: Likewise.
428 * common/config/mcore/mcore-common.c: Likewise.
429 * common/config/microblaze/microblaze-common.c: Likewise.
430 * common/config/mips/mips-common.c: Likewise.
431 * common/config/mmix/mmix-common.c: Likewise.
432 * common/config/mn10300/mn10300-common.c: Likewise.
433 * common/config/nios2/nios2-common.c: Likewise.
434 * common/config/pa/pa-common.c: Likewise.
435 * common/config/pdp11/pdp11-common.c: Likewise.
436 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
437 * common/config/riscv/riscv-common.c: Likewise.
438 * common/config/rs6000/rs6000-common.c: Likewise.
439 * common/config/rx/rx-common.c: Likewise.
440 * common/config/s390/s390-common.c: Likewise.
441 * common/config/sh/sh-common.c: Likewise.
442 * common/config/sparc/sparc-common.c: Likewise.
443 * common/config/tilegx/tilegx-common.c: Likewise.
444 * common/config/tilepro/tilepro-common.c: Likewise.
445 * common/config/v850/v850-common.c: Likewise.
446 * common/config/visium/visium-common.c: Likewise.
447 * common/config/xstormy16/xstormy16-common.c: Likewise.
448 * common/config/xtensa/xtensa-common.c: Likewise.
449 * invoke.texi (-fomit-frame-pointer): Update documentation.
451 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
453 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
455 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
457 * cfg.c (scale_bbs_frequencies_int,
458 cale_bbs_frequencies_gcov_type): Remove.
459 * cfg.h (scale_bbs_frequencies_int,
460 cale_bbs_frequencies_gcov_type): Remove.
462 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
464 * tree-ssa-loop-manip.c
465 (scale_dominated_blocks_in_loop): Update to profile counts.
466 (tree_transform_and_unroll_loop): Likewise.
468 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
470 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
471 scale_bbs_frequencies_int.
473 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
475 * final.c (compute_alignments): Use counts rather than frequencies.
477 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
479 * cfgloopanal.c: Include sreal.h
480 (average_num_loop_insns): Use counts and sreal for accounting.
482 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
484 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
487 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
488 Kito Cheng <kito.cheng@gmail.com>
490 * config/nds32/constraints.md: Provide more constraints.
491 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
492 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
493 support constraints usage.
495 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
497 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
499 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
500 Kito Cheng <kito.cheng@gmail.com>
502 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
503 * config/nds32/nds32.opt: Refine the layout.
504 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
505 TARGET_EXT_STRING): Support new options.
506 * config/nds32/nds32.h: Likewise.
507 * config/nds32/nds32.md: Likewise.
508 * config/nds32/nds32-predicates.c: Likewise.
509 * config/nds32/constraints.md: Likewise.
510 * common/config/nds32/nds32-common.c: Likewise.
512 2017-11-16 Julia Koval <julia.koval@intel.com>
515 * config/i386/gfniintrin.h: Add sse check.
516 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
518 2017-11-16 Julia Koval <julia.koval@intel.com>
520 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
521 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
522 (ix86_handle_option): Handle -mavx512vbmi2.
523 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
524 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
525 * config/i386/i386-c.c (__AVX512VBMI2__): New.
526 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
527 (ix86_valid_target_attribute_inner_p): Ditto.
528 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
529 * config/i386/i386.opt (mavx512vbmi2): New option.
530 * doc/invoke.texi: Add new option.
532 [2017-11-16 Julia Koval <julia.koval@intel.com>
534 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
535 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
536 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
537 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
538 _mm512_gf2p8mul_epi8): New intrinsics.
539 * config/i386/i386-builtin-types.def
540 (V64QI_FTYPE_V64QI_V64QI): New type.
541 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
542 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
543 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
544 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
545 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
546 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
548 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
550 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
551 explicitly as a stream of bytes.
553 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
555 * config/rs6000/altivec.h (vec_xst_be): New #define.
556 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
557 and externalize from *altivec_vperm_<mode>_internal.
558 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
560 (XL_BE_V8HI): Likewise.
561 (XL_BE_V4SI): Likewise.
562 (XL_BE_V4SI): Likewise.
563 (XL_BE_V2DI): Likewise.
564 (XL_BE_V4SF): Likewise.
565 (XL_BE_V2DF): Likewise.
566 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
567 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
568 all array entries with these keys: VSX_BUILTIN_VEC_XL,
569 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
570 VSX_BUILTIN_VEC_XST_BE.
571 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
572 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
574 (altivec_init_builtins): Replace conditional calls to def_builtin
575 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
576 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
577 with unconditional calls. Remove calls to def_builtin for
578 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
579 __builtin_vec_xst_be.
580 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
581 to define_expand, and add alternate RTL generation for P8.
582 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
584 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
585 add alternate RTL generation for P8.
586 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
587 vsx_ld_elemrev_v16qi.
588 (vsx_st_elemrev_v8hi): Convert define_insn
589 to define_expand, and add alternate RTL generation for P8.
590 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
592 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
593 add alternate RTL generation for P8.
594 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
595 vsx_st_elemrev_v16qi.
597 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
600 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
601 TARGET_AVX512ER check.
602 (ix86_option_override_internal): Set MASK_VZEROUPPER if
603 neither -mzeroupper nor -mno-zeroupper is used and
604 TARGET_EMIT_VZEROUPPER is set.
605 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
606 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
608 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
610 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
611 folding of vector compares.
612 (fold_build_vec_cmp): New helper function.
613 (fold_compare_helper): New helper function.
614 (builtin_function_type): Add compare builtins to the list of functions
615 having unsigned arguments. Cosmetic updates to comment indentation.
616 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
617 the not+eq combination.
619 2017-11-15 Bin Cheng <bin.cheng@arm.com>
621 PR tree-optimization/82726
622 PR tree-optimization/70754
623 * tree-predcom.c (order_drefs_by_pos): New function.
624 (combine_chains): Move code setting has_max_use_after to...
625 (try_combine_chains): ...here. New parameter. Sort combined chains
626 according to position information.
627 (tree_predictive_commoning_loop): Update call to above function.
628 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
630 2017-11-15 Bin Cheng <bin.cheng@arm.com>
632 PR tree-optimization/82726
634 2017-01-23 Bin Cheng <bin.cheng@arm.com>
636 PR tree-optimization/70754
637 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
638 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
639 combined stmt before it if not NULL.
640 (combine_chains): Process refs reversely and compute dominance point
644 2017-02-23 Bin Cheng <bin.cheng@arm.com>
646 PR tree-optimization/79663
647 * tree-predcom.c (combine_chains): Process refs in reverse order
648 only for ZERO length chains, and add explaining comment.
650 2017-11-15 Tamar Christina <tamar.christina@arm.com>
652 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
653 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
654 * doc/invoke.texi (ARM Options): Add armv8.3-a.
656 2017-11-15 Tamar Christina <tamar.christina@arm.com>
658 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
660 2017-11-15 Martin Liska <mliska@suse.cz>
662 * tree-cfg.c (pass_warn_function_return::execute):
663 Compare warn_return_type for greater than zero.
665 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
669 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
670 to return true on Xeon and not on Xeon Phi.
671 (ix86_check_avx256_register): Changed to ...
672 (ix86_check_avx_upper_register): ... this. Add extra check for
673 VALID_AVX512F_REG_OR_XI_MODE.
674 (ix86_avx_u128_mode_needed): Changed
675 ix86_check_avx256_register to ix86_check_avx_upper_register.
676 (ix86_check_avx256_stores): Changed to ...
677 (ix86_check_avx_upper_stores): ... this. Changed
678 ix86_check_avx256_register to ix86_check_avx_upper_register.
679 (ix86_avx_u128_mode_after): Changed
680 avx_reg256_found to avx_upper_reg_found. Changed
681 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
682 (ix86_avx_u128_mode_entry): Changed
683 ix86_check_avx256_register to ix86_check_avx_upper_register.
684 (ix86_avx_u128_mode_exit): Ditto.
685 * config/i386/i386.h: (host_detect_local_cpu): New define.
687 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
689 * config/arm/xgene1.md (xgene1): Split into automatons
690 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
691 (xgene1_f_load): Adjust reservations and/or types.
692 (xgene1_f_store): Likewise.
693 (xgene1_load_pair): Likewise.
694 (xgene1_store_pair): Likewise.
695 (xgene1_fp_load1): Likewise.
696 (xgene1_load1): Likewise.
697 (xgene1_store1): Likewise.
698 (xgene1_move): Likewise.
699 (xgene1_alu): Likewise.
700 (xgene1_simd): Likewise.
701 (xgene1_bfm): Likewise.
702 (xgene1_neon_load1): Likewise.
703 (xgene1_neon_store1): Likewise.
704 (xgene1_neon_logic): Likewise.
705 (xgene1_neon_st1): Likewise.
706 (xgene1_neon_ld1r): Likewise.
707 (xgene1_alu_cond): Added.
708 (xgene1_shift_reg): Likwise.
709 (xgene1_bfx): Likewise.
710 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
712 2017-11-15 Jakub Jelinek <jakub@redhat.com>
715 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
717 (can_widen_mult_without_libcall): New function.
718 (expand_mul_overflow): If only checking unsigned mul overflow,
719 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
720 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
721 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
722 (expand_DIVMOD): Formatting fix.
723 * expmed.h (expand_mult): Add NO_LIBCALL argument.
724 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
725 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
727 PR tree-optimization/82977
728 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
729 constructed temporary to strlen_to_stridx.put.
731 2017-11-15 Martin Liska <mliska@suse.cz>
733 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
734 * configure: Regenerate.
736 2017-11-15 Martin Liska <mliska@suse.cz>
739 * config/sh/sh-mem.cc: Use proper probability for
742 2017-11-14 Jeff Law <law@redhat.com>
744 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
745 the red zone for stack_clash_protection_final_dynamic_probe targets
746 when the total dynamic stack size is zero bytes.
748 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
749 blocks is post order.
751 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
753 * dumpfile.h (TDF_COMPARE_DEBUG): New.
754 * final.c (rest_of_clean_state): Set it for the
755 -fcompare-debug dump.
756 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
757 class when TDF_COMPARE_DEBUG is set.
759 * dwarf2out.c (gen_producer_string): Discard
762 2017-11-15 Joseph Myers <joseph@codesourcery.com>
765 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
766 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
767 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
768 (__TGMATH_REAL_2_3): Remove macros.
769 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
770 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
772 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
774 (remquo): Define using __TGMATH_REAL_3.
776 2017-11-14 Jeff Law <law@redhat.com>
778 * vr-values.c: New file with contents extracted from tree-vrp.c.
779 * Makefile.in (OBJS): Add vr-values.o
780 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
781 (set_value_range, set_and_canonicalize_value_range): Likewise.
782 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
783 (value_range_constant_singleton, symbolic_range_p): Likewise.
784 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
785 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
786 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
787 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
788 (find_case_label_range, find_case_label_index): Likewise.
789 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
790 (range_int_cst_singleton_p, value_inside_range): Likewise.
791 (get_single_symbol): Likewise.
792 (switch_update): Move structure definition here.
793 (to_remove_edges, to_update_switch_stmts): Provide externs.
794 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
795 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
796 (vrp_val_is_min, set_value_range): Likewise.
797 (set_and_canonicalize_value_range, copy_value_range): Likewise.
798 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
799 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
800 (range_is_nonnull, range_int_cst_p): Likewwise.
801 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
802 (get_single_symbol, operand_less_p): Likewise
803 (compare_values_warnv, compare_values): Likewise.
804 (value_inside_range, value_range_constant_singleton): Likewise.
805 (zero_nonzero_bitgs_from_vr): Likewise.
806 (extract_range_from_binary_expr_1): Likewise.
807 (overflow_comparison_p): Likewise.
808 (to_remove_edges, to_update_switch_stmts): Likewise.
809 (find_case_label-index, find_case_label_range): Likewise.
810 (switch_update, set_value_range_to_nonnegative): Remove.
811 (set_value_range_to_truthvalue): Likewise.
812 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
813 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
814 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
815 (find_case_label_ranges, test_for_singularity): Likewise.
816 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
817 (x_vr_values): Move to its remaining use site.
819 2017-11-10 Jeff Law <law@redhat.com>
821 * vr-values.h (VR_INITIALIZER): Move #define here.
822 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
823 * Makefile.in (OBJS): Add tree-evrp.o
824 * tree-vrp.h (assert_info): Move structure definition here.
825 (set_value_range_to_varying): Prototype.
826 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
827 (infer_value_range, register_edge_assert_for): Likewise.
828 (stmt_interesting_for_vrp): Likewise.
829 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
830 (set_value_range_to_varying): No longer static.
831 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
832 (infer_value_range, register_edge_assert_for): Likewise.
834 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
836 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
837 generate the XXBRD instruction.
839 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
840 (rs6000_builtin_type_compatible): Treat _Float128 and long double
841 as being compatible if -mabi=ieeelongdouble.
842 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
843 to setup float128 built-ins with hardware support.
844 (BU_FLOAT128_HW_2): Likewise.
845 (BU_FLOAT128_HW_3): Likewise.
846 (BU_FLOAT128_HW_VSX_1): Likewise.
847 (BU_FLOAT128_HW_VSX_2): Likewise.
848 (scalar_extract_expq): Change float128 built-in functions to
849 accommodate having both KFmode and TFmode functions. Use the
850 KFmode variant as the default.
851 (scalar_extract_sigq): Likewise.
852 (scalar_test_neg_qp): Likewise.
853 (scalar_insert_exp_q): Likewise.
854 (scalar_insert_exp_qp): Likewise.
855 (scalar_test_data_class_qp): Likewise.
856 (sqrtf128_round_to_odd): Delete processing the round to odd
857 built-in functions as special built-in functions, and define them
858 as float128 built-ins. Use the KFmode variant as the default.
859 (truncf128_round_to_odd): Likewise.
860 (addf128_round_to_odd): Likewise.
861 (subf128_round_to_odd): Likewise.
862 (mulf128_round_to_odd): Likewise.
863 (divf128_round_to_odd): Likewise.
864 (fmaf128_round_to_odd): Likewise.
865 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
866 support for KFmode and TFmode xststdcqp calls.
867 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
868 point, switch the built-in handlers for the get/set float128
869 exponent, get float128 mantissa, float128 test built-ins, and the
870 float128 round to odd built-in functions. Eliminate creating the
871 float128 round to odd built-in functions as special built-ins.
872 (rs6000_init_builtins): Eliminate special creation of the float128
873 round to odd built-in functions.
874 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
875 function insns to support both TFmode and KFmode variants.
876 (xsxsigqp_<mode>): Likewise.
877 (xsiexpqpf_<mode>): Likewise.
878 (xsiexpqp_<mode>): Likewise.
879 (xststdcqp_<mode>): Likewise.
880 (xststdcnegqp_<mode>): Likewise.
881 (xststdcqp_<mode>): Likewise.
883 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
885 * tree-ssa-threadupdate.c (compute_path_counts): Remove
886 unused path_in_freq_ptr parameter.
887 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
889 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
891 * ipa-inline.c (edge_badness): Dump sreal frequency.
892 (compute_inlined_call_time): Match natural implementaiton ...
893 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
894 forgotten division by CGRAPH_FREQ_BASE.
896 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
898 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
899 Solaris 11. Update comment.
900 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
902 * config/sol2.h (STARTFILE_SPEC): Likewise.
903 * configure: Regenerate.
905 2017-11-14 Carl Love <cel@us.ibm.com>
907 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
908 le_ and be_ prefixes to swap* variables. Remove
909 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
912 2017-11-14 Jason Merrill <jason@redhat.com>
914 Support GTY((cache)) on hash_map.
915 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
916 (ggc_cache_remove): Override it instead of ggc_mx.
917 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
918 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
919 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
920 (simple_cache_map_traits): Override maybe_mx.
921 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
922 (hash_map): Friend gt_cleare_cache.
923 (gt_cleare_cache): New.
924 * tree.h (tree_cache_traits): New hash_map traits class.
925 (tree_cache_map): New typedef.
927 2017-11-14 Richard Biener <rguenther@suse.de>
929 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
930 paramter and handling.
931 (cleanup_control_flow_bb): Likewise.
932 (cleanup_control_flow_pre): New helper performing a DFS walk
933 to call cleanup_control_flow_bb in PRE order.
934 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
935 via cleanup_control_flow_pre.
937 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
939 * config/aarch64/aarch64-simd.md
940 (aarch64_simd_bsl<mode>_internal): Remove DImode.
941 (*aarch64_simd_bsl<mode>_alt): Likewise.
942 (aarch64_simd_bsldi_internal): New.
943 (aarch64_simd_bsldi_alt): Likewise.
945 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
947 * tracer.c (better_p): Do not compare frequencies.
948 * reg-stack.c (better_edge): Likewise.
949 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
952 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
954 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
955 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
956 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
957 * lto-streamer-in.c (input_function): Use update_max_bb_count.
958 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
959 * predict.c (maybe_hot_frequency_p): Inline to ...
960 (maybe_hot_count_p): ... here; rewrite to counts.
961 (counts_to_freqs): Rename to ...
962 (update_max_bb_count): ... this one.
963 (expensive_function_p): Use counts.
964 (estimate_bb_frequencies): Update.
965 (rebuild_frequencies): Update.
966 * predict.h (counts_to_freqs): Rename to ...
967 (update_max_bb_count): ... this one.
968 * profile.c (compute_branch_probabilities): Add debug info
969 * tree-inline.c (expand_call_inline): Update debug info.
970 (optimize_inline_calls): Use update_max_bb_count..
971 (tree_function_versioning): Use update_max_bb_count..
972 * value-prof.c (gimple_value_profile_transformations):
973 Do not use update_max_bb_count.
975 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
977 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
978 always use frequencies.
980 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
982 * bb-reorder.c: Remove frequencies from comments.
983 (better_edge_p): Use profile counts.
984 (find_traces): Dump profile counts.
985 (rotate_loop): Use profile counts.
986 (find_traces_1_round): Likewise.
987 (connect_better_edge_p): Use counts instead of probabilities for
989 (copy_bb_p): Drop early check for non-0 frequency.
990 (sanitize_hot_paths): Update comments.
992 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
994 * ipa-split.c (struct split_point): Add count.
995 (consider_split): Do not compute incoming frequency; compute incoming
996 count and store it to split_point.
997 (split_function): Set count of the call to split part correctly.
999 2017-11-13 Carl Love <cel@us.ibm.com>
1001 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1003 2017-11-13 Tom Tromey <tom@tromey.com>
1005 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1007 2017-11-13 Carl Love <cel@us.ibm.com>
1009 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1010 Add support for builtins:
1011 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1012 vector {un,}signed {char,int,short},
1013 vector {un,}signed {char,int,short}) arguments.
1014 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1015 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1016 Add BU_P9V_AV_2 expansions for the builtins.
1017 * config/rs6000/altivec.h (vec_first_match_index,
1018 vec_first_mismatch_index, vec_first_match_or_eos_index,
1019 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1020 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1021 new extern declaration.
1022 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1023 * config/rs6000/vsx.md (first_match_index_<mode>,
1024 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1025 first_mismatch_or_eos_index_<mode>): Add define expand.
1026 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1027 * doc/extend.texi: Update the built-in documenation file for the new
1030 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1032 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1033 into the min/max operations for _Float<N> and _Float<N>X types.
1035 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1038 * dwarf2out.c (do_eh_frame): New static variable.
1039 (dwarf2out_begin_prologue): Set it.
1040 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1042 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1044 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1046 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1048 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1051 * debug.h (dwarf2out_do_eh_frame): Declare.
1052 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1053 (dwarf2out_do_frame): Use it.
1054 (dwarf2out_do_cfi_asm): Likewise.
1055 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1056 (dwarf2out_assembly_start): Likewise.
1057 (dwarf2out_begin_prologue): Fix comment.
1058 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1059 if the target needs either debug or unwind DWARF2 info.
1060 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1061 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1062 -ftrapping-math, -ftrapv and -fwrapv.
1064 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1066 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1067 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1068 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1069 (estimate_edge_size_and_time): Likewise.
1070 (ipa_merge_fn_summary_after_inlining): Likewise.
1071 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1072 (compute_uninlined_call_time): Use sreal_frequency.
1073 (compute_inlined_call_time): Likewise.
1074 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1075 * profile-count.c: Include sreal.h.
1076 (profile_count::to_sreal_scale): New.
1077 * profile-count.h: Forward declare sreal.
1078 (profile_count::to_sreal_scale): Declare.
1080 2017-11-13 Nathan Sidwell <nathan@acm.org>
1082 * diagnostic.c (maybe_line_and_column): New.
1083 (diagnostic_get_location_text): Use it.
1084 (diagnostic_report_current_module): Likewise.
1085 (test_diagnostic_get_location_text): Add tests.
1087 2017-11-13 Luis Machado <luis.machado@linaro.org>
1089 * doc/md.texi (Specifying processor pipeline description): Fix
1090 incorrect latency for the div instruction example.
1092 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1094 PR tree-optimization/78821
1095 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1096 that bit_not_p is the same.
1097 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1098 (split_group): Count precisely bit_not_p bits in each statement.
1099 (invert_op): New function.
1100 (imm_store_chain_info::output_merged_store): Use invert_op to
1101 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1102 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1104 2017-11-13 Martin Liska <mliska@suse.cz>
1106 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1107 (struct source_info): Likewise.
1108 (add_branch_counts): Likewise.
1109 (add_line_counts): Likewise.
1110 (function_summary): Likewise.
1111 (output_intermediate_line): Likewise.
1112 (generate_results): Likewise.
1114 2017-11-13 Martin Liska <mliska@suse.cz>
1116 * gcov.c (struct block_info): Remove typedef for block_t.
1117 (struct line_info): Likewise.
1118 (line_info::has_block): Likewise.
1119 (EXIT_BLOCK): Likewise.
1120 (unblock): Likewise.
1121 (circuit): Likewise.
1122 (get_cycles_count): Likewise.
1123 (process_file): Likewise.
1124 (read_graph_file): Likewise.
1125 (solve_flow_graph): Likewise.
1126 (find_exception_blocks): Likewise.
1127 (add_line_counts): Likewise.
1128 (accumulate_line_info): Likewise.
1129 (output_line_details): Likewise.
1131 2017-11-13 Martin Liska <mliska@suse.cz>
1133 * gcov.c (struct arc_info): Remove typedef for arc_t.
1134 (struct line_info): Likewise.
1135 (add_branch_counts): Likewise.
1136 (output_branch_count): Likewise.
1137 (function_info::~function_info): Likewise.
1138 (circuit): Likewise.
1139 (output_intermediate_line): Likewise.
1140 (read_graph_file): Likewise.
1141 (solve_flow_graph): Likewise.
1142 (find_exception_blocks): Likewise.
1143 (add_line_counts): Likewise.
1144 (accumulate_line_info): Likewise.
1145 (output_line_details): Likewise.
1146 (output_function_details): Likewise.
1148 2017-11-13 Martin Liska <mliska@suse.cz>
1150 * gcov.c (struct function_info): Remove typedef for function_t.
1151 (struct source_info): Likewise.
1152 (source_info::get_functions_at_location): Likewise.
1153 (solve_flow_graph): Likewise.
1154 (find_exception_blocks): Likewise.
1155 (add_line_counts): Likewise.
1156 (output_intermediate_file): Likewise.
1157 (process_file): Likewise.
1158 (generate_results): Likewise.
1159 (release_structures): Likewise.
1160 (read_graph_file): Likewise.
1161 (read_count_file): Likewise.
1162 (accumulate_line_counts): Likewise.
1163 (output_lines): Likewise.
1165 2017-11-13 Martin Liska <mliska@suse.cz>
1167 * gcov.c (function_info::function_info): Remove num_counts
1168 and add vector<gcov_type>.
1169 (function_info::~function_info): Use the vector.
1170 (process_file): Likewise.
1171 (read_graph_file): Likewise.
1172 (read_count_file): Likewise.
1173 (solve_flow_graph): Likewise.
1175 2017-11-13 Martin Liska <mliska@suse.cz>
1177 * gcov.c (function_info::is_artificial): New function.
1178 (process_file): Erase all artificial early.
1179 (generate_results): Skip as all artificial are already
1182 2017-11-13 Martin Liska <mliska@suse.cz>
1184 * gcov.c (read_graph_file): Store to global vector of functions.
1185 (read_count_file): Iterate the vector.
1186 (process_file): Likewise.
1187 (generate_results): Likewise.
1188 (release_structures): Likewise.
1190 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1192 PR tree-optimization/82954
1193 * gimple-ssa-store-merging.c
1194 (imm_store_chain_info::coalesce_immediate_stores): If
1195 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1197 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1199 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1200 Upddate call to ENDIAN_LANE_N.
1201 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1202 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1203 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1204 and use aarch64_endian_lane_rtx.
1205 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1207 2017-11-12 Tom de Vries <tom@codesourcery.com>
1209 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1211 2017-11-12 Tom de Vries <tom@codesourcery.com>
1213 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1215 2017-11-12 Tom de Vries <tom@codesourcery.com>
1217 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1219 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1220 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1221 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1223 2017-11-11 Martin Sebor <msebor@redhat.com>
1226 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1229 * prefic.c (translate_name): Replace strncpy with memcpy to
1230 avoid -Wstringop-truncation.
1232 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1234 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1236 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1238 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1240 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1242 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1245 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1247 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1248 enable generating XXBRH if the value is in a vector register.
1249 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1250 value is in a vector register.
1251 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1252 register to register bswap64's instead of doing the GPR sequence
1253 used on previous machines.
1254 (bswapdi2_xxbrd): New insn.
1255 (bswapdi2_reg): Disallow on ISA 3.0.
1256 (register to register bswap64 splitter): Do not split the insn on
1257 ISA 3.0 systems that use XXBRD.
1259 2017-11-10 Martin Sebor <msebor@redhat.com>
1262 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1264 (find_subframework_file): Same.
1266 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1268 * auto-profile.c (afdo_indirect_call): Drop frequency.
1269 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1270 (cgraph_node::create_edge): Drop frequency argument.
1271 (cgraph_node::create_indirect_edge): Drop frequency argument.
1272 (cgraph_edge::make_speculative): Drop frequency arguments.
1273 (cgraph_edge::resolve_speculation): Do not update frequencies
1274 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1275 (cgraph_node::dump): Check consistency in IPA mode.
1276 (cgraph_edge::maybe_hot_p): Use IPA counter.
1277 (cgraph_edge::verify_count_and_frequency): Rename to ...
1278 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1279 (cgraph_node::verify_node): Update.
1280 * cgraph.h (struct cgraph_edge): Drop frequency.
1281 (cgraph_edge::frequency): New function.
1282 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1284 (cgraph_edge::rebuild_edges): Likewise.
1285 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1286 (duplicate_thunk_for_node): Do not pass frequency.
1287 (cgraph_node::create_clone): Scale only counts.
1288 (cgraph_node::create_virtual_clone): Do not pass frequency.
1289 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1290 (cgraph_node::create_version_clone): Do not pass frequency.
1291 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1292 (cgraph_node::expand_thunk): Do not pass frequency.
1293 (cgraph_node::create_wrapper): Do not pass frequency.
1294 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1296 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1297 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1298 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1299 (gather_caller_stats): Use frequency function.
1300 (ipcp_cloning_candidate_p): Use frequency function.
1301 (ipcp_propagate_stage): Use frequency function.
1302 (get_info_about_necessary_edges): Use frequency function.
1303 (update_profiling_info): Update only IPA profile.
1304 (update_specialized_profile): Use frequency functoin.
1305 (perhaps_add_new_callers): Update only IPA profile.
1306 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1307 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1308 (dump_ipa_call_summary): Use frequency function.
1309 (estimate_edge_size_and_time): Use frequency function.
1310 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1311 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1312 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1313 (update_noncloned_counts): ... ths one; scale counts only.
1314 (clone_inlined_nodes): Do not scale frequency.
1315 (inline_call): Do not pass frequency.
1316 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1317 (compute_inlined_call_time): Use IPA profile.
1318 (want_inline_small_function_p): Use IPA profile.
1319 (want_inline_self_recursive_call_p): Use IPA profile.
1320 (edge_badness): Use IPA profile.
1321 (lookup_recursive_calls): Use IPA profile.
1322 (recursive_inlining): Do not pass frequency.
1323 (resolve_noninline_speculation): Do not update frequency.
1324 (inline_small_functions): Collect max of IPA profile.
1325 (dump_overall_stats): Dump IPA porfile.
1326 (dump_inline_stats): Dump IPA porfile.
1327 (ipa_inline): Collect IPA stats.
1328 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1329 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1330 (ipa_propagate_frequency): Use frequency function.
1331 (ipa_profile): Cleanup.
1332 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1333 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1334 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1335 (input_node): Do not stream frequency.
1336 (input_edge): Do not stream frequency.
1337 (merge_profile_summaries): Scale only IPA profiles.
1338 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1339 * predict.c (drop_profile): Do not recompute frequency.
1340 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1341 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1342 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1343 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1344 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1345 * tree-inline.c (copy_bb): Do not scale frequency.
1346 (expand_call_inline): Do not scale frequency.
1347 (tree_function_versioning): Do not scale frequency.
1348 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1350 2017-11-10 Julia Koval <julia.koval@intel.com>
1352 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1353 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1354 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1355 (_mm256_maskz_gf2p8affine_epi64_epi8)
1356 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1357 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1358 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1359 (__builtin_ia32_vgf2p8affineqb_v32qi)
1360 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1361 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1363 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1367 (arm_option_override): Refactor.
1368 (arm_option_reconfigure_globals): New.
1369 (arm_options_perform_arch_sanity_checks): New.
1370 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1372 (arm_options_perform_arch_sanity_checks): Likewise
1374 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1376 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1377 (power9-qpmul): New.
1378 * rs6000/rs6000.md ("type" attr): Add qmul.
1379 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1380 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1381 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1383 2017-11-10 Martin Sebor <msebor@redhat.com>
1386 * builtins.c (compute_objsize): Handle arrays that
1387 compute_builtin_object_size likes to fail for. Make extern.
1388 * builtins.h (compute_objsize): Declare.
1389 (check_strncpy_sizes): New function.
1390 (expand_builtin_strncpy): Call check_strncpy_sizes.
1391 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1392 -Wstringop-truncation.
1393 (gimple_fold_builtin_strncat): Same.
1394 * gimple.c (gimple_build_call_from_tree): Set call location.
1395 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1396 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1397 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1398 (handle_builtin_strlen): Use strlen_to_stridx.
1399 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1401 Use strlen_to_stridx.
1402 (pass_strlen::execute): Release strlen_to_stridx.
1403 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1404 (-Wstringop-truncation): Document new option.
1406 2017-11-10 Martin Liska <mliska@suse.cz>
1408 PR gcov-profile/82702
1409 * gcov.c (main): Handle intermediate files in a different
1411 (get_gcov_intermediate_filename): New function.
1412 (output_gcov_file): Remove support of intermediate files.
1413 (generate_results): Allocate intermediate file.
1414 (release_structures): Clean-up properly fn_end.
1415 (output_intermediate_file): Start iterating with line 1.
1417 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1419 PR tree-optimization/82929
1420 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1421 ops_swapped_p non-static data member.
1422 (store_immediate_info::store_immediate_info): Clear it.
1423 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1424 ops set ops_swapped_p.
1425 (count_multiple_uses): Handle ops_swapped_p.
1427 2017-11-10 Martin Liska <mliska@suse.cz>
1429 * coverage.c (coverage_init): Stream information about
1430 support of has_unexecuted_blocks.
1431 * doc/gcov.texi: Document that.
1432 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1433 * gcov.c (read_graph_file): Likewise.
1434 (output_line_beginning): Fix a small issue with
1437 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1439 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1440 reference of trivial component.
1442 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1445 * gimple-ssa-store-merging.c
1446 (pass_store_merging::terminate_all_aliasing_chains): For
1447 gimple_store_p stmts also call refs_output_dependent_p.
1449 PR rtl-optimization/82913
1450 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1453 2017-11-09 Jeff Law <law@redhat.com>
1455 * vr-values.h: New file with vr_values class.
1456 * tree-vrp.c: Include vr-values.h
1457 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1458 data objects into the vr_values class.
1459 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1460 (get_value_range): Make it a member function within vr_values class.
1461 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1462 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1463 (op_with_constant_singleton_value_range): Likewise.
1464 (extract_range_for_var_from_comparison_expr): Likewise.
1465 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1466 (extract_range_from_binary_expr): Likewise.
1467 (extract_range_from_unary_expr): Likewise.
1468 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1469 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1470 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1471 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1472 (compare_name_with_value, compare_names): Likewise.
1473 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1474 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1475 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1476 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1477 (extract_range_from_phi_node): Likewise.
1478 (simplify_truth_ops_using_ranges): Likewise.
1479 (simplify_div_or_mod_using_ranges): Likewise.
1480 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1481 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1482 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1483 (simplify_float_conversion_using_ranges): Likewise.
1484 (simplify_internal_call_using_ranges): Likewise.
1485 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1486 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1487 poitner via x_vr_values for calls into gimple folder.
1488 (vrp_initialize_lattice): Make this the vr_values ctor.
1489 (vrp_free_lattice): Make this the vr_values dtor.
1490 (set_vr_value): New function.
1491 (class vrp_prop): Add vr_values data member. Add various member
1492 functions as well as member functions that delegate to vr_values.
1493 (check_array_ref): Make a member function within vrp_prop class.
1494 (search_for_addr_array, vrp_initialize): Likewise.
1495 (vrp_finalize): Likewise. Revamp to avoid direct access to
1496 vr_value, values_propagated, etc.
1497 (check_array_bounds): Extract vrp_prop class instance pointer from
1498 walk info structure. Use it to call member functions.
1499 (check_all_array_refs): Make a member function within vrp_prop class.
1500 Smuggle class instance pointer via walk info structure.
1501 (x_vr_values): New local static.
1502 (vrp_valueize): Use x_vr_values to get class instance.
1503 (vr_valueize_1): Likewise.
1504 (class vrp_folder): Add vr_values data member. Add various member
1505 functions as well as member functions that delegate to vr_values.
1506 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1507 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1508 class instance from vr_values. Use it to call member functions.
1509 (vrp_dom_walker): Add vr_values data member.
1510 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1511 instance via x_vr_values.
1512 (identify_jump_threads): Accept vr_values as argument. Store
1513 it into the walker structure.
1514 (evrp_dom_walker): Add vr_values class data member. Add various
1516 (evrp_dom_walker::try_find_new_range): Use vr_values data
1517 member to access the memory allocator.
1518 (evrp_dom_walker::before_dom_children): Store vr_values class
1519 instance into the vrp_folder class.
1520 (evrp_dom_walker::push_value_range): Rework to avoid direct
1521 access to num_vr_values and vr_value.
1522 (evrp_dom_walker::pop_value_range): Likewise.
1523 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1524 Use vr_values to get to dump_all_value_ranges member function.
1525 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1526 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1527 Pass vr_values class instance down to identify_jump_threads.
1528 Remove call to vrp_free_lattice.
1529 (debug_all_value_ranges): Remove.
1531 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1532 (vrp_folder): Likewise.
1534 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1535 Get it from the existing bitmap instead.
1536 (vrp_intersect_ranges_1): Likewise.
1538 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1540 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1542 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1543 set bit_not_p to it.
1544 (imm_store_chain_info::coalesce_immediate_stores): Break group
1545 if bit_not_p is different.
1546 (count_multiple_uses, split_group,
1547 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1548 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1549 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1550 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1551 to store_immediate_info ctor.
1553 2017-11-09 Jim Wilson <jimw@sifive.com>
1555 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1556 (scan_prog_file): Likewise.
1558 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1560 * bb-reorder.c (max_entry_frequency): Remove.
1561 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1562 connect_traces, push_to_next_round_p): Remove prototypes.
1563 (find_traces_1_round): Use counts only.
1564 (push_to_next_round_p): Likewise.
1565 (find_traces): Likewise.
1566 (rotate_loop): Likewise.
1567 (find_traces_1_round): Likewise.
1568 (connect_traces): Likewise.
1569 (edge_order): Likewise.
1571 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1573 * config/arm/arm.c (output_return_instruction): Add comments to
1574 indicate requirement for cmse_nonsecure_entry return to account
1575 for the size of clearing instruction output here.
1576 (thumb_exit): Likewise.
1577 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1578 return in hardfloat mode.
1580 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1582 * config/rs6000/rs6000.c (machine_function): Add a bool,
1583 "toc_is_wrapped_separately".
1584 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1585 if it wasn't explicitly set or unset, we are optimizing for speed, and
1586 doing separate shrink-wrapping.
1587 (rs6000_get_separate_components): Enable the TOC component if
1588 saving the TOC register in the prologue.
1589 (rs6000_components_for_bb): Handle the TOC component.
1590 (rs6000_emit_prologue_components): Store the TOC register where needed.
1591 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1592 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1594 2017-11-09 Martin Jambor <mjambor@suse.cz>
1596 * ipa-param-manipulation.c: New file.
1597 * ipa-param-manipulation.h: Likewise.
1598 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1599 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1600 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1601 (ipa_parm_adjustment): Likewise.
1602 (ipa_parm_adjustment_vec): Likewise.
1603 (ipa_get_vector_of_formal_parms): Moved declaration to
1604 ipa-param-manipulation.h.
1605 (ipa_get_vector_of_formal_parm_types): Likewise.
1606 (ipa_modify_formal_parameters): Likewise.
1607 (ipa_modify_call_arguments): Likewise.
1608 (ipa_combine_adjustments): Likewise.
1609 (ipa_dump_param_adjustments): Likewise.
1610 (ipa_modify_expr): Likewise.
1611 (ipa_get_adjustment_candidate): Likewise.
1612 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1613 ipa-param-manipulation.c.
1614 (ipa_get_vector_of_formal_parm_types): Likewise.
1615 (ipa_modify_formal_parameters): Likewise.
1616 (ipa_modify_call_arguments): Likewise.
1617 (ipa_modify_expr): Likewise.
1618 (get_ssa_base_param): Likewise.
1619 (ipa_get_adjustment_candidate): Likewise.
1620 (index_in_adjustments_multiple_times_p): Likewise.
1621 (ipa_combine_adjustments): Likewise.
1622 (ipa_dump_param_adjustments): Likewise.
1623 * tree-sra.c: Also include ipa-param-manipulation.h
1624 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1627 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1628 Alan Hayward <alan.hayward@arm.com>
1629 David Sherwood <david.sherwood@arm.com>
1631 * doc/sourcebuild.texi (vect_masked_store): Document.
1633 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1634 Alan Hayward <alan.hayward@arm.com>
1635 David Sherwood <david.sherwood@arm.com>
1637 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1639 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1640 Alan Hayward <alan.hayward@arm.com>
1641 David Sherwood <david.sherwood@arm.com>
1643 * doc/sourcebuild.texi (vect_variable_length): Document.
1645 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1646 Alan Hayward <alan.hayward@arm.com>
1647 David Sherwood <david.sherwood@arm.com>
1649 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1651 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1652 Alan Hayward <alan.hayward@arm.com>
1653 David Sherwood <david.sherwood@arm.com>
1655 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1657 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1658 Alan Hayward <alan.hayward@arm.com>
1659 David Sherwood <david.sherwood@arm.com>
1661 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1662 previously undocumented selectors.
1663 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1665 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1667 * doc/rtl.texi (const_vector): Say that elements can be
1668 const_wide_ints too.
1669 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1670 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1671 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1672 * optabs.c (expand_vector_broadcast): Likewise.
1674 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1675 Alan Hayward <alan.hayward@arm.com>
1676 David Sherwood <david.sherwood@arm.com>
1678 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1679 scaled index even if the unscaled address was invalid.
1680 Don't increase the complexity of using a scale in that case.
1682 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1683 Alan Hayward <alan.hayward@arm.com>
1684 David Sherwood <david.sherwood@arm.com>
1686 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1687 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1688 UNITS_PER_WORD bytes.
1689 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1690 based on REGMODE_NATURAL_SIZE of the inner mode.
1691 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1692 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1693 * expmed.c (lowpart_bit_field_p): Divide the value up into
1694 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1695 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1696 whether something is likely to occupy more than one register.
1698 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1701 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1702 profile_count::adjust_for_ipa_scaling.
1703 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1704 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1706 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1708 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1710 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1711 (split_group): Add total_orig and total_new arguments, estimate the
1712 number of statements related to the store group without store merging
1713 and with store merging.
1714 (imm_store_chain_info::output_merged_store): Adjust split_group
1715 callers, punt if estimated number of statements with store merging
1716 is not smaller than estimated number of statements without it.
1718 (handled_load): Remove has_single_use checks.
1719 (pass_store_merging::process_store): Likewise.
1721 2017-11-09 Richard Biener <rguenther@suse.de>
1723 PR tree-optimization/82902
1724 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1726 2017-11-09 Martin Liska <mliska@suse.cz>
1729 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1732 2017-11-09 Martin Liska <mliska@suse.cz>
1734 PR tree-optimization/82669
1735 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1737 2017-11-09 Martin Liska <mliska@suse.cz>
1739 PR gcov-profile/48463
1740 * coverage.c (coverage_begin_function): Output also end locus
1741 of a function and information whether the function is
1743 * gcov-dump.c (tag_function): Parse and print the information.
1744 * gcov.c (INCLUDE_MAP): Add include.
1745 (INCLUDE_SET): Likewise.
1746 (struct line_info): Move earlier in the source file because
1747 of vector<line_info> in function_info structure.
1748 (line_info::line_info): Likewise.
1749 (line_info::has_block): Likewise.
1750 (struct source_info): Add new member index.
1751 (source_info::get_functions_at_location): New function.
1752 (function_info::group_line_p): New function.
1753 (output_intermediate_line): New function.
1754 (output_intermediate_file): Use the mentioned function.
1755 (struct function_start): New.
1756 (struct function_start_pair_hash): Likewise.
1757 (process_file): Add code that identifies group functions.
1758 Assign lines either to global or function scope.
1759 (generate_results): Skip artificial functions.
1760 (find_source): Assign index for each source file.
1761 (read_graph_file): Read new flag artificial and end_line.
1762 (add_line_counts): Assign it either to global of function scope.
1763 (accumulate_line_counts): Isolate core of the function to
1764 accumulate_line_info and call it for both function and global
1766 (accumulate_line_info): New function.
1767 (output_line_beginning): Fix GNU coding style.
1768 (print_source_line): New function.
1769 (output_line_details): Likewise.
1770 (output_function_details): Likewise.
1771 (output_lines): Iterate both source (global) scope and function
1773 (struct function_line_start_cmp): New class.
1774 * doc/gcov.texi: Reflect changes in documentation.
1776 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1779 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1780 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1781 and similarly for not instead of neg.
1783 2017-11-08 Andi Kleen <ak@linux.intel.com>
1785 * config/i386/i386.opt: Add -mforce-indirect-call.
1786 * config/i386/predicates.md: Check for flag_force_indirect_call.
1787 * doc/invoke.texi: Document -mforce-indirect-call
1789 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1791 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1793 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1794 config/riscv/riscv.c (predict.h): New include.
1795 (riscv_slow_unaligned_access_p): No longer static.
1796 (riscv_block_move_straight): Add require.
1797 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1799 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1802 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1803 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1804 nonimmediate_operand predicate for operand 1 instead of
1807 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1809 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1811 * config/aarch64/constraints.md (Uml): New constraint.
1812 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1815 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1817 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1818 of two vec_duplicates into a vec_concat.
1820 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1822 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1823 Simplify vec_merge of vec_duplicate and vec_concat.
1824 * config/aarch64/constraints.md (Utq): New constraint.
1825 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1828 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1830 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1831 Simplify vec_merge of vec_duplicate and const_vector.
1832 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1834 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1835 mode iterator. Update predicate on operand 1 to
1836 handle non-const_vec constants. Delete constraints.
1837 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1839 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1841 PR tree-optimization/78821
1842 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1844 (store_operand_info::store_operand_info): Initialize it to false.
1845 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1846 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1847 store in the group, and if chain_info is non-NULL, to ignore altogether
1849 (compatible_load_p): Fail if bit_not_p does not match.
1850 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1851 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1852 (pass_store_merging::process_store): Adjust
1853 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1854 call terminate_all_aliasing_chains newly when adding a store into
1855 a chain with non-NULL chain_info.
1857 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1859 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1861 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1863 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1865 (aarch64_layout_frame): Initialise emit_frame_chain.
1866 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1867 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1869 2017-11-08 Martin Liska <mliska@suse.cz>
1871 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1872 of gimple_call_internal_p.
1874 2017-11-07 Tom de Vries <tom@codesourcery.com>
1876 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1878 2017-11-07 Tom de Vries <tom@codesourcery.com>
1880 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1883 2017-11-08 Martin Liska <mliska@suse.cz>
1886 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1888 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1890 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1892 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1895 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1896 (*cmp<mode>_ccz_1): New insn with $k alternative.
1899 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1900 RTX_COMM_COMPARE as commutative as well.
1901 (ix86_binary_operator_ok): Formatting fix.
1902 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1903 *<code><mode>3<mask_name><round_saeonly_name>,
1904 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1905 *<s>mul<mode>3_highpart<mask_name>,
1906 *vec_widen_umult_even_v16si<mask_name>,
1907 *vec_widen_umult_even_v8si<mask_name>,
1908 *vec_widen_umult_even_v4si<mask_name>,
1909 *vec_widen_smult_even_v16si<mask_name>,
1910 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1911 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1912 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1913 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1914 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1915 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1916 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1917 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1918 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1919 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1920 ix86_binary_operator_ok. Formatting fixes.
1921 (*<plusminus_insn><mode>3<mask_name><round_name>,
1922 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1925 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1927 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1928 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1929 eq and ne if TARGET_ISEL.
1930 (cmp): New code_iterator.
1931 (UNS, UNSU_, UNSIK): New code_attrs.
1932 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1933 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1935 ("eq<mode>3"): ... this.
1936 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1938 ("ne<mode>3"): ... this.
1940 2017-11-07 Julia Koval <julia.koval@intel.com>
1943 * common/config/i386/i386-common.c
1944 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1945 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1946 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1947 * config/i386/i386.opt: Ditto.
1948 * config/i386/i386.c (ix86_target_string): Ditto.
1949 (ix86_option_override_internal): Ditto.
1950 (ix86_init_mpx_builtins): Move MPX to args2.
1951 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1952 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1953 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1954 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1955 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1956 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1957 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1959 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1962 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1963 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1964 (zero-extendsidi peephole2): Remove peephole.
1966 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1969 * stor-layout.c: Include attribs.h.
1970 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1971 explicit lookup of "aligned" attribute.
1973 2017-11-07 Andrew Waterman <andrew@sifive.com>
1975 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1976 (riscv_expand_block_move): Likewise.
1977 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1979 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1980 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1981 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1982 (riscv_adjust_block_mem): Likewise.
1983 (riscv_block_move_loop): Likewise.
1984 (riscv_expand_block_move): Likewise.
1985 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1987 2017-11-07 Michael Clark <michaeljclark@mac.com>
1989 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1990 (MUSL_DYNAMIC_LINKER): Likewise.
1992 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1994 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1995 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1998 2017-11-07 Richard Biener <rguenther@suse.de>
2000 * match.pd: Fix build.
2002 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2003 Jackson Woodruff <jackson.woodruff@arm.com>
2005 PR tree-optimization/71026
2006 * match.pd: Canonicalize negate in division.
2008 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2011 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2013 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2015 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2016 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2018 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2020 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2021 non-scalar integral types.
2022 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2023 (-(A-B), -(~A)): New transformations.
2025 2017-11-07 Tom de Vries <tom@codesourcery.com>
2027 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2028 semicolon after "do {} while (0)".
2029 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2030 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2031 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2032 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2033 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2035 2017-11-07 Tom de Vries <tom@codesourcery.com>
2037 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2038 after "do {} while (0)".
2039 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2040 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2041 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2042 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2043 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2045 2017-11-07 Tom de Vries <tom@codesourcery.com>
2048 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2050 (arm_rtx_costs_internal): Add missing semicolon after
2051 HANDLE_NARROW_SHIFT_ARITH call.
2053 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2055 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2056 disable isel if it was not set explicitly.
2058 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2060 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
2062 (add_type_attribute) likewise.
2064 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2066 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2067 of struct ix86_frame.
2068 (ix86_initial_elimination_offset): Likewise.
2069 (ix86_expand_split_stack_prologue): Likewise.
2071 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2073 * tree-vrp.h (enum value_range_type): Update stale comment.
2075 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2076 Alan Hayward <alan.hayward@arm.com>
2077 David Sherwood <david.sherwood@arm.com>
2079 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2080 (aarch64_expand_vec_perm_const): Take the number of units too.
2081 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2082 (aarch64_expand_vec_perm_const): Likewise.
2083 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2084 (vec_perm<mode>): Update accordingly.
2086 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2087 Alan Hayward <alan.hayward@arm.com>
2088 David Sherwood <david.sherwood@arm.com>
2090 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2091 Take the number of units too.
2092 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2093 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2094 but check for a vector mode before rather than after the call.
2095 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2096 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2097 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2098 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2099 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2100 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2101 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2102 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2103 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2104 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2105 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2106 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2107 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2108 (aarch64_sqdmull2_n<mode>): Update accordingly.
2110 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2111 Alan Hayward <alan.hayward@arm.com>
2112 David Sherwood <david.sherwood@arm.com>
2114 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2115 the number of units too.
2116 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2117 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2118 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2119 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2120 (vec_store_lanesxi<mode>): Update accordingly.
2122 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2123 Alan Hayward <alan.hayward@arm.com>
2124 David Sherwood <david.sherwood@arm.com>
2126 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2127 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2128 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2129 of units rather than the mode.
2130 * config/aarch64/iterators.md (nunits): New mode attribute.
2131 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2132 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2133 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2134 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2135 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2136 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2137 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2138 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2139 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2140 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2141 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2142 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2143 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2144 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2145 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2146 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2147 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2148 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2149 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2150 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2151 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2152 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2153 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2154 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2155 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2156 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2157 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2158 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2159 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2160 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2161 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2162 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2163 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2164 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2165 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2166 (aarch64_simd_vec_setv2di): Likewise.
2168 2017-11-06 Carl Love <cel@us.ibm.com>
2170 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2172 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2173 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2175 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2177 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2178 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2179 Add power 8 macro expansions.
2180 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2181 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2182 power 8 instructions. (VSX_XXBR): Add iterator.
2184 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2186 * config/arm/arm.md (predicable_short_it): Change default to "no",
2187 improve documentation, remove uses that are identical to the default.
2188 (enabled_for_depr_it): Rename to enabled_for_short_it.
2189 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2190 Remove default uses.
2191 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2192 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2193 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2194 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2196 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2199 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2200 float128 helper macros, which are no longer used after deleting
2201 the old 'q' built-in functions, and moving the round to odd
2202 built-in functions to being special built-in functions.
2203 (BU_FLOAT128_2): Likewise.
2204 (BU_FLOAT128_1_HW): Likewise.
2205 (BU_FLOAT128_2_HW): Likewise.
2206 (BU_FLOAT128_3_HW): Likewise.
2207 (FABSQ): Delete old 'q' built-in functions.
2208 (COPYSIGNQ): Likewise.
2209 (SQRTF128_ODD): Move round to odd built-in functions to be
2210 special built-in functions, so that we can handle
2211 -mabi=ieeelongdouble.
2212 (TRUNCF128_ODD): Likewise.
2213 (ADDF128_ODD): Likewise.
2214 (SUBF128_ODD): Likewise.
2215 (MULF128_ODD): Likewise.
2216 (DIVF128_ODD): Likewise.
2217 (FMAF128_ODD): Likewise.
2218 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2219 built-in names to 'f128'.
2220 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2221 old 'q' built-in functions, as the machine independent code for
2222 'f128' built-in functions handles this.
2223 (rs6000_expand_builtin): Add expansion for float128 round to odd
2224 functions, keying off on -mabi=ieeelongdouble of whether to use
2225 the KFmode or TFmode variant.
2226 (rs6000_init_builtins): Initialize the _Float128 round to odd
2228 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2229 _Float128 'q' built-in functions are now mapped into the new
2230 'f128' built-in functions.
2232 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2234 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2235 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2237 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2240 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2241 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2242 * toplev.c: Include "ipa-fnsummary.h".
2243 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2245 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2247 PR tree-optimization/82838
2248 * gimple-ssa-store-merging.c
2249 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2250 on a separate gimple_seq which is then appended to seq.
2252 2017-11-06 Jeff Law <law@redhat.com>
2255 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2256 (get_probe_interval): New functions.
2257 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2258 (ix86_adjust_stack_and_probe): Likewise.
2259 (output_adjust_stack_and_probe): Likewise.
2260 (ix86_emit_probe_stack_range): Likewise.
2261 (ix86_expand_prologue): Likewise.
2263 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2265 PR tree-optimization/82816
2266 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2267 if the modes of the two types are the same.
2268 (convert_plusminus_to_widen): Likewise.
2270 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2272 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2274 (usadv16qi): New define_expand.
2275 (usadv8hi): New define_expand.
2277 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2280 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2282 (inline_transform): Likewise.
2283 * predict.c (counts_to_freqs): Remove useless conditional.
2284 * profile-count.h (profile_count::apply_scale): Move sanity check.
2285 * tree-inline.c (copy_bb): Always scale.
2286 (copy_cfg_body): Likewise.
2288 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2291 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2293 (*cmp_ite0): Add enabled_for_depr_it attribute.
2294 (*cmp_ite1): Likewise.
2296 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2298 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2301 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2303 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2304 return the value by pointer.
2305 (extract_range_from_multiplicative_op_1): Update accordingly.
2306 Return as soon as an operation fails.
2308 2017-11-05 Tom de Vries <tom@codesourcery.com>
2311 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2312 (DEF_SANITIZER_BUILTIN): ... here.
2313 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2314 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2316 2017-11-05 Tom de Vries <tom@codesourcery.com>
2319 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2321 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2322 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2323 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2325 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2326 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2327 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2329 2017-11-05 Tom de Vries <tom@codesourcery.com>
2332 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2335 2017-11-04 Michael Clark <michaeljclark@mac.com>
2337 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2338 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2340 (*addsi3_extended): Likewise.
2341 (*addsi3_extended2): Likewise.
2342 (<optab>si3): Likewise.
2343 (<optab>di3): Likewise.
2344 (<optab><mode>3): Likewise.
2345 (<*optabe>si3_internal): Likewise.
2346 (zero_extendqi<SUPERQI:mode>2): Likewise.
2347 (*add<mode>hi3): Likewise.
2348 (*xor<mode>hi3): Likewise.
2349 (<optab>di3): Likewise.
2350 (*<optab>si3_extend): Likewise.
2351 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2352 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2353 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2355 2017-11-04 Andrew Waterman <andrew@sifive.com>
2357 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2358 TARGET_STRICT_ALIGN based upon -mtune argument.
2360 2017-11-04 Andrew Waterman <andrew@sifive.com>
2362 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2364 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2366 config/i386/i386.c (choose_basereg): Use optional scratch
2367 register and add assertion.
2368 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2369 needed, and don't allocate stack.
2370 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2371 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2372 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2374 2017-11-03 Jeff Law <law@redhat.com>
2376 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2377 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2378 to probe at the start of a noreturn function.
2380 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2382 PR tree-optimization/78821
2383 * gimple-ssa-store-merging.c: Update the file comment.
2384 (MAX_STORE_ALIAS_CHECKS): Define.
2385 (struct store_operand_info): New type.
2386 (store_operand_info::store_operand_info): New constructor.
2387 (struct store_immediate_info): Add rhs_code and ops data members.
2388 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2389 and op1r arguments to the ctor, initialize corresponding data members.
2390 (struct merged_store_group): Add load_align_base and load_align
2392 (merged_store_group::merged_store_group): Initialize them.
2393 (merged_store_group::do_merge): Update them.
2394 (merged_store_group::apply_stores): Pick the constant for
2395 encode_tree_to_bitpos from one of the two operands, or skip
2396 encode_tree_to_bitpos if neither operand is a constant.
2397 (class pass_store_merging): Add process_store method decl. Remove
2398 bool argument from terminate_all_aliasing_chains method decl.
2399 (pass_store_merging::terminate_all_aliasing_chains): Remove
2400 var_offset_p argument and corresponding handling.
2401 (stmts_may_clobber_ref_p): New function.
2402 (compatible_load_p): New function.
2403 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2404 if there is overlap and rhs_code is not INTEGER_CST. For
2405 non-overlapping stores terminate group if rhs is not mergeable.
2406 (get_alias_type_for_stmts): Change first argument from
2407 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2408 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2409 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2411 (get_location_for_stmts): Change first argument from
2412 auto_vec<gimple *> & to vec<gimple *> &.
2413 (struct split_store): Remove orig_stmts data member, add orig_stores.
2414 (split_store::split_store): Create orig_stores rather than orig_stmts.
2415 (find_constituent_stmts): Renamed to ...
2416 (find_constituent_stores): ... this. Change second argument from
2417 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2418 to info structures rather than the statements.
2419 (split_group): Rename ALLOW_UNALIGNED argument to
2420 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2421 it. Adjust find_constituent_stores caller.
2422 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2423 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2424 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2425 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2426 (mem_valid_for_store_merging): New function.
2427 (handled_load): New function.
2428 (pass_store_merging::process_store): New method.
2429 (pass_store_merging::execute): Use process_store method. Adjust
2430 terminate_all_aliasing_chains caller.
2432 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2434 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2435 Return true for more constants, symbols and label references.
2436 (aarch64_valid_floating_const): Remove unused function.
2438 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2441 * config/aarch64/aarch64.c (aarch64_layout_frame):
2442 Undo forcing of LR at bottom of frame.
2444 2017-11-03 Jeff Law <law@redhat.com>
2447 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2448 for int_registers_saved.
2450 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2451 extracted from tree-ssa-dom.c.
2452 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2453 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2454 (record_equivalences_from_incoming_edge): Add additional argument
2455 to single_pred_edge_ignoring_loop_edges call.
2456 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2457 (uncprop_dom_walker::before_dom_children): Add additional argument
2458 to single_pred_edge_ignoring_loop_edges call.
2459 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2460 single_pred_edge_ignoring_loop_edges rather than open coding.
2461 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2463 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2465 * match.pd (-(-A)): Rewrite.
2467 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2469 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2470 (rs6000_emit_int_cmove): New declaration.
2471 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2472 (rs6000_emit_sISEL): Delete.
2473 (rs6000_emit_int_cmove): Make non-static.
2474 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2475 instead of rs6000_emit_sISEL.
2477 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2479 * asan.c (create_cond_insert_point): Maintain profile.
2480 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2482 * basic-block.h (struct basic_block_def): Remove frequency.
2483 (EDGE_FREQUENCY): Use to_frequency
2484 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2486 (find_traces): Update to use to_frequency.
2487 (find_traces_1_round): Likewise; use only IPA counts.
2488 (bb_to_key): Likewise.
2489 (connect_traces): Use IPA counts only.
2490 (copy_bb_p): Update to use to_frequency.
2491 (fix_up_crossing_landing_pad): Likewise.
2492 (sanitize_hot_paths): Likewise.
2493 * bt-load.c (basic_block_freq): Likewise.
2494 * cfg.c (init_flow): Set count_max to uninitialized.
2495 (check_bb_profile): Remove frequencies; check counts.
2496 (dump_bb_info): Do not dump frequencies.
2497 (update_bb_profile_for_threading): Update counts only.
2498 (scale_bbs_frequencies_int): Likewise.
2499 (MAX_SAFE_MULTIPLIER): Remove.
2500 (scale_bbs_frequencies_gcov_type): Update counts only.
2501 (scale_bbs_frequencies_profile_count): Update counts only.
2502 (scale_bbs_frequencies): Update counts only.
2503 * cfg.h (struct control_flow_graph): Add count-max.
2504 (update_bb_profile_for_threading): Update prototype.
2505 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2506 (find_many_sub_basic_blocks): Likewise.
2507 * cfgcleanup.c (try_forward_edges): Likewise.
2508 (try_crossjump_to_edge): Likewise.
2509 * cfgexpand.c (expand_gimple_cond): Likewise.
2510 (expand_gimple_tailcall): Likewise.
2511 (construct_init_block): Likewise.
2512 (construct_exit_block): Likewise.
2513 * cfghooks.c (verify_flow_info): Check consistency of counts.
2514 (dump_bb_for_graph): Do not dump frequencies.
2515 (split_block_1): Do not update frequencies.
2516 (split_edge): Do not update frequencies.
2517 (make_forwarder_block): Do not update frequencies.
2518 (duplicate_block): Do not update frequencies.
2519 (account_profile_record): Do not update frequencies.
2520 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2521 for global heuristics.
2522 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2523 (expected_loop_iterations_unbounded): Use counts only.
2524 * cfgloopmanip.c (scale_loop_profile): Simplify.
2525 (create_empty_loop_on_edge): Simplify
2527 (duplicate_loop_to_header_edge): Simplify
2528 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2529 (update_br_prob_note): Take care of removing note when profile
2531 (relink_block_chain): Do not dump frequency.
2532 (rtl_account_profile_record): Use to_frequency.
2533 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2534 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2535 (cgraph_update_edges_for_call_stmt_node): Likewise.
2536 (cgraph_edge::verify_count_and_frequency): Update.
2537 (cgraph_node::verify_node): Temporarily disable frequency verification.
2538 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2539 to_cgraph_frequency.
2540 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2541 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2543 (cgraph_node::expand_thunk): Update profile.
2544 * except.c (dw2_build_landing_pads): Do not update frequency.
2545 * final.c (compute_alignments): Use to_frequency.
2546 (dump_basic_block_info): Do not dump frequency.
2547 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2548 (dump_gimple_bb_header): Do not dump frequency.
2549 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2551 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2552 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2553 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2554 (init_before_recovery): Do not update frequency.
2555 (sched_create_recovery_edges): Do not update frequency.
2556 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2557 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2558 (ipa_cp_c_finalize): Set max_count to uninitialized.
2559 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2560 (param_change_prob): Use counts.
2561 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2563 * ipa-split.c (consider_split): Use to_frequency.
2564 (split_function): Use to_frequency.
2565 * ira-build.c (loop_compare_func): Likewise.
2566 (mark_loops_for_removal): Likewise.
2567 (mark_all_loops_for_removal): Likewise.
2568 * loop-doloop.c (doloop_modify): Do not update frequency.
2569 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2571 * lto-streamer-in.c (input_function): Update count_max.
2572 * omp-expand.c (expand_omp_taskreg): Update count_max.
2573 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2574 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2575 (maybe_hot_count_p): Use ipa counts only.
2576 (maybe_hot_bb_p): Simplify.
2577 (maybe_hot_edge_p): Simplify.
2578 (probably_never_executed): Do not take frequency argument.
2579 (probably_never_executed_bb_p): Do not pass frequency.
2580 (probably_never_executed_edge_p): Likewise.
2581 (combine_predictions_for_bb): Check that profile is nonzero.
2582 (propagate_freq): Do not set frequency.
2583 (drop_profile): Simplify.
2584 (counts_to_freqs): Simplify.
2585 (expensive_function_p): Use to_frequency.
2586 (propagate_unlikely_bbs_forward): Simplify.
2587 (determine_unlikely_bbs): Simplify.
2588 (estimate_bb_frequencies): Add hack to silence graphite issues.
2589 (compute_function_frequency): Use ipa counts.
2590 (pass_profile::execute): Update.
2591 (rebuild_frequencies): Use counts only.
2592 (force_edge_cold): Use counts only.
2593 * profile-count.c (profile_count::dump): Dump new count types.
2594 (profile_count::differs_from_p): Check compatiblity.
2595 (profile_count::to_frequency): New function.
2596 (profile_count::to_cgraph_frequency): New function.
2597 * profile-count.h (struct function): Declare.
2598 (enum profile_quality): Add profile_guessed_local and
2599 profile_guessed_global0.
2600 (class profile_proability): Decrease number of bits to 29;
2601 update from_reg_br_prob_note and to_reg_br_prob_note.
2602 (class profile_count: Update comment; decrease number of bits
2603 to 61. Check compatibility.
2604 (profile_count::compatible_p): New private member function.
2605 (profile_count::ipa_p): New member function.
2606 (profile_count::operator<): Handle global zero correctly.
2607 (profile_count::operator>): Handle global zero correctly.
2608 (profile_count::operator<=): Handle global zero correctly.
2609 (profile_count::operator>=): Handle global zero correctly.
2610 (profile_count::nonzero_p): New member function.
2611 (profile_count::force_nonzero): New member function.
2612 (profile_count::max): New member function.
2613 (profile_count::apply_scale): Handle IPA scalling.
2614 (profile_count::guessed_local): New member function.
2615 (profile_count::global0): New member function.
2616 (profile_count::ipa): New member function.
2617 (profile_count::to_frequency): Declare.
2618 (profile_count::to_cgraph_frequency): Declare.
2619 * profile.c (OVERLAP_BASE): Delete.
2620 (compute_frequency_overlap): Delete.
2621 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2622 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2623 * sched-ebb.c (rank): Use counts only.
2624 * shrink-wrap.c (handle_simple_exit): Use counts only.
2625 (try_shrink_wrapping): Use counts only.
2626 (place_prologue_for_one_component): Use counts only.
2627 * tracer.c (find_best_predecessor): Use to_frequency.
2628 (find_trace): Use to_frequency.
2629 (tail_duplicate): Use to_frequency.
2630 * trans-mem.c (expand_transaction): Do not update frequency.
2631 * tree-call-cdce.c: Do not update frequency.
2632 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2633 (gimple_merge_blocks): Likewise.
2634 (gimple_split_edge): Likewise.
2635 (gimple_duplicate_sese_region): Likewise.
2636 (gimple_duplicate_sese_tail): Likewise.
2637 (move_sese_region_to_fn): Likewise.
2638 (gimple_account_profile_record): Likewise.
2639 (insert_cond_bb): Likewise.
2640 * tree-complex.c (expand_complex_div_wide): Likewise.
2641 * tree-eh.c (lower_resx): Update profile.
2642 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2644 (initialize_cfun): Do not initialize frequencies
2645 (freqs_to_counts): Delete.
2646 (copy_cfg_body): Ignore count parameter.
2647 (copy_body): Update.
2648 (expand_call_inline): Update count_max.
2649 (optimize_inline_calls): Update count_max.
2650 (tree_function_versioning): Update count_max.
2651 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2652 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2654 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2655 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2656 (try_peel_loop): Likewise.
2657 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2659 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2660 (tree_transform_and_unroll_loop): Do not use frequencies
2661 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2662 Use reliable prediction only.
2663 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2664 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2665 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2666 probability scaling.
2667 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2668 not update frequency
2669 (any_remaining_duplicated_blocks): Likewise.
2670 (update_profile): Likewise.
2671 (estimated_freqs_path): Delete.
2672 (freqs_to_counts_path): Delete.
2673 (clear_counts_path): Delete.
2674 (ssa_fix_duplicate_block_edges): Likewise.
2675 (duplicate_thread_path): Likewise.
2676 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2677 * tree-tailcall.c (decrease_profile): Do not update frequency.
2678 (eliminate_tail_call): Likewise.
2679 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2680 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2681 (optimize_mask_stores): Likewise.
2682 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2683 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2684 (ubsan_expand_ptr_ifn): Update profile.
2685 * value-prof.c (gimple_ic): Simplify.
2686 * value-prof.h (gimple_ic): Update prototype.
2687 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2688 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2690 (want_inline_self_recursive_call_p): Likewise.
2691 (resolve_noninline_speculation): Only cummulate defined counts.
2692 (inline_small_functions): Use nonzero_p.
2693 (ipa_inline): Do not access freed node.
2695 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2697 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2698 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2700 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2702 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2703 non-legitimate address.
2705 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2707 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2708 (*lt0_<mode>di, *lt0_<mode>si): New.
2710 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2712 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2713 TARGET_PAIRED_FLOAT.
2715 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2716 Jim Wilson <jim.wilson@linaro.org>
2718 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2719 * config/aarch64/aarch64-tune.md: Regenerated.
2720 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2721 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2723 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2725 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2726 (arc_expand_prologue): Restore blink for millicode.
2727 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2729 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2732 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2733 gen_vec_duplicate after forcing the scalar into a register.
2735 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2737 * combine (try_combine): Print the insns input to try_combine to the
2740 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2743 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2744 Remove second argument from aarch64_process_target_attr call.
2745 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2747 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2749 (aarch64_handle_attr_arch): Remove second argument.
2750 (aarch64_handle_attr_cpu): Ditto.
2751 (aarch64_handle_attr_tune): Ditto.
2752 (aarch64_handle_attr_isa_flags): Ditto.
2753 (aarch64_process_one_target_attr): Ditto.
2754 (aarch64_process_target_attr): Ditto.
2755 (aarch64_option_valid_attribute_p): Remove second argument.
2756 on aarch64_process_target_attr call.
2758 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2760 * diagnostic.c: Include "selftest-diagnostic.h".
2761 (selftest::assert_location_text): New function.
2762 (selftest::test_diagnostic_get_location_text): New function.
2763 (selftest::diagnostic_c_tests): Call it.
2765 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2767 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2768 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2769 (class selftest::test_diagnostic_context): Move to...
2770 * selftest-diagnostic.c: New file.
2771 * selftest-diagnostic.h: New file.
2773 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2775 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2776 offset range for FT32B.
2777 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2778 * config/ft32/ft32.md: Add TARGET_NOPM.
2779 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2780 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2782 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2784 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2786 2017-11-02 Jeff Law <law@redhat.com>
2788 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2789 virtual keyword on FINAL OVERRIDE members.
2791 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2792 virtuals together. Add virtual destructor.
2793 (substitute_and_fold_engine): Similarly.
2795 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2797 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2799 2017-11-02 Richard Biener <rguenther@suse.de>
2801 PR tree-optimization/82795
2802 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2804 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2806 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2808 (gcc_GAS_CHECK_FEATURE): Remove.
2809 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2810 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2811 * configure: Regenerate.
2813 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2815 * config/arc/arc.c (hwloop_optimize): Account for empty
2818 2017-11-02 Richard Biener <rguenther@suse.de>
2821 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2822 Truncate ARRAY_REF index and element size.
2824 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2826 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2828 2017-11-01 Jeff Law <law@redhat.com>
2830 * tree-ssa-ccp.c (ccp_folder): New class derived from
2831 substitute_and_fold_engine.
2832 (ccp_folder::get_value): New member function.
2833 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2834 (ccp_fold_stmt): Remove prototype.
2835 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2836 * tree-ssa-copy.c (copy_folder): New class derived from
2837 substitute_and_fold_engine.
2838 (copy_folder::get_value): Renamed from get_value.
2839 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2840 * tree-vrp.c (vrp_folder): New class derived from
2841 substitute_and_fold_engine.
2842 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2843 (vrp_folder::get_value): New member function.
2844 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2845 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2846 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2847 provide a class interface to folder/substitute routines.
2848 (ssa_prop_fold_stmt_fn): Remove typedef.
2849 (ssa_prop_get_value_fn): Likewise.
2850 (subsitute_and_fold): Remove prototype.
2851 (replace_uses_in): Likewise.
2852 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2853 Renamed from replace_uses_in. Call the virtual member function
2854 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2855 (substitute_and_fold_dom_walker): Remove initialization of
2856 data member entries for calbacks. Add substitute_and_fold_engine
2857 member and initialize it.
2858 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2859 member functions for get_value, replace_phi_args_in c
2860 replace_uses_in, and fold_stmt calls.
2861 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2862 substitute_and_fold. Remove assert. Update ctor call.
2864 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2865 (ssa_prop_visit_phi_fn): Likewise.
2866 (class ssa_propagation_engine): New class to provide an interface
2868 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2870 (ssa_prop_visit_phi): Likewise.
2871 (ssa_propagation_engine::simulate_stmt): Moved into class.
2872 Call visit_phi/visit_stmt from the class rather than via
2873 file scoped static variables.
2874 (ssa_propagation_engine::simulate_block): Moved into class.
2875 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2876 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2877 set file scoped statics for the visit_stmt/visit_phi callbacks.
2878 * tree-complex.c (complex_propagate): New class derived from
2879 ssa_propagation_engine.
2880 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2881 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2882 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2884 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2885 ssa_propagation_engine.
2886 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2887 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2888 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2889 * tree-ssa-copy.c (copy_prop): New class derived from
2890 ssa_propagation_engine.
2891 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2892 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2893 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2894 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2895 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2896 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2897 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2899 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2901 PR rtl-optimization/82778
2902 PR rtl-optimization/82597
2903 * compare-elim.c (struct comparison): Add in_a_setter field.
2904 (find_comparison_dom_walker::before_dom_children): Remove killed
2905 bitmap and df_simulate_find_defs call, instead walk the defs.
2906 Compute last_setter and initialize in_a_setter. Merge definitions
2907 with first initialization for a few variables.
2908 (try_validate_parallel): Use insn_invalid_p instead of
2909 recog_memoized. Return insn rather than just the pattern.
2910 (try_merge_compare): Fix up comment. Don't uselessly test if
2911 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2913 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2916 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2917 Alan Hayward <alan.hayward@arm.com>
2918 David Sherwood <david.sherwood@arm.com>
2920 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2921 aarch64_hard_regno_nregs to get the number of registers
2924 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2925 Alan Hayward <alan.hayward@arm.com>
2926 David Sherwood <david.sherwood@arm.com>
2928 * config/aarch64/constraints.md (Upl): Rename to...
2930 * config/aarch64/aarch64.md
2931 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2934 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2935 Alan Hayward <alan.hayward@arm.com>
2936 David Sherwood <david.sherwood@arm.com>
2938 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2939 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2942 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2943 Alan Hayward <alan.hayward@arm.com>
2944 David Sherwood <david.sherwood@arm.com>
2946 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2947 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2948 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2950 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2951 of a single element.
2952 * config/aarch64/iterators.md: Add a comment above the permute
2953 unspecs to say that they are generated directly by
2954 aarch64_expand_vec_perm_const.
2955 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2957 2017-11-01 Nathan Sidwell <nathan@acm.org>
2959 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2961 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2963 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2964 and medany code models, and describe what they do.
2966 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2968 Revert accidental duplicate:
2970 * combine.c (can_change_dest_mode): Reject changes in
2971 REGMODE_NATURAL_SIZE.
2973 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2975 PR rtl-optimization/64682
2976 PR rtl-optimization/69567
2977 PR rtl-optimization/69737
2978 PR rtl-optimization/82683
2979 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2980 register mentioned in the note, drop the note, unless it came from I3,
2981 in which case it should go to I3 again.
2983 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2985 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2986 and return false if not.
2987 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2989 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2991 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2992 range is known to be empty.
2994 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2995 Alan Hayward <alan.hayward@arm.com>
2996 David Sherwood <david.sherwood@arm.com>
2998 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2999 and CONST_VECTOR_NUNITS instead of computing the number of units from
3000 the byte sizes of the vector and element.
3001 (simplify_binary_operation_1): Likewise.
3002 (simplify_const_binary_operation): Likewise.
3003 (simplify_ternary_operation): Likewise.
3005 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3006 Alan Hayward <alan.hayward@arm.com>
3007 David Sherwood <david.sherwood@arm.com>
3009 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3010 (int_mem_offset): ...this new function.
3011 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3012 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3013 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3016 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3017 Alan Hayward <alan.hayward@arm.com>
3018 David Sherwood <david.sherwood@arm.com>
3020 * lower-subreg.c (interesting_mode_p): New function.
3021 (compute_costs, find_decomposable_subregs, decompose_register)
3022 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3023 (resolve_clobber, dump_choices): Use it.
3025 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3026 Alan Hayward <alan.hayward@arm.com>
3027 David Sherwood <david.sherwood@arm.com>
3029 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3031 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3032 Alan Hayward <alan.hayward@arm.com>
3033 David Sherwood <david.sherwood@arm.com>
3035 * alias.c (find_base_value, find_base_term): Only process integer
3036 truncations. Check the precision rather than the size.
3038 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3039 Alan Hayward <alan.hayward@arm.com>
3040 David Sherwood <david.sherwood@arm.com>
3042 * machmode.h (is_narrower_int_mode): New function
3043 * optabs.c (expand_float, expand_fix): Use it.
3044 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3046 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3047 Alan Hayward <alan.hayward@arm.com>
3048 David Sherwood <david.sherwood@arm.com>
3050 * rtl.h (narrower_subreg_mode): New function.
3051 * ira-color.c (update_costs_from_allocno): Use it.
3053 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3054 Alan Hayward <alan.hayward@arm.com>
3055 David Sherwood <david.sherwood@arm.com>
3057 * optabs-query.h (convert_optab_p): New function, split out from...
3058 (convert_optab_handler): ...here.
3059 (widening_optab_handler): Delete.
3060 (find_widening_optab_handler): Remove permit_non_widening parameter.
3061 (find_widening_optab_handler_and_mode): Likewise. Provide an
3062 override that operates on mode class wrappers.
3063 * optabs-query.c (widening_optab_handler): Delete.
3064 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3065 parameter. Assert that the two modes are the same class and that
3066 the "from" mode is narrower than the "to" mode. Use
3067 convert_optab_handler instead of widening_optab_handler.
3068 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3069 instead of widening_optab_handler.
3070 * expr.c (expand_expr_real_2): Update calls to
3071 find_widening_optab_handler.
3072 * optabs.c (expand_widen_pattern_expr): Likewise.
3073 (expand_binop_directly): Take the insn_code as a parameter.
3074 (expand_binop): Only call find_widening_optab_handler for
3075 conversion optabs; use optab_handler otherwise. Update calls
3076 to find_widening_optab_handler and expand_binop_directly.
3077 Use convert_optab_handler instead of widening_optab_handler.
3078 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3079 find_widening_optab_handler and use scalar_mode rather than
3081 (convert_plusminus_to_widen): Likewise.
3083 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3084 Alan Hayward <alan.hayward@arm.com>
3085 David Sherwood <david.sherwood@arm.com>
3087 * machmode.h (fixed_size_mode): New class.
3088 * rtl.h (get_pool_mode): Return fixed_size_mode.
3089 * gengtype.c (main): Add fixed_size_mode.
3090 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3091 (get_raw_arg_mode): Likewise.
3092 * doc/tm.texi: Regenerate.
3093 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3094 * targhooks.c (default_get_reg_raw_mode): Likewise.
3095 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3096 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3097 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3098 (msp430_get_raw_result_mode): Likewise.
3099 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3100 * dbxout.c (dbxout_parms): Require fixed-size modes.
3101 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3102 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3103 * omp-low.c (lower_oacc_reductions): Likewise.
3104 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3105 (simplify_subreg): Update accordingly.
3106 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3107 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3108 that aren't fixed-size.
3109 (get_pool_mode): Return a fixed_size_mode.
3110 (output_constant_pool_2): Take a fixed_size_mode.
3112 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3113 Alan Hayward <alan.hayward@arm.com>
3114 David Sherwood <david.sherwood@arm.com>
3116 * doc/rtl.texi (vec_series): Document.
3117 (const): Say that the operand can be a vec_series.
3118 * rtl.def (VEC_SERIES): New rtx code.
3119 * rtl.h (const_vec_series_p_1): Declare.
3120 (const_vec_series_p): New function.
3121 * emit-rtl.h (gen_const_vec_series): Declare.
3122 (gen_vec_series): Likewise.
3123 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3124 (gen_vec_series): Likewise.
3125 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3126 * simplify-rtx.c (simplify_unary_operation): Handle negations
3128 (simplify_binary_operation_series): New function.
3129 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3130 (test_vector_ops_series): New function.
3131 (test_vector_ops): Call it.
3132 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3133 gen_const_vec_series.
3134 (altivec_lvsr): Likewise.
3135 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3137 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3138 Alan Hayward <alan.hayward@arm.com>
3139 David Sherwood <david.sherwood@arm.com>
3141 * doc/rtl.texi (const): Update description of address constants.
3142 Say that vector constants are allowed too.
3143 * common.md (E, F): Use CONSTANT_P instead of checking for
3145 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3146 checking for CONST_VECTOR.
3147 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3149 * expr.c (expand_expr_real_2): Check for vector modes instead
3150 of checking for CONST_VECTOR.
3151 * rtl.h (const_vec_p): New function.
3152 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3153 (unwrap_const_vec_duplicate): Handle them here too.
3155 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3156 David Malcolm <dmalcolm@redhat.com>
3157 Alan Hayward <alan.hayward@arm.com>
3158 David Sherwood <david.sherwood@arm.com>
3160 * rtl.h (vec_duplicate_p): New function.
3161 * selftest-rtl.c (assert_rtx_eq_at): New function.
3162 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3163 (assert_rtx_eq_at): Declare.
3164 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3165 * selftest-run-tests.c (selftest::run_tests): Call it.
3166 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3167 (simplify_unary_operation_1): Recursively handle vector duplicates.
3168 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3170 (simplify_subreg): Handle subregs of vector duplicates.
3171 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3172 (selftest::simplify_rtx_c_tests): New functions.
3174 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3175 Alan Hayward <alan.hayward@arm.com>
3176 David Sherwood <david.sherwood@arm.com>
3178 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3179 (gen_vec_duplicate): Likewise.
3180 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3182 (gen_const_vector): ...here.
3183 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3184 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3185 whose elements are all equal.
3186 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3187 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3188 (simplify_relational_operation): Likewise.
3189 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3191 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3192 (aarch64_expand_vector_init): Likewise.
3193 * config/arm/arm.c (neon_vdup_constant): Likewise.
3194 (neon_expand_vector_init): Likewise.
3195 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3196 (arm_block_set_unaligned_vect): Likewise.
3197 (arm_block_set_aligned_vect): Likewise.
3198 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3199 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3200 (expand_vec_perm_even_odd_pack): Likewise.
3201 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3202 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3203 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3204 gen_const_vec_duplicate.
3205 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3206 * config/mips/mips.c (mips_gen_const_int_vector): Use
3207 gen_const_vec_duplicate.
3208 (mips_expand_vector_init): Use CONST0_RTX.
3209 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3210 (define_split): Use gen_const_vec_duplicate.
3211 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3212 (define_split): Use gen_const_vec_duplicate.
3213 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3214 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3215 * config/spu/spu.c (spu_const): Likewise.
3217 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3218 Alan Hayward <alan.hayward@arm.com>
3219 David Sherwood <david.sherwood@arm.com>
3221 * combine.c (can_change_dest_mode): Reject changes in
3222 REGMODE_NATURAL_SIZE.
3224 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3226 * configure.ac (--enable-libssp): New.
3227 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3228 trying to determine target-specific default. Adjust indentation.
3229 * configure: Regenerated.
3230 * doc/install.texi (Configuration): Expand --disable-libssp
3233 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3235 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3238 2017-10-31 Martin Jambor <mjambor@suse.cz>
3241 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3243 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3245 * auto-profile.c (autofdo_source_profile::read): Use
3246 UNKNOWN_LOCATION rather than 0.
3247 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3248 (warning_at): ...this overload.
3249 (warning_at_rich_loc_n): Rename to...
3250 (warning_n): ...this overload.
3251 (error_at_rich_loc): Rename to...
3252 (error_at): ...this overload.
3253 (pedwarn_at_rich_loc): Rename to...
3254 (pedwarn): ...this overload.
3255 (permerror_at_rich_loc): Rename to...
3256 (permerror): ...this overload.
3257 (inform_at_rich_loc): Rename to...
3258 (inform): ...this overload.
3259 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3260 (diagnostic_n_impl_richloc): Rename to...
3261 (diagnostic_n_impl): ...this rich_location *-based decl.
3262 (inform_at_rich_loc): Rename to...
3263 (inform): ...this, and add an assertion.
3264 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3265 (warning_at_rich_loc): Rename to...
3266 (warning_at): ...this, and add an assertion.
3267 (warning_at_rich_loc_n): Rename to...
3268 (warning_n): ...this, and add an assertion.
3269 (warning_n): Update location_t-based implementation for removal of
3270 location_t-based diagnostic_n_impl.
3271 (pedwarn_at_rich_loc): Rename to...
3272 (pedwarn): ...this, and add an assertion.
3273 (permerror_at_rich_loc): Rename to...
3274 (permerror): ...this, and add an assertion.
3275 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3276 (error_at_rich_loc): Rename to...
3277 (error_at): ...this, and add an assertion.
3278 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3279 (driver::do_spec_on_infiles): Likewise.
3280 * substring-locations.c (format_warning_va): Update for renaming
3281 of inform_at_rich_loc.
3283 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3285 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3286 _Float<N>X built-in functions so that the variant without the
3287 "__builtin_" prefix is only enabled for the GNU C and Objective C
3288 languages when they are in non-strict ANSI/ISO mode.
3289 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3290 * target.def (floatn_builtin_p): Add a target hook to control
3291 whether _Float<N> and _Float<N>X built-in functions without the
3292 "__builtin_" prefix are enabled, and return true for C and
3293 Objective C in the default hook. Include langhooks.h in
3295 * targhooks.h (default_floatn_builtin_p): Likewise.
3296 * targhooks.c (default_floatn_builtin_p): Likewise.
3297 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3298 floatn_builtin_p target hook.
3299 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3301 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3302 Eric Botcazou <ebotcazou@adacore.com>
3304 PR rtl-optimization/81803
3305 * lra-constraints.c (curr_insn_transform): Also reload the whole
3306 register for a strict subreg no wider than a word if this is for
3307 a WORD_REGISTER_OPERATIONS target.
3309 2017-10-31 Jason Merrill <jason@redhat.com>
3311 * gdbinit.in: Skip over inlines from timevar.h.
3313 2017-10-31 Martin Liska <mliska@suse.cz>
3315 * doc/gcov.texi: Document new option.
3316 * gcov.c (print_usage): Likewise print it.
3317 (process_args): Support the argument.
3318 (format_count): New function.
3319 (format_gcov): Use the function.
3321 2017-10-31 Martin Liska <mliska@suse.cz>
3323 * gcov.c (struct name_map): do not use typedef.
3324 Define operator== and operator<.
3325 (name_search): Remove.
3326 (name_sort): Remove.
3327 (main): Do not allocate names.
3328 (process_file): Add vertical space.
3329 (generate_results): Use std::find.
3330 (release_structures): Do not release memory.
3331 (find_source): Use std::find.
3333 2017-10-31 Martin Liska <mliska@suse.cz>
3335 * gcov.c (struct line_info): Remove it's typedef.
3336 (line_info::line_info): Add proper ctor.
3337 (line_info::has_block): Do not use a typedef.
3338 (struct source_info): Do not use typedef.
3339 (circuit): Likewise.
3340 (get_cycles_count): Likewise.
3341 (output_intermediate_file): Iterate via vector iterator.
3342 (add_line_counts): Use std::vector methods.
3343 (accumulate_line_counts): Likewise.
3344 (output_lines): Likewise.
3346 2017-10-31 Martin Liska <mliska@suse.cz>
3348 * gcov.c (struct source_info): Remove typedef.
3349 (source_info::source_info): Add proper ctor.
3350 (accumulate_line_counts): Use struct, not it's typedef.
3351 (output_gcov_file): Likewise.
3352 (output_lines): Likewise.
3353 (main): Do not allocate an array.
3354 (output_intermediate_file): Use size of vector container.
3355 (process_file): Resize the vector.
3356 (generate_results): Do not preallocate, use newly added vector
3358 (release_structures): Do not release sources.
3359 (find_source): Use vector methods.
3360 (add_line_counts): Do not use typedef.
3362 2017-10-31 Martin Liska <mliska@suse.cz>
3364 * doc/gcov.texi: Document that.
3365 * gcov.c (add_line_counts): Mark lines with a non-executed
3367 (output_line_beginning): Handle such lines.
3368 (output_lines): Pass new argument.
3369 (output_intermediate_file): Print it in intermediate format.
3371 2017-10-31 Martin Liska <mliska@suse.cz>
3373 * color-macros.h: New file.
3374 * diagnostic-color.c: Factor out color related to macros to
3376 * doc/gcov.texi: Document -k option.
3377 * gcov.c (INCLUDE_STRING): Include string.h.
3378 (print_usage): Add -k option.
3379 (process_args): Parse it.
3380 (pad_count_string): New function.
3381 (output_line_beginning): Likewise.
3382 (DEFAULT_LINE_START): New macro.
3383 (output_lines): Support color output.
3385 2017-10-31 Martin Liska <mliska@suse.cz>
3387 PR gcov-profile/82633
3388 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3389 their interaction with GCOV infrastructure.
3390 * configure.ac: Add -fkeep-{inline,static}-functions to
3392 * configure: Regenerate.
3394 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3397 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3399 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3402 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3403 into a register if it does not fit into an immediate offset field.
3405 2017-10-31 Olivier Hainque <hainque@adacore.com>
3407 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3409 2017-10-31 Julia Koval <julia.koval@intel.com>
3411 * config.gcc: Add gfniintrin.h.
3412 * config/i386/gfniintrin.h: New.
3413 * config/i386/i386-builtin-types.def
3414 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3415 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3416 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3417 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3418 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3419 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3420 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3421 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3422 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3423 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3424 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3425 * config/i386/immintrin.h: Include gfniintrin.h.
3426 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3428 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3430 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3432 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3434 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3435 (arm_ashldi3_1bit): Remove pattern.
3436 (ashrdi3): Remove shift by 1 expansion.
3437 (arm_ashrdi3_1bit): Remove pattern.
3438 (lshrdi3): Remove shift by 1 expansion.
3439 (arm_lshrdi3_1bit): Remove pattern.
3440 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3441 cost of ashldi3 by 1.
3442 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3443 (<shift>di3_neon): Likewise.
3445 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3447 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3448 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3449 and (*aarch64_simd_mov<VQ:mode>).
3450 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3451 pattern alternative.
3452 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3453 attributes to match pattern alternative.
3455 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3457 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3458 * config/rs6000/emmintrin.h: New file.
3459 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3461 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3463 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3464 * (movdi_vfp_cortexa8): Remove pattern.
3466 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3468 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3470 (Specific, alpha*-dec-osf5.1): Remove.
3471 (Specific, mips-sgi-irix5): Remove.
3472 (Specific, mips-sgi-irix6): Remove.
3474 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3477 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3478 arguments to clear_bit_region_be.
3480 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3482 * gimplify.c: Include memmodel.h.
3484 2017-10-30 Martin Jambor <mjambor@suse.cz>
3486 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3487 condition whether loop should be executed at all.
3489 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3491 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3492 gimple folding of vec_madd() intrinsics.
3493 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3494 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3495 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3497 2017-10-30 Richard Biener <rguenther@suse.de>
3499 PR tree-optimization/82762
3501 2017-10-23 Richard Biener <rguenther@suse.de>
3503 PR tree-optimization/82129
3505 2017-08-01 Richard Biener <rguenther@suse.de>
3507 PR tree-optimization/81181
3508 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3509 (compute_antic): ... end of iteration here.
3511 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3513 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3514 -std=iso9899:2017 and -std=gnu17.
3515 * doc/standards.texi (C Language): Document C17 support.
3516 * doc/cpp.texi (Overview): Mention -std=c17.
3517 (Standard Predefined Macros): Document C11 and C17 values of
3518 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3519 * doc/extend.texi (Inline): Do not list individual options for
3520 standards newer than C99.
3521 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3523 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3526 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3528 * asan.c (asan_finish_file): Align asan globals array by shadow
3531 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3534 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3535 (struct store_immediate_info): Add bitregion_start and bitregion_end
3537 (store_immediate_info::store_immediate_info): Add brs and bre
3538 arguments and initialize bitregion_{start,end} from those.
3539 (struct merged_store_group): Add bitregion_start, bitregion_end,
3540 align_base and mask fields. Drop unnecessary struct keyword from
3541 struct store_immediate_info. Add do_merge method.
3542 (clear_bit_region_be): Use memset instead of loop storing zeros.
3543 (merged_store_group::do_merge): New method.
3544 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3545 stores as long as the surrounding bitregions have no gaps.
3546 (merged_store_group::merge_overlapping): Use do_merge.
3547 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3548 is byte aligned, rather than requiring that start and width are
3549 byte aligned. Drop unnecessary struct keyword from
3550 struct store_immediate_info. Allocate and populate also mask array.
3551 Make start of the arrays relative to bitregion_start rather than
3552 start and size them according to bitregion_{end,start} difference.
3553 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3554 struct store_immediate_info.
3555 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3556 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3557 struct keyword from struct store_immediate_info.
3558 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3559 between stores as long as the surrounding bitregions have no gaps.
3561 (struct split_store): Add orig non-static data member.
3562 (split_store::split_store): Initialize orig to false.
3563 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3564 if there is exactly a single original stmt. Change stmts argument
3565 to pointer from reference, if NULL, don't push anything to it. Add
3566 first argument, use it to optimize skipping over orig stmts that
3567 are known to be before bitpos already. Simplify.
3568 (split_group): Return unsigned int count how many stores are or
3569 would be needed rather than a bool. Add allow_unaligned argument.
3570 Change split_stores argument from reference to pointer, if NULL,
3571 only do a dry run computing how many stores would be produced.
3572 Rewritten algorithm to use both alignment and misalign if
3573 !allow_unaligned and handle bitfield stores with gaps.
3574 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3575 from bitregion_start instead of start. Compute allow_unaligned
3576 here, if true, do 2 split_group dry runs to compute which one
3577 produces fewer stores and prefer aligned if equal. Punt if
3578 new count is bigger or equal than original before emitting any
3579 statements, rather than during that. Remove no longer needed
3580 new_ssa_names tracking. Replace num_stmts with
3581 split_stores.length (). Use 32-bit stack allocated entries
3582 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3583 if possible. Handle bitfields with gaps.
3584 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3585 Compute bitregion_{start,end} for the stores and construct
3586 store_immediate_info with that. Formatting fixes.
3588 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3591 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3592 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3594 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3596 * gimplify.c: Include tm_p.h.
3598 * common.opt (gcoff): Re-add as ignored option.
3599 (gcoff1, gcoff2, gcoff3): Likewise.
3601 * Makefile.in (OBJS): Delete sdbout.o.
3602 (GTFILES): Delete $(srcdir)/sdbout.c.
3603 * debug.h: Delete sdb_debug_hooks.
3604 * final.c: Delete sdbout.h include.
3605 (final_scan_insn): Delete SDB_DEBUG check.
3606 (rest_of_clean_state): Likewise.
3607 * output.h: Delete sdb_begin_function_line.
3610 * toplev.c: Delete sdbout.h include.
3611 (process_options): Delete SDB_DEBUG check.
3612 * tree-core.h (tree_type_common): Delete pointer field of
3614 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3615 TYPE_SYMTAB_POINTER.
3616 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3617 (TYPE_SYMTAB_IS_POINTER): Delete.
3618 (TYPE_SYMTAB_IS_DIE): Renumber.
3619 * xcoffout.c: Refer to former sdbout.c file.
3620 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3622 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3623 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3624 (Debugging Options): Delete -gcoff.
3625 (-gstabs): Delete SDB reference.
3627 (-gcoff@var{level}): Delete.
3628 * doc/passes.texi (Debugging information output): Delete SDB and
3629 sdbout.c references.
3630 * doc/tm.texi: Regenerate.
3631 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3632 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3634 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3635 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3636 (SDB_DEBUGGING_INFO): Delete.
3637 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3638 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3639 * target.def (output_source_filename): Delete COFF reference.
3641 * common.opt (gcoff): Delete.
3642 (gxcoff+): Update Negative chain.
3643 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3645 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3646 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3647 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3649 (expand_function_start): Change sdb reference to past tense.
3650 (expand_function_end): Change sdb reference to past tense.
3651 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3652 * opts.c (debug_type_names): Delete coff entry.
3653 (common_handle_option): Delete OPT_gcoff case.
3654 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3656 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3657 * config/cris/cris.h: Delete SDB reference in comment.
3658 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3659 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3660 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3661 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3663 (ix86_expand_prologue): Likewise.
3664 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3665 * config/ia64/ia64.h: Likewise.
3666 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3667 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3669 * config/mmix/mmix.h: Likewise.
3670 * config/nds32/nds32.c: Likewise.
3671 * config/stormy/storym16.h: Likewise.
3672 * config/visium/visium.h: Likewise.
3673 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3675 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3677 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3678 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3679 FRAME_POINTER_REGNUM point at high end of local var area.
3681 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3683 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3684 Move comment around. Do not reset best_edge for a copiable
3685 destination if the copy would cause a partition change.
3686 (better_edge_p): Remove redundant check.
3688 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3690 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3692 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3694 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3695 for math functions that have _Float<N> and _Float<N>X variants.
3696 (mathfn_built_in_2): Add support for math functions that have
3697 _Float<N> and _Float<N>X variants.
3698 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3699 (expand_builtin_mathfn_ternary): Add support for fma with
3700 _Float<N> and _Float<N>X variants.
3701 (expand_builtin): Likewise.
3702 (fold_builtin_3): Likewise.
3703 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3704 create math function _Float<N> and _Float<N>X variants as external
3706 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3707 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3708 the __builtin_ prefix and if not strict ansi, without the prefix.
3709 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3710 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3711 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3712 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3713 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3714 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3715 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3716 function signatures for fma _Float<N> and _Float<N>X variants.
3717 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3718 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3719 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3720 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3721 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3722 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3723 * gencfn-macros.c (print_case_cfn): Add support for math functions
3724 that have _Float<N> and _Float<N>X variants.
3725 (print_define_operator_list): Likewise.
3726 (fltfn_suffixes): Likewise.
3728 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3729 for math functions that have _Float<N> and _Float<N>X variants.
3730 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3731 and _Float<N>X variants.
3732 (COPYSIGN): Likewise.
3735 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3736 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3738 (integer_valued_read_call_p): Likewise.
3739 * fold-const-call.c (fold_const_call_ss): Likewise.
3740 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3741 _Float<N> and _Float<N>X variants.
3742 (fold_const_call_ssss): Add support for fma _Float<N> and
3743 _Float<N>X variants.
3744 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3745 support for copysign and fma _Float<N> and _Float<N>X variants.
3746 (backprop::process_builtin_call_use): Likewise.
3747 * tree-call-cdce.c (can_test_argument_range); Add support for
3748 sqrt _Float<N> and _Float<N>X variants.
3749 (edom_only_function): Likewise.
3750 (get_no_error_domain): Likewise.
3751 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3752 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3753 copysign _Float<N> and _Float<N>X variants.
3754 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3755 handled by machine independent code.
3756 (FMAF128): Likewise.
3757 * doc/cpp.texi (Common Predefined Macros): Document defining
3758 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3759 fma _Float<N> and _Float<N>X variants.
3761 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3764 * config/i386/i386-modes.def (CCFPU): Remove definition.
3765 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3766 (ix86_cc_modes_compatible): Ditto.
3767 (ix86_expand_carry_flag_compare): Ditto.
3768 (ix86_expand_int_movcc): Ditto.
3769 (ix86_expand_int_addcc): Ditto.
3770 (ix86_reverse_condition): Ditto.
3771 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3772 Return true/false for unordered/ordered fp comparisons.
3773 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3774 (ix86_prepare_fp_compare_args): Update for rename.
3775 (ix86_expand_fp_compare): Update for rename. Generate unordered
3776 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3777 (ix86_expand_sse_compare_and_jump): Ditto.
3778 * config/i386/predicates.md (fcmov_comparison_operator):
3779 Remove CCFPU mode handling.
3780 (ix86_comparison_operator): Ditto.
3781 (ix86_carry_flag_operator): Ditto.
3782 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3783 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3784 (*cmpu<mode>_cc_i387): Ditto.
3785 (FPCMP): Remove mode iterator.
3786 (unord): Remove mode attribute.
3787 (unord_subst): New define_subst transformation
3788 (unord): New define_subst attribute.
3790 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3791 (*cmpi<unord>xf_i387): Ditto.
3792 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3793 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3794 using unord_subst transformation.
3795 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3796 (round_saeonly): Also handle CCFP mode.
3797 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3798 Remove UNSPEC_SAHF unspec handling.
3800 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3802 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3804 2017-10-27 Jeff Law <law@redhat.com>
3806 * gimple-ssa-sprintf.c: Include domwalk.h.
3807 (class sprintf_dom_walker): New class, derived from dom_walker.
3808 (sprintf_dom_walker::before_dom_children): New function.
3809 (struct call_info): Moved into sprintf_dom_walker class
3810 (compute_formath_length, handle_gimple_call): Likewise.
3811 (sprintf_length::execute): Call the dominator walker rather
3812 than walking the statements.
3814 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3815 gimple statement locations.
3816 (check_array_bounds): Corresponding changes. Get the statement's
3817 location directly from wi->stmt.
3819 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3822 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3824 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3826 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3827 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3829 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3832 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3833 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3834 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3835 maybe_get_pool_constant.
3836 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3839 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3841 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3842 2.26 caveat. Update gas and gld versions.
3843 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3846 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3848 * cgraph.h (set_malloc_flag): Declare.
3849 * cgraph.c (set_malloc_flag_1): New function.
3850 (set_malloc_flag): Likewise.
3851 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3852 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3854 (read_ipa_call_summary): Add support for reading is_return_callee.
3855 (write_ipa_call_summary): Stream is_return_callee.
3856 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3857 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3858 ipa-prop.h, ipa-fnsummary.h.
3859 (pure_const_names): Change to static.
3860 (malloc_state_e): Define.
3861 (malloc_state_names): Define.
3862 (funct_state_d): Add field malloc_state.
3863 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3864 (check_retval_uses): New function.
3865 (malloc_candidate_p): Likewise.
3866 (analyze_function): Add support for malloc attribute.
3867 (pure_const_write_summary): Stream malloc_state.
3868 (pure_const_read_summary): Add support for reading malloc_state.
3869 (dump_malloc_lattice): New function.
3870 (propagate_malloc): New function.
3871 (warn_function_malloc): New function.
3872 (ipa_pure_const::execute): Call propagate_malloc and
3873 ipa_free_fn_summary.
3874 (pass_local_pure_const::execute): Add support for malloc attribute.
3875 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3876 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3878 2017-10-27 Martin Liska <mliska@suse.cz>
3880 PR gcov-profile/82457
3881 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3882 for fork-like functions to be properly instrumented.
3884 2017-10-27 Richard Biener <rguenther@suse.de>
3887 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3888 info when we redirected EH.
3890 2017-10-26 Michael Collison <michael.collison@arm.com>
3892 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3894 (<optab>_trunchf<GPI:mode>2: New pattern.
3895 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3896 * config/aarch64/iterators.md (wv): New mode attribute.
3897 (vf, VF): New mode attributes.
3898 (vgp, VGP): New mode attributes.
3899 (s): Update attribute with SImode and DImode prefixes.
3901 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3903 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3904 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3905 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3906 (nios2_option_overide): Initialize it. Don't allow R0-relative
3907 addressing with PIC.
3908 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3909 (nios2_symbolic_constant_p): Likewise.
3910 (nios2_legitimate_address_p): Likewise.
3911 (nios2_r0rel_section_name_p): New.
3912 (nios2_symbol_ref_in_r0rel_data_p): New.
3913 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3914 (r0rel_constant_p): New.
3915 (nios2_print_operand_address): Handle r0rel_constant_p.
3916 (nios2_cdx_narrow_form_p): Likewise.
3917 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3918 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3919 (Nios II Options): Document -mr0rel-sec.
3921 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3923 * config/nios2/nios2.c: Include xregex.h.
3924 (nios2_gprel_sec_regex): New.
3925 (nios2_option_overide): Initialize it. Don't allow GP-relative
3926 addressing with PIC.
3927 (nios2_small_section_name_p): Check for regex match.
3928 * config/nios2/nios2.opt (mgprel-sec=): New option.
3929 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3930 (Nios II Options): Document -mgprel-sec.
3932 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3934 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3936 2017-10-26 Tom de Vries <tom@codesourcery.com>
3938 PR tree-optimization/82707
3939 * gimple.c (gimple_copy): Fix unsharing of
3940 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3942 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3944 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3945 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3946 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3947 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3948 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3949 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3950 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3951 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3952 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3953 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3954 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3955 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3956 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3957 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3958 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3959 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3960 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3962 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3964 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3966 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3967 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3968 warning to rs6000.c. Remove the Undocumented flag, since it has
3970 (-mabi=ibmlongdouble): Likewise.
3971 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3972 already set, set the default format for long double.
3973 (rs6000_debug_reg_global): Print whether long double is IBM or
3975 (rs6000_option_override_internal): Rework setting long double
3976 format. Only warn if the user is changing the long double default
3977 and they did not use -Wno-psabi.
3978 * doc/invoke.texi (PowerPC options): Update the documentation for
3979 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3981 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3982 Alan Hayward <alan.hayward@arm.com>
3983 David Sherwood <david.sherwood@arm.com>
3985 * rtl.h (wider_subreg_mode): New function.
3986 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3987 rather than an unsigned int *.
3988 * ira-color.c (regno_max_ref_width): Replace with...
3989 (regno_max_ref_mode): ...this new variable.
3990 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3991 Use wider_subreg_mode.
3992 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3993 rather than an unsigned int *.
3994 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3995 (process_alt_operands): Likewise.
3996 (invariant_p): Likewise.
3997 * lra-spills.c (assign_mem_slot): Likewise.
3998 (add_pseudo_to_slot): Likewise.
3999 * lra.c (collect_non_operand_hard_regs): Likewise.
4000 (add_regs_to_insn_regno_info): Likewise.
4001 * reload1.c (regno_max_ref_width): Replace with...
4002 (regno_max_ref_mode): ...this new variable.
4003 (reload): Update accordingly. Update call to
4004 ira_sort_regnos_for_alter_reg.
4005 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4006 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4007 (scan_paradoxical_subregs): Likewise.
4009 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4011 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4012 (aarch64_frame): Add emit_frame_chain boolean.
4013 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4014 Move eh_return case to aarch64_layout_frame.
4015 (aarch64_layout_frame): Initialize emit_frame_chain.
4016 (aarch64_expand_prologue): Use emit_frame_chain.
4018 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4020 * config/aarch64/aarch64.c (aarch64_layout_frame):
4021 Ensure LR is always stored at the bottom of the callee-saves.
4022 Remove rarely used frame layout which saves callee-saves at top of
4023 frame, so the store of LR can be used as a valid probe in all cases.
4025 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4027 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4028 Improve unaligned TImode/TFmode base/offset split.
4030 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4031 Alan Hayward <alan.hayward@arm.com>
4032 David Sherwood <david.sherwood@arm.com>
4034 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4035 * combine.c (find_single_use_1): Likewise.
4036 (expand_field_assignment): Likewise.
4037 (move_deaths): Likewise.
4038 * lra-constraints.c (simplify_operand_subreg): Likewise.
4039 (curr_insn_transform): Likewise.
4040 * lra.c (collect_non_operand_hard_regs): Likewise.
4041 (add_regs_to_insn_regno_info): Likewise.
4042 * rtlanal.c (reg_referenced_p): Likewise.
4043 (covers_regno_no_parallel_p): Likewise.
4045 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4047 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4048 Don't print any bits outside the precision of the value.
4049 * wide-int.cc (test_printing): Add some new tests.
4051 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4053 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4054 supports -xbrace_comment option.
4055 * configure: Regenerate.
4056 * config.in: Regenerate.
4057 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4058 (ASM_CPU_SPEC): Use it.
4060 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4062 * target.def (static_rtx_alignment): New hook.
4063 * targhooks.h (default_static_rtx_alignment): Declare.
4064 * targhooks.c (default_static_rtx_alignment): New function.
4065 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4066 * doc/tm.texi: Regenerate.
4067 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4068 instead of targetm.constant_alignment. Remove call to
4070 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4071 (cris_preferred_mininum_alignment): New function, split out from...
4072 (cris_constant_alignment): ...here.
4073 (cris_static_rtx_alignment): New function.
4074 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4076 (ix86_constant_alignment): ...here.
4077 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4078 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4079 (mmix_static_rtx_alignment): New function.
4080 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4081 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4083 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4086 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4087 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4089 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4091 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4092 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4093 mark as a sign-extending load.
4094 (local_pic_load_u): Define.
4096 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4099 * fold-const.c (operand_equal_for_comparison_p): Also return true
4100 if ARG0 is a simple variant of ARG1 with narrower precision.
4101 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4103 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4105 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4107 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4108 scatter_static, scatter_per_elt.
4109 * x86-tune-costs.h: Add new cost entries.
4111 2017-10-25 Richard Biener <rguenther@suse.de>
4113 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4114 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4115 class pass_fre): Move to ...
4116 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4117 class pass_fre): ... here and adjust for statistics.
4119 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4122 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4123 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4124 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4125 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4128 2017-10-25 Richard Biener <rguenther@suse.de>
4130 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4131 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4132 eliminate_push_avail, eliminate_insert): Move inside...
4133 (class eliminate_dom_walker): ... this class in preparation
4135 (fini_eliminate): Remove by merging with ...
4136 (eliminate): ... this function. Adjust for class changes.
4137 (pass_pre::execute): Remove fini_eliminate call.
4138 (pass_fre::execute): Likewise.
4140 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4143 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4144 (VPERMI2, VPERMI2I): New mode iterators.
4145 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4146 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4148 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4149 mode iterator. Remove 3 old define_insn patterns.
4150 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4151 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4152 VPERMI2 mode iterator, remove the other two expanders.
4153 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4154 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4155 instructions, remove the other two patterns.
4156 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4157 mode iterator, remove the other two patterns.
4158 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4159 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4160 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4161 and adjust argument order accordingly.
4162 (ix86_expand_vec_perm): Adjust caller.
4163 (expand_vec_perm_1): Likewise.
4164 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4165 (expand_vec_perm_vpermt2_vpshub2): ... this.
4166 (ix86_expand_vec_perm_const_1): Adjust caller.
4167 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4170 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4171 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4172 (vec_shl_<mode>): Remove unused expander.
4173 (avx512bw_<shift_insn><mode>3): New define_insn.
4174 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4175 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4177 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4180 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4183 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4185 PR rtl-optimization/82396
4186 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4187 (autopref_multipass_init): Simplify initialization.
4188 (autopref_rank_data): Simplify sort order.
4189 * gcc/sched-int.h (autopref_multipass_data_): Remove
4190 multi_mem_insn_p, min_offset and max_offset.
4192 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4195 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4196 Check special value of flag_omit_frame_pointer.
4197 (aarch64_can_eliminate): Likewise.
4198 (aarch64_override_options_after_change_1): Simplify handling of
4199 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4201 2017-10-24 Richard Biener <rguenther@suse.de>
4203 PR tree-optimization/82697
4204 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4205 zero for conditional load and unconditional store.
4207 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4209 * doc/install.texi: Document bootstrap-cet.
4211 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4214 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4215 ENDBR instruction at function entrance if function is only
4218 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4221 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4222 patterns to better describe from which operation the CF is computed.
4223 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4224 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4225 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4226 is 0, use _0 suffixed expanders instead of emitting a comparison
4229 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4231 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4232 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4234 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4237 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4238 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4239 * loop-iv.c (iv_get_reaching_def): Likewise.
4240 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4241 variable is promoted and the partition contains undefined values.
4243 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4245 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4247 (nios2_address_cost): Define.
4248 (nios2_legitimize_address): Recognize (exp + constant) directly.
4249 (TARGET_ADDRESS_COST): Define.
4251 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4253 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4254 (nios2_symbolic_memory_operand_p): Declare.
4255 (nios2_split_large_constant): Declare.
4256 (nios2_split_symbolic_memory_operand): Declare.
4257 * config/nios2/nios2.c: Adjust includes.
4258 (nios2_symbolic_constant_allowed): New.
4259 (nios2_symbolic_constant_p): New.
4260 (nios2_plus_symbolic_constant_p): New.
4261 (nios2_valid_addr_expr_p): Recognize addresses involving
4263 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4264 (nios2_symbolic_memory_operand_p): New.
4265 (nios2_large_constant_p): New.
4266 (nios2_split_large_constant): New.
4267 (nios2_split_plus_large_constant): New.
4268 (nios2_split_symbolic_memory_operand): New.
4269 (nios2_legitimize_address): Code refactoring. Handle addresses
4270 involving symbolic constants.
4271 (nios2_emit_move_sequence): Likewise.
4272 (nios2_print_operand): Improve error output.
4273 (nios2_print_operand_address): Handle LO_SUM.
4274 (nios2_cdx_narrow_form_p): Likewise.
4275 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4276 operands involving symbolic constants.
4277 (movhi_internal, movsi_internal): Likewise.
4278 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4279 (extendhisi2, extendqi<mode>2): Likewise.
4281 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4283 * tree-pass.h (PROP_rtl_split_insns): Define.
4284 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4286 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4288 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4290 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4293 * target.def (const_not_ok_for_debug_p): Default to
4294 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4295 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4296 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4297 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4298 which targetm.const_not_ok_for_debug_p returned true.
4299 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4301 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4303 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4304 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4305 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4307 (ix86_const_not_ok_for_debug_p): New function.
4308 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4309 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4311 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4314 * system.h: Conditionally include "unique-ptr.h" if
4315 INCLUDE_UNIQUE_PTR is defined.
4316 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4317 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4319 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4321 * config/rl78/rl78.md: New define_expand "subdi3".
4323 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4326 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4327 DF_REF_INSN if DF_REF_INSN_INFO is false.
4329 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4331 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4332 xmm_move instead of sse_move.
4333 (sse_store_index): New function.
4334 (ix86_register_move_cost): Be more sensible about mismatch stall;
4335 model AVX moves correctly; make difference between sse->integer and
4337 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4338 moves; make difference between SSE and AVX.
4339 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4340 and zmm_move. Increase size of sse load and store tables;
4341 add unaligned load and store tables; add ssemmx_to_integer.
4342 * x86-tune-costs.h: Update all entries according to real
4343 move latencies from Agner Fog's manual and chip documentation.
4345 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4348 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4349 * config/i386/constraints.md (Wf): New constraint.
4350 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4351 (cmp<dwi>_doubleword): Removed.
4352 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4353 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4354 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4355 expand with cmp<dwi>_doubleword. For LTU and GEU use
4356 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4358 * common.opt (gcolumn-info): Enable by default.
4359 * doc/invoke.texi (gcolumn-info): Document new default.
4361 2017-10-23 Richard Biener <rguenther@suse.de>
4363 PR tree-optimization/82672
4364 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4365 Fold the stmt if we propagated into it.
4367 2017-10-23 Richard Biener <rguenther@suse.de>
4369 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4370 (bitmap_remove_expr_from_set): ... this. All callers call this
4371 for non-constant values.
4372 (bitmap_set_subtract): Rename to...
4373 (bitmap_set_subtract_expressions): ... this. Adjust and
4375 (bitmap_set_contains_value): Remove superfluous check.
4376 (bitmap_set_replace_value): Inline into single caller ...
4377 (bitmap_value_replace_in_set): ... here and simplify.
4378 (dependent_clean): Merge into ...
4379 (clean): ... this using an overload. Adjust.
4380 (prune_clobbered_mems): Adjust.
4381 (compute_antic_aux): Likewise.
4382 (compute_partial_antic_aux): Likewise.
4384 2017-10-23 Richard Biener <rguenther@suse.de>
4386 PR tree-optimization/82129
4388 2017-08-01 Richard Biener <rguenther@suse.de>
4390 PR tree-optimization/81181
4391 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4392 (compute_antic): ... end of iteration here.
4394 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4396 * target.def (starting_frame_offset): New hook.
4397 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4398 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4399 * doc/tm.texi.in: Regenerate.
4400 * hooks.h (hook_hwi_void_0): Declare.
4401 * hooks.c (hook_hwi_void_0): New function.
4402 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4403 STARTING_FRAME_OFFSET.
4404 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4405 * reload1.c (reload): Likewise.
4406 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4407 instead of STARTING_FRAME_OFFSET.
4408 * function.c (try_fit_stack_local): Likewise.
4409 (assign_stack_local_1): Likewise
4410 (instantiate_virtual_regs): Likewise.
4411 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4412 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4413 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4414 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4415 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4416 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4417 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4418 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4419 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4420 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4421 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4422 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4423 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4424 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4425 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4426 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4427 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4428 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4429 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4430 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4431 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4432 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4433 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4434 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4435 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4436 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4437 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4438 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4439 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4440 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4441 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4442 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4443 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4444 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4445 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4446 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4447 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4448 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4449 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4450 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4451 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4452 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4453 return a HOST_WIDE_INT.
4454 (avr_builtin_setjmp_frame_value): Use it instead of
4455 STARTING_FRAME_OFFSET.
4456 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4457 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4458 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4460 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4461 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4462 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4463 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4464 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4465 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4466 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4467 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4468 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4469 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4470 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4471 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4473 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4474 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4475 * config/mips/mips.c (mips_compute_frame_info): Refer to
4476 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4477 (mips_starting_frame_offset): New function.
4478 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4479 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4480 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4481 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4482 and return a HOST_WIDE_INT.
4483 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4484 (mmix_initial_elimination_offset): Refer to
4485 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4486 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4487 * config/pa/pa.c (pa_starting_frame_offset): New function.
4488 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4489 (pa_expand_prologue): Likewise.
4490 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4491 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4492 !FRAME_GROWS_DOWNWARD handling to...
4493 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4494 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4495 !FRAME_GROWS_DOWNWARD handling to...
4496 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4497 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4498 !FRAME_GROWS_DOWNWARD handling to...
4499 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4500 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4502 (rs6000_starting_frame_offset): New function.
4503 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4504 !FRAME_GROWS_DOWNWARD handling to...
4505 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4506 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4507 !FRAME_GROWS_DOWNWARD handling to...
4508 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4509 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4510 !FRAME_GROWS_DOWNWARD handling to...
4511 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4512 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4513 (rs6000_starting_frame_offset): New function.
4514 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4515 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4516 * config/vax/vax.c (vax_starting_frame_offset): New function.
4517 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4518 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4519 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4520 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4521 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4522 * system.h (STARTING_FRAME_OFFSET): Poison.
4524 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4526 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4527 SCALAR_TYPE_MODE instead of TYPE_MODE.
4529 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4530 Alan Hayward <alan.hayward@arm.com>
4531 David Sherwood <david.sherwood@arm.com>
4533 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4535 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4536 Alan Hayward <alan.hayward@arm.com>
4537 David Sherwood <david.sherwood@arm.com>
4539 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4541 2017-10-23 Richard Biener <rguenther@suse.de>
4543 PR tree-optimization/82129
4544 * tree-ssa-pre.c (bitmap_set_and): Remove.
4545 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4546 canonicalizing expressions in the set to those with lowest
4547 ID rather than taking that from the first edge.
4549 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4551 * combine.c (rtx_equal_for_field_assignment_p): Use
4552 byte_lowpart_offset.
4554 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4555 Alan Hayward <alan.hayward@arm.com>
4556 David Sherwood <david.sherwood@arm.com>
4558 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4559 to a SUBREG_PROMOTED_VAR.
4561 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4562 Alan Hayward <alan.hayward@arm.com>
4563 David Sherwood <david.sherwood@arm.com>
4565 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4566 (expand_debug_source_expr): Likewise.
4567 * combine.c (combine_simplify_rtx): Likewise.
4568 * cse.c (fold_rtx): Likewise.
4569 * optabs.c (expand_float): Likewise.
4570 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4571 (simplify_binary_operation_1): Likewise.
4573 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4574 Alan Hayward <alan.hayward@arm.com>
4575 David Sherwood <david.sherwood@arm.com>
4577 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4578 (record_promoted_value): Likewise.
4579 * expr.c (expand_expr_real_2): Likewise.
4580 * ree.c (update_reg_equal_equiv_notes): Likewise.
4581 (combine_set_extension): Likewise.
4582 * rtlanal.c (low_bitmask_len): Likewise.
4583 * simplify-rtx.c (neg_const_int): Likewise.
4584 (simplify_binary_operation_1): Likewise.
4586 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4587 Alan Hayward <alan.hayward@arm.com>
4588 David Sherwood <david.sherwood@arm.com>
4590 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4591 * regcprop.c (maybe_mode_change): Likewise.
4592 * reload1.c (alter_reg): Likewise.
4594 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4596 * inchash.h (inchash::hash::add_wide_int): New function.
4597 * lto-streamer-out.c (hash_tree): Use it.
4599 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4601 * inchash.h (inchash::hash::add_wide_int): Rename to...
4602 (inchash::hash::add_hwi): ...this.
4603 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4604 (polymorphic_call_target_hasher::hash): Likewise.
4605 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4606 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4607 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4608 * lto-streamer-out.c (hash_tree): Likewise.
4609 * optc-save-gen.awk: Likewise.
4610 * tree.c (add_expr): Likewise.
4612 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4615 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4616 for ordered inequality comparisons even with TARGET_IEEE_FP.
4618 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4621 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4622 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4623 Expand with cmp<dwi>_doubleword.
4625 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4627 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4628 List CET intrinsics.
4629 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4630 specific to -fcf-protection option.
4632 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4634 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4635 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4636 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4637 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4638 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4639 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4640 (extra_objs): Add cet.o for Linux/x86 targets.
4641 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4642 * config/i386/cet.c: New file.
4643 * config/i386/cetintrin.h: Likewise.
4644 * config/i386/t-cet: Likewise.
4645 * config/i386/cpuid.h (bit_SHSTK): New.
4646 (bit_IBT): Likewise.
4647 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4648 pass IBT and SHSTK bits.
4649 * config/i386/i386-builtin-types.def
4650 (VOID_FTYPE_UNSIGNED_PVOID): New.
4651 (VOID_FTYPE_UINT64_PVOID): Likewise.
4652 * config/i386/i386-builtin.def: Add CET intrinsics.
4653 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4654 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4655 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4656 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4658 * config/i386/i386.c (rest_of_insert_endbranch): New.
4659 (pass_data_insert_endbranch): Likewise.
4660 (pass_insert_endbranch): Likewise.
4661 (make_pass_insert_endbranch): Likewise.
4662 (ix86_notrack_prefixed_insn_p): Likewise.
4663 (ix86_target_string): Add -mibt, -mshstk flags.
4664 (ix86_option_override_internal): Add flag_cf_protection
4666 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4667 (ix86_print_operand): Add 'notrack' prefix output.
4668 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4669 (ix86_expand_builtin): Expand CET intrinsics.
4670 (x86_output_mi_thunk): Add 'endbranch' instruction.
4671 * config/i386/i386.h (TARGET_IBT): New.
4672 (TARGET_IBT_P): Likewise.
4673 (TARGET_SHSTK): Likewise.
4674 (TARGET_SHSTK_P): Likewise.
4675 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4676 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4677 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4678 (builtin_setjmp_setup): New pattern.
4679 (builtin_longjmp): Likewise.
4680 (rdssp<mode>): Likewise.
4681 (incssp<mode>): Likewise.
4682 (saveprevssp): Likewise.
4683 (rstorssp): Likewise.
4684 (wrss<mode>): Likewise.
4685 (wruss<mode>): Likewise.
4686 (setssbsy): Likewise.
4687 (clrssbsy): Likewise.
4688 (nop_endbr): Likewise.
4689 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4691 * config/i386/immintrin.h: Include <cetintrin.h>.
4692 * config/i386/linux-common.h
4693 (file_end_indicate_exec_stack_and_cet): New prototype.
4694 (TARGET_ASM_FILE_END): New.
4696 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4698 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4699 latencies instead of having separate table; make difference between
4700 integer and float costs.
4701 * i386.h (processor_costs): Remove scalar_stmt_cost,
4702 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4703 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4705 * x86-tune-costs.h: Remove entries which has been removed in
4706 procesor_costs from all tables; make cond_taken_branch_cost
4707 and cond_not_taken_branch_cost COST_N_INSNS based.
4709 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4711 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4713 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4715 * config/i386/i386.md (isa): Remove fma_avx512f.
4716 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4717 <avx512>_fmadd_<mode>_mask3<round_name>,
4718 <avx512>_fmsub_<mode>_mask<round_name>,
4719 <avx512>_fmsub_<mode>_mask3<round_name>,
4720 <avx512>_fnmadd_<mode>_mask<round_name>,
4721 <avx512>_fnmadd_<mode>_mask3<round_name>,
4722 <avx512>_fnmsub_<mode>_mask<round_name>,
4723 <avx512>_fnmsub_<mode>_mask3<round_name>,
4724 <avx512>_fmaddsub_<mode>_mask<round_name>,
4725 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4726 <avx512>_fmsubadd_<mode>_mask<round_name>,
4727 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4728 (*vec_widen_umult_even_v16si<mask_name>,
4729 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4730 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4732 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4734 * extend.texi: Add 'nocf_check' documentation.
4735 * gimple.texi: Add second parameter to
4736 gimple_build_call_from_tree.
4737 * invoke.texi: Add -fcf-protection documentation.
4738 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4740 2017-10-20 Richard Biener <rguenther@suse.de>
4742 PR tree-optimization/82473
4743 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4744 the largest input type.
4746 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4748 * c-attribs.c (handle_nocf_check_attribute): New function.
4749 (c_common_attribute_table): Add 'nocf_check' handling.
4750 * gimple-parser.c: Add second argument NULL to
4751 gimple_build_call_from_tree.
4752 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4753 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4755 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4756 * common.opt: Add fcf-protection flag.
4757 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4758 * flag-types.h: Add enum cf_protection_level.
4759 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4760 Add 'nocf_check' attribute propagation to gimple call.
4761 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4762 (gimple_build_call_from_tree): Update prototype.
4763 (gimple_call_nocf_check_p): New function.
4764 (gimple_call_set_nocf_check): Likewise.
4765 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4766 * ipa-icf.c: Add nocf_check attribute in statement hash.
4767 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4768 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4769 * toplev.c (process_options): Add flag_cf_protection handling.
4771 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4773 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4775 2017-10-20 Richard Biener <rguenther@suse.de>
4777 PR tree-optimization/82603
4778 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4779 remove false predicated stores.
4781 2017-10-20 Richard Biener <rguenther@suse.de>
4783 * graphite-isl-ast-to-gimple.c
4784 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4785 Remove return value and simplify, dump copied stmt after lhs
4787 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4788 Reduce dump verbosity.
4789 (gsi_insert_earliest): Likewise.
4790 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4791 * graphite.c (print_global_statistics): Adjust dumping.
4792 (print_graphite_scop_statistics): Likewise.
4793 (print_graphite_statistics): Do not dump loops here.
4794 (graphite_transform_loops): But here.
4796 2017-10-20 Nicolas Roche <roche@adacore.com>
4798 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4799 * configure: Regenerate.
4801 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4804 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4805 functions when optimizing replace GIMPLE_RETURN stmts with
4806 calls to __builtin_unreachable ().
4809 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4810 for -fsanitize=thread link of executables.
4811 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4812 link of executables.
4815 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4817 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4818 define_insns for logical vector shifts to use VI248_AVX512BW
4819 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4820 condition, useless isa and prefix attributes. Change the first
4821 2 of these define_insns to ...
4822 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4823 define_insn for avx512vl.
4824 (<shift_insn><mode>3): ... and this, new define_insn without
4825 masking for non-avx512vl.
4828 * config/i386/sse.md (*andnot<mode>3,
4829 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4830 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4831 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4832 not applied use empty suffix even for TARGET_AVX512VL.
4833 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4834 is applied, supply evex,evex or evex,evex,evex instead of just
4837 2017-10-20 Julia Koval <julia.koval@intel.com>
4839 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4840 (OPTION_MASK_ISA_GFNI_UNSET): New.
4841 (ix86_handle_option): Handle OPT_mgfni.
4842 * config/i386/cpuid.h (bit_GFNI): New.
4843 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4844 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4845 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4846 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4847 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4848 * config/i386/i386.opt: Add mgfni.
4850 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4852 * config/msp430/msp430.c (msp430_option_override): Disable
4853 -fdelete-null-pointer-checks.
4854 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4856 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4858 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4859 of x87 and SSE instructions.
4861 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4863 * asan.c (create_cond_insert_point): Do not update edge count.
4864 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4865 (afdo_propagate_circuit): Likewise.
4866 (afdo_calculate_branch_prob): Likewise.
4867 (afdo_annotate_cfg): Likewise.
4868 * basic-block.h (struct edge_def): Remove count.
4869 (edge_def::count): New accessor.
4870 * bb-reorder.c (rotate_loop): Update.
4871 (find_traces_1_round): Update.
4872 (connect_traces): Update.
4873 (sanitize_hot_paths): Update.
4874 * cfg.c (unchecked_make_edge): Update.
4875 (make_single_succ_edge): Update.
4876 (check_bb_profile): Update.
4877 (dump_edge_info): Update.
4878 (update_bb_profile_for_threading): Update.
4879 (scale_bbs_frequencies_int): Update.
4880 (scale_bbs_frequencies_gcov_type): Update.
4881 (scale_bbs_frequencies_profile_count): Update.
4882 (scale_bbs_frequencies): Update.
4883 * cfganal.c (connect_infinite_loops_to_exit): Update.
4884 * cfgbuild.c (compute_outgoing_frequencies): Update.
4885 (find_many_sub_basic_blocks): Update.
4886 * cfgcleanup.c (try_forward_edges): Update.
4887 (try_crossjump_to_edge): Update
4888 * cfgexpand.c (expand_gimple_cond): Update
4889 (expand_gimple_tailcall): Update
4890 (construct_exit_block): Update
4891 * cfghooks.c (verify_flow_info): Update
4892 (redirect_edge_succ_nodup): Update
4893 (split_edge): Update
4894 (make_forwarder_block): Update
4895 (duplicate_block): Update
4896 (account_profile_record): Update
4897 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4898 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4899 * cfgloopmanip.c (scale_loop_profile): Update.
4901 (lv_adjust_loop_entry_edge): Update.
4902 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4903 (force_nonfallthru_and_redirect): Update.
4904 (purge_dead_edges): Update.
4905 (rtl_flow_call_edges_add): Update.
4906 * cgraphunit.c (init_lowered_empty_function): Update.
4907 (cgraph_node::expand_thunk): Update.
4908 * gimple-pretty-print.c (dump_probability): Update.
4909 (dump_edge_probability): Update.
4910 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4911 * haifa-sched.c (sched_create_recovery_edges): Update.
4912 * hsa-gen.c (convert_switch_statements): Update.
4913 * ifcvt.c (dead_or_predicable): Update.
4914 * ipa-inline-transform.c (inline_transform): Update.
4915 * ipa-split.c (split_function): Update.
4916 * ipa-utils.c (ipa_merge_profiles): Update.
4917 * loop-doloop.c (add_test): Update.
4918 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4919 * lto-streamer-in.c (input_cfg): Update.
4920 (input_function): Update.
4921 * lto-streamer-out.c (output_cfg): Update.
4922 * modulo-sched.c (sms_schedule): Update.
4923 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4924 * predict.c (maybe_hot_edge_p): Update.
4925 (unlikely_executed_edge_p): Update.
4926 (probably_never_executed_edge_p): Update.
4927 (dump_prediction): Update.
4928 (drop_profile): Update.
4929 (propagate_unlikely_bbs_forward): Update.
4930 (determine_unlikely_bbs): Update.
4931 (force_edge_cold): Update.
4932 * profile.c (compute_branch_probabilities): Update.
4933 * reg-stack.c (better_edge): Update.
4934 * shrink-wrap.c (handle_simple_exit): Update.
4935 * tracer.c (better_p): Update.
4936 * trans-mem.c (expand_transaction): Update.
4937 (split_bb_make_tm_edge): Update.
4938 * tree-call-cdce.c: Update.
4939 * tree-cfg.c (gimple_find_sub_bbs): Update.
4940 (gimple_split_edge): Update.
4941 (gimple_duplicate_sese_region): Update.
4942 (gimple_duplicate_sese_tail): Update.
4943 (gimple_flow_call_edges_add): Update.
4944 (insert_cond_bb): Update.
4945 (execute_fixup_cfg): Update.
4946 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4947 * tree-complex.c (expand_complex_div_wide): Update.
4948 * tree-eh.c (lower_resx): Update.
4949 (unsplit_eh): Update.
4950 (cleanup_empty_eh_move_lp): Update.
4951 * tree-inline.c (copy_edges_for_bb): Update.
4952 (freqs_to_counts): Update.
4953 (copy_cfg_body): Update.
4954 * tree-ssa-dce.c (remove_dead_stmt): Update.
4955 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4956 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4957 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4958 (unloop_loops): Update.
4959 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4960 * tree-ssa-loop-split.c (connect_loops): Update.
4961 (split_loop): Update.
4962 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4963 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4964 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4965 * tree-ssa-reassoc.c (branch_fixup): Update.
4966 * tree-ssa-tail-merge.c (replace_block_by): Update.
4967 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4968 (compute_path_counts): Update.
4969 (update_profile): Update.
4970 (recompute_probabilities): Update.
4971 (update_joiner_offpath_counts): Update.
4972 (estimated_freqs_path): Update.
4973 (freqs_to_counts_path): Update.
4974 (clear_counts_path): Update.
4975 (ssa_fix_duplicate_block_edges): Update.
4976 (duplicate_thread_path): Update.
4977 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4978 (case_bit_test_cmp): Update.
4979 (collect_switch_conv_info): Update.
4980 (gen_inbound_check): Update.
4981 (do_jump_if_equal): Update.
4982 (emit_cmp_and_jump_insns): Update.
4983 * tree-tailcall.c (decrease_profile): Update.
4984 (eliminate_tail_call): Update.
4985 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4986 (vect_do_peeling): Update.
4987 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4988 * ubsan.c (ubsan_expand_null_ifn): Update.
4989 (ubsan_expand_ptr_ifn): Update.
4990 * value-prof.c (gimple_divmod_fixed_value): Update.
4991 (gimple_mod_pow2): Update.
4992 (gimple_mod_subtract): Update.
4993 (gimple_ic): Update.
4994 (gimple_stringop_fixed_value): Update.
4996 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4999 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5001 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5003 PR rtl-optimization/82395
5004 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5005 based on non_spilled_static_chain_regno_p.
5007 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5009 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5010 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5011 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5013 2017-10-19 Martin Sebor <msebor@redhat.com>
5015 PR tree-optimization/82596
5016 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5018 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5020 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5021 (get_mem_refs_of_builtin_call): Likewise.
5022 * builtins.c (expand_builtin_apply): Adjust call to
5023 allocate_dynamic_stack_space.
5024 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5025 the third argument to allocate_dynamic_stack_space, otherwise -1.
5026 (expand_builtin): Deal with all alloca variants.
5027 (is_inexpensive_builtin): Likewise.
5028 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5029 * calls.c (special_function_p): Deal with all alloca variants.
5030 (initialize_argument_information): Adjust call to
5031 allocate_dynamic_stack_space.
5032 (expand_call): Likewise.
5033 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5034 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5035 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5036 use it for the stack usage computation.
5037 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5038 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5039 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5040 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5041 (in_loop_p): Remove first argument and useless check.
5042 (pass_walloca::execute): Remove useless test and adjust call to above.
5043 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5044 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5045 (gimplify_call_expr): Deal with all alloca variants.
5046 * hsa-gen.c (gen_hsa_alloca): Likewise.
5047 (gen_hsa_insns_for_call): Likewise.
5048 * ipa-pure-const.c (special_builtin_state): Likewise.
5049 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5050 * tree-object-size.c (alloc_object_size): Likewise.
5051 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5052 (call_may_clobber_ref_p_1): Likewise.
5053 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5054 (ccp_fold_stmt): Likewise.
5055 (optimize_stack_restore): Likewise.
5056 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5057 (mark_all_reaching_defs_necessary_1): Likewise.
5058 (propagate_necessity): Likewise.
5059 (eliminate_unnecessary_stmts): Likewise.
5060 * tree.c (build_common_builtin_nodes): Build
5061 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5062 (build_alloca_call_expr): New function.
5063 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5064 (CASE_BUILT_IN_ALLOCA): Likewise.
5065 (build_alloca_call_expr): Declare.
5066 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5068 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5071 * dwarf2out.c (new_die_raw): New static inline function.
5072 (new_die): Use it to create the DIE.
5073 (add_AT_external_die_ref): Likewise.
5074 (clone_die): Likewise.
5075 (clone_as_declaration): Likewise.
5076 (dwarf2out_vms_debug_main_pointer): Likewise.
5077 (base_type_die): Likewise. Remove early return for corner cases.
5078 Do not call add_pubtype on the DIE here.
5079 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5080 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5081 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5082 native order exists for base types, attach the DIE manually and call
5083 add_pubtype on it. Do not equate a reverse order DIE to the type.
5085 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5087 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5088 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5090 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5092 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5094 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5095 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5096 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5097 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5098 * opts.c (sanitizer_opts): Add builtin.
5099 * ubsan.c (instrument_builtin): New function.
5100 (pass_ubsan::execute): Call it.
5101 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5102 * doc/invoke.texi: Document -fsanitize=builtin.
5104 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5105 builtins, store max (log2 (align), 0) into uchar field instead of
5106 align into uptr field.
5107 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5108 store uchar 0 field instead of uptr 0 field.
5109 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5110 instead of passing one address of struct with 2 locations pass
5111 two addresses of structs with 1 location each.
5112 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5113 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5114 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5115 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5116 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5117 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5118 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5119 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5121 2017-10-19 Martin Liska <mliska@suse.cz>
5124 * file-find.c (remove_prefix): Remove.
5125 * file-find.h (remove_prefix): Likewise.
5126 * gcc-ar.c: Remove smartness of lookup.
5128 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5130 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5131 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5132 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5134 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5137 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5138 (setcc + and to xor + setcc): New peephole2.
5140 2017-10-19 Tom de Vries <tom@codesourcery.com>
5142 * doc/sourcebuild.texi (Test Directives, Variants of
5143 dg-require-support): Add dg-require-stack-size.
5145 2017-10-19 Martin Liska <mliska@suse.cz>
5148 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5149 that have a large alignment.
5150 (gimplify_target_expr): Likewise.
5152 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5154 PR rtl-optimization/82602
5155 * ira.c (rtx_moveable_p): Return false for volatile asm.
5157 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5160 * config/i386/i386-modes.def (CCGZ): New CC mode.
5161 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5162 * config/i386/predicates.md (ix86_comparison_operator):
5164 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5165 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5166 with double-word subtraction.
5167 (put_condition_code): Handle CCGZmode.
5169 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5171 * wide-int.cc (debug (const wide_int &)): New.
5172 (debug (const wide_int *)): New.
5173 (debug (const widest_int &)): New.
5174 (debug (const widest_int *)): New.
5176 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5179 * lra-constraints.c (curr_insn_transform): Use non-input operand
5180 instead of output one for matched reload.
5182 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5184 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5185 (tree-ssa-loop-ivopts.h): New header file.
5186 (struct builtin_info): New fields.
5187 (classify_builtin_1): Compute and record base and offset parts for
5188 memset builtin partition by calling strip_offset.
5189 (offset_cmp, fuse_memset_builtins): New functions.
5190 (finalize_partitions): Fuse adjacent memset partitions by calling
5192 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5193 Expose the interface.
5194 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5196 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5198 PR tree-optimization/82574
5199 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5200 that data reference must be executed exactly once per iteration
5201 against the outermost loop in nest.
5202 (classify_partition): Update call to above function.
5204 2017-10-18 Richard Biener <rguenther@suse.de>
5206 PR tree-optimization/82591
5207 * graphite.c (graphite_transform_loops): Move code gen message
5209 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5210 Here. Handle scop_to_isl_ast failing.
5211 (scop_to_isl_ast): Limit the number of ISL operations.
5213 2017-10-18 Richard Biener <rguenther@suse.de>
5215 * graphite-isl-ast-to-gimple.c
5216 (translate_isl_ast_to_gimple::set_rename): Simplify.
5217 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5218 (graphite_copy_stmts_from_block): ... here.
5219 (copy_bb_and_scalar_dependences): Simplify.
5220 (add_parameters_to_ivs_params): Canonicalize.
5221 (generate_entry_out_of_ssa_copies): Simplify.
5222 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5223 by passing in ISL dimension.
5224 (parameter_index_in_region_1): Rename to ...
5225 (parameter_index_in_region): ... this.
5226 (extract_affine): Adjust assert, pass down parameter index.
5227 (add_param_constraints): Use range-info when available.
5228 (build_scop_context): Adjust.
5229 * sese.c (new_sese_info): Adjust.
5230 (free_sese_info): Likewise.
5231 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5232 Remove unused typedefs.
5233 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5235 2017-10-18 Martin Liska <mliska@suse.cz>
5237 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5239 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5241 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5242 (s390_sched_init): Do not reset s390_sched_state if we entered the
5243 current basic block via a fallthru edge and all others are unlikely.
5245 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5247 * config/s390/s390.c (NUM_SIDES): New variable.
5248 (LONGRUNNING_THRESHOLD): New variable.
5249 (LATENCY_FACTOR): New variable.
5250 (s390_sched_score): Decrease score for long-running instructions on
5252 (s390_sched_variable_issue): Perform bookkeeping for long-running
5255 2017-10-18 Richard Biener <rguenther@suse.de>
5257 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5258 Simplify with removal of the parameter rename map.
5259 (set_rename): Likewise.
5260 (should_copy_to_new_region): Likewise.
5261 (graphite_copy_stmts_from_block): Likewise.
5262 (copy_bb_and_scalar_dependences): Remove initialization of
5263 unused copied_bb_map.
5265 (copy_internal_parameters): Likewise.
5266 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5267 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5268 Use INTEGRAL_TYPE_P.
5269 (parameter_index_in_region_1): Rename to ...
5270 (assign_parameter_index_in_region): ... this. Assert we have
5271 a parameter we handle.
5272 (scan_tree_for_params): Adjust.
5273 * sese.h (parameter_rename_map_t): Remove.
5274 (struct sese_info_t): Remove unused parameter_rename_map and
5275 copied_bb_map members.
5276 * sese.c (new_sese_info): Adjust.
5277 (free_sese_info): Likewise.
5279 2017-10-18 Martin Liska <mliska@suse.cz>
5282 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5283 on an abnormal edge.
5285 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5287 * doc/invoke.texi (ffunction-sections and fdata-sections):
5290 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5292 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5293 the use statement can throw internally.
5295 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5297 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5298 any RTX present on the RHS of a SET.
5299 * compare-elim.c (try_eliminate_compare): Restore comment.
5301 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5303 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5304 may be also called on tcc_type nodes.
5305 * langhooks.c (lhd_tree_size): Likewise.
5307 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5309 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5310 format_warning_at_substring.
5311 (maybe_warn): Convert source_range * param to a location_t. Pass
5312 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5313 (format_directive): Remove code to extract source_ranges and
5314 source_range * in favor of just a location_t.
5315 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5317 * substring-locations.c (format_warning_va): Convert
5318 source_range * param to a location_t.
5319 (format_warning_at_substring): Likewise.
5320 * substring-locations.h (format_warning_va): Likewise.
5321 (format_warning_at_substring): Likewise.
5323 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5325 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5327 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5328 and scatter/gather ops.
5330 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5331 vec_gather_load and vec_scatter_store.
5332 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5333 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5334 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5335 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5336 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5337 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5339 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5341 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5342 Detect FTST instruction and handle its register pops. Only pop
5343 second operand if can_pop_second_op is true.
5344 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5345 set can_pop_second_op to false in the compare_for_stack_reg call.
5347 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5348 output_fp_compare for stack register operands.
5349 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5350 instructions here. Do not emit stack register pops here. Assert
5351 that FCOMPP pops next to top stack register. Rewrite function.
5353 2017-10-17 Nathan Sidwell <nathan@acm.org>
5356 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5357 use DECL_ASSEMBLER_NAME_RAW.
5360 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5363 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5364 Wilco Dijkstra <wilco.dijkstra@arm.com>
5366 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5367 converstion to Pmode from the buf_addr.
5369 2017-10-17 Richard Biener <rguenther@suse.de>
5371 * graphite-dependences.c (scop_get_reads_and_writes): Change
5372 output parameters to references.
5374 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5376 PR 71026/tree-optimization
5377 * fold-const.c (distribute_real_division): Removed.
5378 (fold_binary_loc): Remove calls to distribute_real_divison.
5380 2017-10-17 Richard Biener <rguenther@suse.de>
5382 * graphite-scop-detection.c
5383 (scop_detection::stmt_has_simple_data_refs_p): Always use
5384 the full nest as region.
5385 (try_generate_gimple_bb): Likewise.
5386 * sese.c (scalar_evolution_in_region): Simplify now that
5387 SCEV can handle instantiation in regions.
5388 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5389 in the non-loop part of a function if requested.
5391 2017-10-17 Richard Biener <rguenther@suse.de>
5393 PR tree-optimization/82563
5394 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5396 (graphite_regenerate_ast_isl): Call it.
5397 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5399 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5401 PR tree-optimization/82549
5402 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5403 Formatting fixes. Instead of calling make_bit_field_ref with negative
5406 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5408 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5409 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5410 _mm_maskz_reduce_ss): New.
5411 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5412 __builtin_ia32_reducess_mask): Ditto..
5413 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5414 * config/i386/sse.md (reduces<mode>): Renamed to ...
5415 (reduces<mode><mask_scalar_name>): ... this.
5416 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5418 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5419 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5421 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5423 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5424 * selftest-run-tests.c (selftest::run_tests): Call
5425 selftest::unique_ptr_tests_cc_tests.
5426 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5427 * unique-ptr-tests.cc: New file.
5429 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5432 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5434 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5435 (make_hard_regno_born, make_hard_regno_dead): Update
5436 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5438 2017-10-16 Jeff Law <law@redhat.com>
5440 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5442 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5444 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5446 2017-10-16 Olivier Hainque <hainque@adacore.com>
5448 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5449 with_cpu if we were configured for an e500v2 target cpu name.
5451 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5453 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5454 * doc/invoke.texi: Document +nodsp as a valid extension for
5457 2017-10-16 Martin Liska <mliska@suse.cz>
5459 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5460 (test_set_range): Likewise.
5461 (test_range_functions): Rename to ...
5462 (test_bit_in_range): ... this.
5463 (sbitmap_c_tests): Add new test.
5465 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5467 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5469 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5470 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5472 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5474 * config/aarch64/aarch64-builtins.c
5475 (aarch64_types_quadopu_lane_qualifiers): New.
5476 (TYPES_QUADOPU_LANE): New.
5477 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5478 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5479 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5480 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5481 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5482 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5483 (Vdottype, DOTPROD): New.
5484 (sur): Add SDOT and UDOT.
5486 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5488 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5489 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5490 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5492 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5493 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5494 Enable TARGET_DOTPROD.
5495 (cortex-a75.cortex-a55): Likewise.
5496 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5498 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5500 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5501 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5503 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5505 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5506 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5507 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5508 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5509 * config/arm/types.md (neon_dot, neon_dot_q): New.
5510 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5512 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5514 * config/arm/arm.h (TARGET_DOTPROD): New.
5515 * config/arm/arm.c (arm_arch_dotprod): New.
5516 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5517 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5518 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5519 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5520 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5521 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5522 * doc/invoke.texi (armv8.2-a): Document dotprod
5524 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5526 * i386.c (ix86_vec_cost): New function.
5527 (ix86_rtx_costs): Handle vector operations better.
5528 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5529 * x86-tune-costs.h: Add new costs to all tables.
5531 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5533 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5535 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5536 divsd, sqrtss and sqrtsd
5537 * x86-tune-costs.h: Add new entries to all costs.
5538 (znver1_cost): Fix to match real instruction latencies.
5540 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5541 Michael Collison <michael.collison@arm.com>
5543 * compare-elim.c: Include emit-rtl.h.
5544 (can_merge_compare_into_arith): New function.
5545 (try_validate_parallel): Likewise.
5546 (try_merge_compare): Likewise.
5547 (try_eliminate_compare): Call the above when no previous clobber
5549 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5552 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5556 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5557 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5561 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5562 to be any operand_equal_p operands. For & (B - 1) require
5563 B to be power of 2. Recognize
5564 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5566 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5569 * optabs.c (expand_memory_blockage): Fix call of
5570 targetm.have_memory_blockage.
5572 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5575 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5576 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5577 objects to extra_objs instead of overwriting it.
5579 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5581 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5582 Use any_fp_register_operand as operand[3] predicate. Simplify
5583 equality test for operands[2] and operands[4] memory location.
5584 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5585 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5586 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5587 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5588 any_fp_register_operand as operand[1] predicate. Simplify
5589 equality test for operands[0] and operands[3] memory location.
5590 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5591 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5592 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5594 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5596 * target-insns.def: Add memory_blockage.
5597 * optabs.c (expand_memory_blockage): New function.
5598 (expand_asm_memory_barrier): Rename ...
5599 (expand_asm_memory_blockage): ... to this.
5600 (expand_mem_thread_fence): Call expand_memory_blockage
5601 instead of expand_asm_memory_barrier.
5602 (expand_mem_singnal_fence): Ditto.
5603 (expand_atomic_load): Ditto.
5604 (expand_atomic_store): Ditto.
5605 * doc/md.texi (Standard Pattern Names For Generation):
5606 Document memory_blockage instruction pattern.
5608 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5610 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5611 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5612 * config/rl78/rl78.md: New define_expand "adddi3".
5614 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5616 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5619 2017-10-13 Jeff Law <law@redhat.com>
5621 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5623 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5626 * internal-fn.c (expand_mul_overflow): If both operands have
5627 the same highpart of -1 or 0 and the topmost bit of lowpart
5628 is different, overflow is if res <= 0 rather than res < 0.
5630 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5632 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5633 TARGET_P9_VECTOR code for unaligned_load case.
5635 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5637 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5639 2017-10-13 Nathan Sidwell <nathan@acm.org>
5641 * tree-core.h (tree_contains_struct): Make bool.
5642 * tree.c (tree_contains_struct): Likewise.
5643 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5644 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5645 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5646 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5647 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5649 2017-10-13 Richard Biener <rguenther@suse.de>
5651 * graphite-isl-ast-to-gimple.c
5652 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5653 parameters and dominance check.
5654 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5655 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5656 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5657 Do not update SSA form here or do intermediate IL verification.
5658 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5659 (graphite_initialize): Remove check on the number of loops in
5660 the function and inline into graphite_transform_loops.
5661 (graphite_finalize): Inline into graphite_transform_loops.
5662 (graphite_transform_loops): Perform SSA update and IL verification
5664 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5666 2017-10-13 Richard Biener <rguenther@suse.de>
5668 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5669 graphite_expression_type_precision): Avoid global constructor
5671 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5672 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5673 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5674 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5675 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5676 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5678 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5681 * config/i386/i386.h (ix86_red_zone_size): New.
5682 * config/i386/i386.md (push peephole2s): Replace
5683 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5685 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5686 Alan Hayward <alan.hayward@arm.com>
5687 David Sherwood <david.sherwood@arm.com>
5689 * combine.c (can_change_dest_mode): Reject changes in
5690 REGMODE_NATURAL_SIZE.
5692 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5693 Alan Hayward <alan.hayward@arm.com>
5694 David Sherwood <david.sherwood@arm.com>
5696 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5697 (expand_debug_source_expr): Likewise.
5698 * combine.c (combine_simplify_rtx): Likewise.
5699 * cse.c (fold_rtx): Likewise.
5700 * fwprop.c (canonicalize_address): Likewise.
5701 * targhooks.c (default_shift_truncation_mask): Likewise.
5703 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5704 Alan Hayward <alan.hayward@arm.com>
5705 David Sherwood <david.sherwood@arm.com>
5707 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5708 (widened_mode): Likewise.
5709 (expand_unop): Likewise.
5710 * ree.c (transform_ifelse): Likewise.
5711 (merge_def_and_ext): Likewise.
5712 (combine_reaching_defs): Likewise.
5713 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5715 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5716 Alan Hayward <alan.hayward@arm.com>
5717 David Sherwood <david.sherwood@arm.com>
5719 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5720 * combine.c (gen_lowpart_for_combine): Likewise.
5721 * dwarf2out.c (rtl_for_decl_location): Likewise.
5722 * final.c (alter_subreg): Likewise.
5723 * rtlhooks.c (gen_lowpart_general): Likewise.
5724 (gen_lowpart_if_possible): Likewise.
5726 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5727 Alan Hayward <alan.hayward@arm.com>
5728 David Sherwood <david.sherwood@arm.com>
5730 * calls.c (expand_call): Use subreg_lowpart_offset.
5731 * cse.c (cse_insn): Likewise.
5732 * regcprop.c (copy_value): Likewise.
5733 (copyprop_hardreg_forward_1): Likewise.
5735 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5738 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5739 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5740 =Q constraints to +Q and into insn condition add check
5741 that operands[0] and operands[1] are equal.
5742 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5743 =Q constraints to +Q and into insn condition add check
5744 that operands[0] is equal to either operands[1] or operands[2].
5747 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5748 instead of handling MINUS_EXPR twice (once for each argument),
5749 canonicalize operand order and handle just once, use rtype where
5750 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5753 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5754 any values of __C while still being pattern recognizable as a simple
5757 2017-10-13 Richard Biener <rguenther@suse.de>
5759 PR tree-optimization/82451
5761 2017-10-02 Richard Biener <rguenther@suse.de>
5763 PR tree-optimization/82355
5764 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5765 a mapping for the enclosing loop but avoid generating one for
5767 (copy_bb_and_scalar_dependences): Remove premature codegen
5768 error on PHIs in blocks duplicated into multiple places.
5769 * graphite-scop-detection.c
5770 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5771 in the region use it as loop and nest to analyze the DR in.
5772 (try_generate_gimple_bb): Likewise.
5773 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5774 (add_loop_constraints): For blocks in a loop not in the region
5775 create a dimension with a single iteration.
5776 * sese.h (gbb_loop_at_index): Remove assert.
5778 * cfgloop.c (loop_preheader_edge): For the loop tree root
5779 return the single successor of the entry block.
5780 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5781 Reset the SCEV hashtable and niters.
5782 * graphite-scop-detection.c
5783 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5784 assert that we only have POLYNOMIAL_CHREC that vary in loops
5785 contained in the region.
5786 (scop_detection::graphite_can_represent_expr): Adjust.
5787 (scop_detection::stmt_has_simple_data_refs_p): For loops
5788 not in the region set loop to NULL. The nest is now the
5789 entry edge to the region.
5790 (try_generate_gimple_bb): Likewise.
5791 * sese.c (scalar_evolution_in_region): Adjust for
5792 instantiate_scev change.
5793 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5794 Make nest parameter the edge into the region.
5795 (create_data_ref): Likewise.
5796 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5797 entry edge into a region and adjust instantiate_scev calls.
5798 (create_data_ref): Likewise.
5799 (graphite_find_data_references_in_stmt): Likewise.
5800 (find_data_references_in_stmt): Pass the loop preheader edge
5801 from the nest argument.
5802 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5803 parameter the edge into the region.
5804 (instantiate_parameters): Use the loop preheader edge as entry.
5805 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5807 (get_instantiated_value_entry): Make instantiate_below parameter
5808 the edge into the region.
5809 (instantiate_scev_name): Likewise. Adjust dominance checks,
5810 when we cannot use loop-based instantiation instantiate by
5811 walking use-def chains.
5812 (instantiate_scev_poly): Adjust.
5813 (instantiate_scev_binary): Likewise.
5814 (instantiate_scev_convert): Likewise.
5815 (instantiate_scev_not): Likewise.
5816 (instantiate_array_ref): Remove.
5817 (instantiate_scev_3): Likewise.
5818 (instantiate_scev_2): Likewise.
5819 (instantiate_scev_1): Likewise.
5820 (instantiate_scev_r): Do not blindly handle N-operand trees.
5821 Do not instantiate array-refs. Handle all constants and invariants.
5822 (instantiate_scev): Make instantiate_below parameter
5823 the edge into the region.
5824 (resolve_mixers): Use the loop preheader edge for the region
5825 parameter to instantiate_scev_r.
5826 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5828 2017-10-13 Richard Biener <rguenther@suse.de>
5830 PR tree-optimization/82525
5831 * graphite-isl-ast-to-gimple.c
5832 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5834 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5835 Fail code generation when we cannot represent the isl integer.
5836 (binary_op_to_tree): Elide modulo operations that are no-ops
5837 in the type we code generate. Remove now superfluous code
5840 2017-10-13 Richard Biener <rguenther@suse.de>
5842 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5843 (scop_detection::harmful_loop_in_region): Remove premature
5844 IV type restriction.
5845 (scop_detection::graphite_can_represent_scev): We can handle
5846 pointer IVs just fine.
5848 2017-10-13 Alan Modra <amodra@gmail.com>
5850 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5851 "Clobbers and Scratch Registers". Add paragraph on
5852 alternative to clobbers for scratch registers and OpenBLAS
5855 2017-10-13 Alan Modra <amodra@gmail.com>
5857 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5858 example of a memory input for a string of known length. Move
5859 commentary out of table. Add a number of new examples
5860 covering array memory inputs.
5862 2017-10-12 Martin Liska <mliska@suse.cz>
5864 PR tree-optimization/82493
5865 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5866 (test_range_functions): New function.
5867 (sbitmap_c_tests): Likewise.
5868 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5869 * selftest.h (sbitmap_c_tests): New function.
5871 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5873 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5875 * config/rs6000/amo.h: Fix spacing issue.
5877 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5880 * config/i386/i386.md (*ashl<mode>3_mask_1,
5881 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5882 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5885 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5887 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5888 (profile_probability): Set max_probability
5889 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5891 * profile-count.c (profile_probability::differs_from_p): Do not
5892 rely on max_probaiblity == 10000
5894 2017-10-12 Jeff Law <law@redhat.com>
5896 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5899 2017-10-12 Martin Sebor <msebor@redhat.com>
5903 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5904 (handle_alias_pairs): Call it.
5905 * common.opt (-Wattribute-alias): New option.
5906 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5907 * doc/invoke.texi (-Wattribute-alias): Document.
5909 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5912 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5914 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5916 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5917 (make_hard_regno_born, make_hard_regno_dead): Update
5918 bb_killed_pseudos and bb_gen_pseudos.
5920 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5922 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5924 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5926 * config/alpha/alpha.c (alpha_split_conditional_move):
5927 Use std::swap instead of manually swapping.
5928 (alpha_stdarg_optimize_hook): Ditto.
5929 (alpha_canonicalize_comparison): Ditto.
5931 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5933 * tree-loop-distribution.c (struct builtin_info): New struct.
5934 (struct partition): Refactor fields into struct builtin_info.
5935 (partition_free): Free struct builtin_info.
5936 (build_size_arg_loc, build_addr_arg_loc): Delete.
5937 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5938 information from struct builtin_info.
5939 (find_single_drs): New function refactored from classify_partition.
5940 Also moved builtin validity checks to this function.
5941 (compute_access_range, alloc_builtin): New functions.
5942 (classify_builtin_st, classify_builtin_ldst): New functions.
5943 (classify_partition): Refactor code into functions find_single_drs,
5944 classify_builtin_st and classify_builtin_ldst.
5945 (distribute_loop): Don't do runtime alias check when distributing
5947 (find_seed_stmts_for_distribution): New function.
5948 (pass_loop_distribution::execute): Refactor code finding seed
5949 stmts into above function. Support distribution for the innermost
5950 two-level loop nest. Adjust dump information.
5952 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5954 * tree-loop-distribution.c: Adjust the general comment.
5955 (NUM_PARTITION_THRESHOLD): New macro.
5956 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5957 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5958 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5959 in call to build_partition_graph.
5960 (finalize_partitions): New parameter. Make loop distribution more
5961 conservative by fusing more partitions.
5962 (distribute_loop): Don't do runtime alias check in case of loop nest
5964 (find_seed_stmts_for_distribution): New function.
5965 (prepare_perfect_loop_nest): New function.
5966 (pass_loop_distribution::execute): Refactor code finding seed stmts
5967 and loop nest into above functions. Support loop nest distribution.
5968 Adjust dump information accordingly.
5970 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5972 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5973 and set PTYPE_SEQUENTIAL for merged partition.
5975 2017-10-12 Richard Biener <rguenther@suse.de>
5977 PR tree-optimization/69728
5979 2017-09-19 Richard Biener <rguenther@suse.de>
5981 PR tree-optimization/69728
5982 * graphite-sese-to-poly.c (schedule_error): New global.
5983 (add_loop_schedule): Handle empty domain by failing the
5985 (build_original_schedule): Handle schedule_error.
5987 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5988 domain by returning an unchanged schedule.
5990 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5992 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5993 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5995 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5997 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6000 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6003 * expr.c (store_field): Don't optimize away bitsize == 0 store
6004 from CALL_EXPR with addressable return type.
6006 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6008 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6009 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6010 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6011 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6012 TARGET_ISEL instead of TARGET_ISEL<sel>.
6014 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6016 * config/rs6000/rs6000.c
6017 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6019 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6021 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6023 (reg_or_cint_operand): Fix comment.
6024 (reg_or_zero_operand): New predicate.
6025 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6026 * config/rs6000/rs6000.c (output_isel): Delete.
6027 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6028 instead of reg_or_cint_operand. Output instruction directly (not via
6030 (isel_unsigned_<mode>): Ditto.
6031 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6032 gpc_reg_operand. Add an instruction alternative for this. Output
6033 instruction directly.
6034 (*isel_reversed_unsigned_<mode>): Ditto.
6036 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6038 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6039 (TARGET_CANONICALIZE_COMPARISON): Define.
6041 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6044 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6045 Check whether the dest is REG before adding REG_EQUIV note.
6047 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6050 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6052 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6053 (make_hard_regno_born, make_hard_regno_dead): Update
6054 bb_killed_pseudos and bb_gen_pseudos.
6056 2017-10-11 Nathan Sidwell <nathan@acm.org>
6058 * incpath.h (enum incpath_kind): Name enum, prefix values.
6059 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6060 * incpath.c (heads, tails): Use INC_MAX.
6061 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6062 (merge_include_chains, split_quote_chain,
6063 register_include_chains): Update incpath_kind names.
6064 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6065 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6067 (add_framework_path, darwin_register_objc_includes): Likewise.
6068 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6070 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6072 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6073 Do not use float_operator operator predicate.
6074 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6075 * config/i386/predicates.md (float_operator): Remove predicate.
6077 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6079 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6080 (*jccxf_i387): Ditto.
6081 (*jcc<mode>_i387): Ditto.
6082 (*jccu<mode>_i387): Ditto.
6083 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6084 (*jcc_*_i387 splitters): Remove.
6085 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6086 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6087 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6090 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6092 * profile-count.h (slow_safe_scale_64bit): New function.
6093 (safe_scale_64bit): New inline.
6094 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6095 * profile-count.c: Include wide-int.h
6096 (slow_safe_scale_64bit): New.
6098 2017-10-11 Nathan Sidwell <nathan@acm.org>
6100 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6101 HAS_DECL_ASSEMBLER_NAME_P.
6102 * gimple-expr.c (gimple_decl_printable_name: Check
6103 HAS_DECL_ASSEMBLER_NAME_P too.
6104 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6105 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6106 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6107 * passes.c (rest_of_decl_compilation): Check
6108 HAS_DECL_ASSEMBLER_NAME_P too.
6109 * recog.c (verify_changes): Likewise.
6110 * tree-pretty-print.c (dump_decl_name): Likewise.
6111 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6113 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6114 (SET_DECL_ASSEMBLER_NAME): Use it.
6115 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6116 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6117 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6119 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6121 * config.gcc (i386, x86_64): Add extra objects.
6122 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6123 (ix86_min_insn_size): Declare.
6124 (ix86_issue_rate): Declare.
6125 (ix86_adjust_cost): Declare.
6126 (ia32_multipass_dfa_lookahead): Declare.
6127 (ix86_macro_fusion_p): Declare.
6128 (ix86_macro_fusion_pair_p): Declare.
6129 (ix86_bd_has_dispatch): Declare.
6130 (ix86_bd_do_dispatch): Declare.
6131 (ix86_core2i7_init_hooks): Declare.
6132 (ix86_atom_sched_reorder): Declare.
6133 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6134 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6135 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6136 (rip_relative_addr_p): Rename to ...
6137 (ix86_rip_relative_addr_p): ... this one; export.
6138 (memory_address_length): Update.
6139 (ix86_issue_rate): Move to x86-tune-sched.c.
6140 (ix86_flags_dependent): Move to x86-tune-sched.c.
6141 (ix86_agi_dependent): Move to x86-tune-sched.c.
6142 (exact_dependency_1): Move to x86-tune-sched.c.
6143 (exact_store_load_dependency): Move to x86-tune-sched.c.
6144 (ix86_adjust_cost): Move to x86-tune-sched.c.
6145 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6146 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6147 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6148 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6149 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6150 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6151 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6152 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6153 (min_insn_size): Rename to ...
6154 (ix86_min_insn_size): ... this one; export.
6155 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6156 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6157 (core2i7_first_cycle_multipass_backtrack): Move to
6158 x86-tune-sched-core.c.
6159 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6160 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6161 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6162 (ix86_avoid_jump_mispredicts): Update.
6163 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6164 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6165 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6166 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6167 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6168 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6169 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6170 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6171 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6172 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6173 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6174 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6175 (BIG): Move to ix86-tune-sched-bd.c.
6176 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6177 (enum insn_path): Move to ix86-tune-sched-bd.c.
6178 (get_mem_group): Move to ix86-tune-sched-bd.c.
6179 (is_cmp): Move to ix86-tune-sched-bd.c.
6180 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6181 (is_branch): Move to ix86-tune-sched-bd.c.
6182 (is_prefetch): Move to ix86-tune-sched-bd.c.
6183 (init_window): Move to ix86-tune-sched-bd.c.
6184 (allocate_window): Move to ix86-tune-sched-bd.c.
6185 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6186 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6187 (process_end_window): Move to ix86-tune-sched-bd.c.
6188 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6189 (find_constant): Move to ix86-tune-sched-bd.c.
6190 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6191 (has_immediate): Move to ix86-tune-sched-bd.c.
6192 (get_insn_path): Move to ix86-tune-sched-bd.c.
6193 (get_insn_group): Move to ix86-tune-sched-bd.c.
6194 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6195 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6196 (add_insn_window): Move to ix86-tune-sched-bd.c.
6197 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6198 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6199 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6200 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6201 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6202 (do_dispatch): Move to ix86-tune-sched-bd.c.
6203 (has_dispatch): Move to ix86-tune-sched-bd.c.
6204 * config/i386/t-i386: Add new object files.
6205 * config/i386/x86-tune-costs.h: New file.
6206 * config/i386/x86-tune-sched-atom.c: New file.
6207 * config/i386/x86-tune-sched-bd.c: New file.
6208 * config/i386/x86-tune-sched-core.c: New file.
6209 * config/i386/x86-tune-sched.c: New file.
6211 2017-10-11 Liu Hao <lh_mouse@126.com>
6213 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6214 the generic version below instead.
6215 (should_colorize): Recognize Windows consoles as terminals
6217 * pretty-print.c [__MINGW32__] (write_all): New function.
6218 [__MINGW32__] (find_esc_head): Likewise.
6219 [__MINGW32__] (find_esc_terminator): Likewise.
6220 [__MINGW32__] (eat_esc_sequence): Likewise.
6221 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6223 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6226 2017-10-11 Richard Biener <rguenther@suse.de>
6228 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6229 Properly call analyze_scalar_evolution with the loop of the stmt.
6231 2017-10-11 Richard Biener <rguenther@suse.de>
6233 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6234 * tree-core.h (tree_base): Add chrec_var union member.
6235 * tree.h (CHREC_VAR): Remove.
6236 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6237 * tree-chrec.h (build_polynomial_chrec): Adjust.
6238 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6239 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6241 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6243 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6244 * match.pd: ... here.
6245 ((T) X == (T) Y): Relax condition.
6247 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6249 PR tree-optimization/82472
6250 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6252 (break_alias_scc_partitions): Update postorder number.
6254 2017-10-11 Martin Liska <mliska@suse.cz>
6257 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6259 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6261 2017-10-11 Martin Liska <mliska@suse.cz>
6263 * print-rtl.c (print_insn): Move declaration of idbuf
6264 to same scope as name.
6266 2017-10-11 Martin Liska <mliska@suse.cz>
6271 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6274 2017-10-11 Martin Liska <mliska@suse.cz>
6277 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6280 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6282 * config.gcc (default_gnu_indirect_function): Default to yes for
6283 arm*-*-linux* with glibc.
6285 2017-10-11 Richard Biener <rguenther@suse.de>
6287 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6288 default-defs and types we do not want to analyze.
6289 (interpret_loop_phi): Replace unreachable code with an assert.
6290 (compute_scalar_evolution_in_loop): Remove and inline ...
6291 (analyze_scalar_evolution_1): ... here, replacing condition with
6292 what makes the intent clearer. Remove handling of cases
6293 get_scalar_evolution now handles.
6295 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6297 PR rtl-optimization/81434
6298 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6299 comment for main loop. In sched_group_found if, also add checks for
6300 pass and min_cost_group.
6302 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6304 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6305 (rs6000_insn_cost): New function.
6306 * config/rs6000/rs6000.md (cost): New attribute.
6308 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6309 H.J. Lu <hongjiu.lu@intel.com>
6313 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6314 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6315 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6316 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6317 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6318 OPTION_MASK_ISA_MMX and some other bit set require both
6319 mmx and the ISAs without the mmx bit.
6320 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6321 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6322 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6323 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6324 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6325 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6326 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6327 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6328 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6329 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6330 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6331 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6332 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6333 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6334 Add OPTION_MASK_ISA_MMX.
6336 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6338 * config.gcc (armv7*-*-freebsd*): New target.
6339 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6341 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6343 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6344 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6347 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6349 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6350 a template parameter.
6351 (WIDE_INT_REF_FOR): Update accordingly.
6352 * tree.h (wi::int_traits <const_tree>): Delete.
6353 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6354 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6355 (wi::tree_to_wide_ref): New typedef.
6356 (wi::to_wide): New function.
6357 * calls.c (get_size_range): Use wi::to_wide when operating on
6359 * cgraph.c (cgraph_node::create_thunk): Likewise.
6360 * config/i386/i386.c (ix86_data_alignment): Likewise.
6361 (ix86_local_alignment): Likewise.
6362 * dbxout.c (stabstr_O): Likewise.
6363 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6364 * expr.c (const_vector_from_tree): Likewise.
6365 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6366 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6367 (fold_negate_expr_1, int_const_binop_1, const_binop)
6368 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6369 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6370 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6371 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6372 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6373 (fold_not_const, round_up_loc): Likewise.
6374 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6375 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6376 (alloca_call_type): Likewise.
6377 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6378 * godump.c (go_output_typedef): Likewise.
6379 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6380 * internal-fn.c (get_min_precision): Likewise.
6381 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6382 * ipa-polymorphic-call.c
6383 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6384 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6385 (ipa_modify_call_arguments): Likewise.
6386 * match.pd: Likewise.
6387 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6388 * print-tree.c (print_node_brief, print_node): Likewise.
6389 * stmt.c (expand_case): Likewise.
6390 * stor-layout.c (layout_type): Likewise.
6391 * tree-affine.c (tree_to_aff_combination): Likewise.
6392 * tree-cfg.c (group_case_labels_stmt): Likewise.
6393 * tree-data-ref.c (dr_analyze_indices): Likewise.
6394 (prune_runtime_alias_test_list): Likewise.
6395 * tree-dump.c (dequeue_and_dump): Likewise.
6396 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6397 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6398 * tree-pretty-print.c (dump_generic_node): Likewise.
6399 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6400 (simple_iv_with_niters): Likewise.
6401 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6402 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6403 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6404 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6405 (refine_value_range_using_guard, number_of_iterations_ne_max)
6406 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6407 (get_cst_init_from_scev, record_nonwrapping_iv)
6408 (scev_var_range_cant_overflow): Likewise.
6409 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6410 * tree-ssa-pre.c (compute_avail): Likewise.
6411 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6412 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6413 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6414 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6415 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6416 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6417 (dump_case_nodes, try_switch_expansion): Likewise.
6418 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6419 (vect_do_peeling): Likewise.
6420 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6421 * tree-vect-stmts.c (vectorizable_load): Likewise.
6422 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6423 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6424 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6425 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6426 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6427 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6428 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6429 (evrp_dom_walker::before_dom_children): Likewise.
6430 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6431 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6432 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6433 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6434 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6435 (get_range_pos_neg): Likewise.
6436 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6437 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6438 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6439 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6440 * config/avr/avr.c (avr_fold_builtin): Likewise.
6441 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6442 * config/msp430/msp430.c (msp430_attr): Likewise.
6443 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6444 * config/powerpcspe/powerpcspe-c.c
6445 (altivec_resolve_overloaded_builtin): Likewise.
6446 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6447 (rs6000_expand_ternop_builtin): Likewise.
6448 * config/rs6000/rs6000-c.c
6449 (altivec_resolve_overloaded_builtin): Likewise.
6450 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6451 (rs6000_expand_ternop_builtin): Likewise.
6452 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6454 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6456 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6457 when copying loop nest with only one inner loop.
6459 2017-10-10 Richard Biener <rguenther@suse.de>
6461 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6462 blocks if SCEV is active.
6463 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6465 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6466 (scev_initialize): Assert we are not yet initialized.
6468 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6470 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6471 inner loop's exit stmt by making it always exit the loop, otherwise
6472 we would generate an infinite empty loop.
6474 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6476 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6477 renaming variables in new preheader if it's deleted.
6479 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6481 * tree-loop-distribution.c (struct partition): Remove unused field
6482 loops of the structure.
6483 (partition_alloc, partition_free): Ditto.
6484 (build_rdg_partition_for_vertex): Ditto.
6486 2017-10-09 Jeff Law <law@redhat.com>
6488 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6489 return type to match prototype and documentation.
6491 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6493 * config/rs6000/rs6000.c (processor_costs): Move to ...
6494 * config/rs6000/rs6000.h: ... here.
6495 (rs6000_cost): Declare.
6497 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6499 * except.c (setjmp_fn): New global variable.
6500 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6501 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6502 if DONT_USE_BUILTIN_SETJMP is defined.
6504 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6506 * target.def (insn_cost): New hook.
6507 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6508 * doc/tm.texi: Regenerate.
6509 * rtlanal.c (insn_cost): Use the new hook.
6511 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6513 * combine.c (combine_validate_cost): Compute the new insn_cost,
6514 not just pattern_cost.
6515 (try_combine): Adjust comment.
6517 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6519 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6521 * combine.c (uid_insn_cost): Adjust comment.
6522 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6524 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6525 * dse.c (find_shift_sequence): Ditto.
6526 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6527 (bb_valid_for_noce_process_p): Use pattern_cost.
6528 * rtl.h (insn_rtx_cost): Delete.
6529 (pattern_cost): New prototype.
6530 (insn_cost): New prototype.
6531 * rtlanal.c (insn_rtx_cost): Rename to...
6532 (pattern_cost): ... this.
6535 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6537 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6538 (*jcc<mode>_0_r_i387): Ditto.
6539 (*jccxf_r_i387): Ditto.
6540 (*jcc<mode>_r_i387): Ditto.
6541 (*jccu<mode>_r_i387): Ditto.
6542 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6543 (*jcc): Rename from *jcc_1.
6545 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6547 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6548 deferred rescans after the lvx/stvx recombination pre-pass.
6550 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6552 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6553 memory operation instruction support.
6554 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6555 (rs6000-ibm-aix[789]*): Likewise.
6556 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6557 Document new functions.
6559 2017-10-09 Richard Biener <rguenther@suse.de>
6561 PR tree-optimization/82397
6562 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6563 equality only for semantically equal trees.
6565 2017-10-09 Richard Biener <rguenther@suse.de>
6567 PR tree-optimization/82449
6568 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6569 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6570 allow constant addresses.
6571 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6574 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6576 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6579 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6582 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6585 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6588 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6590 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6593 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6594 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6596 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6598 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6599 (WI_BINARY_PREDICATE_RESULT): Likewise.
6600 (wi::binary_traits::operator_result): New type.
6601 (wi::binary_traits::predicate_result): Likewise.
6602 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6603 (generic_wide_int::operator==, generic_wide_int::operator!=)
6604 (generic_wide_int::operator&, generic_wide_int::and_not)
6605 (generic_wide_int::operator|, generic_wide_int::or_not)
6606 (generic_wide_int::operator^, generic_wide_int::operator+
6607 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6609 (operator~, unary operator-, operator==, operator!=, operator&)
6610 (operator|, operator^, operator+, binary operator-, operator*): New
6612 * expr.c (get_inner_reference): Use wi::bit_and_not.
6613 * fold-const.c (fold_binary_loc): Likewise.
6614 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6615 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6616 (bit_value_binop): Likewise.
6617 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6618 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6619 (extract_range_from_binary_expr_1): Likewise.
6620 (masked_increment): Likewise.
6621 (simplify_bit_ops_using_ranges): Likewise.
6623 2017-10-09 Martin Jambor <mjambor@suse.cz>
6626 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6627 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6628 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6630 (hsa_fixup_mov_insn_type): New function.
6631 (hsa_op_with_type::get_in_type): Use it.
6632 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6633 immediates in an assert.
6634 (hsa_op_with_type::extend_int_to_32bit): New method.
6635 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6636 types. Convert to dest type if necessary.
6637 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6638 (reg_for_gimple_ssa): Pass false as min32int to
6639 hsa_type_for_scalar_tree_type.
6640 (gen_hsa_addr): Fixup type when creating addresable temporary.
6641 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6642 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6643 necessary. Call hsa_fixup_mov_insn_type.
6644 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6645 extend operands and convert to dest type if necessary.
6646 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6647 to dest type if necessary.
6648 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6649 if conversion nt necessary and size matches.
6650 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6651 to dest type if necessary.
6652 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6653 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6655 (gen_hsa_clrsb): Likewise.
6656 (gen_hsa_ffs): Likewise.
6657 (gen_hsa_divmod): Extend operands and convert to dest type if
6659 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6661 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6663 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6664 Remove empty default arguments. Use a brace block as output
6666 (conditional return): Ditto.
6668 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6669 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6671 (group_ending_nop): Ditto.
6672 (doloop_end): Ditto.
6673 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6674 (splitters for those): Ditto.
6676 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6678 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6679 a conditional jump (and the compare for it) so that pc_rtx is the
6681 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6682 for the deleted and renamed ctr<mode>_internal[234] patterns.
6683 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6684 Delete second conditional return pattern.
6685 (ctr<mode>_internal2): Delete this second bdnz pattern.
6686 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6687 (ctr<mode>_internal4): Delete this second bdz pattern.
6689 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6691 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6692 (always_initialized_rtx_for_ssa_name_p): New predicate.
6693 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6694 (finish_out_of_ssa): Free new field of SA.
6695 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6696 * tree-ssa-coalesce.c: Include tree-ssa.h.
6697 (get_parm_default_def_partitions): Remove extern keyword.
6698 (get_undefined_value_partitions): New function.
6699 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6700 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6702 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6704 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6706 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6708 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6710 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6711 for targets that preffer 128bit.
6713 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6715 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6717 2017-10-08 Olivier Hainque <hainque@adacore.com>
6719 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6720 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6722 (thumb_set_return_address): Likewise.
6724 2017-10-08 Olivier Hainque <hainque@adacore.com>
6726 * common/config/arm/arm-common.c (arm_except_unwind_info):
6727 Handle DWARF2_UNWIND_INFO.
6729 2017-10-07 Michael Collison <michael.collison@arm.com>
6731 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6734 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6736 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6737 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6738 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6739 defined, force the creation of a new block for a dispatch label.
6741 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6743 * invoke.texi (Wsuggest-attribute=cold): Document.
6744 * common.opt (Wsuggest-attribute=cold): New
6745 * ipa-pure-const.c (warn_function_cold): New function.
6746 * predict.c (compute_function_frequency): Use it.
6747 * predict.h (warn_function_cold): Declare.
6749 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6751 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6754 2017-10-06 Martin Liska <mliska@suse.cz>
6756 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6757 keyword for member functions.
6758 (struct sanopt_tree_couple): New struct.
6759 (struct sanopt_tree_couple_hash): New function.
6760 (struct sanopt_ctx): Add new hash_map.
6761 (has_dominating_ubsan_ptr_check): New function.
6762 (record_ubsan_ptr_check_stmt): Likewise.
6763 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6764 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6765 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6767 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6770 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6771 aarch64_simd_valid_immediate on CONST_VECTORs.
6772 (aarch64_reg_or_bic_imm): Likewise.
6774 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6776 PR rtl-optimization/82396
6777 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6779 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6781 * graphite-dependences.c (scop_get_reads): Move code to...
6782 (scop_get_must_writes): Move code to...
6783 (scop_get_may_writes): Move code to...
6784 (scop_get_reads_and_writes): ... here.
6785 (scop_get_dependences): Call scop_get_reads_and_writes.
6787 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6789 PR tree-optimization/82434
6790 * fold-const.h (can_native_encode_type_p,
6791 can_native_encode_string_p): Remove.
6792 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6793 don't encode anything, just return what would be otherwise returned.
6794 (native_encode_fixed, native_encode_complex, native_encode_vector):
6796 (native_encode_string): Likewise. Inline by hand
6797 can_native_encode_string_p.
6798 (can_native_encode_type_p): Remove.
6799 (can_native_encode_string_p): Remove.
6800 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6801 STRING_CSTs using can_native_encode_string_p, test all
6802 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6803 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6804 argument from native_encode_expr.
6805 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6806 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6809 2017-10-06 Richard Biener <rguenther@suse.de>
6811 PR tree-optimization/82397
6812 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6813 operand_equal_p but rely on data_ref_compare_tree for detecting
6815 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6816 to match up with dr_group_sort_cmp.
6818 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6821 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6823 * config/s390/s390-builtin-types.def: Regenerate.
6825 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6828 * config/s390/s390-builtin-types.def: Regenerate.
6829 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6830 Change flag from B_VXE to B_VX.
6831 (s390_vec_min_dbl): Remove B_VXE flag.
6833 2017-10-06 Richard Biener <rguenther@suse.de>
6835 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6836 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6837 translate_isl_ast_to_gimple::is_valid_rename,
6838 translate_isl_ast_to_gimple::get_rename,
6839 translate_isl_ast_to_gimple::get_def_bb_for_const,
6840 translate_isl_ast_to_gimple::get_new_name,
6841 translate_isl_ast_to_gimple::collect_all_ssa_names,
6842 translate_isl_ast_to_gimple::copy_loop_phi_args,
6843 translate_isl_ast_to_gimple::collect_all_ssa_names,
6844 translate_isl_ast_to_gimple::copy_loop_phi_args,
6845 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6846 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6847 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6848 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6849 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6850 translate_isl_ast_to_gimple::copy_cond_phi_args,
6851 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6852 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6853 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6854 translate_isl_ast_to_gimple::rename_uses,
6855 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6856 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6857 (set_rename_for_each_def): Likewise.
6858 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6859 here. Handle rewriting SCEV analyzable uses here.
6860 (copy_bb_and_scalar_dependences): Generate code for PHI
6862 (graphite_regenerate_ast_isl): Adjust.
6863 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6864 (add_write, add_read): New functions.
6865 (build_cross_bb_scalars_def): Use it and simplify.
6866 (build_cross_bb_scalars_use): Likewise.
6867 (graphite_find_cross_bb_scalar_vars): Inline into...
6868 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6869 simulating out-of-SSA. Compute liveout and add dependencies.
6870 (build_scops): Force an empty entry block.
6871 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6873 (sese_build_liveouts): Declare.
6874 (sese_trivially_empty_bb_p): Likewise.
6875 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6876 compute liveout and debug_liveout.
6877 (sese_bad_liveouts_use): Remove.
6878 (sese_reset_debug_liveouts_bb): Likewise.
6879 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6880 (sese_build_liveouts): Build liveout and debug_liveout and store
6882 (new_sese_info): Adjust.
6883 (free_sese_info): Likewise.
6884 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6885 do not build liveout here.
6886 (move_sese_in_condition): Adjust region entry.
6887 (scev_analyzable_p): Match up with chrec_apply requirements.
6888 (sese_trivially_empty_bb_p): New.
6889 * tree-into-ssa.c (get_reaching_def): Properly support generating
6890 default-defs for incremental rewrite of anonymous names.
6892 2017-10-06 Richard Biener <rguenther@suse.de>
6894 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6895 precision do not perform modulo reduction.
6897 2017-10-06 Richard Biener <rguenther@suse.de>
6899 PR tree-optimization/82436
6900 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6901 conservatively choose the vectorization factor when checking
6902 whether we can perform the required load permutation.
6903 (vect_transform_slp_perm_load): Assert when we may not fail.
6905 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6907 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6908 message for incompatible -msdata=* and -mcall-* options.
6910 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6912 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6913 rate for post-reload scheduling.
6915 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6917 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6919 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6921 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6922 to improve monte carlo in scimark.
6924 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6926 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6927 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6928 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6929 pentium4_cost, nocona_cost): Set reassociation width to 1.
6930 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6931 width to 2 for fp operations and 1 otherwise.
6932 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6934 (atom_cost): Set reassociation width to 2.
6935 (slm_cost, generic_cost): Set fp reassociation width
6936 to 2 and 1 otherwise.
6937 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6938 (core_cost): Set fp reassociation width to 4 and vector to 2.
6939 (ix86_reassociation_width): Rewrite using cost table; special case
6940 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6941 and TARGET_AVX128_OPTIMAL.
6942 * config/i386/i386.h (processor_costs): Add
6943 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6944 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6945 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6946 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6947 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6948 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6950 2017-10-05 Nathan Sidwell <nathan@acm.org>
6952 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6954 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6956 * config/arm/arm.c (arm_test_fpu_data): New.
6957 (arm_run_selftests): Call arm_test_fpu_data.
6959 2017-10-04 Nathan Sidwell <nathan@acm.org>
6961 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6963 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6965 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6967 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6968 check type for aarch64_simd_valid_immediate.
6969 (aarch64_output_simd_mov_immediate): Update prototype.
6970 (aarch64_simd_valid_immediate): Update prototype.
6971 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6972 support for ORR-immediate.
6973 (and<mode>3): modified pattern to add support for BIC-immediate.
6974 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6975 now checks for valid immediate for BIC and ORR based on new enum
6977 (aarch64_output_simd_mov_immediate): Function now used to output
6978 BIC/ORR imm as well based on new enum argument.
6979 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6981 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6982 (aarch64_reg_or_bic_imm): Likewise.
6984 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6986 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6987 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6989 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6993 PR rtl-optimization/82396
6994 * haifa-sched.c (autopref_multipass_init): Simplify
6996 (autopref_rank_data): Simplify sort order.
6997 * sched-int.h (autopref_multipass_data_): Remove
6998 multi_mem_insn_p, min_offset and max_offset.
7000 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7002 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7004 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7006 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7009 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7011 * doc/sourcebuild.texi: Document vect_long_mult.
7013 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7015 PR tree-optimization/82413
7016 * fold-const.c (build_range_check): Use widest_int when comparing
7017 the maximum ETYPE value with HIGH.
7019 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7021 PR rtl-optimization/82396
7022 * haifa-sched.c (autopref_multipass_init): Simplify
7024 (autopref_rank_data): Simplify sort order.
7025 * sched-int.h (autopref_multipass_data_): Remove
7026 multi_mem_insn_p, min_offset and max_offset.
7028 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7030 PR tree-optimization/82381
7031 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7032 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7035 PR tree-optimization/82374
7036 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7037 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7038 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7039 current_function_decl to the new decl.
7041 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7043 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7044 helper macro for IEEE float128 hardware built-in functions.
7045 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7047 (TRUNCF128_ODD): Likewise.
7048 (ADDF128_ODD): Likewise.
7049 (SUBF128_ODD): Likewise.
7050 (MULF128_ODD): Likewise.
7051 (DIVF128_ODD): Likewise.
7052 (FMAF128_ODD): Likewise.
7053 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7054 UNSPEC_TRUNC_ROUND_TO_ODD.
7055 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7056 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7057 floating point round to odd instructions.
7058 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7059 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7060 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7061 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7062 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7063 (trunc<mode>sf2_hw): Change the truncate with round to odd
7064 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7065 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7066 to odd hardware instructions.
7067 (sub<mode>3_odd): Likewise.
7068 (mul<mode>3_odd): Likewise.
7069 (div<mode>3_odd): Likewise.
7070 (sqrt<mode>2_odd): Likewise.
7071 (fma<mode>4_odd): Likewise.
7072 (fms<mode>4_odd): Likewise.
7073 (nfma<mode>4_odd): Likewise.
7074 (nfms<mode>4_odd): Likewise.
7075 (trunc<mode>df2_odd): Change the truncate with round to odd
7076 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7078 * doc/extend.texi (PowerPC built-in functions): Update documentation
7079 for existing IEEE float128-bit built-in functions. Add built-in
7080 functions that generate the IEEE 128-bit floating point round to
7083 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7085 PR rtl-optimization/77729
7086 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7087 to (X&(C1&~C2))|C2 transformations.
7089 2017-10-03 Martin Jambor <mjambor@suse.cz>
7091 PR tree-optimization/82363
7092 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7093 mismatch, mark lacc written regardless of racc.
7095 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7097 PR tree-optimization/82381
7098 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7099 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7100 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7101 fallthrough into reassoc_stmt_dominates_stmt_p.
7104 * combine.c (combine_instructions): Don't combine in unreachable
7107 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7110 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7111 function to not use the have_cpu variable. Do not set cpu_index,
7112 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7114 (rs6000_valid_attribute_p): Remove duplicate initializations of
7115 old_optimize and func_optimize.
7116 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7117 (rs6000_activate_target_options): Make global.
7118 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7121 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7123 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7124 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7125 Set *poffset to 0 and *psize and *pmax_size to -1 if
7126 *poffset + *psize overflows in HOST_WIDE_INT.
7128 PR tree-optimization/82387
7129 PR tree-optimization/82388
7130 PR tree-optimization/82389
7131 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7132 instead of live_bytes non-NULL.
7134 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7137 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7140 2017-10-02 Richard Biener <rguenther@suse.de>
7142 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7143 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7144 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7146 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7148 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7149 requested precision matches the type's.
7150 * calls.c (alloc_max_size): Calculate the new candidate size as
7151 a widest_int and use wi::to_widest when comparing it with the
7152 current candidate size.
7153 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7154 zero rather than integer_zero_node.
7155 * match.pd: Check for a no-op conversion before using wi::add
7156 rather than after. Use tree_to_uhwi when summing small shift
7157 counts into an unsigned int.
7159 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7160 Alan Hayward <alan.hayward@arm.com>
7161 David Sherwood <david.sherwood@arm.com>
7164 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7165 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7166 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7167 POINTER_AND_FP_REGS.
7169 2017-10-02 Richard Biener <rguenther@suse.de>
7171 PR tree-optimization/82355
7172 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7173 a mapping for the enclosing loop but avoid generating one for
7175 (copy_bb_and_scalar_dependences): Remove premature codegen
7176 error on PHIs in blocks duplicated into multiple places.
7177 * graphite-scop-detection.c
7178 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7179 in the region use it as loop and nest to analyze the DR in.
7180 (try_generate_gimple_bb): Likewise.
7181 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7182 (add_loop_constraints): For blocks in a loop not in the region
7183 create a dimension with a single iteration.
7184 * sese.h (gbb_loop_at_index): Remove assert.
7186 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7188 * omp-expand.c (adjust_context_scope): New function.
7189 (expand_parallel_call): Call adjust_context_scope.
7191 2017-10-01 Jeff Law <law@redhat.com>
7193 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7194 dom_opt_dom_walker class with direct access to private members.
7195 Add comments. Call test_for_singularity.
7196 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7197 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7198 m_dummy_cond anymore.
7199 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7201 (pass_dominator:execute): Build the dummy_cond here and pass it
7202 to the dom_opt_dom_walker ctor.
7203 (test_for_singularity): New function.
7205 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7206 Maya Rashish <coypu@sdf.org>
7208 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7209 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7210 (alpha*-*-netbsd*) Use nbsd_tm_file.
7211 (arm*-*-netbsdelf*) Likewise.
7212 (i[34567]86-*-netbsdelf*) Likewise.
7213 (x86_64-*-netbsd*) Likewise.
7214 (mips*-*-netbsd*) Likewise.
7215 (powerpc-*-netbsd*) Likewise.
7216 (sh*-*-netbsd*) Likewise.
7217 (sparc-*-netbsdelf*) Likewise.
7218 (sparc64-*-netbsd*) Likewise.
7219 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7221 (vax-*-netbsdelf*) Likewise.
7222 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7223 (UINT_FAST8_TYPE) Likewise.
7224 (INT_FAST16_TYPE) Check CHAR_FAST16.
7225 (UINT_FAST16_TYPE) Likewise.
7227 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7230 * config/i386/i386.md
7231 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7232 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7233 *divmodsi4_zext_2): New define_insn_and_split.
7234 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7235 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7236 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7237 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7238 New define_insn_and_split.
7239 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7240 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7241 operands[1] having DImode when mode is SImode.
7243 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7244 always SImode for DIV and MOD in REG_EQUAL notes.
7246 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7249 * match.pd: Fix handling of NaNs in pattern.
7251 2017-09-29 Jeff Law <law@redhat.com>
7253 * sbitmap.c (bitmap_bit_in_range_p): New function.
7254 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7255 * tree-ssa-dse.c (live_bytes_read): New function.
7256 (dse_classify_store): Ignore reads of dead bytes.
7258 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7259 typos and whitespace errors.
7260 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7261 * config/s390/s390.c (s390_emit_prologue): Likewise.
7263 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7266 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7267 with a symbol for LRA.
7269 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7271 PR rtl-optimization/82338
7272 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7274 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7276 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7277 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7278 (qsort_chk): Declare.
7279 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7280 (qsort_chk): New function.
7282 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7284 PR tree-optimization/82337
7285 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7286 phi definition if the PHI result appears in an abnormal PHI.
7287 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7288 basis appears in an abnormal PHI.
7290 2017-09-29 Richard Biener <rguenther@suse.de>
7292 * graphite-isl-ast-to-gimple.c
7293 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7294 (binary_op_to_tree): Use it.
7295 (get_rename_from_scev): Likewise.
7296 (copy_loop_phi_nodes): Likewise.
7297 (copy_bb_and_scalar_dependences): Likewise.
7298 (translate_pending_phi_nodes): Likewise.
7300 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7303 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7304 for movabsq $(i32 << shift), r64.
7306 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7308 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7309 index when encoding %esp as %rsp to avoid 0x67 prefix.
7311 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7313 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7314 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7316 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7318 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7319 Extensions with more than 16 double VFP registers.
7320 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7321 to_clear_mask and all code related to it. Replace the remaining
7322 entry by a sbitmap and adapt code accordingly.
7324 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7326 * brig-builtins.def: Change pure attributes to const.
7328 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7330 * config.gcc (default_gnu_indirect_function): Default to yes for
7331 sparc*-*-linux* with glibc.
7333 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7335 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7336 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7337 when creating .init_array and .fini_array sections with priority
7340 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7343 * config/aarch64/aarch64.c
7344 (aarch64_builtin_support_vector_misalignment): Always return false
7345 when misalignment is unknown.
7347 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7349 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7350 this function to return false if the definition used by the swap
7351 instruction is artificial, or if the memory address from which the
7352 constant value is loaded is not represented by a base address held
7353 in a register or if the base address register is a frame or stack
7354 pointer. Additionally, return false if the base address of the
7355 loaded constant is a SYMBOL_REF but is not considered to be a
7357 (replace_swapped_load_constant): New function.
7358 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7359 loaded constant vector with a load of a swapped constant vector.
7361 2017-09-27 Carl Love <cel@us.ibm.com>
7363 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7364 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7365 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7368 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7370 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7371 first, always call autopref_rank_data otherwise.
7373 2017-09-27 Richard Biener <rguenther@suse.de>
7375 * graphite-scop-detection.c (find_scop_parameters): Move
7376 loop bound handling ...
7377 (gather_bbs::before_dom_children): ... here, avoiding the need
7378 to build scop_info->loop_nest.
7379 (record_loop_in_sese): Remove.
7380 * sese.h (sese_info_t::loop_nest): Remove.
7381 * sese.c (new_sese_info): Do not allocate loop_nest.
7382 (free_sese_info): Do not free loop_nest.
7384 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7387 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7388 lhs from calls if the lhs has addressable type.
7390 2017-09-27 Richard Biener <rguenther@suse.de>
7392 * graphite.h (scop::max_alias_set): New member.
7393 * graphite-scop-detection.c: Remove references to non-existing
7394 --param in comments.
7395 (build_alias_sets): Record the maximum alias set used for drs.
7396 (build_scops): Support zero as unlimited for
7397 --param graphite-max-arrays-per-scop.
7398 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7400 (build_poly_sr_1): ... here. Compute alias set based on the
7401 maximum alias set used for drs rather than
7402 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7404 2017-09-27 Richard Biener <rguenther@suse.de>
7406 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7407 --param loop-block-tile-size=0 to disable tiling.
7409 2017-09-27 Richard Biener <rguenther@suse.de>
7411 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7412 (graphite-max-nb-scop-params): Document special value zero.
7413 * domwalk.h (dom_walker::STOP): New symbolical constant.
7414 (dom_walker::dom_walker): Add optional parameter for bb to
7416 (dom_walker::~dom_walker): Declare.
7417 (dom_walker::before_dom_children): Document STOP return value.
7418 (dom_walker::m_user_bb_to_rpo): New member.
7419 (dom_walker::m_bb_to_rpo): Likewise.
7420 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7421 mapping here if not provided by the user.
7422 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7423 provided by the user.
7424 (dom_walker::STOP): Define.
7425 (dom_walker::walk): Do not compute bb to RPO mapping here.
7426 Support STOP return value from before_dom_children to stop
7428 * graphite-optimize-isl.c (optimize_isl): If the schedule
7429 is the same still generate code if -fgraphite-identity
7430 or -floop-parallelize-all are given.
7431 * graphite-scop-detection.c: Include cfganal.h.
7432 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7434 (gather_bbs::before_dom_children): Return STOP for BBs
7436 (build_scops): Compute bb to RPO mapping and pass it to
7437 the domwalk. Treat --param graphite-max-nb-scop-params=0
7438 as not limiting the number of params.
7439 * graphite.c (graphite_initialize): Remove limit on the
7440 number of basic-blocks in a function.
7441 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7442 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7443 default value of 10.
7445 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7447 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7448 Adjust code to eliminate needing to do the shift right 32-bits
7449 operation after XSCVDPSPN.
7451 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7453 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7454 ((X / Y) != 0 -> X >= Y): Likewise.
7456 2017-09-26 Carl Love <cel@us.ibm.com>
7458 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7459 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7460 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7461 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7462 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7463 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7464 definitions and overloading.
7465 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7466 statement for P9V_BUILTIN_XST_LEN_R.
7467 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7468 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7469 define_expand and define_insn for the instructions and builtins.
7470 * doc/extend.texi: Update the built-in documentation file for the new
7472 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7473 define_insn for the instructions
7475 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7478 * gcc/config/netbsd-protos.h: New file.
7479 * gcc/config/netbsd.c: New file.
7480 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7481 * gcc/config/t-netbsd: New file.
7482 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7483 (tmake_file) Add t-netbsd.
7484 (extra_objs) Add netbsd.o.
7486 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7490 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7492 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7494 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7495 sign extension from a vector register to a GPR by doing a 32-bit
7496 direct move and then an EXTSW.
7497 (extendsi<mode>2 splitter): Likewise.
7498 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7499 right or vector extract after doing XSCVDPSPN. Use
7500 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7502 (movdi_from_sf_zero_ext): Likewise.
7503 (reload_gpr_from_vsxsf): Likewise.
7504 (p8_mfvsrd_4_disf): Delete, no longer used.
7505 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7506 then needing to move the SFmode to a GPR to use the XSCVDPSP
7507 instruction instead of FRSP and XSCVDPSPN.
7508 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7509 it is adjacent to the other XSCVSPDP insns.
7510 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7511 SFmode to be in traditional Altivec registers.
7512 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7513 (vsx_xscvspdpn): Likewise.
7514 (vsx_xscvspdpn_scalar): Likewise.
7516 2017-09-26 Martin Jambor <mjambor@suse.cz>
7518 * tree-sra.c (compare_access_positions): Put integral types first,
7519 stabilize sorting of integral types, remove conditions putting
7520 non-full-precision integers last.
7521 (sort_and_splice_var_accesses): Disable scalarization if a
7522 non-integert would be represented by a non-full-precision integer.
7524 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7526 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7527 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7528 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7529 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7530 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7531 conditionals inside the function instead of around it. Call
7532 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7533 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7535 2017-09-26 Richard Biener <rguenther@suse.de>
7537 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7539 (scop_detection::build_scop_breadth): ... this. Removed.
7540 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7541 (scop_detection::harmful_stmt_in_bb): Likewise.
7542 (scop_detection::graphite_can_represent_stmt): Likewise.
7543 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7544 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7545 (scop_detection::can_represent_loop_1): ... this. Removed.
7546 (scop_detection::harmful_loop_in_region): Simplify after inlining
7547 the above and remove more quadraticness.
7548 (build_scops): Adjust.
7549 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7552 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7555 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7556 REGNO (base) == SP_REG if base is a REG.
7559 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7560 if it is different SSA_NAME.
7561 (optimize_range_tests_cmp_bitwise): New function.
7562 (optimize_range_tests): Call it.
7564 2017-09-26 Richard Biener <rguenther@suse.de>
7566 PR tree-optimization/82321
7567 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7568 for the def being inside the loop.
7570 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7572 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7574 * config/s390/s390-builtins.def: Fix constraint on op4.
7576 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7578 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7579 independent expanders.
7580 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7581 ("vec_ordered", "vec_unordered"): New expanders.
7583 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7585 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7588 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7590 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7591 vec_unpacks_lo_v16qi.
7592 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7594 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7596 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7597 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7598 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7600 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7602 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7604 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7606 ("*vec_slb<mode>"): New insn pattern.
7607 ("vec_shr_<mode>"): New expander.
7608 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7609 and force the shift count operand to V16QImode.
7610 ("vec_srb<mode>"): Set shift count mode to V16QI.
7612 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7614 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7615 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7616 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7618 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7621 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7623 2017-09-26 Richard Biener <rguenther@suse.de>
7625 PR tree-optimization/82320
7626 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7629 2017-09-25 Jeff Law <law@redhat.com>
7631 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7632 prototype for new argument.
7633 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7634 mostly extracted from rs6000_emit_allocate_stack.
7635 (rs6000_emit_probe_stack_range_stack_clash): New function.
7636 (rs6000_emit_allocate_stack): Call
7637 rs6000_emit_probe_stack_range_stack_clash as needed.
7638 (rs6000_emit_probe_stack_range): Add additional argument
7639 to call to gen_probe_stack_range{si,di}.
7640 (output_probe_stack_range): New.
7641 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7642 (output_probe_stack_range_stack_clash): New.
7643 (rs6000_emit_prologue): Emit notes into dump file as requested.
7644 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7645 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7646 Add additional operand and pass it to output_probe_stack_range.
7648 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7650 PR tree-optimization/82163
7651 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7652 (checking_verify_loop_closed_ssa): New parameter.
7653 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7654 (check_loop_closed_ssa_stmt): Delete.
7655 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7656 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7657 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7660 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7662 * brig-builtins.def: Treat HSAIL barrier builtins as
7663 setjmp/longjump style functions.
7665 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7667 * target.def (constant_alignment): New hook.
7668 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7669 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7670 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7671 * doc/tm.texi: Regenerate.
7672 * targhooks.h (default_constant_alignment): Declare.
7673 (constant_alignment_word_strings): Likewise.
7674 * targhooks.c (default_constant_alignment): New function.
7675 (constant_alignment_word_strings): Likewise.
7676 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7677 instead of CONSTANT_ALIGNMENT.
7678 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7679 (force_const_mem): Likewise.
7680 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7681 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7682 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7683 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7684 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7686 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7687 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7688 constant_alignment_word_strings.
7689 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7690 (CONSTANT_ALIGNMENT): Likewise.
7691 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7692 (arm_constant_alignment): New function.
7693 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7694 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7695 constant_alignment_word_strings.
7696 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7697 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7698 constant_alignment_word_strings.
7699 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7700 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7701 (cris_constant_alignment): New function.
7702 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7703 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7704 (epiphany_constant_alignment): New function.
7705 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7706 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7707 constant_alignment_word_strings.
7708 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7709 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7710 constant_alignment_word_strings.
7711 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7712 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7713 constant_alignment_word_strings.
7714 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7715 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7716 * config/i386/i386.c (ix86_constant_alignment): Make static.
7717 Use the same interface as the target hook.
7718 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7719 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7720 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7721 constant_alignment_word_strings.
7722 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7723 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7724 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7725 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7726 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7727 constant_alignment_word_strings.
7728 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7729 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7730 constant_alignment_word_strings.
7731 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7732 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7733 constant_alignment_word_strings.
7734 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7735 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7737 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7738 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7739 * config/mips/mips.c (mips_constant_alignment): New function.
7740 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7741 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7742 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7743 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7744 (mmix_constant_alignment): Make static. Use the same interface
7746 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7747 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7748 constant_alignment_word_strings.
7749 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7750 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7751 constant_alignment_word_strings.
7752 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7753 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7754 constant_alignment_word_strings.
7755 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7756 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7757 (rs6000_constant_alignment): New function.
7758 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7759 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7760 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7761 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7762 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7763 (rs6000_constant_alignment): New function.
7764 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7765 * config/s390/s390.c (s390_constant_alignment): New function.
7766 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7767 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7768 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7769 constant_alignment_word_strings.
7770 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7771 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7772 (sparc_constant_alignment): New function.
7773 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7774 * config/spu/spu.c (spu_constant_alignment): New function.
7775 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7776 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7777 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7778 constant_alignment_word_strings.
7779 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7780 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7781 constant_alignment_word_strings.
7782 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7783 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7784 constant_alignment_word_strings.
7785 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7786 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7787 (visium_constant_alignment): New function.
7788 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7789 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7790 (xtensa_constant_alignment): New function.
7791 * system.h (CONSTANT_ALIGNMENT): Poison.
7793 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7795 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7796 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7797 (rs6000_builtin_valid_without_lhs): New helper function.
7798 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7799 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7801 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7803 * target.h (vec_perm_indices): Use unsigned short rather than
7805 (auto_vec_perm_indices): Likewise.
7806 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7807 Use unsigned int rather than unsigned char.
7808 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7810 2017-09-25 Richard Biener <rguenther@suse.de>
7812 * cfgloop.h (sort_sibling_loops): Declare.
7813 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7814 (sort_sibling_loops): New function sorting the sibling loop list
7816 * graphite.c (graphite_transform_loops): Sort sibling loops.
7818 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7820 * target.def (vec_perm_const_ok): Change sel parameter to
7822 * optabs-query.c (can_vec_perm_p): Update accordingly.
7823 * doc/tm.texi: Regenerate.
7824 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7825 auto_vec_perm_indices and remove separate nelt field.
7826 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7827 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7828 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7829 (aarch64_expand_vec_perm_const): Update accordingly.
7830 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7831 to vec_perm_indices.
7832 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7833 auto_vec_perm_indices and remove separate nelt field.
7834 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7835 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7836 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7838 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7839 to vec_perm_indices.
7840 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7841 sel to vec_perm_indices.
7842 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7843 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7844 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7846 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7849 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7852 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7853 on the FUNCTION_DECL function context if it has a DIE that is a
7856 2017-09-25 Richard Biener <rguenther@suse.de>
7858 PR tree-optimization/82285
7859 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7862 2017-09-25 Tom de Vries <tom@codesourcery.com>
7866 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7869 2017-09-25 Richard Biener <rguenther@suse.de>
7871 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7872 ISL errors other than isl_error_quota happen. Dump if the
7873 schedule is the same.
7874 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7875 errors instead of aborting inside ISL.
7877 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7880 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7881 of libgcc_eh for m64.
7882 * config/i386/darwin64.h: Likewise.
7884 2017-09-25 Richard Biener <rguenther@suse.de>
7887 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7888 attribute for incomplete types nor twice for complete ones.
7890 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7893 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7894 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7897 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7900 * config/i386/i386.opt (mprefer-avx256): Use
7901 ix86_target_flags variable.
7902 * config/i386/i386.c (ix86_target_string): Move
7903 -mprefer-avx256 to flag2_opts.
7905 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7908 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7909 and x != -1 | y != -1 into (x & y) != -1.
7911 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7913 * config.gcc: Add new case statement to set
7914 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7915 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7916 s390x-*-linux* case statements. Added aarch64 to the list of
7917 supported architectures.
7919 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7921 PR tree-optimization/82289
7922 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7923 STMT_VINFO_RELEVANT_P.
7925 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7926 Alan Hayward <alan.hayward@arm.com>
7927 David Sherwood <david.sherwood@arm.com>
7929 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7930 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7931 (extract_range_from_binary_expr_1): Don't call
7932 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7934 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7935 Alan Hayward <alan.hayward@arm.com>
7936 David Sherwood <david.sherwood@arm.com>
7938 * target.def (preferred_vector_alignment): New hook.
7939 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7941 * doc/tm.texi: Regenerate.
7942 * targhooks.h (default_preferred_vector_alignment): Declare.
7943 * targhooks.c (default_preferred_vector_alignment): New function.
7944 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7946 (DR_TARGET_ALIGNMENT): New macro.
7947 (aligned_access_p): Update commentary.
7948 (vect_known_alignment_in_bytes): New function.
7949 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7951 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7952 Calculate the misalignment based on the target alignment rather than
7954 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7955 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7956 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7957 the target alignment, rather than masking the element misalignment
7958 with the number of elements in a vector. Also use the target
7959 alignment when calculating the maximum number of peels.
7960 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7961 instead of TYPE_ALIGN_UNIT.
7962 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7963 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7964 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7965 (vect_create_data_ref_ptr): Likewise.
7966 (vect_setup_realignment): Realign by ANDing with
7967 -DR_TARGET_MISALIGNMENT.
7968 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7969 the number of peels based on DR_TARGET_ALIGNMENT.
7970 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7971 with the guaranteed alignment boundary when deciding whether
7973 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7974 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7975 (ensure_base_align): Remove stmt_info parameter. Get the
7976 target base alignment from DR_TARGET_ALIGNMENT.
7977 (vectorizable_store): Update call accordingly. Interpret
7978 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7980 (vectorizable_load): Likewise.
7982 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7983 Alan Hayward <alan.hayward@arm.com>
7984 David Sherwood <david.sherwood@arm.com>
7986 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7987 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7988 (vect_enhance_data_refs_alignment): Likewise.
7990 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7992 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7993 error. Only quit immediately if parsing is complete.
7994 (BEGIN): Initialize fatal_err and parse_done.
7995 (begin fpu, end fpu): Check number of arguments.
7996 (begin arch, end arch): Likewise.
7997 (begin cpu, end cpu): Likewise.
7998 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7999 (optalias): Likewise.
8001 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8003 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8004 * config/arm/arm-isa.h: Delete. Move definitions to ...
8005 * arm-cpus.in: ... here. Use new feature and fgroup values.
8006 * config/arm/arm.c (arm_option_override): Use lower case for feature
8008 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8009 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8010 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8012 (print_isa_bits_for): New function.
8013 (gen_isa): New function.
8014 (gen_comm_data): Use print_isa_bits_for.
8015 (define feature): New keyword.
8016 (define fgroup): New keyword.
8017 * config/arm/t-arm (TM_H): Remove.
8018 (GTM_H): Add arm-isa.h.
8019 (arm-isa.h): Add rule to generate file.
8020 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8021 case for feature bit names.
8023 2017-09-22 Richard Biener <rguenther@suse.de>
8025 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8027 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8028 print of no dependency loops ...
8029 * graphite.c (graphite_transform_loops): ... here.
8030 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8032 (same_close_phi_node, remove_duplicate_close_phi,
8033 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8034 (canonicalize_loop_closed_ssa): ... here and simplify.
8035 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8036 (optimize_isl): Use dump_printf_loc to tell when we stopped
8037 optimizing because of an ISL timeout.
8039 2017-09-22 Richard Biener <rguenther@suse.de>
8041 PR tree-optimization/82291
8042 * tree-if-conv.c (predicate_mem_writes): Make sure to
8043 remove writes in blocks predicated with false.
8045 2017-09-22 Richard Biener <rguenther@suse.de>
8047 * sese.c: Include cfganal.h.
8048 (if_region_set_false_region): Remove.
8049 (create_if_region_on_edge): Likewise.
8050 (move_sese_in_condition): Re-implement without destroying
8053 2017-09-22 Richard Biener <rguenther@suse.de>
8055 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8056 Verify both BBs contain loop PHI nodes before dispatching to
8058 (graphite_regenerate_ast_isl): Do not recompute dominators,
8059 do not verify three times. Restructure for clarity.
8060 * graphite-scop-detection.c (same_close_phi_node,
8061 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8062 defined_in_loop_p, canonicalize_loop_closed_ssa,
8063 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8064 checking and SSA rewrite, move to ...
8065 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8066 (graphite_initialize): Do not pass in ctx, do not reset the
8067 SCEV cache, compute only dominators.
8068 (graphite_transform_loops): Allocate ISL ctx after
8069 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8070 Maintain post-dominators only around build_scops.
8071 * sese.c (if_region_set_false_region): Make static. Free
8072 and recompute dominators.
8073 (move_sese_in_condition): Assert we don't get called with
8074 post-dominators computed.
8075 * sese.h (if_region_set_false_region): Remove.
8077 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8079 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8080 mode attribute for TARGET_AVX512VL.
8082 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8084 * config/i386/i386.opt (mprefer-avx256): New option.
8085 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8087 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8088 for TARGET_PREFER_AVX256.
8089 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8091 2017-09-21 Jeff Law <law@redhat.com>
8093 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8094 Fix dump output if the only stack space is for pushed registers.
8096 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8098 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8101 2017-09-21 Martin Sebor <msebor@redhat.com>
8104 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8105 code (in C++) or code that triggers warnings.
8107 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8109 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8111 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8113 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8114 * sched-rgn.c: Ditto.
8115 * sel-sched-ir.c: Ditto.
8117 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8119 * toplev.h (set_random_seed): Adjust return type.
8120 * toplev.c (init_local_tick): Move eager initialization of random_seed
8121 to get_random_seed. Adjust comment.
8122 (init_random_seed): Inline to get_random_seed, delete.
8123 (get_random_seed): Initialize random_seed lazily.
8124 (set_random_seed): Do not return previous value.
8125 (print_switch_value): Do not call get_random_seed.
8127 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8129 * cgraph.c (delete_function_version): New, broken out from...
8130 (cgraph_node::delete_function_version): ...here. Rename to
8131 cgraph_node::delete_function_version_by_decl. Update all uses.
8132 (cgraph_node::remove): Call delete_function_version.
8134 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8137 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8138 VAR_DECLs to which addressable non-volatile parameters are mapped
8139 and for id->retvar after the return value assignment. Clear
8140 id->retval and id->retbnd after inlining.
8142 2017-09-21 Richard Biener <rguenther@suse.de>
8144 PR tree-optimization/82276
8145 PR tree-optimization/82244
8146 * tree-vrp.c (build_assert_expr_for): Set
8147 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8149 (remove_range_assertions): Revert earlier change.
8151 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8154 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8156 2017-09-21 Richard Biener <rguenther@suse.de>
8158 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8159 Restore valid IL after code generation errors.
8160 * graphite.c (graphite_transform_loops): Diagnose code
8161 generation issues as MSG_MISSED_OPTIMIZATION and continue
8162 with processing SCOPs.
8164 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8165 Alan Hayward <alan.hayward@arm.com>
8166 David Sherwood <david.sherwood@arm.com>
8168 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8169 rather than choosing between plus_constant and gen_rtx_<CODE>.
8170 * expr.c (emit_push_insn): Likewise.
8171 (expand_expr_real_2): Likewise.
8173 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8174 Alan Hayward <alan.hayward@arm.com>
8175 David Sherwood <david.sherwood@arm.com>
8177 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8179 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8180 Alan Hayward <alan.hayward@arm.com>
8181 David Sherwood <david.sherwood@arm.com>
8183 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8184 calling tree_to_uhwi.
8186 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8187 Alan Hayward <alan.hayward@arm.com>
8188 David Sherwood <david.sherwood@arm.com>
8190 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8191 INTEGER_CST rather than a negative test for ADDR_EXPR.
8193 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8194 Alan Hayward <alan.hayward@arm.com>
8195 David Sherwood <david.sherwood@arm.com>
8197 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8198 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8200 2017-09-21 Richard Biener <rguenther@suse.de>
8202 PR tree-optimization/71351
8203 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8204 graphite_create_new_loop_guard): Remove, fold remaining parts
8206 (translate_isl_ast_node_for): ... here and simplify.
8208 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8211 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8212 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8213 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8214 alternative always use QI mode, for -Os imov (=R,R) alternative
8215 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8218 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8219 Jeff Law <law@redhat.com>
8221 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8222 (allocate_stack_space): New function, partially extracted from
8224 (s390_emit_prologue): Track offset to most recent stack probe.
8225 Code to allocate space moved into allocate_stack_space.
8226 Dump actions when no stack is allocated.
8227 (s390_prologue_plus_offset): New function.
8228 (s390_emit_stack_probe): Likewise.
8230 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8232 * common.opt (Wa, Wl, Wp, g, gz=): Add
8234 (gno-column-info): Remove.
8235 (gcolumn-info): Drop RejectNegative.
8237 (gno-record-gcc-switches): Remove.
8238 (grecord-gcc-switches): Drop RejectNegative.
8239 (gno-split-dwarf): Remove.
8240 (gsplit-dwarf): Drop RejectNegative.
8241 (gno-strict-dwarf): Remove.
8242 (gstrict-dwarf): Drop RejectNegative.
8243 * config/darwin.opt (gfull, gused): Add RejectNegative.
8244 * dwarf2out.c (gen_producer_string): Drop
8245 gno-record-gcc-switches handler.
8246 * optc-gen.awk: Add g to prefixes with negative forms.
8247 * opts-common.c (remapping_prefix_p): New.
8248 (find_opt): Check it.
8249 (generate_canonical_option): Test g prefix.
8250 (option_map): Add -gno- mapping.
8251 (add_misspelling_candidates): Check remapping_prefix_p.
8253 2017-09-20 Jeff Law <law@redhat.com>
8255 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8256 thinko in stack clash protection support.
8258 * explow.c (compute_stack_clash_protection_loop_data): Use
8259 CONST_INT_P instead of explicit test. Verify object is a
8260 CONST_INT_P before looking at INTVAL.
8261 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8262 instead of explicit test.
8264 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8267 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8268 address instead of to r1 and r11.
8270 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8272 * config.gcc: Support "knm".
8273 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8274 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8276 * config/i386/i386.c (m_KNM): Define.
8277 (processor_target_table): Add "knm".
8279 (ix86_option_override_internal): Add "knm".
8280 (ix86_issue_rate): Add PROCESSOR_KNM.
8281 (ix86_adjust_cost): Ditto.
8282 (ia32_multipass_dfa_lookahead): Ditto.
8283 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8284 (fold_builtin_cpu): Add M_INTEL_KNM.
8285 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8286 (processor_type): Add PROCESSOR_KNM.
8287 * config/i386/x86-tune.def: Add m_KNM.
8288 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8290 2017-09-20 Richard Biener <rguenther@suse.de>
8292 PR tree-optimization/80213
8293 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8294 are allowed in empty BBs as well.
8295 (canonicalize_loop_closed_ssa): Also look for other complex
8297 (scop_detection::get_sese): Include the loop-closed PHI block
8299 (scop_detection::merge_sese): Remove code adding extra blocks.
8300 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8301 (build_scops): Assert the final returned scop is invalid.
8303 2017-09-20 Richard Biener <rguenther@suse.de>
8305 PR tree-optimization/82264
8306 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8308 (vn_phi_lookup): Likewise.
8309 (vn_phi_insert): Likewise.
8311 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8313 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8314 that fits into uhwi or shwi, add DW_AT_const_value regardless
8315 of early_dwarf without going through RTL, using add_AT_unsigned
8318 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8319 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8320 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8321 (DEBUG_MACRO_SECTION): Likewise.
8322 (DEBUG_LTO_MACRO_SECTION): Likewise.
8323 (DEBUG_STR_DWO_SECTION): Likewise.
8324 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8325 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8326 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8327 (DEBUG_LTO_LINE_STR_SECTION): Define.
8328 (init_sections_and_labels): Initialize debug_line_str_section
8329 variable. Initialize debug_loc_section for -gdwarf-5 to
8330 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8332 2017-09-20 Richard Biener <rguenther@suse.de>
8334 * graphite-sese-to-poly.c (extract_affine): Properly handle
8335 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8337 2017-09-20 Richard Biener <rguenther@suse.de>
8339 PR tree-optimization/81373
8340 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8341 Force SESE live-out defs to be handled even if they are
8344 2017-09-19 Jeff Law <law@redhat.com>
8346 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8347 nothing for stack adjustments with REG_STACK_CHECK.
8348 * sched-deps.c (parse_add_or_inc): Reject insns with
8349 REG_STACK_CHECK from dependency breaking.
8350 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8351 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8352 * reg-notes.def (STACK_CHECK): New note.
8354 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8355 (ix86_expand_prologue): Dump stack clash info as needed.
8356 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8358 * function.c (dump_stack_clash_frame_info): New function.
8359 * function.h (dump_stack_clash_frame_info): Prototype.
8360 (enum stack_clash_probes): New enum.
8362 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8363 flag_stack_clash_protection.
8364 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8365 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8366 (arm_frame_pointer_required): Likewise.
8367 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8368 (ia64_expand_prologue): Likewise.
8369 * config/mips/mips.c (mips_expand_prologue): Likewise.
8370 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8371 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8372 (sparc_flat_expand_prologue): Likewise.
8373 * config/spu/spu.c (spu_expand_prologue): Likewise.
8375 * explow.c: Include "params.h".
8376 (anti_adjust_stack_and_probe_stack_clash): New function.
8377 (get_stack_check_protect): Likewise.
8378 (compute_stack_clash_protection_loop_data): Likewise.
8379 (emit_stack_clash_protection_loop_start): Likewise.
8380 (emit_stack_clash_protection_loop_end): Likewise.
8381 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8382 Use anti_adjust_stack_and_probe_stack_clash.
8383 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8384 (emit_stack_clash_protection_loop_start): Likewise.
8385 (emit_stack_clash_protection_loop_end): Likewise.
8386 * rtl.h (get_stack_check_protect): Prototype.
8387 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8388 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8389 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8391 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8393 * doc/tm.texi: Rebuilt.
8394 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8395 get_stack_check_protect.
8396 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8397 * config/arm/arm.c (arm_expand_prologue): Likewise.
8398 (arm_frame_pointer_required): Likewise.
8399 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8400 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8401 * config/mips/mips.c (mips_expand_prologue): Likewise.
8402 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8403 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8404 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8405 (sparc_flat_expand_prologue): Likewise.
8407 * common.opt (-fstack-clash-protection): New option.
8408 * flag-types.h (enum stack_check_type): Note difference between
8409 -fstack-check= and -fstack-clash-protection.
8410 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8411 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8412 * toplev.c (process_options): Issue warnings/errors for cases
8413 not handled with -fstack-clash-protection.
8414 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8415 (-fstack-check): Note additional problem with -fstack-check=generic.
8416 Note that -fstack-check is primarily for Ada and refer users
8417 to -fstack-clash-protection for stack-clash-protection.
8418 Document new params for stack clash protection.
8420 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8422 * config/i386/i386.c (ix86_split_long_move): Do not handle
8423 address used for LEA in a special way.
8425 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8427 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8429 2017-09-19 Martin Sebor <msebor@redhat.com>
8432 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8433 of incompatible types.
8435 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8437 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8438 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8439 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8440 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8442 2017-09-19 Richard Biener <rguenther@suse.de>
8444 PR tree-optimization/82244
8445 * tree-vrp.c (remove_range_assertions): Do not propagate
8446 a constant to abnormals but replace the assert with a copy.
8448 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8450 PR rtl-optimization/57878
8451 PR rtl-optimization/68988
8452 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8453 avoidance test involving non_reload_pseudos. Move frequency test
8454 below the general fragmentation avoidance test.
8456 2017-09-19 Richard Biener <rguenther@suse.de>
8458 PR tree-optimization/69728
8459 * graphite-sese-to-poly.c (schedule_error): New global.
8460 (add_loop_schedule): Handle empty domain by failing the
8462 (build_original_schedule): Handle schedule_error.
8464 2017-09-19 Richard Biener <rguenther@suse.de>
8466 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8467 Do not iterate to sibling loops but only to siblings of inner
8470 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8473 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8474 that the registers are different.
8476 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8478 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8479 to processor_model and "amdfam17h" to arch_names_table.
8480 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8482 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8485 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8487 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8488 Alan Hayward <alan.hayward@arm.com>
8489 David Sherwood <david.sherwood@arm.com>
8491 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8493 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8495 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8496 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8497 vect_schedule_slp_instance.
8498 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8499 Update call to vect_slp_analyze_node_operations. Simplify return
8501 (vect_slp_analyze_bb_1): Update call accordingly.
8502 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8503 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8504 (vect_schedule_slp): Update call accordingly.
8506 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8507 Alan Hayward <alan.hayward@arm.com>
8508 David Sherwood <david.sherwood@arm.com>
8510 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8511 with types that aren't in fact scalar.
8513 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8515 * tree-vect-slp.c (vect_record_max_nunits): New function,
8517 (vect_build_slp_tree_1): ...here.
8518 (vect_build_slp_tree_2): Call it for phis too.
8520 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8522 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8523 to vect_get_vec_def_for_operand when getting the mask operand.
8525 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8526 Alan Hayward <alan.hayward@arm.com>
8527 David Sherwood <david.sherwood@arm.com>
8529 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8532 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8533 Alan Hayward <alan.hayward@arm.com>
8534 David Sherwood <david.sherwood@arm.com>
8536 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8537 calculation for vector booleans.
8539 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8540 Alan Hayward <alan.hayward@arm.com>
8541 David Sherwood <david.sherwood@arm.com>
8543 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8545 (vect_transform_stmt): ...here.
8546 (vect_analyze_stmt): Use it instead of calling
8547 vectorizable_live_operation directly.
8549 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8551 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8552 non-SIMT targets in acc vector loops.
8554 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8556 * configure.ac: Add arc and check if assembler supports gdwarf2.
8557 * configure: Regenerate.
8559 2017-09-18 Richard Biener <rguenther@suse.de>
8561 PR tree-optimization/82220
8562 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8563 epilogue niters from the min_profitable_iters compute.
8565 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8568 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8569 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8570 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8572 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8575 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8576 switching to a new text section.
8578 2017-09-18 Richard Biener <rguenther@suse.de>
8580 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8582 (build_alias_set): Reject aliases with no access function.
8584 2017-09-18 Richard Biener <rguenther@suse.de>
8586 PR tree-optimization/79622
8587 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8589 (build_cross_bb_scalars_use): Likewise.
8591 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8593 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8595 2017-09-18 Alan Modra <amodra@gmail.com>
8598 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8599 stack_pointer_rtx for count 0. Update comments. Break up
8600 large rtl expression.
8602 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8604 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8605 Increase to 20 bytes.
8606 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8607 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8608 or avx version of the stub.
8610 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8613 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8614 compute the minimum stack alignment. Also update preferred stack
8615 boundary for leaf functions.
8617 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8619 PR tree-optimization/82228
8620 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8623 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8625 * common/config/nds32/nds32-common.c
8626 (nds32_option_optimization_table): Refine formatting.
8627 (nds32_option_optimization_table): Use -fsched-pressure and
8628 -fomit-frame-pointer for specific optimization level.
8630 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8632 * config/nds32/nds32.c: Refine formatting and comments.
8633 * config/nds32/nds32.h: Likewise.
8634 * config/nds32/nds32.md: Likewise.
8635 * config/nds32/nds32-cost.c: Likewise.
8636 * config/nds32/nds32-isr.c: Likewise.
8637 * config/nds32/nds32-md-auxiliary.c: Likewise.
8638 * config/nds32/nds32-multiple.md: Likewise.
8639 * config/nds32/nds32-predicates.c: Likewise.
8641 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8642 Jakub Jelinek <jakub@redhat.com>
8644 Add support for -std=c++2a.
8645 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8647 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8649 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8652 * doc/extend.texi (Common Function Attributes): Add
8653 references to ARM, AArch64, and S/390 specific attributes.
8654 (Function Specific Option Pragmas): Add AArch64 and S/390
8655 to list of back ends that support the target pragma.
8657 2017-09-15 Nathan Sidwell <nathan@acm.org>
8659 * doc/standards.texi: Fix C++17 description. Update URLs for
8662 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8664 * common.opt (Wcast-align=strict): New warning option.
8665 * doc/invoke.texi: Document -Wcast-align=strict.
8667 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8669 * cgraph.h (cgraph_thunk_info): Add comments.
8670 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8671 assert for VIRTUAL_* arguments stricter.
8673 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8675 PR tree-optimization/71026
8676 * match.pd: Move RDIV patterns from fold-const.c
8677 * fold-const.c (distribute_real_division): Removed.
8678 (fold_binary_loc): Remove calls to distribute_real_divison.
8680 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8682 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8683 c++1z and gnu++1z as deprecated. Change other references to
8684 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8685 Change -Wc++1z-compat to -Wc++17-compat.
8686 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8687 * dwarf2out.c (highest_c_language): Handle C++17.
8688 (gen_compile_unit_die): Likewise.
8690 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8692 PR rtl-optimization/82192
8693 * combine.c (make_extraction): Don't look through non-paradoxical
8694 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8697 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8698 Alan Hayward <alan.hayward@arm.com>
8699 David Sherwood <david.sherwood@arm.com>
8701 * target.def (function_arg_offset): New hook.
8702 * targhooks.h (default_function_arg_offset): Declare.
8703 * targhooks.c (default_function_arg_offset): New function.
8704 * function.c (locate_and_pad_parm): Use
8705 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8706 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8707 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8708 * doc/tm.texi: Regenerate.
8709 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8710 * config/spu/spu.c (spu_function_arg_offset): New function.
8711 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8712 * system.h (FUNCTION_ARG_OFFSET): Poison.
8714 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8715 Alan Hayard <alan.hayward@arm.com>
8716 David Sherwood <david.sherwood@arm.com>
8718 * target.def (truly_noop_truncation): New hook.
8719 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8720 than TRULY_NOOP_TRUNCATION.
8721 * hooks.h (hook_bool_uint_uint_true): Declare.
8722 * hooks.c (hook_bool_uint_uint_true): New function.
8723 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8724 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8725 * doc/tm.texi: Regenerate.
8726 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8727 rather than TRULY_NOOP_TRUNCATION in comments.
8728 (simplify_comparison): Likewise.
8729 (record_truncated_value): Likewise.
8730 * expmed.c (extract_bit_field_1): Likewise.
8731 (extract_split_bit_field): Likewise.
8732 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8733 instead of TRULY_NOOP_TRUNCATION.
8734 * function.c (assign_parm_setup_block): Likewise.
8735 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8736 * rtlhooks.c: Include target.h.
8737 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8738 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8739 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8740 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8741 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8742 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8743 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8744 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8745 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8746 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8747 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8748 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8749 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8750 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8751 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8752 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8753 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8754 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8755 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8756 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8757 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8758 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8759 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8760 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8761 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8762 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8763 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8764 rather than TRULY_NOOP_TRUNCATION in comments.
8765 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8766 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8767 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8768 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8769 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8770 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8771 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8772 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8773 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8774 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8775 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8776 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8777 rather than TRULY_NOOP_TRUNCATION in comments.
8778 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8779 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8780 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8781 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8782 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8783 TRULY_NOOP_TRUNCATION condition.
8784 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8785 (TRULY_NOOP_TRUNCATION): Delete.
8786 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8787 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8788 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8789 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8790 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8791 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8792 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8793 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8794 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8795 rather than TRULY_NOOP_TRUNCATION in comments.
8796 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8797 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8798 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8799 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8800 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8801 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8803 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8806 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8807 (*cmp_ior): Likewise.
8808 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8809 (*ior_scc_scc_cmp): Likewise.
8810 (*and_scc_scc): Likewise.
8811 (*and_scc_scc_cmp): Likewise.
8813 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8814 Alan Hayard <alan.hayward@arm.com>
8815 David Sherwood <david.sherwood@arm.com>
8817 * target.def (can_change_mode_class): New hook.
8818 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8819 (hard_regno_nregs): Likewise.
8820 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8821 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8822 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8823 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8824 (LOAD_EXTEND_OP): Update accordingly.
8825 * doc/tm.texi: Regenerate.
8826 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8827 CANNOT_CHANGE_MODE_CLASS.
8828 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8829 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8830 * combine.c (simplify_set): Update accordingly.
8831 * emit-rtl.c (validate_subreg): Likewise.
8832 * recog.c (general_operand): Likewise.
8833 * regcprop.c (mode_change_ok): Likewise.
8834 * reload1.c (choose_reload_regs): Likewise.
8835 (inherit_piecemeal_p): Likewise.
8836 * rtlanal.c (simplify_subreg_regno): Likewise.
8837 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8838 instead of CANNOT_CHANGE_MODE_CLASS.
8839 (reload_cse_simplify_operands): Likewise.
8840 * reload.c (push_reload): Use targetm.can_change_mode_class
8841 instead of CANNOT_CHANGE_MODE_CLASS.
8842 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8843 REG_CANNOT_CHANGE_MODE_P.
8844 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8845 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8846 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8847 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8848 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8849 (arm_can_change_mode_class): New function.
8850 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8851 than CANNOT_CHANGE_MODE_CLASS in comments.
8852 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8853 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8854 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8855 (ix86_can_change_mode_class): ...this new function, inverting the
8856 sense of the return value.
8857 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8858 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8859 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8860 (ia64_can_change_mode_class): New function.
8861 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8862 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8863 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8864 (m32c_can_change_mode_class): ...this new function, inverting the
8865 sense of the return value.
8866 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8867 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8868 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8869 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8870 (mips_can_change_mode_class): ...this new function, inverting the
8871 sense of the return value.
8872 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8873 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8874 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8875 (msp430_can_change_mode_class): New function.
8876 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8877 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8878 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8879 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8880 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8881 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8882 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8883 (pa_cannot_change_mode_class): Replace with...
8884 (pa_can_change_mode_class): ...this new function, inverting the
8885 sense of the return value.
8886 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8887 than CANNOT_CHANGE_MODE_CLASS in comments.
8888 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8889 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8890 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8891 (pdp11_cannot_change_mode_class): Replace with...
8892 (pdp11_can_change_mode_class): ...this new function, inverting the
8893 sense of the return value.
8894 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8895 * config/powerpcspe/powerpcspe-protos.h
8896 (rs6000_cannot_change_mode_class_ptr): Delete.
8897 * config/powerpcspe/powerpcspe.c
8898 (rs6000_cannot_change_mode_class_ptr): Delete.
8899 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8900 (rs6000_option_override_internal): Assign to
8901 targetm.can_change_mode_class instead of
8902 rs6000_cannot_change_mode_class_ptr.
8903 (rs6000_cannot_change_mode_class): Replace with...
8904 (rs6000_can_change_mode_class): ...this new function, inverting the
8905 sense of the return value.
8906 (rs6000_debug_cannot_change_mode_class): Replace with...
8907 (rs6000_debug_can_change_mode_class): ...this new function.
8908 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8909 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8910 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8911 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8912 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8914 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8915 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8916 (rs6000_option_override_internal): Assign to
8917 targetm.can_change_mode_class instead of
8918 rs6000_cannot_change_mode_class_ptr.
8919 (rs6000_cannot_change_mode_class): Replace with...
8920 (rs6000_can_change_mode_class): ...this new function, inverting the
8921 sense of the return value.
8922 (rs6000_debug_cannot_change_mode_class): Replace with...
8923 (rs6000_debug_can_change_mode_class): ...this new function.
8924 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8925 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8926 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8927 (s390_can_change_mode_class): ...this new function, inverting the
8928 sense of the return value.
8929 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8930 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8931 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8932 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8933 (sh_cannot_change_mode_class): Replace with...
8934 (sh_can_change_mode_class): ...this new function, inverting the
8935 sense of the return value.
8936 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8937 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8938 (sparc_can_change_mode_class): New function.
8939 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8940 * config/spu/spu.c (spu_can_change_mode_class): New function.
8941 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8942 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8943 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8944 (visium_can_change_mode_class): New function.
8945 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8947 2017-09-15 Richard Biener <rguenther@suse.de>
8949 PR tree-optimization/82217
8950 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8951 but not undefined case.
8953 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8956 * postreload.c (reload_cse_simplify_operands): Skip
8957 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8959 2017-09-15 Richard Biener <rguenther@suse.de>
8961 PR tree-optimization/68823
8962 * graphite-scop-detection.c (build_alias_set): If we have a
8963 possible dependence check whether we can handle them by just
8964 looking at the DRs DR_ACCESS_FNs.
8965 (build_scops): If build_alias_set fails, fail the SCOP.
8967 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8969 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8970 to support float128 built-in functions that require the ISA 3.0
8972 (BU_FLOAT128_3_HW): Likewise.
8973 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8975 (FMAF128): Likewise.
8977 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8978 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8979 floating point instructions.
8980 (rs6000_invalid_builtin): Likewise.
8981 (rs6000_builtin_mask_names): Likewise.
8982 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8983 (RS6000_BTM_FLOAT128_HW): Likewise.
8984 (RS6000_BTM_COMMON): Likewise.
8985 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8987 * doc/extend.texi (RS/6000 built-in functions): Document the
8988 IEEE 128-bit floating point square root and fused multiply-add
8991 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8993 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8994 reg (r2) isn't in the set of registers defined in the prologue.
8996 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8997 Alan Hayward <alan.hayward@arm.com>
8998 David Sherwood <david.sherwood@arm.com>
9000 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9001 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9002 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9003 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9004 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9006 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9007 max_vectorization_factor.
9008 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9010 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9011 Alan Hayward <alan.hayward@arm.com>
9012 David Sherwood <david.sherwood@arm.com>
9014 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9015 (vect_worthwhile_without_simd_p): Declare.
9016 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9017 (vectorizable_reduction): Use it.
9018 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9019 (vectorizable_operation): Likewise.
9021 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9022 Alan Hayward <alan.hayward@arm.com>
9023 David Sherwood <david.sherwood@arm.com>
9025 * tree-vectorizer.h (vect_get_num_copies): New function.
9026 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9027 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9028 (vectorizable_induction): Likewise.
9029 (vectorizable_live_operation): Likewise.
9030 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9031 (vectorizable_bswap): Likewise.
9032 (vectorizable_call): Likewise.
9033 (vectorizable_conversion): Likewise.
9034 (vectorizable_assignment): Likewise.
9035 (vectorizable_shift): Likewise.
9036 (vectorizable_operation): Likewise.
9037 (vectorizable_store): Likewise.
9038 (vectorizable_load): Likewise.
9039 (vectorizable_condition): Likewise.
9040 (vectorizable_comparison): Likewise.
9041 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9043 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9044 Alan Hayward <alan.hayward@arm.com>
9045 David Sherwood <david.sherwood@arm.com>
9047 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9048 of vect_init_vector.
9050 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9051 Alan Hayward <alan.hayward@arm.com>
9052 David Sherwood <david.sherwood@arm.com>
9054 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9055 an inline wrapper that provides a location.
9056 (gimple_build_vector): Likewise.
9057 * gimple-fold.c (gimple_build_vector_from_val): New function.
9058 (gimple_build_vector): Likewise.
9059 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9060 functions to build the initial value. Always return a gimple value.
9061 (get_initial_defs_for_reduction): Likewise. Only compute
9063 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9064 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9065 (vectorizable_induction): Use gimple_build_vector rather than
9068 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9069 Alan Hayward <alan.hayward@arm.com>
9070 David Sherwood <david.sherwood@arm.com>
9072 * target.h (vec_perm_indices): New typedef.
9073 (auto_vec_perm_indices): Likewise.
9074 * optabs-query.h: Include target.h
9075 (can_vec_perm_p): Take a vec_perm_indices *.
9076 * optabs-query.c (can_vec_perm_p): Likewise.
9077 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9078 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9079 * tree-vect-generic.c (lower_vec_perm): Likewise.
9080 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9081 (vect_grouped_load_supported): Likewise.
9082 (vect_shift_permute_load_chain): Likewise.
9083 (vect_permute_store_chain): Use auto_vec_perm_indices.
9084 (vect_permute_load_chain): Likewise.
9085 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9086 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9087 Update uses of can_vec_perm_p.
9088 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9089 mode with a number of elements. Take a vec_perm_indices *.
9090 (vect_create_epilog_for_reduction): Update accordingly.
9091 Use auto_vec_perm_indices.
9092 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9093 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9094 (vect_transform_slp_perm_load): Likewise.
9095 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9096 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9097 (vect_gen_perm_mask_checked): Likewise.
9098 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9099 (vect_gen_perm_mask_checked): Likewise.
9100 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9101 (vectorizable_store): Likewise.
9102 (vectorizable_load): Likewise.
9103 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9104 (vectorizable_bswap): Likewise.
9106 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9107 Alan Hayward <alan.hayward@arm.com>
9108 David Sherwood <david.sherwood@arm.com>
9110 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9111 * tree.c (build_vector): Likewise.
9112 (build_vector_from_ctor): Update accordingly.
9113 (build_vector_from_val): Likewise.
9114 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9115 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9116 * tree-vect-generic.c (add_rshift): Likewise.
9117 (expand_vector_divmod): Likewise.
9118 (optimize_vector_constructor): Likewise.
9119 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9120 (vect_transform_slp_perm_load): Likewise.
9121 (vect_schedule_slp_instance): Likewise.
9122 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9123 (vectorizable_call): Likewise.
9124 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9125 * expmed.c (make_tree): Likewise.
9126 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9127 a vector passed to build_vector.
9128 (fold_convert_const): Likewise.
9129 (exact_inverse): Likewise.
9130 (fold_ternary_loc): Likewise.
9131 (fold_relational_const): Likewise.
9132 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9133 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9134 (const_unop): Likewise. Store the reduction accumulator in a
9135 variable rather than an array.
9136 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9137 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9138 the new vector, rather than constructing it after the input arrays.
9139 (native_interpret_vector): Use auto_vec<tree> when building
9140 a vector passed to build_vector. Add elements in order.
9141 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9142 auto_vec<tree> when building a vector passed to build_vector.
9143 (vect_create_epilog_for_reduction): Likewise.
9144 (vectorizable_induction): Likewise.
9145 (get_initial_def_for_reduction): Likewise. Fix indentation of
9147 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9149 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9150 passed to build_vector.
9152 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9153 Alan Hayward <alan.hayward@arm.com>
9154 David Sherwood <david.sherwood@arm.com>
9156 * tree-core.h (tree_base::u): Add an "nelts" field.
9157 (tree_vector): Use VECTOR_CST_NELTS as the length.
9158 * tree.c (tree_size): Likewise.
9159 (make_vector): Initialize VECTOR_CST_NELTS.
9160 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9161 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9162 TYPE_VECTOR_SUBPARTS.
9163 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9164 as the number of units, setting it from VECTOR_CST_NELTS.
9165 (const_vector_from_tree): Likewise.
9166 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9167 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9168 (fold_negate_expr_1): Likewise.
9169 (fold_convert_const): Likewise.
9170 (const_binop): Likewise. Differentiate the number of output and
9172 (const_unop): Likewise.
9173 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9174 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9175 in cases that did the opposite.
9177 2017-09-14 Richard Biener <rguenther@suse.de>
9179 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9182 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9184 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9186 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9189 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9190 if and where to split a bb, except for splitting before debug insn
9191 sequences followed by non-label real insn. Delete debug insns
9192 in between basic blocks.
9194 * combine.c (make_compound_operation_int): Formatting fixes.
9196 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9197 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9198 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9199 * config/sol2.h (LINK_EH_SPEC): Likewise.
9200 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9201 * config/s390/linux.h (LINK_SPEC): Likewise.
9202 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9203 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9204 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9205 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9206 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9207 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9208 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9209 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9211 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9213 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9215 (ENDFILE_LINUX_SPEC): Likewise.
9216 (LINK_EH_SPEC): Likewise.
9217 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9218 (LINK_OS_LINUX_SPEC32): Likewise.
9219 (LINK_OS_LINUX_SPEC64): Likewise.
9220 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9221 (LINK_OS_LINUX_SPEC): Likewise.
9223 2017-09-13 Martin Liska <mliska@suse.cz>
9226 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9227 CASE_HIGH is NULL_TREE.
9229 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9230 Alan Hayward <alan.hayward@arm.com>
9231 David Sherwood <david.sherwood@arm.com>
9233 * target.def (secondary_memory_needed): New hook.
9234 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9235 instead of SECONDARY_MEMORY_NEEDED.
9236 (secondary_memory_needed_mode): Likewise.
9237 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9238 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9239 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9240 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9241 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9242 * doc/tm.texi: Regenerate.
9243 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9244 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9245 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9246 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9247 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9248 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9249 mode argument first and change the reg_class arguments to reg_class_t.
9250 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9251 Make static. Update the call to inline_secondary_memory_needed.
9252 (ix86_register_move_cost): Update the call to
9253 inline_secondary_memory_needed.
9254 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9255 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9257 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9258 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9260 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9261 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9262 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9263 and match hook interface. Add comment from mips.h.
9264 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9265 * config/mmix/mmix.md (truncdfsf2): Refer to
9266 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9268 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9269 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9270 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9271 (pa_secondary_memory_needed): New function.
9272 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9273 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9274 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9275 (pdp11_secondary_memory_needed): Make static and match hook interface.
9276 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9277 * config/powerpcspe/powerpcspe-protos.h
9278 (rs6000_secondary_memory_needed_ptr): Delete.
9279 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9281 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9282 (rs6000_option_override_internal): Assign to
9283 targetm.secondary_memory_needed rather than
9284 rs6000_secondary_memory_needed_ptr.
9285 (rs6000_secondary_memory_needed): Match hook interface.
9286 (rs6000_debug_secondary_memory_needed): Likewise.
9287 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9288 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9289 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9290 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9291 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9292 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9294 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9295 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9296 (rs6000_option_override_internal): Assign to
9297 targetm.secondary_memory_needed rather than
9298 rs6000_secondary_memory_needed_ptr.
9299 (rs6000_secondary_memory_needed): Match hook interface.
9300 (rs6000_debug_secondary_memory_needed): Likewise.
9301 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9302 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9303 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9304 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9305 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9306 (sparc_secondary_memory_needed): New function.
9307 * lra-constraints.c (check_and_process_move): Refer to
9308 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9310 (curr_insn_transform): Likewise.
9311 (process_alt_operands): Use targetm.secondary_memory_needed
9312 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9313 (check_secondary_memory_needed_p): Likewise.
9314 (choose_split_class): Likewise.
9315 * reload.c: Unconditionally include code that was previously
9316 conditional on SECONDARY_MEMORY_NEEDED.
9317 (push_secondary_reload): Use targetm.secondary_memory_needed
9318 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9319 (push_reload): Likewise.
9320 * reload1.c: Unconditionally include code that was previously
9321 conditional on SECONDARY_MEMORY_NEEDED.
9322 (choose_reload_regs): Use targetm.secondary_memory_needed
9323 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9324 (gen_reload): Likewise.
9325 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9327 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9328 Alan Hayward <alan.hayward@arm.com>
9329 David Sherwood <david.sherwood@arm.com>
9331 * target.def (secondary_memory_needed_mode): New hook:
9332 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9333 * targhooks.h (default_secondary_memory_needed_mode): New function.
9334 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9335 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9336 * doc/tm.texi: Regenerate.
9337 * lra-constraints.c (check_and_process_move): Use
9338 targetm.secondary_memory_needed_mode instead of
9339 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9340 (curr_insn_transform): Likewise.
9341 * reload.c (get_secondary_mem): Likewise.
9342 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9343 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9345 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9346 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9347 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9348 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9349 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9351 * config/powerpcspe/powerpcspe-protos.h
9352 (rs6000_secondary_memory_needed_mode): Delete.
9353 * config/powerpcspe/powerpcspe.c
9354 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9355 (rs6000_secondary_memory_needed_mode): Make static.
9356 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9357 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9359 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9361 (rs6000_secondary_memory_needed_mode): Make static.
9362 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9363 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9364 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9365 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9366 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9368 (sparc_secondary_memory_needed_mode): New function.
9369 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9371 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9373 * config/aarch64/constraints.md (Umq): New constraint.
9374 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9376 (mov<mode>): Update condition.
9378 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9380 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9381 when bitposition is the same.
9383 2017-09-13 Richard Biener <rguenther@suse.de>
9385 * dwarf2out.c (output_die_symbol): Remove.
9386 (output_die): Do not output a DIEs symbol.
9388 2017-09-13 Richard Biener <rguenther@suse.de>
9391 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9392 default-def to avoid breaking iterator update with the weird
9393 interaction with cgraph_update_edges_for_call_stmt_node.
9395 2017-09-13 Richard Biener <rguenther@suse.de>
9397 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9398 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9399 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9400 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9402 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9404 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9405 Disable pc relative literal load irrespective of
9406 TARGET_FIX_ERR_A53_84341 for default.
9408 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9410 * config/sparc/sparc.c (output_return): Output the source location of
9411 the insn in the delay slot, if any.
9412 (output_sibcall): Likewise.
9414 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9417 * common.opt (-static-pie): New alias.
9418 (shared): Negate static-pie.
9419 (-no-pie): Update help text.
9421 (static-pie): New option.
9422 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9423 -static-pie support.
9424 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9425 (LINK_EH_SPEC): Likewise.
9426 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9427 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9428 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9429 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9430 (init_gcc_specs): Likewise.
9431 (init_spec): Likewise.
9432 (display_help): Update help message for -pie.
9433 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9436 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9438 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9439 (movdi_aarch64): Likewise.
9440 (movti_aarch64): Likewise.
9442 2017-09-12 Simon Wright <simon@pushface.org>
9445 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9446 calculation of the minor version, always output as 0.
9448 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9451 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9452 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9453 on it early, rather than manual conversion late. For
9454 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9455 instead of performing manual conversion.
9457 2017-09-12 Carl Love <cel@us.ibm.com>
9459 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9460 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9462 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9463 VMULOSW): Add definitions.
9464 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9465 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9466 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9467 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9468 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9470 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9472 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9474 (movti_aarch64): Likewise.
9475 (movdf_aarch64): Likewise.
9476 (movtf_aarch64): Likewise.
9477 (load_pairdi): Likewise.
9478 (store_pairdi): Likewise.
9479 (load_pairdf): Likewise.
9480 (store_pairdf): Likewise.
9481 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9482 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9483 (ldr_got_small_<mode>): Likewise.
9484 (ldr_got_small_28k_<mode>): Likewise.
9485 (ldr_got_tiny): Likewise.
9486 * config/aarch64/iterators.md (ldst_sz): New.
9487 (ldpstp_sz): Likewise.
9488 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9490 (thunderx_load): Split load_8 to load_16.
9491 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9493 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9494 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9495 (xgene1_store_pair): Split store_8 to store_16.
9496 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9497 (falkor_st_0_st_sd): Split store_8 to store_16.
9499 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9501 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9502 and store1/2/3/4 to store_4/8/12/16.
9503 * config/aarch64/aarch64.md: Update for rename.
9504 * config/arm/arm.md: Likewise.: Likewise.
9505 * config/arm/arm.c: Likewise.
9506 * config/arm/thumb1.md: Likewise.
9507 * config/arm/thumb2.md: Likewise.
9508 * config/arm/vfp.md: Likewise.
9509 * config/arm/arm-generic.md: Likewise.
9510 * config/arm/arm1020e.md: Likewise.
9511 * config/arm/arm1026ejs.md: Likewise.
9512 * config/arm/arm1136jfs.md: Likewise.
9513 * config/arm/arm926ejs.md: Likewise.
9514 * config/arm/cortex-a15.md: Likewise.
9515 * config/arm/cortex-a17.md: Likewise.
9516 * config/arm/cortex-a5.md: Likewise.
9517 * config/arm/cortex-a53.md: Likewise.
9518 * config/arm/cortex-a57.md: Likewise.
9519 * config/arm/cortex-a7.md: Likewise.
9520 * config/arm/cortex-a8.md: Likewise.
9521 * config/arm/cortex-a9.md: Likewise.
9522 * config/arm/cortex-m4.md: Likewise.
9523 * config/arm/cortex-m7.md: Likewise.
9524 * config/arm/cortex-r4.md: Likewise.
9525 * config/arm/exynos-m1.md: Likewise.
9526 * config/arm/fa526.md: Likewise.
9527 * config/arm/fa606te.md: Likewise.
9528 * config/arm/fa626te.md: Likewise.
9529 * config/arm/fa726te.md: Likewise.
9530 * config/arm/fmp626.md: Likewise.
9531 * config/arm/iwmmxt.md: Likewise.
9532 * config/arm/ldmstm.md: Likewise.
9533 * config/arm/marvell-pj4.md: Likewise.
9534 * config/arm/xgene1.md: Likewise.
9535 * config/aarch64/thunderx.md: Likewise.
9536 * config/aarch64/thunderx2t99.md: Likewise.
9537 * config/aarch64/falkor.md: Likewise.
9539 2017-09-12 Martin Liska <mliska@suse.cz>
9541 * attribs.c (private_lookup_attribute): New function.
9542 * attribs.h (private_lookup_attribute): Declared here.
9543 (lookup_attribute): Called from this place.
9545 2017-09-12 Richard Biener <rguenther@suse.de>
9547 PR tree-optimization/82157
9548 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9549 stmts with side-effects.
9551 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9552 Alan Hayward <alan.hayward@arm.com>
9553 David Sherwood <david.sherwood@arm.com>
9555 * target.def (hard_regno_nregs): New hook.
9556 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9557 * targhooks.h (default_hard_regno_nregs): Declare.
9558 * targhooks.c (default_hard_regno_nregs): New function.
9559 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9560 (TARGET_HARD_REGNO_NREGS): ...this hook.
9561 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9562 (CLASS_MAX_NREGS): Likewise.
9563 * doc/tm.texi: Regenerate.
9564 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9565 instead of HARD_REGNO_NREGS.
9566 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9567 HARD_REGNO_NREGS in the comment.
9568 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9569 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9570 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9571 Return an unsigned int.
9572 (TARGET_HARD_REGNO_NREGS): Redefine.
9573 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9574 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9575 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9576 (arc_hard_regno_nregs): New function.
9577 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9578 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9579 (arm_hard_regno_nregs): New function.
9580 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9581 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9582 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9583 (TARGET_HARD_REGNO_NREGS): Redefine.
9584 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9585 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9586 (HARD_REGNO_NREGS): Delete.
9587 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9588 (cr16_hard_regno_nregs): New function.
9589 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9590 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9591 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9592 (cris_hard_regno_nregs): New function.
9593 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9594 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9595 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9596 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9597 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9598 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9599 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9600 (frv_hard_regno_nregs): Make static. Take and return an
9602 (frv_class_max_nregs): Remove outdated copy of documentation.
9603 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9604 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9605 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9606 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9607 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9608 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9609 (TARGET_HARD_REGNO_NREGS): Redefine.
9610 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9611 (CLASS_MAX_NREGS): Update comment.
9612 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9613 (ia64_hard_regno_nregs): New function.
9614 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9615 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9616 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9617 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9618 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9620 (m32c_hard_regno_nregs): Likewise. Make static.
9621 (TARGET_HARD_REGNO_NREGS): Redefine.
9622 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9623 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9624 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9625 (m68k_hard_regno_nregs): New function.
9626 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9627 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9628 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9629 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9630 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9631 Take and return an unsigned int.
9632 (TARGET_HARD_REGNO_NREGS): Redefine.
9633 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9634 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9635 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9636 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9637 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9638 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9639 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9640 (msp430_hard_regno_nregs): Make static. Take and return an
9642 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9643 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9644 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9645 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9646 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9647 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9648 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9649 (TARGET_HARD_REGNO_NREGS): Redefine.
9650 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9651 (PA_HARD_REGNO_NREGS): ...this.
9652 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9653 (PA_HARD_REGNO_NREGS): ...this.
9654 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9655 (pa_hard_regno_nregs): New function.
9656 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9657 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9658 (pdp11_hard_regno_nregs): New function.
9659 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9660 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9661 (rs6000_hard_regno_nregs_hook): New function.
9662 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9663 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9664 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9665 Take and return an unsigned int. Move earlier in file.
9666 (TARGET_HARD_REGNO_NREGS): Redefine.
9667 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9668 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9669 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9670 (rl78_hard_regno_nregs): Make static. Take and return an
9672 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9673 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9674 (rs6000_hard_regno_nregs_hook): New function.
9675 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9676 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9677 (TARGET_HARD_REGNO_NREGS): Redefine.
9678 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9679 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9680 instead of HARD_REGNO_NREGS.
9681 (s390_hard_regno_nregs): New function.
9682 (s390_hard_regno_mode_ok): Add comment from s390.h.
9683 (TARGET_HARD_REGNO_NREGS): Redefine.
9684 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9685 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9686 (sh_hard_regno_nregs): New function.
9687 (sh_pass_in_reg_p): Use it.
9688 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9689 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9690 (sparc_hard_regno_nregs): New function.
9691 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9692 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9693 (spu_function_arg_advance): Use it, supplying a valid register number.
9694 (TARGET_HARD_REGNO_NREGS): Redefine.
9695 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9696 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9697 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9698 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9699 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9700 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9701 (CLASS_MAX_NREGS): Remove copy of old documentation.
9702 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9703 (visium_hard_regno_nregs): New function.
9704 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9705 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9706 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9707 xtensa_hard_regno_nregs): New function.
9708 * system.h (HARD_REGNO_NREGS): Poison.
9710 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9712 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9713 hard_regno_nregs instead of HARD_REGNO_NREGS.
9714 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9715 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9716 (c6x_expand_epilogue): Likewise.
9717 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9718 (frv_read_iacc_argument): Likewise.
9719 * config/sh/sh.c: Include regs.h.
9720 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9721 (regs_used): Likewise.
9722 (output_stack_adjust): Likewise.
9723 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9724 * expmed.c: Include regs.h.
9725 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9726 * ree.c: Include regs.h.
9727 (combine_reaching_defs): Use hard_regno_nregs instead of
9729 (add_removable_extension): Likewise.
9731 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9733 * regs.h (hard_regno_nregs): Turn into a function.
9734 (end_hard_regno): Update accordingly.
9735 * caller-save.c (setup_save_areas): Likewise.
9736 (save_call_clobbered_regs): Likewise.
9737 (replace_reg_with_saved_mem): Likewise.
9738 (insert_restore): Likewise.
9739 (insert_save): Likewise.
9740 * combine.c (can_change_dest_mode): Likewise.
9741 (move_deaths): Likewise.
9742 (distribute_notes): Likewise.
9743 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9744 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9745 (rs6000_split_multireg_move): Likewise.
9746 (rs6000_register_move_cost): Likewise.
9747 (rs6000_memory_move_cost): Likewise.
9748 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9749 (rs6000_split_multireg_move): Likewise.
9750 (rs6000_register_move_cost): Likewise.
9751 (rs6000_memory_move_cost): Likewise.
9752 * cselib.c (cselib_reset_table): Likewise.
9753 (cselib_lookup_1): Likewise.
9754 * emit-rtl.c (set_mode_and_regno): Likewise.
9755 * function.c (aggregate_value_p): Likewise.
9756 * ira-color.c (setup_profitable_hard_regs): Likewise.
9757 (check_hard_reg_p): Likewise.
9758 (calculate_saved_nregs): Likewise.
9759 (assign_hard_reg): Likewise.
9760 (improve_allocation): Likewise.
9761 (calculate_spill_cost): Likewise.
9762 * ira-emit.c (modify_move_list): Likewise.
9763 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9764 (ira_hard_reg_in_set_p): Likewise.
9765 * ira.c (setup_reg_mode_hard_regset): Likewise.
9766 (clarify_prohibited_class_mode_regs): Likewise.
9767 (check_allocation): Likewise.
9768 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9769 (lra_setup_reg_renumber): Likewise.
9770 (setup_try_hard_regno_pseudos): Likewise.
9771 (spill_for): Likewise.
9772 (assign_hard_regno): Likewise.
9773 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9774 * lra-constraints.c (in_class_p): Likewise.
9775 (lra_constraint_offset): Likewise.
9776 (simplify_operand_subreg): Likewise.
9777 (lra_constraints): Likewise.
9778 (split_reg): Likewise.
9779 (split_if_necessary): Likewise.
9780 (invariant_p): Likewise.
9781 (inherit_in_ebb): Likewise.
9782 * lra-lives.c (process_bb_lives): Likewise.
9783 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9784 (get_hard_regs): Likewise.
9785 (do_remat): Likewise.
9786 * lra-spills.c (assign_spill_hard_regs): Likewise.
9787 * mode-switching.c (create_pre_exit): Likewise.
9788 * postreload.c (reload_combine_recognize_pattern): Likewise.
9789 * recog.c (peep2_find_free_register): Likewise.
9790 * regcprop.c (kill_value_regno): Likewise.
9791 (set_value_regno): Likewise.
9792 (copy_value): Likewise.
9793 (maybe_mode_change): Likewise.
9794 (find_oldest_value_reg): Likewise.
9795 (copyprop_hardreg_forward_1): Likewise.
9796 * regrename.c (check_new_reg_p): Likewise.
9797 (regrename_do_replace): Likewise.
9798 * reload.c (push_reload): Likewise.
9799 (combine_reloads): Likewise.
9800 (find_dummy_reload): Likewise.
9801 (operands_match_p): Likewise.
9802 (find_reloads): Likewise.
9803 (find_equiv_reg): Likewise.
9804 (reload_adjust_reg_for_mode): Likewise.
9805 * reload1.c (count_pseudo): Likewise.
9806 (count_spilled_pseudo): Likewise.
9807 (find_reg): Likewise.
9808 (clear_reload_reg_in_use): Likewise.
9809 (free_for_value_p): Likewise.
9810 (allocate_reload_reg): Likewise.
9811 (choose_reload_regs): Likewise.
9812 (reload_adjust_reg_for_temp): Likewise.
9813 (emit_reload_insns): Likewise.
9814 (delete_output_reload): Likewise.
9815 * rtlanal.c (subreg_get_info): Likewise.
9816 * sched-deps.c (sched_analyze_reg): Likewise.
9817 * sel-sched.c (init_regs_for_mode): Likewise.
9818 (mark_unavailable_hard_regs): Likewise.
9819 (choose_best_reg_1): Likewise.
9820 (verify_target_availability): Likewise.
9821 * valtrack.c (dead_debug_insert_temp): Likewise.
9822 * var-tracking.c (track_loc_p): Likewise.
9823 (emit_note_insn_var_location): Likewise.
9824 * varasm.c (make_decl_rtl): Likewise.
9825 * reginfo.c (choose_hard_reg_mode): Likewise.
9826 (init_reg_modes_target): Refer directly to
9827 this_target_regs->x_hard_regno_nregs.
9829 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9831 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9832 instead of hard_regno_nregs.
9834 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9836 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9837 end_hard_regno instead of hard_regno_nregs.
9838 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9839 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9840 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9841 * ira-color.c (improve_allocation): Likewise.
9842 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9843 * lra-lives.c (mark_regno_live): Likewise.
9844 (mark_regno_dead): Likewise.
9845 * lra-remat.c (operand_to_remat): Likewise.
9846 * lra.c (collect_non_operand_hard_regs): Likewise.
9847 * postreload.c (reload_combine_note_store): Likewise.
9848 (move2add_valid_value_p): Likewise.
9849 * reload.c (regno_clobbered_p): Likewise.
9851 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9853 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9855 * config/v850/v850.c (v850_reorg): Likewise.
9856 * reload.c (refers_to_regno_for_reload_p): Likewise.
9857 (find_equiv_reg): Likewise.
9858 * reload1.c (reload_reg_reaches_end_p): Likewise.
9860 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9862 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9864 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9865 * config/arm/arm.c (output_move_neon): Likewise.
9866 (arm_attr_length_move_neon): Likewise.
9867 (neon_split_vcombine): Likewise.
9868 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9869 (c6x_mark_reg_written): Likewise.
9870 (c6x_dwarf_register_span): Likewise.
9871 * config/i386/i386.c (ix86_save_reg): Likewise.
9872 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9873 (rws_access_reg): Likewise.
9874 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9875 * mode-switching.c (create_pre_exit): Likewise.
9876 * ree.c (combine_reaching_defs): Likewise.
9877 (add_removable_extension): Likewise.
9878 * regcprop.c (find_oldest_value_reg): Likewise.
9879 (copyprop_hardreg_forward_1): Likewise.
9880 * reload.c (reload_inner_reg_of_subreg): Likewise.
9881 (push_reload): Likewise.
9882 (combine_reloads): Likewise.
9883 (find_dummy_reload): Likewise.
9884 (reload_adjust_reg_for_mode): Likewise.
9885 * reload1.c (find_reload_regs): Likewise.
9886 (forget_old_reloads_1): Likewise.
9887 (reload_reg_free_for_value_p): Likewise.
9888 (reload_adjust_reg_for_temp): Likewise.
9889 (emit_reload_insns): Likewise.
9890 (delete_output_reload): Likewise.
9891 * sel-sched.c (choose_best_reg_1): Likewise.
9892 (choose_best_pseudo_reg): Likewise.
9894 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9895 Alan Hayward <alan.hayward@arm.com>
9896 David Sherwood <david.sherwood@arm.com>
9898 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9899 * target.def (slow_unaligned_access): New hook.
9900 * targhooks.h (default_slow_unaligned_access): Declare.
9901 * targhooks.c (default_slow_unaligned_access): New function.
9902 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9903 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9904 * doc/tm.texi: Regenerate.
9905 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9906 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9907 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9909 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9910 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9912 (rs6000_slow_unaligned_access): New function.
9913 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9914 (expand_block_compare): Likewise.
9915 (expand_strn_compare): Likewise.
9916 (rs6000_rtx_costs): Likewise.
9917 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9918 (riscv_slow_unaligned_access): Likewise.
9919 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9920 (riscv_slow_unaligned_access_p): ...this and make static.
9921 (riscv_option_override): Update accordingly.
9922 (riscv_slow_unaligned_access): New function.
9923 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9924 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9925 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9926 (rs6000_slow_unaligned_access): New function.
9927 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9928 (rs6000_rtx_costs): Likewise.
9929 * config/rs6000/rs6000-string.c (expand_block_compare)
9930 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9931 of SLOW_UNALIGNED_ACCESS.
9932 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9933 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9934 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9935 of SLOW_UNALIGNED_ACCESS.
9936 * expmed.c (simple_mem_bitfield_p): Likewise.
9937 * expr.c (alignment_for_piecewise_move): Likewise.
9938 (emit_group_load_1): Likewise.
9939 (emit_group_store): Likewise.
9940 (copy_blkmode_from_reg): Likewise.
9941 (emit_push_insn): Likewise.
9942 (expand_assignment): Likewise.
9943 (store_field): Likewise.
9944 (expand_expr_real_1): Likewise.
9945 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9946 * lra-constraints.c (simplify_operand_subreg): Likewise.
9947 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9948 * gimple-ssa-store-merging.c: Likewise in block comment at start
9950 * tree-ssa-strlen.c: Include target.h.
9951 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9952 of SLOW_UNALIGNED_ACCESS.
9953 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9955 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9957 PR rtl-optimization/82185
9958 * expmed.c (emit_store_flag_int): Only test tem if it has been
9961 2017-09-12 Richard Biener <rguenther@suse.de>
9964 * match.pd ((FTYPE) N CMP CST): Fix typo.
9966 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9968 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9970 (mips_near_type_p): Add 'short_call' attribute as a synonym
9972 * doc/extend.texi (short_call): Document new function attribute.
9974 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9977 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9978 assertion check that in the condition.
9979 (get_atomic_generic_size): Likewise. Before testing if parameter
9980 has pointer type, if it has array type, call for C++
9981 default_conversion to perform array-to-pointer conversion.
9983 2017-09-12 Richard Biener <rguenther@suse.de>
9985 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9986 for operations we cannot scalarize.
9988 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9990 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9991 vectors heap vectors. Clean up comments.
9992 Make visited_bbs a reference.
9993 (profitable_jump_thread_path): Make GC
9994 vectors heap vectors. Clean up comments.
9996 (convert_and_register_jump_thread_path): Make GC vectors heap
9998 (check_subpath_and_update_thread_path): Same. Clean up comments.
9999 Make visited_bbs a reference.
10000 (handle_phi): Abstract common code to to
10001 register_jump_thread_path_if_profitable.
10002 Rename VAR_BB to DEF_BB.
10004 Make GC vectors heap vectors.
10005 Make visited_bbs a reference.
10006 (handle_assignment): Same.
10007 (register_jump_thread_path_if_profitable): New.
10008 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10010 Make GC vectors heap vectors. Clean up comments.
10011 Make visited_bbs a reference.
10012 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10013 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10016 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10019 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10020 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10022 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10024 Revert r251800 and r251799.
10026 2017-09-11 Martin Jambor <mjambor@suse.cz>
10029 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10030 arguments in advance.
10031 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10032 use it to find predecessor edges.
10033 (naive_outof_ssa): Collect vector of predecessors.
10035 2017-09-08 Jason Merrill <jason@redhat.com>
10037 PR c++/70029 - ICE with ref-qualifier and -flto
10038 * langhooks.h (struct lang_hooks_for_types): Add
10039 copy_lang_qualifiers.
10040 * attribs.c (build_type_attribute_qual_variant): Use it.
10041 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10043 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10044 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10046 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10049 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10050 (*mulsi3_sp64): New instruction.
10051 (mulsi3): New expander.
10053 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10055 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10057 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10059 * sancov.c: Include memmodel.h.
10061 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10064 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10067 2017-09-07 Carl Love <cel@us.ibm.com>
10069 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10070 the sldi instruction.
10072 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10074 * sancov.c: Include tm_p.h.
10076 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10079 * output.h (switch_to_other_text_partition): New declaration.
10080 * varasm.c (switch_to_other_text_partition): New function.
10081 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10082 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10083 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10084 to the other text partition before emitting LCL label and switch back
10085 after emitting the word after it.
10087 2017-09-07 Richard Biener <rguenther@suse.de>
10089 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10090 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10091 critical edges here, after loop init.
10092 (pass_data_pre): Remove PROP_no_crit_edges flags.
10093 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10094 for valueization of call args to avoid leaking VN_TOP.
10095 (visit_use): Assert we do not visit default defs.
10096 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10097 Use error_mark_node to more easily detect leaking VN_TOP.
10098 All default-defs are varying, not VN_TOP. Mark them visited.
10099 (run_scc_vn): Make code match comment.
10101 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10103 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10104 OPTION_MASK_FLOAT128_KEYWORD.
10105 (POWERPC_MASKS): Likewise.
10106 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10107 support for the -mfloat128-type option, and make -mfloat128
10108 default on PowerPC Linux systems. Define or undefine
10109 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10110 Define __float128 to be __ieee128 if IEEE 128-bit support is
10111 enabled, or undefine it.
10112 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10113 Delete defining __FLOAT128_TYPE__.
10114 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10115 -mfloat128-type option and make -mfloat128 default on PowerPC
10117 (TARGET_FLOAT128_TYPE): Likewise.
10118 (-mfloat128-type): Likewise.
10119 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10120 Delete the -mfloat128-type option and make -mfloat128 default on
10121 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10122 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10123 128-bit floating point is enabled. Change tests from using
10124 -mfloat128-type to -mfloat128.
10125 (rs6000_mangle_type): Use the correct mangling for the __float128
10126 type even if normal long double is restricted to 64-bits.
10127 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10129 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10130 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10131 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10133 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10134 documentation for -mfloat128.
10136 2017-09-06 Olivier Hainque <hainque@adacore.com>
10138 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10140 2017-09-06 Wish Wu <wishwu007@gmail.com>
10141 Jakub Jelinek <jakub@redhat.com>
10143 * asan.c (initialize_sanitizer_builtins): Add
10144 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10145 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10146 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10147 BT_FN_VOID_UINT64_PTR variables.
10148 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10149 (BT_FN_VOID_UINT16_UINT16): Likewise.
10150 (BT_FN_VOID_UINT32_UINT32): Likewise.
10151 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10152 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10153 (BT_FN_VOID_UINT64_PTR): Likewise.
10154 * common.opt (flag_sanitize_coverage): New variable.
10155 (fsanitize-coverage=trace-pc): Remove.
10156 (fsanitize-coverage=): Add.
10157 * flag-types.h (enum sanitize_coverage_code): New enum.
10158 * fold-const.c (fold_range_test): Disable non-short-circuit
10159 optimization if flag_sanitize_coverage.
10160 (fold_truth_andor): Likewise.
10161 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10162 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10163 (coverage_sanitizer_opts): New array.
10164 (get_closest_sanitizer_option): Add OPTS argument, handle also
10165 OPT_fsanitize_coverage_.
10166 (parse_sanitizer_options): Adjusted to also handle
10167 OPT_fsanitize_coverage_.
10168 (common_handle_option): Add OPT_fsanitize_coverage_.
10169 * sancov.c (instrument_comparison, instrument_switch): New function.
10170 (sancov_pass): Add trace-cmp support.
10171 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10172 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10173 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10174 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10175 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10176 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10177 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10178 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10179 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10180 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10182 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10184 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10185 error. Only quit immediately if parsing is complete.
10186 (BEGIN): Initialize fatal_err and parse_done.
10187 (begin fpu, end fpu): Check number of arguments.
10188 (begin arch, end arch): Likewise.
10189 (begin cpu, end cpu): Likewise.
10190 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10191 (optalias): Likewise.
10193 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10195 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10196 * config/arm/arm-isa.h: Delete. Move definitions to ...
10197 * arm-cpus.in: ... here. Use new feature and fgroup values.
10198 * config/arm/arm.c (arm_option_override): Use lower case for feature
10200 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10201 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10202 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10204 (print_isa_bits_for): New function.
10205 (gen_isa): New function.
10206 (gen_comm_data): Use print_isa_bits_for.
10207 (define feature): New keyword.
10208 (define fgroup): New keyword.
10209 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10210 (arm-isa.h): Add rule to generate file.
10211 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10212 case for feature bit names.
10214 2017-09-06 Richard Biener <rguenther@suse.de>
10216 * tree-ssa-pre.c (NECESSARY): Remove.
10217 (create_expression_by_pieces): Do not touch pass-local flags.
10218 (insert_into_preds_of_block): Likewise.
10219 (do_pre_regular_insertion): Likewise.
10220 (eliminate_insert): Likewise.
10221 (eliminate_dom_walker::before_dom_children): Likewise.
10222 (fini_eliminate): Do not look at inserted_exprs.
10223 (mark_operand_necessary): Remove.
10224 (remove_dead_inserted_code): Replace with simple work-list
10225 algorithm based on inserted_exprs and SSA uses.
10226 (pass_pre::execute): Re-order fini_eliminate and
10227 remove_dead_inserted_code.
10229 2017-09-06 Olivier Hainque <hainque@adacore.com>
10231 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10232 for VxWorks 7. Adjust surrounding comments.
10234 2017-09-06 Richard Biener <rguenther@suse.de>
10236 * gimple-ssa-strength-reduction.c
10237 (find_candidates_dom_walker::before_dom_children): Also allow
10240 2017-09-06 Richard Biener <rguenther@suse.de>
10242 PR tree-optimization/82108
10243 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10244 for gap in the non-permutation SLP case.
10246 2017-09-06 Martin Jambor <mjambor@suse.cz>
10248 PR tree-optimization/82078
10249 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10250 add_access_to_work_queue...
10251 (build_accesses_from_assign): ...here.
10252 (propagate_all_subaccesses): Make sure racc is the group
10253 representative, if there is one.
10255 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10257 PR middle-end/82095
10258 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10261 2017-09-06 Richard Biener <rguenther@suse.de>
10263 * gimple-ssa-strength-reduction.c
10264 (find_candidates_dom_walker::before_doom_children): Use a
10265 type and not a mode check.
10267 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10270 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10272 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10274 2017-09-05 Jeff Law <law@redhat.com>
10276 PR tree-optimization/64910
10277 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10278 cases where we have 3 or more operands.
10280 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10282 PR middle-end/81768
10283 * omp-low.c (lower_omp_for): Recompute tree invariant if
10284 gimple_omp_for_initial/final is ADDR_EXPR.
10286 PR middle-end/81768
10287 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10288 into gimple val before gimplification fo the COND_EXPR.
10290 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10292 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10293 REGION_COPY argument.
10294 (thread_through_all_blocks): Remove unused argument to
10295 duplicate_thread_path.
10297 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10298 Alan Hayward <alan.hayward@arm.com>
10299 David Sherwood <david.sherwood@arm.com>
10301 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10302 Take a scalar_mode rather than a machine_mode.
10303 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10304 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10305 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10306 (aarch64_gen_adjusted_ldpstp): Likewise.
10307 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10309 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10310 Alan Hayward <alan.hayward@arm.com>
10311 David Sherwood <david.sherwood@arm.com>
10313 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10314 Take a scalar_int_mode instead of a machine_mode.
10315 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10316 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10317 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10318 (aarch64_simd_attr_length_rglist): Delete.
10319 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10320 a scalar_int_mode instead of a machine_mode.
10321 (aarch64_add_offset): Likewise.
10322 (aarch64_internal_mov_immediate): Likewise
10323 (aarch64_add_constant_internal): Likewise.
10324 (aarch64_add_constant): Likewise.
10325 (aarch64_movw_imm): Likewise.
10326 (aarch64_rtx_arith_op_extract_p): Likewise.
10327 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10328 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10329 Remove assert that the mode isn't a vector.
10330 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10331 (aarch64_expand_mov_immediate): Update calls after above changes.
10332 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10333 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10334 (aarch64_move_imm): Likewise.
10335 (aarch64_can_const_movi_rtx_p): Likewise.
10336 (aarch64_strip_extend): Likewise.
10337 (aarch64_extr_rtx_p): Likewise.
10338 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10339 a CONST_INT when the mode parameter is VOIDmode.
10340 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10342 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10344 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10345 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10346 (bitwise_type_for_mode): Update accordingly.
10348 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10350 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10351 * stor-layout.c (mode_for_size_tree): Likewise.
10352 (mode_for_array): Update accordingly.
10353 (layout_decl): Likewise.
10354 (compute_record_mode): Likewise. Only set the mode once.
10356 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10358 * target.def (get_mask_mode): Change return type to opt_mode.
10360 * doc/tm.texi: Regenerate.
10361 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10362 * targhooks.c (default_get_mask_mode): Likewise.
10363 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10364 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10365 targetm.get_mask_mode.
10366 * tree.c (build_truth_vector_type): Likewise.
10368 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10370 * machmode.h (mode_for_vector): Return an opt_mode.
10371 * stor-layout.c (mode_for_vector): Likewise.
10372 (mode_for_int_vector): Update accordingly.
10373 (layout_type): Likewise.
10374 * config/i386/i386.c (emit_memmov): Likewise.
10375 (ix86_expand_set_or_movmem): Likewise.
10376 (ix86_expand_vector_init): Likewise.
10377 (ix86_get_mask_mode): Likewise.
10378 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10380 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10381 * expmed.c (extract_bit_field_1): Likewise.
10382 * expr.c (expand_expr_real_2): Likewise.
10383 * optabs-query.c (can_vec_perm_p): Likewise.
10384 (can_vec_mask_load_store_p): Likewise.
10385 * optabs.c (expand_vec_perm): Likewise.
10386 * targhooks.c (default_get_mask_mode): Likewise.
10387 * tree-vect-stmts.c (vectorizable_store): Likewise.
10388 (vectorizable_load): Likewise.
10389 (get_vectype_for_scalar_type_and_size): Likewise.
10391 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10393 * machmode.h (mode_for_int_vector): New function.
10394 * stor-layout.c (mode_for_int_vector): Likewise.
10395 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10396 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10397 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10398 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10399 (s390_expand_vcond): Likewise.
10401 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10403 * machmode.h (opt_machine_mode): New type.
10404 (opt_mode<T>): Allow construction from anything that can be
10406 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10407 (mode_for_size): Return an opt_machine_mode.
10408 * stor-layout.c (mode_for_size): Likewise.
10409 (mode_for_size_tree): Update call accordingly.
10410 (bitwise_mode_for_mode): Likewise.
10411 (make_fract_type): Likewise.
10412 (make_accum_type): Likewise.
10413 * caller-save.c (replace_reg_with_saved_mem): Update call
10415 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10416 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10417 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10418 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10419 * expmed.c (extract_bit_field_1): Likewise.
10420 * reload.c (get_secondary_mem): Likewise.
10421 * varasm.c (assemble_integer): Likewise.
10422 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10425 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10427 * machmode.h (decimal_float_mode_for_size): New function.
10428 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10429 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10430 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10431 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10432 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10433 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10435 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10437 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10438 (get_builtin_sync_mode): Likewise.
10439 (expand_ifn_atomic_compare_exchange): Likewise.
10440 (expand_builtin_atomic_clear): Likewise.
10441 (expand_builtin_atomic_test_and_set): Likewise.
10442 (fold_builtin_atomic_always_lock_free): Likewise.
10443 * calls.c (compute_argument_addresses): Likewise.
10444 (emit_library_call_value_1): Likewise.
10445 (store_one_arg): Likewise.
10446 * combine.c (combine_instructions): Likewise.
10447 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10448 * config/arm/arm.c (arm_function_value): Likewise.
10449 (aapcs_allocate_return_reg): Likewise.
10450 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10451 * config/i386/i386.c (construct_container): Likewise.
10452 (ix86_gimplify_va_arg): Likewise.
10453 (ix86_expand_sse_cmp): Likewise.
10454 (emit_memmov): Likewise.
10455 (emit_memset): Likewise.
10456 (expand_small_movmem_or_setmem): Likewise.
10457 (ix86_expand_pextr): Likewise.
10458 (ix86_expand_pinsr): Likewise.
10459 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10460 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10462 * config/mips/mips.c (mips_function_value_1) Likewise.
10463 (mips_block_move_straight): Likewise.
10464 (mips_expand_ins_as_unaligned_store): Likewise.
10465 * config/powerpcspe/powerpcspe.c
10466 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10467 (rs6000_darwin64_record_arg_flush): Likewise.
10468 * config/rs6000/rs6000.c
10469 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10470 (rs6000_darwin64_record_arg_flush): Likewise.
10471 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10472 (sparc_function_value_1): Likewise.
10473 * config/spu/spu.c (adjust_operand): Likewise.
10474 (spu_emit_branch_or_set): Likewise.
10475 (arith_immediate_p): Likewise.
10476 * emit-rtl.c (gen_lowpart_common): Likewise.
10477 * expr.c (expand_expr_real_1): Likewise.
10478 * function.c (assign_parm_setup_block): Likewise.
10479 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10480 * reload1.c (alter_reg): Likewise.
10481 * stor-layout.c (mode_for_vector): Likewise.
10482 (layout_type): Likewise.
10484 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10486 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10487 (spu_convert_move): Likewise.
10488 * lower-subreg.c (resolve_simple_move): Likewise.
10490 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10493 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10494 define_insn to a define_expand.
10495 (altivec_vsum2sws_direct): New define_insn.
10496 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10498 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10500 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10503 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10508 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10509 realign stack if stack alignment needed is less than incoming
10512 2017-09-05 Marek Polacek <polacek@redhat.com>
10515 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10518 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10520 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10522 2017-09-05 Richard Biener <rguenther@suse.de>
10524 PR tree-optimization/82084
10525 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10527 2017-09-05 Richard Biener <rguenther@suse.de>
10529 PR tree-optimization/82102
10530 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10532 2017-09-05 Martin Liska <mliska@suse.cz>
10534 PR tree-optimization/82032
10535 * tree-cfg.c (generate_range_test): New function.
10536 * tree-cfg.h (generate_range_test): Declared here.
10537 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10538 (cleanup_control_expr_graph): Use it.
10539 * tree-switch-conversion.c (try_switch_expansion): Remove
10541 (emit_case_nodes): Use generate_range_test.
10543 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10546 * config/i386/i386.md (*<btsc><mode>_mask): Add
10547 TARGET_USE_BT to insn constraint.
10548 (*btr<mode>_mask): Ditto.
10550 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10552 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10553 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10555 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10558 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10559 TARGET_NEON and TARGET_IWMMXT.
10560 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10561 TARGET_NEON and TARGET_IWMMXT.
10562 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10564 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10566 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10567 (ix86_rewrite_tls_address): Ditto.
10568 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10569 (ix86_rewrite_tls_address_1): Ditto.
10570 (ix86_rewrite_tls_address): Ditto.
10571 * config/i386/predicates.md (tls_address_pattern): New predicate.
10572 * config/i386/i386.md (TLS address splitter): New splitter.
10574 2017-09-04 Richard Biener <rguenther@suse.de>
10576 PR tree-optimization/82084
10577 * fold-const.h (can_native_encode_string_p): Declare.
10578 * fold-const.c (can_native_encode_string_p): Factor out from ...
10579 (native_encode_string): ... here.
10580 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10581 vectorizing stores from constants we later cannot handle.
10583 2017-09-04 Marek Polacek <polacek@redhat.com>
10586 * doc/invoke.texi: Update -Wtautological-compare documentation.
10588 2017-09-04 Jeff Law <law@redhat.com>
10590 PR tree-optimization/64910
10591 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10592 swap the first and last operand if the last is a constant.
10594 2017-09-04 Marek Polacek <polacek@redhat.com>
10597 * convert.c (do_narrow): When sanitizing signed integer overflows,
10598 bail out for signed types.
10599 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10601 2017-09-04 Richard Biener <rguenther@suse.de>
10603 PR tree-optimization/82060
10604 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10605 Move devirtualization after stmt folding and before EH/AB/noreturn
10606 cleanup to get the stmt refs canonicalized. Use a bool instead
10607 of gimple_modified_p since that doesn't work for NOPs. Schedule
10608 NOPs generated by folding for removal.
10610 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10611 Alan Hayward <alan.hayward@arm.com>
10612 David Sherwood <david.sherwood@arm.com>
10614 * coretypes.h (pad_direction): New enum.
10615 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10616 (FUNCTION_ARG_PADDING): Likewise.
10617 * target.def (function_arg_padding): New hook.
10618 * targhooks.h (default_function_arg_padding): Declare.
10619 * targhooks.c (default_function_arg_padding): New function.
10620 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10621 (TARGET_FUNCTION_ARG_PADDING): ...this.
10622 * doc/tm.texi: Regenerate.
10623 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10624 instead of direction.
10625 (compute_argument_addresses): Likewise.
10626 (load_register_parameters): Likewise.
10627 (emit_library_call_value_1): Likewise.
10628 (store_one_arg): Use targetm.calls.function_arg_padding instead
10629 of FUNCTION_ARG_PADDING.
10630 (must_pass_in_stack_var_size_or_pad): Likewise.
10631 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10632 (emit_group_store): Likewise.
10633 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10634 instead of FUNCTION_ARG_PADDING.
10635 (emit_push_insn): Likewise, and propagate enum change throughout
10637 * function.h (direction): Delete.
10638 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10640 * function.c (assign_parm_find_stack_rtl): Likewise.
10641 (assign_parm_setup_block_p): Likewise.
10642 (assign_parm_setup_block): Likewise.
10643 (gimplify_parameters): Likewise.
10644 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10645 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10647 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10648 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10649 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10650 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10651 (aarch64_function_arg_padding): ...this new function.
10652 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10653 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10654 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10655 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10656 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10657 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10658 (arm_pad_arg_upward): Replace with...
10659 (arm_function_arg_padding): ...this new function.
10660 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10662 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10663 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10664 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10665 (ia64_hpux_function_arg_padding): Replace with...
10666 (ia64_function_arg_padding): ...this new function. Use pad_direction
10667 instead of direction. Check for TARGET_HPUX.
10668 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10669 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10670 (iq2000_function_arg_padding): New function.
10671 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10672 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10673 (mips_function_arg_padding): ...this new function.
10674 (mips_pad_reg_upward): Update accordingly.
10675 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10676 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10677 targetm.calls.function_arg_padding.
10678 (FUNCTION_ARG_PADDING): Delete.
10679 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10680 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10681 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10682 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10683 (nios2_block_reg_padding): Return pad_direction instead of direction.
10684 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10685 instead of direction.
10686 (nios2_function_arg_padding): Likewise. Make static.
10687 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10688 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10689 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10690 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10691 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10692 (pa_function_arg_padding): Make static. Return pad_direction instead
10694 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10695 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10696 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10697 instead of direction. Use targetm.calls.function_arg_padding.
10698 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10699 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10700 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10701 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10702 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10704 (function_arg_padding): Rename to...
10705 (rs6000_function_arg_padding): ...this. Make static. Return
10706 pad_direction instead of direction.
10707 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10708 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10709 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10710 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10711 instead of direction. Use targetm.calls.function_arg_padding.
10712 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10713 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10714 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10715 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10716 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10717 (function_arg_padding): Rename to...
10718 (rs6000_function_arg_padding): ...this. Make static. Return
10719 pad_direction instead of direction.
10720 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10721 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10722 * config/s390/s390.c (s390_function_arg_padding): New function.
10723 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10724 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10725 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10726 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10727 (function_arg_padding): Rename to...
10728 (sparc_function_arg_padding): ...this. Make static. Return
10729 pad_direction instead of direction.
10730 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10731 * config/spu/spu.c (spu_function_arg_padding): New function.
10732 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10733 * system.h (FUNCTION_ARG_PADDING): Poison.
10735 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10736 Alan Hayward <alan.hayward@arm.com>
10737 David Sherwood <david.sherwood@arm.com>
10739 * target.def (modes_tieable_p): New hook.
10740 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10741 (TARGET_MODES_TIEABLE_P): ...this.
10742 * doc/tm.texi.in: Regenerate.
10743 * hooks.h (hook_bool_mode_mode_true): Declare.
10744 * hooks.c (hook_bool_mode_mode_true): New function.
10745 * combine.c (subst): Use targetm.modes_tieable_p instead of
10747 * dse.c (find_shift_sequence): Likewise.
10748 * expmed.c (extract_low_bits): Likewise.
10749 * lower-subreg.c: Include target.h.
10750 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10752 * rtlanal.c (rtx_cost): Likewise.
10753 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10754 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10755 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10756 (TARGET_MODES_TIEABLE_P): Redefine.
10757 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10758 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10759 (TARGET_MODES_TIEABLE_P): Redefine.
10760 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10761 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10762 (arc_modes_tieable_p): New function.
10763 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10764 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10765 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10766 (arm_modes_tieable_p): Make static.
10767 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10768 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10769 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10770 (TARGET_MODES_TIEABLE_P): Redefine.
10771 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10772 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10773 (TARGET_MODES_TIEABLE_P): Redefine.
10774 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10775 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10776 (cr16_modes_tieable_p): New function.
10777 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10778 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10779 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10780 (TRULY_NOOP_TRUNCATION): Update comment.
10781 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10782 (TRULY_NOOP_TRUNCATION): Update comment.
10783 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10784 (frv_modes_tieable_p): New function.
10785 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10786 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10787 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10788 (TARGET_MODES_TIEABLE_P): Redefine.
10789 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10790 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10791 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10792 (TARGET_MODES_TIEABLE_P): Redefine.
10793 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10794 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10795 (ia64_modes_tieable_p): New function.
10796 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10797 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10798 (iq2000_modes_tieable_p): New function.
10799 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10800 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10801 (lm32_modes_tieable_p): New function.
10802 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10803 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10804 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10805 (TARGET_MODES_TIEABLE_P): Redefine.
10806 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10807 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10808 (m32r_modes_tieable_p): New function.
10809 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10810 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10811 (m68k_modes_tieable_p): New function.
10812 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10813 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10814 (mcore_modes_tieable_p): New function.
10815 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10816 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10818 (TARGET_MODES_TIEABLE_P): Redefine.
10819 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10820 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10821 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10822 (TARGET_MODES_TIEABLE_P): Redefine.
10823 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10824 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10825 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10826 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10827 (mn10300_modes_tieable_p): ...this and make static.
10828 (TARGET_MODES_TIEABLE_P): Redefine.
10829 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10830 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10831 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10832 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10833 (msp430_modes_tieable_p): Make static.
10834 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10835 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10836 (TARGET_MODES_TIEABLE_P): Redefine.
10837 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10838 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10839 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10840 (TARGET_MODES_TIEABLE_P): Redefine.
10841 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10842 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10843 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10844 (TARGET_MODES_TIEABLE_P): Redefine.
10845 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10846 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10847 (pdp11_modes_tieable_p): New function.
10848 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10849 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10850 (rs6000_modes_tieable_p): New function.
10851 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10852 * config/powerpcspe/powerpcspe.md: Update comment.
10853 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10854 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10855 (TARGET_MODES_TIEABLE_P): Redefine.
10856 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10857 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10858 (rl78_modes_tieable_p): New function.
10859 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10860 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10861 (rs6000_modes_tieable_p): New function.
10862 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10863 * config/rs6000/rs6000.md: Update comment.
10864 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10865 * config/rx/rx.c (rx_modes_tieable_p): New function.
10866 (TARGET_MODES_TIEABLE_P): Redefine.
10867 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10868 * config/s390/s390.c (s390_modes_tieable_p): New function.
10869 (TARGET_MODES_TIEABLE_P): Redefine.
10870 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10871 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10872 (sh_modes_tieable_p): New function.
10873 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10874 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10875 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10876 (sparc_modes_tieable_p): Make static.
10877 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10878 * config/spu/spu.c (spu_modes_tieable_p): New function.
10879 (TARGET_MODES_TIEABLE_P): Redefine.
10880 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10881 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10882 (TARGET_MODES_TIEABLE_P): Redefine.
10883 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10884 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10885 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10886 * config/v850/v850.c (v850_modes_tieable_p): New function.
10887 (TARGET_MODES_TIEABLE_P): Redefine.
10888 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10889 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10890 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10891 (visium_modes_tieable_p): New function.
10892 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10893 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10894 (xtensa_modes_tieable_p): New function.
10895 * system.h (MODES_TIEABLE_P): Poison.
10897 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10898 Alan Hayward <alan.hayward@arm.com>
10899 David Sherwood <david.sherwood@arm.com>
10901 * target.def (hard_regno_mode_ok): New hook.
10902 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10903 (TARGET_HARD_REGNO_MODE_OK): ...this.
10904 * doc/tm.texi.in: Regenerate.
10905 * hooks.h (hook_bool_uint_mode_true): Declare.
10906 * hooks.c (hook_bool_uint_mode_true): New function.
10907 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10908 HARD_REGNO_MODE_OK.
10909 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10910 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10911 instead of HARD_REGNO_MODE_OK.
10912 * caller-save.c: Include target.h.
10913 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10914 HARD_REGNO_MODE_OK.
10915 * combine.c (can_combine_p): Likewise.
10916 (combinable_i3pat): Likewise.
10917 (can_change_dest_mode): Likewise.
10918 * expr.c (init_expr_target): Likewise.
10919 (convert_move): Likewise.
10920 (convert_modes): Likewise.
10921 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10922 (setup_prohibited_mode_move_regs): Likewise.
10923 * ira.h (target_ira): Likewise.
10924 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10925 * lra-constraints.c (process_alt_operands): Likewise.
10926 (split_reg): Likewise.
10927 * recog.c (peep2_find_free_register): Likewise.
10928 * ree.c (combine_reaching_defs): Likewise.
10929 * regcprop.c (maybe_mode_change): Likewise.
10930 * reginfo.c (init_reg_sets_1): Likewise.
10931 (choose_hard_reg_mode): Likewise.
10932 (simplifiable_subregs): Likewise.
10933 * regrename.c (check_new_reg_p): Likewise.
10934 * reload.c (find_valid_class): Likewise.
10935 (find_valid_class_1): Likewise.
10936 (reload_inner_reg_of_subreg): Likewise.
10937 (push_reload): Likewise.
10938 (combine_reloads): Likewise.
10939 (find_dummy_reload): Likewise.
10940 (find_reloads): Likewise.
10941 * reload1.c (find_reg): Likewise.
10942 (set_reload_reg): Likewise.
10943 (allocate_reload_reg): Likewise.
10944 (choose_reload_regs): Likewise.
10945 (reload_adjust_reg_for_temp): Likewise.
10946 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10947 (simplify_subreg_regno): Likewise.
10948 * sel-sched.c (init_regs_for_mode): Likewise.
10949 * varasm.c (make_decl_rtl): Likewise.
10950 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10951 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10952 HARD_REGNO_MODE_OK.
10953 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10954 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10955 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10956 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10957 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10958 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10959 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10960 (arc_mode_class): Delete.
10961 (HARD_REGNO_MODE_OK): Delete.
10962 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10963 (arc_hard_regno_mode_ok): Rename old array to...
10964 (arc_hard_regno_mode_ok_modes): ...this.
10965 (arc_conditional_register_usage): Update accordingly.
10966 (arc_mode_class): Make static.
10967 (arc_hard_regno_mode_ok): New function.
10968 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10969 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10970 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10971 (arm_hard_regno_mode_ok): Make static.
10972 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10973 HARD_REGNO_MODE_OK.
10974 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10975 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10976 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10978 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10979 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10980 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10981 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10982 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10983 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10984 * config/bfin/predicates.md (valid_reg_operand): Use
10985 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10986 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10987 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10988 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10989 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10990 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10991 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10992 (cr16_hard_regno_mode_ok): Make static and return a bool.
10993 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10994 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10995 (cris_hard_regno_mode_ok): New function.
10996 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10997 (epiphany_mode_class): Delete.
10998 (HARD_REGNO_MODE_OK): Delete.
10999 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11000 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11001 (hard_regno_mode_ok): Rename to...
11002 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11003 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11004 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11005 HARD_REGNO_MODE_OK.
11006 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11007 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11008 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11009 (frv_hard_regno_mode_ok): Make static and return a bool.
11010 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11011 HARD_REGNO_MODE_OK.
11012 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11013 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11014 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11015 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11017 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11018 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11019 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11020 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11022 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11023 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11024 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11025 (ia64_hard_regno_mode_ok): New function.
11026 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11027 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11028 (iq2000_hard_regno_mode_ok): New function.
11029 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11030 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11031 (lm32_hard_regno_mode_ok): New function.
11032 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11033 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11034 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11035 instead of HARD_REGNO_MODE_OK.
11036 (m32c_hard_regno_ok): Rename to...
11037 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11038 (m32c_cannot_change_mode_class): Update accordingly.
11039 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11040 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11041 (m32r_mode_class): Delete.
11042 (HARD_REGNO_MODE_OK): Delete.
11043 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11044 (m32r_hard_regno_mode_ok): Rename to...
11045 (m32r_hard_regno_modes): ...this.
11046 (m32r_mode_class): Make static.
11047 (m32r_hard_regno_mode_ok): New function.
11048 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11049 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11050 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11051 (m68k_hard_regno_mode_ok): Make static.
11052 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11053 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11054 (mcore_hard_regno_mode_ok): New function.
11055 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11056 (HARD_REGNO_MODE_OK): Delete.
11057 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11059 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11060 (microblaze_hard_regno_mode_ok): New function.
11061 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11062 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11063 (mips_hard_regno_mode_ok): Delete.
11064 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11065 (mips_hard_regno_mode_ok_p): ...this and make static.
11066 (mips_hard_regno_mode_ok_p): Rename to...
11067 (mips_hard_regno_mode_ok_uncached): ...this.
11068 (mips_hard_regno_mode_ok): New function.
11069 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11070 of HARD_REGNO_MODE_OK.
11071 (mips_option_override): Update after above name changes.
11072 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11073 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11074 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11075 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11076 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11077 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11078 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11079 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11080 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11081 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11082 (msp430_hard_regno_mode_ok): Make static and return a bool.
11083 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11084 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11085 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11087 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11088 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11089 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11090 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11091 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11092 (PA_HARD_REGNO_MODE_OK): ...this
11093 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11094 (PA_HARD_REGNO_MODE_OK): ...this.
11095 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11096 (pa_hard_regno_mode_ok): New function.
11097 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11098 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11099 (pdp11_hard_regno_mode_ok): New function.
11100 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11101 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11103 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11105 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11106 (rs6000_hard_regno_mode_ok): Rename to...
11107 (rs6000_hard_regno_mode_ok_uncached): ...this.
11108 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11109 (rs6000_hard_regno_mode_ok): New function.
11110 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11111 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11112 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11113 (riscv_hard_regno_mode_ok): ...this and make static.
11114 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11115 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11116 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11117 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11118 (rl78_hard_regno_mode_ok): Make static and return bool.
11119 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11120 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11122 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11123 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11124 (rs6000_hard_regno_mode_ok): Rename to...
11125 (rs6000_hard_regno_mode_ok_uncached): ...this.
11126 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11127 (rs6000_hard_regno_mode_ok): New function.
11128 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11129 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11130 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11131 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11132 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11133 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11134 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11135 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11136 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11137 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11138 (sh_hard_regno_mode_ok): Make static.
11139 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11140 instead of HARD_REGNO_MODE_OK.
11141 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11142 (sparc_mode_class): Delete.
11143 (HARD_REGNO_MODE_OK): Delete.
11144 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11145 (hard_regno_mode_classes): Make static.
11146 (sparc_mode_class): Likewise.
11147 (sparc_hard_regno_mode_ok): New function.
11148 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11149 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11150 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11152 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11153 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11154 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11155 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11156 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11157 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11158 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11159 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11160 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11161 (visium_hard_regno_mode_ok): New function.
11162 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11163 instead of HARD_REGNO_MODE_OK.
11164 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11165 (HARD_REGNO_MODE_OK): Delete.
11166 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11167 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11168 (xtensa_option_override): Update accordingly.
11169 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11170 (xtensa_hard_regno_mode_ok): New function.
11171 * system.h (HARD_REGNO_MODE_OK): Poison.
11173 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11174 Alan Hayward <alan.hayward@arm.com>
11175 David Sherwood <david.sherwood@arm.com>
11177 * target.def (hard_regno_call_part_clobbered): New hook.
11178 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11179 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11180 * doc/tm.texi: Regenerate.
11181 * hooks.h (hook_bool_uint_mode_false): Declare.
11182 * hooks.c (hook_bool_uint_mode_false): New function.
11183 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11184 * cselib.c (cselib_process_insn): Use
11185 targetm.hard_regno_call_part_clobbered instead of
11186 HARD_REGNO_CALL_PART_CLOBBERED.
11187 * ira-conflicts.c (ira_build_conflicts): Likewise.
11188 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11189 * lra-constraints.c (need_for_call_save_p): Likewise.
11190 * lra-lives.c: Include target.h.
11191 (check_pseudos_live_through_calls): Use
11192 targetm.hard_regno_call_part_clobbered instead of
11193 HARD_REGNO_CALL_PART_CLOBBERED.
11194 * regcprop.c: Include target.h.
11195 (copyprop_hardreg_forward_1): Use
11196 targetm.hard_regno_call_part_clobbered instead of
11197 HARD_REGNO_CALL_PART_CLOBBERED.
11198 * reginfo.c (choose_hard_reg_mode): Likewise.
11199 * regrename.c (check_new_reg_p): Likewise.
11200 * reload.c (find_equiv_reg): Likewise.
11201 * reload1.c (emit_reload_insns): Likewise.
11202 * sched-deps.c (deps_analyze_insn): Likewise.
11203 * sel-sched.c (init_regs_for_mode): Likewise.
11204 (mark_unavailable_hard_regs): Likewise.
11205 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11206 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11207 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11209 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11210 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11211 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11213 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11215 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11216 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11217 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11219 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11220 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11221 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11223 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11224 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11226 * config/powerpcspe/powerpcspe.c
11227 (rs6000_hard_regno_call_part_clobbered): New function.
11228 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11229 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11230 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11232 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11233 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11234 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11236 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11237 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11238 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11240 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11241 Alan Hayward <alan.hayward@arm.com>
11242 David Sherwood <david.sherwood@arm.com>
11244 * rtl.h (subreg_memory_offset): Declare.
11245 * emit-rtl.c (subreg_memory_offset): New function.
11246 * expmed.c (store_bit_field_1): Use it.
11247 * expr.c (undefined_operand_subword_p): Likewise.
11248 * simplify-rtx.c (simplify_subreg): Likewise.
11250 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11252 PR rtl-optimization/57448
11255 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11256 using atomic_load pattern.
11257 (expand_atomic_store): Likewise.
11259 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11262 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11263 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11264 handling. Use replace_call_with_value with NULL instead of
11265 gsi_replace, unlink_stmt_vdef and release_defs.
11267 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11270 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11272 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11275 * rtl.h (emit_library_call_value_1): Declare.
11276 (emit_library_call): Replace declaration with a series of overloads.
11277 Remove the parameter count argument.
11278 (emit_library_call_value): Likewise.
11279 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11280 with an "rtx_mode_t *".
11281 (emit_library_call_value): Delete.
11282 (emit_library_call): Likewise.
11283 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11284 (asan_emit_allocas_unpoison): Likewise.
11285 * builtins.c (expand_builtin_powi): Likewise.
11286 (expand_asan_emit_allocas_unpoison): Likewise.
11287 * cfgexpand.c (expand_main_function): Likewise.
11288 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11289 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11290 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11291 * config/arm/arm.c (arm_trampoline_init): Likewise.
11292 (arm_call_tls_get_addr): Likewise.
11293 (arm_expand_divmod_libfunc): Likewise.
11294 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11295 (smulsi3_highpart): Likewise.
11296 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11297 (c6x_expand_compare): Likewise.
11298 (c6x_expand_movmem): Likewise.
11299 * config/frv/frv.c (frv_trampoline_init): Likewise.
11300 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11301 (ix86_expand_divmod_libfunc): Likewise.
11302 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11303 (ia64_expand_compare): Likewise.
11304 (ia64_profile_hook): Likewise.
11305 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11306 (nonlocal_goto): Likewise.
11307 (restore_stack_nonlocal): Likewise.
11308 * config/m32r/m32r.c (block_move_call): Likewise.
11309 (m32r_trampoline_init): Likewise.
11310 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11311 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11312 (m68k_call_m68k_read_tp): Likewise.
11313 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11314 (microblaze_expand_divide): Likewise.
11315 * config/mips/mips.h (mips_args): Likewise.
11316 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11317 (MIPS_ICACHE_SYNC): Likewise.
11318 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11319 (nios2_trampoline_init): Likewise.
11320 * config/pa/pa.c (hppa_tls_call): Likewise.
11321 (pa_trampoline_init): Likewise.
11322 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11323 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11324 (expand_strn_compare): Likewise.
11325 (rs6000_generate_compare): Likewise.
11326 (rs6000_expand_float128_convert): Likewise.
11327 (output_profile_hook): Likewise.
11328 (rs6000_trampoline_init): Likewise.
11329 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11330 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11331 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11332 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11333 (rs6000_generate_compare): Likewise.
11334 (rs6000_expand_float128_convert): Likewise.
11335 (output_profile_hook): Likewise.
11336 (rs6000_trampoline_init): Likewise.
11337 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11338 * config/sh/sh.c (sh_trampoline_init): Likewise.
11339 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11340 (sparc_emit_float_lib_cmp): Likewise.
11341 (sparc32_initialize_trampoline): Likewise.
11342 (sparc64_initialize_trampoline): Likewise.
11343 (sparc_profile_hook): Likewise.
11344 * config/spu/spu.c (ea_load_store): Likewise.
11345 * config/spu/spu.md (floatunssidf2): Likewise.
11346 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11347 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11348 * config/visium/visium.c (expand_block_move_4): Likewise.
11349 (expand_block_move_2): Likewise.
11350 (expand_block_move_1): Likewise.
11351 (expand_block_set_4): Likewise.
11352 (expand_block_set_2): Likewise.
11353 (expand_block_set_1): Likewise.
11354 (visium_trampoline_init): Likewise.
11355 (visium_profile_hook): Likewise.
11356 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11357 (xtensa_setup_frame_addresses): Likewise.
11358 (xtensa_trampoline_init): Likewise.
11359 * except.c (sjlj_emit_function_enter): Likewise.
11360 (sjlj_emit_function_exit): Likewise.
11361 * explow.c (allocate_dynamic_stack_space): Likewise.
11362 (probe_stack_range): Likewise.
11363 * expr.c (convert_mode_scalar): Likewise.
11364 * optabs.c (expand_binop): Likewise.
11365 (expand_twoval_binop_libfunc): Likewise.
11366 (expand_unop): Likewise.
11367 (prepare_cmp_insn): Likewise.
11368 (prepare_float_lib_cmp): Likewise.
11369 (expand_float): Likewise.
11370 (expand_fix): Likewise.
11371 (expand_fixed_convert): Likewise.
11372 (maybe_emit_sync_lock_test_and_set): Likewise.
11373 (expand_atomic_compare_and_swap): Likewise.
11374 (expand_mem_thread_fence): Likewise.
11375 (expand_atomic_fetch_op): Likewise.
11377 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11379 * doc/generic.texi (OpenACC): Adjust URL.
11380 * doc/invoke.texi (C Dialect Options): Ditto.
11382 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11384 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11385 predicate for operand 1. Add (m,<S>) constraint.
11386 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11387 Prevent memory operand 1 with register operand 2.
11389 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11391 PR rtl-optimization/82024
11392 * combine.c (try_combine): If the combination result is a PARALLEL,
11393 and we only need to retain the SET in there that would be placed
11394 at I2, check that we can place that at I3 instead, before doing so.
11396 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11399 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11401 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11402 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11405 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11406 Jeff Law <law@redhat.com>
11408 * varasm.c (bss_initializer_p): Do not put constants into .bss
11409 (categorize_decl_for_section): Handle bss_initializer_p returning
11410 false when DECL_INITIAL is NULL.
11412 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11415 * config/s390/s390.c (s390_can_inline_p): New function.
11417 2017-09-01 Jeff Law <law@redhat.com>
11419 PR tree-optimization/82052
11420 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11421 Always initialize the returned slot after a hash table miss
11422 when INSERT is true.
11424 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11426 * config/s390/s390.md (mem_signal_fence): Remove.
11427 * doc/md.texi (mem_signal_fence): Remove.
11428 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11430 * target-insns.def (mem_signal_fence): Remove.
11432 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11435 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11438 * asan.c (create_odr_indicator): Strip name encoding from assembler
11439 name before appending it after __odr_asan_.
11441 2017-09-01 Martin Liska <mliska@suse.cz>
11443 PR tree-optimization/82059
11444 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11445 frequency only when an edge is redirected.
11447 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11449 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11450 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11451 (arc_conditional_register_usage): Remove ARC600 lp_count
11453 (arc_file_start): Emit Tag_ARC_CPU_variation.
11454 (arc_can_use_doloop_p): New conditions to use ZOLs.
11455 (hwloop_fail): New function.
11456 (hwloop_optimize): Likewise.
11457 (hwloop_pattern_reg): Likewise.
11458 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11459 (arc_reorg_loops): New function, calls reorg_loops.
11460 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11461 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11463 (arc_loop_hazard): Remove function, functionality moved into
11465 (arc_hazard): Remove arc_loop_hazard call.
11466 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11467 into hwloop_optimize.
11468 (arc_label_align): Remove ZOL handling.
11469 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11470 * config/arc/arc.md (doloop_begin): Remove pattern.
11471 (doloop_begin_i): Likewise.
11472 (doloop_end_i): Likewise.
11473 (doloop_fallback): Likewise.
11474 (doloop_fallback_m): Likewise.
11475 (doloop_end): Reimplement expand.
11476 (arc_lp): New pattern for LP instruction.
11477 (loop_end): New pattern.
11478 (loop_fail): Likewise.
11479 (decrement_and_branch_until_zero): Likewise.
11480 * config/arc/arc.opt (mlpc-width): New option.
11481 * doc/invoke.texi (mlpc-width): Document option.
11483 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11485 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11486 (arc_ccfsm_advance): Fix checking for delay slots.
11487 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11489 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11491 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11493 (movhi_insn): Update store instruction constraint which are saving
11494 6-bit short immediates.
11495 (movsi_insn): Consider also short scaled load operations.
11496 (zero_extendhisi2_i): Use Usd constraint instead of T.
11497 (extendhisi2_i): Add q constraint.
11498 (arc_clzsi2): Add type and length attributes.
11499 (arc_ctzsi2): Likewise.
11500 * config/arc/constraints.md (Usc): Update constraint, the
11501 assembler can parse two relocations for a single instruction.
11503 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11505 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11506 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11508 2017-08-31 Olivier Hainque <hainque@adacore.com>
11510 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11511 match as powerpc-wrs-vxworks*.
11513 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11515 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11516 register constraint for by-element operand.
11517 (aarch64_mls_elt_merge<mode>): Likewise.
11519 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11521 * config/arc/arc.c (arc_can_follow_jump): Check for short
11524 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11526 * config.gcc: Use g.opt for arc.
11527 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11528 functionality moved to ...
11529 (legitimate_scaled_address_p): New function, ...here.
11530 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11531 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11532 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11534 (arc_override_options): Handle G option.
11535 (arc_output_pic_addr_const): Correct function definition.
11536 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11537 (arc_decl_anon_ns_mem_p): Delete.
11538 (arc_in_small_data_p): Overhaul this function to take into
11539 consideration the value given via G option.
11540 (arc_rewrite_small_data_1): Renamed and corrected old
11541 arc_rewrite_small_data function.
11542 (arc_rewrite_small_data): New function.
11543 (small_data_pattern): Don't use pic_offset_table_rtx.
11544 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11545 * config/arc/simdext.md (movmisalignv2hi): Use
11546 prepare_move_operands function.
11548 (movmisalign*): Likewise.
11549 * doc/invoke.texi (ARC options): Document -G option.
11551 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11553 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11555 * config/arc/arc.c (arc_print_operand): Output scalled address for
11556 sdata whenever is possible.
11557 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11558 load/stores are available.
11559 (compact_sda_memory_operand): Check for the alignment required by
11560 code density instructions.
11561 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11563 * config/arc/constraints.md (Usd): Update constraint.
11564 (Us0): New constraint.
11565 (Usc): Update constraint.
11567 2017-08-31 Richard Biener <rguenther@suse.de>
11569 PR middle-end/82054
11570 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11571 function only once.
11573 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11575 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11576 Resize type_signature.
11578 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11579 Alan Hayward <alan.hayward@arm.com>
11580 David Sherwood <david.sherwood@arm.com>
11582 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11583 subregs whose inner modes can be stored in GPRs.
11584 (aarch64_classify_index): Likewise.
11586 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11587 Alan Hayward <alan.hayward@arm.com>
11588 David Sherwood <david.sherwood@arm.com>
11590 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11591 (V_INT_EQUIV): ...this.
11592 (v_cmp_result): Rename to...
11593 (v_int_equiv): ...this.
11594 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11595 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11596 (copysign<mode>3): Likewise.
11597 (aarch64_simd_bsl<mode>_internal): Likewise.
11598 (aarch64_simd_bsl<mode>): Likewise.
11599 (vec_cmp<mode><mode>): Likewise.
11600 (vcond<mode><mode>): Likewise.
11601 (vcond<v_cmp_mixed><mode>): Likewise.
11602 (vcondu<mode><v_cmp_mixed>): Likewise.
11603 (aarch64_cm<optab><mode>): Likewise.
11604 (aarch64_cmtst<mode>): Likewise.
11605 (aarch64_fac<optab><mode>): Likewise.
11606 (vec_perm_const<mode>): Likewise.
11607 (vcond_mask_<mode><v_cmp_result>): Rename to...
11608 (vcond_mask_<mode><v_int_equiv>): ...this.
11609 (vec_cmp<mode><v_cmp_result>): Rename to...
11610 (vec_cmp<mode><v_int_equiv>): ...this.
11612 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11613 Alan Hayward <alan.hayward@arm.com>
11614 David Sherwood <david.sherwood@arm.com>
11616 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11618 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11619 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11620 (UNSPEC_LD4_DREG): New unspecs.
11621 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11622 (aarch64_ld2<mode>_dreg_be): Replace with...
11623 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11625 (aarch64_ld3<mode>_dreg_le)
11626 (aarch64_ld3<mode>_dreg_be): Replace with...
11627 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11629 (aarch64_ld4<mode>_dreg_le)
11630 (aarch64_ld4<mode>_dreg_be): Replace with...
11631 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11634 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11636 PR tree-optimization/81987
11637 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11638 insert an initializer in a location not dominated by the stride
11641 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11643 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11644 on the entire header of the finally block in the fallthru case.
11646 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11648 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11650 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11652 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11653 rs6000_emit_move_from_cr and call renamed function.
11654 (rs6000_emit_prologue): Call renamed functions.
11655 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11656 movesi_from_cr, remove volatile CRs.
11658 2017-08-30 Jon Beniston <jon@beniston.com>
11659 Richard Biener <rguenther@suse.de>
11661 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11662 of VECTOR_MODE_P check.
11663 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11664 element vector types.
11666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11668 * df.h (df_read_modify_subreg_p): Remove in favor of...
11669 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11670 const_rtx instead of an rtx.
11671 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11672 * df-problems.c (df_word_lr_mark_ref): Likewise.
11673 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11674 (mark_pseudo_reg_dead): Likewise.
11675 (mark_ref_dead): Likewise.
11676 * reginfo.c (init_subregs_of_mode): Likewise.
11677 * sched-deps.c (sched_analyze_1): Likewise.
11678 * df-scan.c (df_def_record_1): Likewise.
11679 (df_uses_record): Likewise.
11680 (df_read_modify_subreg_p): Remove in favor of...
11681 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11682 const_rtx instead of an rtx.
11684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11685 Alan Hayward <alan.hayward@arm.com>
11686 David Sherwood <david.sherwood@arm.com>
11688 * rtl.h (partial_subreg_p): New function.
11689 * caller-save.c (save_call_clobbered_regs): Use it.
11690 * calls.c (expand_call): Likewise.
11691 * combine.c (combinable_i3pat): Likewise.
11692 (simplify_set): Likewise.
11693 (make_extraction): Likewise.
11694 (make_compound_operation_int): Likewise.
11695 (gen_lowpart_or_truncate): Likewise.
11696 (force_to_mode): Likewise.
11697 (make_field_assignment): Likewise.
11698 (reg_truncated_to_mode): Likewise.
11699 (record_truncated_value): Likewise.
11700 (move_deaths): Likewise.
11701 * cse.c (record_jump_cond): Likewise.
11702 (cse_insn): Likewise.
11703 * cselib.c (cselib_lookup_1): Likewise.
11704 * expmed.c (extract_bit_field_using_extv): Likewise.
11705 * function.c (assign_parm_setup_reg): Likewise.
11706 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11707 * ira-build.c (create_insn_allocnos): Likewise.
11708 * lra-coalesce.c (merge_pseudos): Likewise.
11709 * lra-constraints.c (match_reload): Likewise.
11710 (simplify_operand_subreg): Likewise.
11711 (curr_insn_transform): Likewise.
11712 * lra-lives.c (process_bb_lives): Likewise.
11713 * lra.c (new_insn_reg): Likewise.
11714 (lra_substitute_pseudo): Likewise.
11715 * regcprop.c (mode_change_ok): Likewise.
11716 (maybe_mode_change): Likewise.
11717 (copyprop_hardreg_forward_1): Likewise.
11718 * reload.c (push_reload): Likewise.
11719 (find_reloads): Likewise.
11720 (find_reloads_subreg_address): Likewise.
11721 * reload1.c (alter_reg): Likewise.
11722 (eliminate_regs_1): Likewise.
11723 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11725 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11727 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11728 back to if statements, including unpack.
11730 2017-08-30 Martin Liska <mliska@suse.cz>
11732 PR inline-asm/82001
11733 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11735 (func_checker::compare_asm_inputs_outputs): ... this function.
11736 (func_checker::compare_gimple_asm): Use the function to compare
11737 also ASM constrains.
11738 * ipa-icf-gimple.h: Rename the function.
11740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11741 Alan Hayward <alan.hayward@arm.com>
11742 David Sherwood <david.sherwood@arm.com>
11744 * coretypes.h (complex_mode): New type.
11745 * gdbhooks.py (build_pretty_printer): Handle it.
11746 * machmode.h (complex_mode): New class.
11747 (complex_mode::includes_p): New function.
11748 (is_complex_int_mode): Likewise.
11749 (is_complex_float_mode): Likewise.
11750 * genmodes.c (get_mode_class): Handle complex mode classes.
11751 * function.c (expand_function_end): Use is_complex_int_mode.
11753 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11754 Alan Hayward <alan.hayward@arm.com>
11755 David Sherwood <david.sherwood@arm.com>
11757 * coretypes.h (scalar_mode_pod): New typedef.
11758 * gdbhooks.py (build_pretty_printer): Handle it.
11759 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11760 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11761 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11762 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11763 as_a <scalar_mode>.
11765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11766 Alan Hayward <alan.hayward@arm.com>
11767 David Sherwood <david.sherwood@arm.com>
11769 * machmode.h (mode_for_vector): Take a scalar_mode instead
11771 * stor-layout.c (mode_for_vector): Likewise.
11772 * explow.c (promote_mode): Use as_a <scalar_mode>.
11773 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11775 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11776 Alan Hayward <alan.hayward@arm.com>
11777 David Sherwood <david.sherwood@arm.com>
11779 * target.def (preferred_simd_mode): Take a scalar_mode
11780 instead of a machine_mode.
11781 * targhooks.h (default_preferred_simd_mode): Likewise.
11782 * targhooks.c (default_preferred_simd_mode): Likewise.
11783 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11784 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11785 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11786 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11787 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11788 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11789 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11790 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11792 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11793 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11794 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11795 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11796 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11797 * doc/tm.texi: Regenerate.
11798 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11802 Alan Hayward <alan.hayward@arm.com>
11803 David Sherwood <david.sherwood@arm.com>
11805 * target.def (scalar_mode_supported_p): Take a scalar_mode
11806 instead of a machine_mode.
11807 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11808 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11809 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11810 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11811 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11812 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11813 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11814 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11815 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11816 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11817 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11818 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11819 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11820 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11822 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11823 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11824 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11825 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11826 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11828 * doc/tm.texi: Regenerate.
11830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11831 Alan Hayward <alan.hayward@arm.com>
11832 David Sherwood <david.sherwood@arm.com>
11834 * coretypes.h (opt_scalar_mode): New typedef.
11835 * gdbhooks.py (build_pretty_printers): Handle it.
11836 * machmode.h (mode_iterator::get_2xwider): Add overload for
11838 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11840 * expr.c (convert_mode_scalar): Likewise.
11841 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11842 * optabs.c (expand_float): Likewise.
11843 (expand_fix): Likewise.
11844 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11847 Alan Hayward <alan.hayward@arm.com>
11848 David Sherwood <david.sherwood@arm.com>
11850 * optabs.c (expand_float): Explicitly check for scalars before
11851 using a branching expansion.
11852 (expand_fix): Likewise.
11854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11855 Alan Hayward <alan.hayward@arm.com>
11856 David Sherwood <david.sherwood@arm.com>
11858 * expr.c (convert_mode): Split scalar handling out into...
11859 (convert_mode_scalar): ...this new function. Treat the modes
11862 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11863 Alan Hayward <alan.hayward@arm.com>
11864 David Sherwood <david.sherwood@arm.com>
11866 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11868 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11871 Alan Hayward <alan.hayward@arm.com>
11872 David Sherwood <david.sherwood@arm.com>
11874 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11875 rather than a machine_mode.
11876 (fixed_from_string): Likewise.
11877 (fixed_convert): Likewise.
11878 (fixed_convert_from_int): Likewise.
11879 (fixed_convert_from_real): Likewise.
11880 (real_convert_from_fixed): Likewise.
11881 * fixed-value.c (fixed_from_double_int): Likewise.
11882 (fixed_from_string): Likewise.
11883 (fixed_convert): Likewise.
11884 (fixed_convert_from_int): Likewise.
11885 (fixed_convert_from_real): Likewise.
11886 (real_convert_from_fixed): Likewise.
11887 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11890 Alan Hayward <alan.hayward@arm.com>
11891 David Sherwood <david.sherwood@arm.com>
11893 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11894 of separate mode class checks. Do not allow vector modes here.
11895 (immed_wide_int_const): Use as_a <scalar_mode>.
11896 * explow.c (trunc_int_for_mode): Likewise.
11897 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11898 (wi::shwi): Likewise.
11899 (wi::min_value): Likewise.
11900 (wi::max_value): Likewise.
11901 * dwarf2out.c (loc_descriptor): Likewise.
11902 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11903 for CONST_WIDE_INT.
11905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11906 Alan Hayward <alan.hayward@arm.com>
11907 David Sherwood <david.sherwood@arm.com>
11909 * tree.h (SCALAR_TYPE_MODE): New macro.
11910 * expr.c (expand_expr_addr_expr_1): Use it.
11911 (expand_expr_real_2): Likewise.
11912 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11913 (fold_convert_const_fixed_from_int): Likewise.
11914 (fold_convert_const_fixed_from_real): Likewise.
11915 (native_encode_fixed): Likewise
11916 (native_encode_complex): Likewise
11917 (native_encode_vector): Likewise.
11918 (native_interpret_fixed): Likewise.
11919 (native_interpret_real): Likewise.
11920 (native_interpret_complex): Likewise.
11921 (native_interpret_vector): Likewise.
11922 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11923 (simd_clone_adjust_argument_types): Likewise.
11924 (simd_clone_init_simd_arrays): Likewise.
11925 (simd_clone_adjust): Likewise.
11926 * stor-layout.c (layout_type): Likewise.
11927 * tree.c (build_minus_one_cst): Likewise.
11928 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11929 * tree-inline.c (estimate_move_cost): Likewise.
11930 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11931 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11932 (vectorizable_reduction): Likewise.
11933 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11934 (vect_recog_mixed_size_cond_pattern): Likewise.
11935 (check_bool_pattern): Likewise.
11936 (adjust_bool_pattern): Likewise.
11937 (search_type_for_mask_1): Likewise.
11938 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11939 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11940 (vectorizable_load): Likewise.
11941 (vectorizable_store): Likewise.
11942 * ubsan.c (ubsan_encode_value): Likewise.
11943 * varasm.c (output_constant): Likewise.
11945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11946 Alan Hayward <alan.hayward@arm.com>
11947 David Sherwood <david.sherwood@arm.com>
11949 * coretypes.h (scalar_mode): New class.
11950 * machmode.h (scalar_mode): Likewise.
11951 (scalar_mode::includes_p): New function.
11952 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11953 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11954 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11955 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11956 * expmed.c (store_bit_field_1): Likewise.
11957 (extract_bit_field_1): Likewise.
11958 * expr.c (write_complex_part): Likewise.
11959 (read_complex_part): Likewise.
11960 (emit_move_complex_push): Likewise.
11961 (expand_expr_real_2): Likewise.
11962 * function.c (assign_parm_setup_reg): Likewise.
11963 (assign_parms_unsplit_complex): Likewise.
11964 * optabs.c (expand_binop): Likewise.
11965 * rtlanal.c (subreg_get_info): Likewise.
11966 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11967 * varasm.c (output_constant_pool_2): Likewise.
11969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11970 Alan Hayward <alan.hayward@arm.com>
11971 David Sherwood <david.sherwood@arm.com>
11973 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11975 (expmed_mult_highpart_optab): Likewise.
11976 (expmed_mult_highpart): Likewise.
11978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11979 Alan Hayward <alan.hayward@arm.com>
11980 David Sherwood <david.sherwood@arm.com>
11982 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11983 instead of a machine_mode.
11984 (builtin_memset_read_str): Likewise.
11985 * builtins.c (c_readstr): Likewise.
11986 (builtin_memcpy_read_str): Likewise.
11987 (builtin_strncpy_read_str): Likewise.
11988 (builtin_memset_read_str): Likewise.
11989 (builtin_memset_gen_str): Likewise.
11990 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11991 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11992 instead of a machine_mode.
11993 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11995 (make_extraction): Likewise.
11996 (try_widen_shift_mode): Take and return scalar_int_modes instead
11998 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11999 a scalar_int_mode instead of a machine_mode.
12000 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12001 (avr_addr_space_pointer_mode): Likewise.
12002 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12003 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12004 (msp430_unwind_word_mode): Likewise.
12005 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12006 (spu_addr_space_pointer_mode): Likewise.
12007 (spu_addr_space_address_mode): Likewise.
12008 (spu_libgcc_cmp_return_mode): Likewise.
12009 (spu_libgcc_shift_count_mode): Likewise.
12010 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12011 (rl78_addr_space_pointer_mode): Likewise.
12012 (fl78_unwind_word_mode): Likewise.
12013 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12015 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12016 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12017 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12018 (mips_valid_pointer_mode): Likewise.
12019 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12020 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12021 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12023 (ft32_addr_space_address_mode): Likewise.
12024 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12025 scalar_int_mode instead of a machine_mode.
12026 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12028 (m32c_addr_space_address_mode): Likewise.
12029 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12030 (rs6000_eh_return_filter_mode): Likewise.
12031 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12032 (rs6000_eh_return_filter_mode): Likewise.
12033 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12034 (s390_libgcc_shift_count_mode): Likewise.
12035 (s390_unwind_word_mode): Likewise.
12036 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12038 * target.def (mode_rep_extended): Likewise.
12039 (valid_pointer_mode): Likewise.
12040 (addr_space.valid_pointer_mode): Likewise.
12041 (eh_return_filter_mode): Return a scalar_int_mode rather than
12043 (libgcc_cmp_return_mode): Likewise.
12044 (libgcc_shift_count_mode): Likewise.
12045 (unwind_word_mode): Likewise.
12046 (addr_space.pointer_mode): Likewise.
12047 (addr_space.address_mode): Likewise.
12048 * doc/tm.texi: Regenerate.
12049 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12051 (do_jump): Use scalar_int_mode for local variables.
12052 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12053 rather than a machine_mode.
12054 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12055 (scompare_loc_descriptor_wide): Likewise.
12056 (scompare_loc_descriptor_narrow): Likewise.
12057 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12059 * except.c (sjlj_emit_dispatch_table): Likewise.
12060 (expand_builtin_eh_copy_values): Likewise.
12061 * explow.c (convert_memory_address_addr_space_1): Likewise.
12062 Take a scalar_int_mode rather than a machine_mode.
12063 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12064 than a machine_mode.
12065 (memory_address_addr_space): Use scalar_int_mode for local variables.
12066 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12067 rather than a machine_mode.
12068 * expmed.c (mask_rtx): Likewise.
12069 (init_expmed_one_conv): Likewise.
12070 (expand_mult_highpart_adjust): Likewise.
12071 (extract_high_half): Likewise.
12072 (expmed_mult_highpart_optab): Likewise.
12073 (expmed_mult_highpart): Likewise.
12074 (expand_smod_pow2): Likewise.
12075 (expand_sdiv_pow2): Likewise.
12076 (emit_store_flag_int): Likewise.
12077 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12079 (extract_low_bits): Likewise.
12080 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12082 * expr.c (pieces_addr::adjust): Likewise.
12083 (can_store_by_pieces): Likewise.
12084 (store_by_pieces): Likewise.
12085 (clear_by_pieces_1): Likewise.
12086 (expand_expr_addr_expr_1): Likewise.
12087 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12088 (expand_expr_real_1): Likewise.
12089 (try_casesi): Likewise.
12090 * final.c (shorten_branches): Likewise.
12091 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12092 type of "mode" to machine_mode.
12093 * internal-fn.c (expand_arith_overflow_result_store): Take a
12094 scalar_int_mode rather than a machine_mode.
12095 (expand_mul_overflow): Use scalar_int_mode for local variables.
12096 * loop-doloop.c (doloop_modify): Likewise.
12097 (doloop_optimize): Likewise.
12098 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12099 than a machine_mode.
12100 (expand_doubleword_shift_condmove): Likewise.
12101 (expand_doubleword_shift): Likewise.
12102 (expand_doubleword_clz): Likewise.
12103 (expand_doubleword_popcount): Likewise.
12104 (expand_doubleword_parity): Likewise.
12105 (expand_absneg_bit): Use scalar_int_mode for local variables.
12106 (prepare_float_lib_cmp): Likewise.
12107 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12108 rather than a machine_mode.
12109 (convert_memory_address_addr_space): Likewise.
12110 (get_mode_bounds): Likewise.
12111 (get_address_mode): Return a scalar_int_mode rather than a
12113 * rtlanal.c (get_address_mode): Likewise.
12114 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12115 than a machine_mode.
12116 * targhooks.c (default_mode_rep_extended): Likewise.
12117 (default_valid_pointer_mode): Likewise.
12118 (default_addr_space_valid_pointer_mode): Likewise.
12119 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12120 than a machine_mode.
12121 (default_libgcc_cmp_return_mode): Likewise.
12122 (default_libgcc_shift_count_mode): Likewise.
12123 (default_unwind_word_mode): Likewise.
12124 (default_addr_space_pointer_mode): Likewise.
12125 (default_addr_space_address_mode): Likewise.
12126 * targhooks.h (default_eh_return_filter_mode): Likewise.
12127 (default_libgcc_cmp_return_mode): Likewise.
12128 (default_libgcc_shift_count_mode): Likewise.
12129 (default_unwind_word_mode): Likewise.
12130 (default_addr_space_pointer_mode): Likewise.
12131 (default_addr_space_address_mode): Likewise.
12132 (default_mode_rep_extended): Take a scalar_int_mode rather than
12134 (default_valid_pointer_mode): Likewise.
12135 (default_addr_space_valid_pointer_mode): Likewise.
12136 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12138 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12139 rather than a machine_mode.
12140 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12141 for local variables.
12142 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12143 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12144 than a machine_mode.
12146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12147 Alan Hayward <alan.hayward@arm.com>
12148 David Sherwood <david.sherwood@arm.com>
12150 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12151 the mode argument to scalar_int_mode.
12152 (do_jump_by_parts_zero_rtx): Likewise.
12153 (do_jump_by_parts_equality_rtx): Likewise.
12154 (do_jump_by_parts_greater): Take a mode argument.
12155 (do_jump_by_parts_equality): Likewise.
12156 (do_jump_1): Update calls accordingly.
12158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12159 Alan Hayward <alan.hayward@arm.com>
12160 David Sherwood <david.sherwood@arm.com>
12162 * is-a.h (safe_dyn_cast): New function.
12163 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12164 (jump_table_for_label): Likewise.
12165 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12166 instead of an rtx_insn *.
12167 (shorten_branches): Use dyn_cast instead of LABEL_P and
12168 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12169 rtx_jump_table_data::get_data_mode.
12170 (final_scan_insn): Likewise.
12172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12173 Alan Hayward <alan.hayward@arm.com>
12174 David Sherwood <david.sherwood@arm.com>
12176 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12177 trying to combine a full-register integer set with a subreg
12180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12181 Alan Hayward <alan.hayward@arm.com>
12182 David Sherwood <david.sherwood@arm.com>
12184 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12185 that is always either address_mode or pointer_mode.
12187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12188 Alan Hayward <alan.hayward@arm.com>
12189 David Sherwood <david.sherwood@arm.com>
12191 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12192 when the two are known to be equal.
12194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12195 Alan Hayward <alan.hayward@arm.com>
12196 David Sherwood <david.sherwood@arm.com>
12198 * simplify-rtx.c (simplify_const_unary_operation): Use
12199 is_a <scalar_int_mode> instead of checking for a nonzero
12200 precision. Forcibly convert op_mode to a scalar_int_mode
12201 in that case. More clearly differentiate the operand and
12202 result modes and use the former when deciding what the value
12203 of a count-bits operation should be. Use is_int_mode instead
12204 of checking for a MODE_INT. Remove redundant check for whether
12205 this mode has a zero precision.
12207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12208 Alan Hayward <alan.hayward@arm.com>
12209 David Sherwood <david.sherwood@arm.com>
12211 * optabs.c (widen_leading): Change the type of the mode argument
12212 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12213 (widen_bswap): Likewise.
12214 (expand_parity): Likewise.
12215 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12216 (expand_ffs): Likewise.
12217 (epand_unop): Check for scalar integer modes before calling the
12220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12221 Alan Hayward <alan.hayward@arm.com>
12222 David Sherwood <david.sherwood@arm.com>
12224 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12226 (expand_expr_real_1): Update call accordingly.
12228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12229 Alan Hayward <alan.hayward@arm.com>
12230 David Sherwood <david.sherwood@arm.com>
12232 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12233 value_mode arguments. Use scalar_int_mode internally.
12234 (store_bit_field_1): Rename the new integer mode from imode
12235 to op0_mode and use it instead of GET_MODE (op0). Update calls
12236 to store_split_bit_field, store_bit_field_using_insv and
12237 store_fixed_bit_field.
12238 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12239 Use scalar_int_mode internally. Use a bit count rather than a mode
12240 when calculating the largest bit size for get_best_mode.
12241 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12242 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12243 Remove assertion that OP0 has a scalar integer mode.
12244 (store_split_bit_field): Add op0_mode and value_mode arguments.
12245 Update calls to extract_fixed_bit_field.
12246 (extract_bit_field_using_extv): Add an op0_mode argument.
12247 Use scalar_int_mode internally.
12248 (extract_bit_field_1): Rename the new integer mode from imode to
12249 op0_mode and use it instead of GET_MODE (op0). Update calls to
12250 extract_split_bit_field, extract_bit_field_using_extv and
12251 extract_fixed_bit_field.
12252 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12253 to extract_split_bit_field and extract_fixed_bit_field_1.
12254 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12255 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12256 on the target mode.
12257 (extract_split_bit_field): Add an op0_mode argument. Update call
12258 to extract_fixed_bit_field.
12260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12261 Alan Hayward <alan.hayward@arm.com>
12262 David Sherwood <david.sherwood@arm.com>
12264 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12265 * explow.c (hard_function_value): Likewise.
12266 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12267 convert_to_mode call outside the loop.
12268 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12269 for the mode iterator. Require the mode specified by max_pieces
12271 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12273 (copy_blkmode_to_reg): Likewise.
12274 (set_storage_via_setmem): Likewise.
12275 * optabs.c (prepare_cmp_insn): Likewise.
12276 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12277 * stor-layout.c (finish_bitfield_representative): Likewise.
12279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12280 Alan Hayward <alan.hayward@arm.com>
12281 David Sherwood <david.sherwood@arm.com>
12283 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12284 * expr.c (convert_move): Use them.
12285 (convert_modes): Likewise.
12286 (store_expr_with_bounds): Likewise.
12288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12289 Alan Hayward <alan.hayward@arm.com>
12290 David Sherwood <david.sherwood@arm.com>
12292 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12293 parameter for the mode of "x". Remove the "known_x", "known_mode"
12294 and "known_ret" arguments. Change the type of the mode argument
12295 to scalar_int_mode.
12296 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12297 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12298 (reg_num_sign_bit_copies_for_combine): Likewise.
12299 * rtlanal.c (nonzero_bits1): Likewise.
12300 (num_sign_bit_copies1): Likewise.
12301 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12302 (reg_num_sign_bit_copies_general): Likewise.
12303 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12304 (reg_nonzero_bits_general): Likewise.
12306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12307 Alan Hayward <alan.hayward@arm.com>
12308 David Sherwood <david.sherwood@arm.com>
12310 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12311 than in subroutines. Return 1 for non-integer modes.
12312 (cached_num_sign_bit_copies): Change the type of the mode parameter
12313 to scalar_int_mode.
12314 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12315 classes. Handle CONST_INT_P first and then check whether X also
12316 has a scalar integer mode. Check the same thing for inner registers
12317 of a SUBREG and for values that are being extended or truncated.
12319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12320 Alan Hayward <alan.hayward@arm.com>
12321 David Sherwood <david.sherwood@arm.com>
12323 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12324 in subroutines. Return the mode mask for non-integer modes.
12325 (cached_nonzero_bits): Change the type of the mode parameter
12326 to scalar_int_mode.
12327 (nonzero_bits1): Likewise. Remove early exit for other mode
12328 classes. Handle CONST_INT_P first and then check whether X
12329 also has a scalar integer mode.
12331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12332 Alan Hayward <alan.hayward@arm.com>
12333 David Sherwood <david.sherwood@arm.com>
12335 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12336 Return a scalar_int_mode and assert that the size is greater than
12338 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12339 (op_by_pieces_d::op_by_pieces_d): Likewise.
12340 (op_by_pieces_d::run): Likewise.
12341 (can_store_by_pieces): Likewise.
12343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12344 Alan Hayward <alan.hayward@arm.com>
12345 David Sherwood <david.sherwood@arm.com>
12347 * combine.c (extract_left_shift): Add a mode argument and update
12349 (make_compound_operation_int): Change the type of the mode parameter
12350 to scalar_int_mode and update the call to extract_left_shift.
12352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12353 Alan Hayward <alan.hayward@arm.com>
12354 David Sherwood <david.sherwood@arm.com>
12356 * combine.c (simplify_and_const_int): Change the type of the mode
12357 parameter to scalar_int_mode.
12358 (simplify_and_const_int_1): Likewise. Update recursive call.
12360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12361 Alan Hayward <alan.hayward@arm.com>
12362 David Sherwood <david.sherwood@arm.com>
12364 * combine.c (simplify_compare_const): Check that the mode is a
12365 scalar_int_mode (rather than VOIDmode) before testing its
12367 (simplify_comparison): Move COMPARISON_P handling out of the
12368 loop and restrict the latter part of the loop to scalar_int_modes.
12369 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12370 and when considering SUBREG_REGs. Use is_int_mode instead of
12371 checking GET_MODE_CLASS against MODE_INT.
12373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12374 Alan Hayward <alan.hayward@arm.com>
12375 David Sherwood <david.sherwood@arm.com>
12377 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12378 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12379 shift_unit_mode and for modes involved in scalar shifts.
12381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12382 Alan Hayward <alan.hayward@arm.com>
12383 David Sherwood <david.sherwood@arm.com>
12385 * combine.c (force_int_to_mode): New function, split out from...
12386 (force_to_mode): ...here. Keep xmode up-to-date and use it
12387 instead of GET_MODE (x).
12389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12390 Alan Hayward <alan.hayward@arm.com>
12391 David Sherwood <david.sherwood@arm.com>
12393 * optabs-query.h (extraction_insn::struct_mode): Change type to
12394 opt_scalar_int_mode and update comment.
12395 (extraction_insn::field_mode): Change type to scalar_int_mode.
12396 (extraction_insn::pos_mode): Likewise.
12397 * combine.c (make_extraction): Update accordingly.
12398 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12399 (get_optab_extraction_insn): Likewise.
12400 * recog.c (simplify_while_replacing): Likewise.
12401 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12402 parameter to opt_scalar_int_mode.
12404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12405 Alan Hayward <alan.hayward@arm.com>
12406 David Sherwood <david.sherwood@arm.com>
12408 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12409 to a scalar_int_mode instead of a machine_mode.
12410 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12411 (get_best_mode): Return a boolean and use a pointer argument to store
12412 the selected mode. Replace the limit mode parameter with a bit limit.
12413 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12414 for the values returned by bit_field_mode_iterator::next_mode.
12415 (store_bit_field): Update call to get_best_mode.
12416 (store_fixed_bit_field): Likewise.
12417 (extract_fixed_bit_field): Likewise.
12418 * expr.c (optimize_bitfield_assignment_op): Likewise.
12419 * fold-const.c (optimize_bit_field_compare): Likewise.
12420 (fold_truth_andor_1): Likewise.
12421 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12422 Update for new type of m_mode.
12423 (get_best_mode): As above.
12425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12426 Alan Hayward <alan.hayward@arm.com>
12427 David Sherwood <david.sherwood@arm.com>
12429 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12430 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12431 (store_bit_field): Check is_a <scalar_int_mode> before calling
12432 strict_volatile_bitfield_p.
12433 (extract_bit_field): Likewise.
12435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12436 Alan Hayward <alan.hayward@arm.com>
12437 David Sherwood <david.sherwood@arm.com>
12439 * target.def (cstore_mode): Return a scalar_int_mode.
12440 * doc/tm.texi: Regenerate.
12441 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12442 * targhooks.h (default_cstore_mode): Likewise.
12443 * targhooks.c (default_cstore_mode): Likewise, using a forced
12445 * expmed.c (emit_cstore): Expect the target of the cstore to be
12448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12449 Alan Hayward <alan.hayward@arm.com>
12450 David Sherwood <david.sherwood@arm.com>
12452 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12454 (niter_desc): Likewise mode.
12455 (iv_analyze): Add a mode parameter.
12457 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12458 and change its type to scalar_int_mode.
12459 * loop-iv.c: Update commentary at head of file.
12460 (iv_constant): Pass the mode paraeter before the rtx it describes
12461 and change its type to scalar_int_mode. Remove VOIDmode handling.
12462 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12463 (iv_extend): Likewise.
12464 (shorten_into_mode): Likewise.
12465 (iv_add): Use scalar_int_mode.
12466 (iv_mult): Likewise.
12467 (iv_shift): Likewise.
12468 (canonicalize_iv_subregs): Likewise.
12469 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12470 it describes and change its mode to scalar_int_mode. Also change
12471 the type of the returned inner_mode to scalar_int_mode.
12472 (get_biv_step): Likewise, turning outer_mode from a pointer
12473 into a direct parameter. Update call to get_biv_step_1.
12474 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12475 iv_constant and get_biv_step.
12476 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12477 and change its type to scalar_int_mode. Don't initialise iv->mode
12478 to VOIDmode and remove later checks for its still being VOIDmode.
12479 Update calls to iv_analyze_op and iv_analyze_expr. Check
12480 is_a <scalar_int_mode> when changing the mode under consideration.
12481 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12482 Update call to iv_analyze_expr.
12483 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12484 inner register is not also a scalar_int_mode. Update call to
12486 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12487 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12488 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12489 separate mode class checks. Update calls to iv_analyze. Remove
12490 fix-up of VOIDmodes after iv_analyze_biv.
12491 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12492 don't have a scalar_int_mode. Update call to biv_p.
12494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12495 Alan Hayward <alan.hayward@arm.com>
12496 David Sherwood <david.sherwood@arm.com>
12498 * cfgexpand.c (convert_debug_memory_address): Use
12499 as_a <scalar_int_mode>.
12500 * combine.c (expand_compound_operation): Likewise.
12501 (make_extraction): Likewise.
12502 (change_zero_ext): Likewise.
12503 (simplify_comparison): Likewise.
12504 * cse.c (cse_insn): Likewise.
12505 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12506 (mem_loc_descriptor): Likewise.
12507 (loc_descriptor): Likewise.
12508 * expmed.c (init_expmed_one_mode): Likewise.
12509 (synth_mult): Likewise.
12510 (emit_store_flag_1): Likewise.
12511 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12512 of a comparison with size.
12513 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12514 (reduce_to_bit_field_precision): Likewise.
12515 * function.c (expand_function_end): Likewise.
12516 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12517 * loop-doloop.c (doloop_modify): Likewise.
12518 * optabs.c (expand_binop): Likewise.
12519 (expand_unop): Likewise.
12520 (expand_copysign_absneg): Likewise.
12521 (prepare_cmp_insn): Likewise.
12522 (maybe_legitimize_operand): Likewise.
12523 * recog.c (const_scalar_int_operand): Likewise.
12524 * rtlanal.c (get_address_mode): Likewise.
12525 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12526 (simplify_cond_clz_ctz): Likewise.
12527 * tree-nested.c (get_nl_goto_field): Likewise.
12528 * tree.c (build_vector_type_for_mode): Likewise.
12529 * var-tracking.c (use_narrower_mode): Likewise.
12531 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12532 Alan Hayward <alan.hayward@arm.com>
12533 David Sherwood <david.sherwood@arm.com>
12535 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12536 * builtins.c (expand_builtin_signbit): Use it.
12537 * cfgexpand.c (expand_debug_expr): Likewise.
12538 * dojump.c (do_jump): Likewise.
12539 (do_compare_and_jump): Likewise.
12540 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12541 * expmed.c (make_tree): Likewise.
12542 * expr.c (expand_expr_real_2): Likewise.
12543 (expand_expr_real_1): Likewise.
12544 (try_casesi): Likewise.
12545 * fold-const-call.c (fold_const_call_ss): Likewise.
12546 * fold-const.c (unextend): Likewise.
12547 (extract_muldiv_1): Likewise.
12548 (fold_single_bit_test): Likewise.
12549 (native_encode_int): Likewise.
12550 (native_encode_string): Likewise.
12551 (native_interpret_int): Likewise.
12552 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12553 * internal-fn.c (expand_addsub_overflow): Likewise.
12554 (expand_neg_overflow): Likewise.
12555 (expand_mul_overflow): Likewise.
12556 (expand_arith_overflow): Likewise.
12557 * match.pd: Likewise.
12558 * stor-layout.c (layout_type): Likewise.
12559 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12560 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12561 * tree-ssanames.c (get_range_info): Likewise.
12562 * tree-switch-conversion.c (array_value_type) Likewise.
12563 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12564 (vect_recog_divmod_pattern): Likewise.
12565 (vect_recog_mixed_size_cond_pattern): Likewise.
12566 * tree-vrp.c (extract_range_basic): Likewise.
12567 (simplify_float_conversion_using_ranges): Likewise.
12568 * tree.c (int_fits_type_p): Likewise.
12569 * ubsan.c (instrument_bool_enum_load): Likewise.
12570 * varasm.c (mergeable_string_section): Likewise.
12571 (narrowing_initializer_constant_valid_p): Likewise.
12572 (output_constant): Likewise.
12574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12575 Alan Hayward <alan.hayward@arm.com>
12576 David Sherwood <david.sherwood@arm.com>
12578 * machmode.h (NARROWEST_INT_MODE): New macro.
12579 * expr.c (alignment_for_piecewise_move): Use it instead of
12580 GET_CLASS_NARROWEST_MODE (MODE_INT).
12581 (push_block): Likewise.
12582 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12584 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12586 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12587 Alan Hayward <alan.hayward@arm.com>
12588 David Sherwood <david.sherwood@arm.com>
12590 * postreload.c (move2add_valid_value_p): Change the type of the
12591 mode parameter to scalar_int_mode.
12592 (move2add_use_add2_insn): Add a mode parameter and use it instead
12594 (move2add_use_add3_insn): Likewise.
12595 (reload_cse_move2add): Update accordingly.
12597 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12598 Alan Hayward <alan.hayward@arm.com>
12599 David Sherwood <david.sherwood@arm.com>
12601 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12603 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12604 * optabs.c (expand_unop): Likewise.
12606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12607 Alan Hayward <alan.hayward@arm.com>
12608 David Sherwood <david.sherwood@arm.com>
12610 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12611 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12612 (popcount_loc_descriptor): Likewise.
12613 (bswap_loc_descriptor): Likewise.
12614 (rotate_loc_descriptor): Likewise.
12615 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12616 calling the functions above.
12618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12619 Alan Hayward <alan.hayward@arm.com>
12620 David Sherwood <david.sherwood@arm.com>
12622 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12624 (try_combine): Likewise.
12625 (simplify_if_then_else): Likewise.
12626 * cse.c (cse_insn): Likewise.
12627 * dwarf2out.c (mem_loc_descriptor): Likewise.
12628 * emit-rtl.c (gen_lowpart_common): Likewise.
12629 * simplify-rtx.c (simplify_truncation): Likewise.
12630 (simplify_binary_operation_1): Likewise.
12631 (simplify_const_relational_operation): Likewise.
12632 (simplify_ternary_operation): Likewise.
12633 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12636 Alan Hayward <alan.hayward@arm.com>
12637 David Sherwood <david.sherwood@arm.com>
12639 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12640 * reload.c (push_reload): Likewise.
12641 (find_reloads): Likewise.
12643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12644 Alan Hayward <alan.hayward@arm.com>
12645 David Sherwood <david.sherwood@arm.com>
12647 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12648 (make_compound_operation_int): Likewise.
12649 (change_zero_ext): Likewise.
12650 * expr.c (convert_move): Likewise.
12651 (convert_modes): Likewise.
12652 * fwprop.c (forward_propagate_subreg): Likewise.
12653 * loop-iv.c (get_biv_step_1): Likewise.
12654 * optabs.c (widen_operand): Likewise.
12655 * postreload.c (move2add_valid_value_p): Likewise.
12656 * recog.c (simplify_while_replacing): Likewise.
12657 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12658 (simplify_binary_operation_1): Likewise. Remove redundant
12659 mode equality check.
12661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12662 Alan Hayward <alan.hayward@arm.com>
12663 David Sherwood <david.sherwood@arm.com>
12665 * combine.c (combine_simplify_rtx): Add checks for
12666 is_a <scalar_int_mode>.
12667 (simplify_if_then_else): Likewise.
12668 (make_field_assignment): Likewise.
12669 (simplify_comparison): Likewise.
12670 * ifcvt.c (noce_try_bitop): Likewise.
12671 * loop-invariant.c (canonicalize_address_mult): Likewise.
12672 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12675 Alan Hayward <alan.hayward@arm.com>
12676 David Sherwood <david.sherwood@arm.com>
12678 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12679 is_a <scalar_int_mode> instead of != BLKmode.
12681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12682 Alan Hayward <alan.hayward@arm.com>
12683 David Sherwood <david.sherwood@arm.com>
12685 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12686 instead of != VOIDmode.
12687 * combine.c (if_then_else_cond): Likewise.
12688 (change_zero_ext): Likewise.
12689 * dwarf2out.c (mem_loc_descriptor): Likewise.
12690 (loc_descriptor): Likewise.
12691 * rtlanal.c (canonicalize_condition): Likewise.
12692 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12695 Alan Hayward <alan.hayward@arm.com>
12696 David Sherwood <david.sherwood@arm.com>
12698 * simplify-rtx.c (simplify_binary_operation_1): Use
12699 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12701 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12702 Alan Hayward <alan.hayward@arm.com>
12703 David Sherwood <david.sherwood@arm.com>
12705 * wide-int.h (int_traits<unsigned char>) New class.
12706 (int_traits<unsigned short>) Likewise.
12707 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12708 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12710 * combine.c (set_nonzero_bits_and_sign_copies): Use
12711 is_a <scalar_int_mode>.
12712 (find_split_point): Likewise.
12713 (combine_simplify_rtx): Likewise.
12714 (simplify_logical): Likewise.
12715 (expand_compound_operation): Likewise.
12716 (expand_field_assignment): Likewise.
12717 (make_compound_operation): Likewise.
12718 (extended_count): Likewise.
12719 (change_zero_ext): Likewise.
12720 (simplify_comparison): Likewise.
12721 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12722 (ucompare_loc_descriptor): Likewise.
12723 (minmax_loc_descriptor): Likewise.
12724 (mem_loc_descriptor): Likewise.
12725 (loc_descriptor): Likewise.
12726 * expmed.c (init_expmed_one_mode): Likewise.
12727 * lra-constraints.c (lra_constraint_offset): Likewise.
12728 * optabs.c (prepare_libcall_arg): Likewise.
12729 * postreload.c (move2add_note_store): Likewise.
12730 * reload.c (operands_match_p): Likewise.
12731 * rtl.h (load_extend_op): Likewise.
12732 * rtlhooks.c (gen_lowpart_general): Likewise.
12733 * simplify-rtx.c (simplify_truncation): Likewise.
12734 (simplify_unary_operation_1): Likewise.
12735 (simplify_binary_operation_1): Likewise.
12736 (simplify_const_binary_operation): Likewise.
12737 (simplify_const_relational_operation): Likewise.
12738 (simplify_subreg): Likewise.
12739 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12740 * var-tracking.c (adjust_mems): Likewise.
12741 (prepare_call_arguments): Likewise.
12743 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12744 Alan Hayward <alan.hayward@arm.com>
12745 David Sherwood <david.sherwood@arm.com>
12747 * machmode.h (is_int_mode): New fuction.
12748 * combine.c (find_split_point): Use it.
12749 (combine_simplify_rtx): Likewise.
12750 (simplify_if_then_else): Likewise.
12751 (simplify_set): Likewise.
12752 (simplify_shift_const_1): Likewise.
12753 (simplify_comparison): Likewise.
12754 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12755 * cse.c (notreg_cost): Likewise.
12756 (cse_insn): Likewise.
12757 * cselib.c (cselib_lookup_1): Likewise.
12758 * dojump.c (do_jump_1): Likewise.
12759 (do_compare_rtx_and_jump): Likewise.
12760 * dse.c (get_call_args): Likewise.
12761 * dwarf2out.c (rtl_for_decl_init): Likewise.
12762 (native_encode_initializer): Likewise.
12763 * expmed.c (emit_store_flag_1): Likewise.
12764 (emit_store_flag): Likewise.
12765 * expr.c (convert_modes): Likewise.
12766 (store_field): Likewise.
12767 (expand_expr_real_1): Likewise.
12768 * fold-const.c (fold_read_from_constant_string): Likewise.
12769 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12770 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12771 * optabs.c (expand_binop): Likewise.
12772 (expand_unop): Likewise.
12773 (expand_abs_nojump): Likewise.
12774 (expand_one_cmpl_abs_nojump): Likewise.
12775 * simplify-rtx.c (mode_signbit_p): Likewise.
12776 (val_signbit_p): Likewise.
12777 (val_signbit_known_set_p): Likewise.
12778 (val_signbit_known_clear_p): Likewise.
12779 (simplify_relational_operation_1): Likewise.
12780 * tree.c (vector_type_mode): Likewise.
12782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12783 Alan Hayward <alan.hayward@arm.com>
12784 David Sherwood <david.sherwood@arm.com>
12786 * machmode.h (smallest_mode_for_size): Fix formatting.
12787 (smallest_int_mode_for_size): New function.
12788 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12789 instead of smallest_mode_for_size.
12790 * combine.c (make_extraction): Likewise.
12791 * config/arc/arc.c (arc_expand_movmem): Likewise.
12792 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12793 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12794 * config/s390/s390.c (s390_expand_insv): Likewise.
12795 * config/sparc/sparc.c (assign_int_registers): Likewise.
12796 * config/spu/spu.c (spu_function_value): Likewise.
12797 (spu_function_arg): Likewise.
12798 * coverage.c (get_gcov_type): Likewise.
12799 (get_gcov_unsigned_t): Likewise.
12800 * dse.c (find_shift_sequence): Likewise.
12801 * expmed.c (store_bit_field_1): Likewise.
12802 * expr.c (convert_move): Likewise.
12803 (store_field): Likewise.
12804 * internal-fn.c (expand_arith_overflow): Likewise.
12805 * optabs-query.c (get_best_extraction_insn): Likewise.
12806 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12807 * stor-layout.c (layout_type): Likewise.
12808 (initialize_sizetypes): Likewise.
12809 * targhooks.c (default_get_mask_mode): Likewise.
12810 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12813 Alan Hayward <alan.hayward@arm.com>
12814 David Sherwood <david.sherwood@arm.com>
12816 * machmode.h (opt_mode::else_blk): New function.
12817 (int_mode_for_mode): Declare.
12818 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12819 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12821 * cfgexpand.c (expand_debug_expr): Likewise.
12822 * combine.c (gen_lowpart_or_truncate): Likewise.
12823 (gen_lowpart_for_combine): Likewise.
12824 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12825 * config/avr/avr.c (avr_to_int_mode): Likewise.
12826 (avr_out_plus_1): Likewise.
12827 (avr_out_plus): Likewise.
12828 (avr_out_round): Likewise.
12829 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12830 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12831 (s390_expand_vcond): Likewise.
12832 * config/spu/spu.c (spu_split_immediate): Likewise.
12833 (spu_expand_mov): Likewise.
12834 * dse.c (get_stored_val): Likewise.
12835 * expmed.c (store_bit_field_1): Likewise.
12836 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12838 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12839 (extract_low_bits): Likewise.
12840 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12841 handling rather than repeating the check.
12842 (emit_group_store): Likewise.
12843 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12844 * optabs.c (expand_absneg_bit): Likewise.
12845 (expand_copysign_absneg): Likewise.
12846 (expand_copysign_bit): Likewise.
12847 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12848 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12849 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12850 * var-tracking.c (prepare_call_arguments): Likewise.
12851 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12852 int_mode_for_mode instead of mode_for_size.
12853 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12856 Alan Hayward <alan.hayward@arm.com>
12857 David Sherwood <david.sherwood@arm.com>
12859 * machmode.h (int_mode_for_size): New function.
12860 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12861 instead of mode_for_size.
12862 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12864 * combine.c (expand_field_assignment): Use int_mode_for_size
12865 instead of mode_for_size.
12866 (make_extraction): Likewise.
12867 (simplify_shift_const_1): Likewise.
12868 (simplify_comparison): Likewise.
12869 * dojump.c (do_jump): Likewise.
12870 * dwarf2out.c (mem_loc_descriptor): Likewise.
12871 * emit-rtl.c (init_derived_machine_modes): Likewise.
12872 * expmed.c (flip_storage_order): Likewise.
12873 (convert_extracted_bit_field): Likewise.
12874 * expr.c (copy_blkmode_from_reg): Likewise.
12875 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12876 * internal-fn.c (expand_mul_overflow): Likewise.
12877 * lower-subreg.c (simple_move): Likewise.
12878 * optabs-libfuncs.c (init_optabs): Likewise.
12879 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12880 * tree.c (vector_type_mode): Likewise.
12881 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12882 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12883 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12884 * tree-vect-stmts.c (vectorizable_load): Likewise.
12885 (vectorizable_store): Likewise.
12887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12888 Alan Hayward <alan.hayward@arm.com>
12889 David Sherwood <david.sherwood@arm.com>
12891 * coretypes.h (pod_mode): New type.
12892 (scalar_int_mode_pod): New typedef.
12893 * machmode.h (pod_mode): New class.
12894 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12895 * genmodes.c (emit_mode_int_n): Update accordingly.
12896 * lower-subreg.h (target_lower_subreg): Change type to
12897 scalar_int_mode_pod.
12898 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12899 scalar_int_mode_pod.
12901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12902 Alan Hayward <alan.hayward@arm.com>
12903 David Sherwood <david.sherwood@arm.com>
12905 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12906 machine_mode to scalar_int_mode.
12907 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12908 (rs6000_option_override_internal): Remove cast to int.
12909 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12910 machine_mode to scalar_int_mode.
12911 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12912 (rs6000_option_override_internal): Remove cast to int.
12913 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12914 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12916 * config/s390/s390.c (s390_expand_builtin): Likewise.
12917 * coretypes.h (scalar_int_mode): New type.
12918 (opt_scalar_int_mode): New typedef.
12919 * machmode.h (scalar_int_mode): New class.
12920 (scalar_int_mode::includes_p): New function.
12921 (byte_mode): Change type to scalar_int_mode.
12922 (word_mode): Likewise.
12923 (ptr_mode): Likewise.
12924 * emit-rtl.c (byte_mode): Likewise.
12925 (word_mode): Likewise.
12926 (ptr_mode): Likewise.
12927 (init_derived_machine_modes): Update accordingly.
12928 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12929 and MODE_PARTIAL_INT.
12930 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12931 opt_scalar_int_mode.
12933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12934 Alan Hayward <alan.hayward@arm.com>
12935 David Sherwood <david.sherwood@arm.com>
12937 * target.def (libgcc_floating_mode_supported_p): Take a
12939 * doc/tm.texi: Regenerate.
12940 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12942 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12943 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12947 Alan Hayward <alan.hayward@arm.com>
12948 David Sherwood <david.sherwood@arm.com>
12950 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12951 * doc/tm.texi: Regenerate.
12952 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12953 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12954 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12955 * targhooks.h (default_floatn_mode): Likewise.
12956 * targhooks.c (default_floatn_mode): Likewise.
12957 * tree.c (build_common_tree_nodes): Update accordingly.
12959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12960 Alan Hayward <alan.hayward@arm.com>
12961 David Sherwood <david.sherwood@arm.com>
12963 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12964 (mode_iterator::iterate_p): Likewise.
12965 (mode_iterator::get_wider): Likewise.
12966 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12969 Alan Hayward <alan.hayward@arm.com>
12970 David Sherwood <david.sherwood@arm.com>
12972 * coretypes.h (opt_scalar_float_mode): New typedef.
12973 * machmode.h (float_mode_for_size): New function.
12974 * emit-rtl.c (double_mode): Delete.
12975 (init_emit_once): Use float_mode_for_size.
12976 * stor-layout.c (layout_type): Likewise.
12977 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12980 Alan Hayward <alan.hayward@arm.com>
12981 David Sherwood <david.sherwood@arm.com>
12983 * output.h (assemble_real): Take a scalar_float_mode.
12984 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12985 * config/arm/arm.md (consttable_4): Likewise.
12986 (consttable_8): Likewise.
12987 (consttable_16): Likewise.
12988 * config/mips/mips.md (consttable_float): Likewise.
12989 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12990 * varasm.c (assemble_real): Take a scalar_float_mode.
12991 (output_constant_pool_2): Update accordingly.
12992 (output_constant): Likewise.
12994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12995 Alan Hayward <alan.hayward@arm.com>
12996 David Sherwood <david.sherwood@arm.com>
12998 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12999 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13000 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13001 (native_encode_real): Likewise.
13002 (native_interpret_real): Likewise.
13003 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13004 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13006 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13007 Alan Hayward <alan.hayward@arm.com>
13008 David Sherwood <david.sherwood@arm.com>
13010 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13011 <scalar_float_mode>. Simplify.
13012 (gen_extend_conv_libfunc): Likewise.
13014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13015 Alan Hayward <alan.hayward@arm.com>
13016 David Sherwood <david.sherwood@arm.com>
13018 * coretypes.h (scalar_float_mode): New type.
13019 * machmode.h (mode_traits::from_int): Use machine_mode if
13020 USE_ENUM_MODES is defined.
13021 (is_a): New function.
13023 (dyn_cast): Likewise.
13024 (scalar_float_mode): New class.
13025 (scalar_float_mode::includes_p): New function.
13026 (is_float_mode): Likewise.
13027 * gdbhooks.py (MachineModePrinter): New class.
13028 (build_pretty_printer): Use it for scalar_float_mode.
13029 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13030 (format_helper::format_helper): Turn into a template.
13031 * genmodes.c (get_mode_class): New function.
13032 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13033 or machine_mode if none.
13034 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13035 as_a <scalar_float_mode>.
13036 * dwarf2out.c (mem_loc_descriptor): Likewise.
13037 (insert_float): Likewise.
13038 (add_const_value_attribute): Likewise.
13039 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13040 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13041 (expand_unop): Update accordingly.
13042 (expand_abs_nojump): Likewise.
13043 (expand_copysign_absneg): Take a scalar_float_mode.
13044 (expand_copysign_bit): Likewise.
13045 (expand_copysign): Update accordingly.
13047 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13048 Alan Hayward <alan.hayward@arm.com>
13049 David Sherwood <david.sherwood@arm.com>
13051 * coretypes.h (opt_mode): New class.
13052 * machmode.h (opt_mode): Likewise.
13053 (opt_mode::else_void): New function.
13054 (opt_mode::require): Likewise.
13055 (opt_mode::exists): Likewise.
13056 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13057 (GET_MODE_2XWIDER_MODE): Likewise.
13058 (mode_iterator::get_wider): Update accordingly.
13059 (mode_iterator::get_2xwider): Likewise.
13060 (mode_iterator::get_known_wider): Likewise, turning into a template.
13061 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13062 forcing a wider mode to exist.
13063 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13064 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13065 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13066 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13067 * lower-subreg.c (init_lower_subreg): Likewise.
13068 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13069 on the final iteration.
13070 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13071 a wider mode exists before asking for a move pattern.
13072 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13073 forcing a wider mode to exist.
13074 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13075 returning false if no such mode exists.
13076 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13077 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13078 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13079 Avoid checking for a MODE_INT if we already know the mode is not a
13081 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13082 forcing a wider mode to exist.
13083 (expmed_mult_highpart_optab): Likewise.
13084 (expmed_mult_highpart): Likewise.
13085 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13087 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13088 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13089 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13090 * internal-fn.c (expand_mul_overflow): Update use of
13091 GET_MODE_2XWIDER_MODE.
13092 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13093 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13094 GET_MODE_WIDER_MODE.
13095 (convert_plusminus_to_widen): Likewise.
13096 * tree-switch-conversion.c (array_value_type): Likewise.
13097 * var-tracking.c (emit_note_insn_var_location): Likewise.
13098 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13099 Return false inside rather than outside the loop if no wider mode
13101 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13102 and GET_MODE_2XWIDER_MODE
13103 (can_compare_p): Use else_void.
13104 * gdbhooks.py (OptMachineModePrinter): New class.
13105 (build_pretty_printer): Use it for opt_mode.
13107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13108 Alan Hayward <alan.hayward@arm.com>
13109 David Sherwood <david.sherwood@arm.com>
13111 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13112 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13115 Alan Hayward <alan.hayward@arm.com>
13116 David Sherwood <david.sherwood@arm.com>
13118 * machmode.h (mode_traits): New structure.
13119 (get_narrowest_mode): New function.
13120 (mode_iterator::start): Likewise.
13121 (mode_iterator::iterate_p): Likewise.
13122 (mode_iterator::get_wider): Likewise.
13123 (mode_iterator::get_known_wider): Likewise.
13124 (mode_iterator::get_2xwider): Likewise.
13125 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13126 (FOR_EACH_MODE): Likewise.
13127 (FOR_EACH_MODE_FROM): Likewise.
13128 (FOR_EACH_MODE_UNTIL): Likewise.
13129 (FOR_EACH_WIDER_MODE): Likewise.
13130 (FOR_EACH_2XWIDER_MODE): Likewise.
13131 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13132 * combine.c (simplify_comparison): Likewise
13133 * config/i386/i386.c (type_natural_mode): Likewise.
13134 * cse.c (cse_insn): Likewise.
13135 * dse.c (find_shift_sequence): Likewise.
13136 * emit-rtl.c (init_derived_machine_modes): Likewise.
13137 (init_emit_once): Likewise.
13138 * explow.c (hard_function_value): Likewise.
13139 * expmed.c (extract_fixed_bit_field_1): Likewise.
13140 (extract_bit_field_1): Likewise.
13141 (expand_divmod): Likewise.
13142 (emit_store_flag_1): Likewise.
13143 * expr.c (init_expr_target): Likewise.
13144 (convert_move): Likewise.
13145 (alignment_for_piecewise_move): Likewise.
13146 (widest_int_mode_for_size): Likewise.
13147 (emit_block_move_via_movmem): Likewise.
13148 (copy_blkmode_to_reg): Likewise.
13149 (set_storage_via_setmem): Likewise.
13150 (compress_float_constant): Likewise.
13151 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13152 * optabs-query.c (get_best_extraction_insn): Likewise.
13153 * optabs.c (expand_binop): Likewise.
13154 (expand_twoval_unop): Likewise.
13155 (expand_twoval_binop): Likewise.
13156 (widen_leading): Likewise.
13157 (widen_bswap): Likewise.
13158 (expand_parity): Likewise.
13159 (expand_unop): Likewise.
13160 (prepare_cmp_insn): Likewise.
13161 (prepare_float_lib_cmp): Likewise.
13162 (expand_float): Likewise.
13163 (expand_fix): Likewise.
13164 (expand_sfix_optab): Likewise.
13165 * postreload.c (move2add_use_add2_insn): Likewise.
13166 * reg-stack.c (reg_to_stack): Likewise.
13167 * reginfo.c (choose_hard_reg_mode): Likewise.
13168 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13169 * stor-layout.c (mode_for_size): Likewise.
13170 (smallest_mode_for_size): Likewise.
13171 (mode_for_vector): Likewise.
13172 (finish_bitfield_representative): Likewise.
13173 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13174 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13175 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13176 * var-tracking.c (prepare_call_arguments): Likewise.
13178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13179 Alan Hayward <alan.hayward@arm.com>
13180 David Sherwood <david.sherwood@arm.com>
13182 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13183 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13184 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13185 * machmode.h (mode_size): Move earlier in file.
13186 (mode_precision): Likewise.
13187 (mode_inner): Likewise.
13188 (mode_nunits): Likewise.
13189 (mode_unit_size): Likewise.
13190 (unit_unit_precision): Likewise.
13191 (mode_wider): Likewise.
13192 (mode_2xwider): Likewise.
13193 (machine_mode): New class.
13194 (mode_to_bytes): New function.
13195 (mode_to_bits): Likewise.
13196 (mode_to_precision): Likewise.
13197 (mode_to_inner): Likewise.
13198 (mode_to_unit_size): Likewise.
13199 (mode_to_unit_precision): Likewise.
13200 (mode_to_nunits): Likewise.
13201 (GET_MODE_SIZE): Use mode_to_bytes.
13202 (GET_MODE_BITSIZE): Use mode_to_bits.
13203 (GET_MODE_PRECISION): Use mode_to_precision.
13204 (GET_MODE_INNER): Use mode_to_inner.
13205 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13206 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13207 (GET_MODE_NUNITS): Use mode_to_nunits.
13208 * system.h (ALWAYS_INLINE): New macro.
13209 * config/powerpcspe/powerpcspe-c.c
13210 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13211 int for arg1_mode and arg2_mode.
13213 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13214 Alan Hayward <alan.hayward@arm.com>
13215 David Sherwood <david.sherwood@arm.com>
13217 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13218 Prefix mode names with E_ in case statements.
13219 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13220 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13221 (aarch64_split_simd_move): Likewise.
13222 (aarch64_gen_storewb_pair): Likewise.
13223 (aarch64_gen_loadwb_pair): Likewise.
13224 (aarch64_gen_store_pair): Likewise.
13225 (aarch64_gen_load_pair): Likewise.
13226 (aarch64_get_condition_code_1): Likewise.
13227 (aarch64_constant_pool_reload_icode): Likewise.
13228 (get_rsqrte_type): Likewise.
13229 (get_rsqrts_type): Likewise.
13230 (get_recpe_type): Likewise.
13231 (get_recps_type): Likewise.
13232 (aarch64_gimplify_va_arg_expr): Likewise.
13233 (aarch64_simd_container_mode): Likewise.
13234 (aarch64_emit_load_exclusive): Likewise.
13235 (aarch64_emit_store_exclusive): Likewise.
13236 (aarch64_expand_compare_and_swap): Likewise.
13237 (aarch64_gen_atomic_cas): Likewise.
13238 (aarch64_emit_bic): Likewise.
13239 (aarch64_emit_atomic_swap): Likewise.
13240 (aarch64_emit_atomic_load_op): Likewise.
13241 (aarch64_evpc_trn): Likewise.
13242 (aarch64_evpc_uzp): Likewise.
13243 (aarch64_evpc_zip): Likewise.
13244 (aarch64_evpc_ext): Likewise.
13245 (aarch64_evpc_rev): Likewise.
13246 (aarch64_evpc_dup): Likewise.
13247 (aarch64_gen_ccmp_first): Likewise.
13248 (aarch64_gen_ccmp_next): Likewise.
13249 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13250 (alpha_emit_xfloating_libcall): Likewise.
13251 (emit_insxl): Likewise.
13252 (alpha_arg_type): Likewise.
13253 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13254 (arc_preferred_simd_mode): Likewise.
13255 (arc_secondary_reload): Likewise.
13256 (get_arc_condition_code): Likewise.
13257 (arc_print_operand): Likewise.
13258 (arc_legitimate_constant_p): Likewise.
13259 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13260 * config/arc/arc.md (casesi_load): Likewise.
13261 (casesi_compact_jump): Likewise.
13262 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13263 (cc_use_register): Likewise.
13264 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13265 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13266 (arm_init_iwmmxt_builtins): Likewise.
13267 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13268 (neon_expand_vector_init): Likewise.
13269 (arm_attr_length_move_neon): Likewise.
13270 (maybe_get_arm_condition_code): Likewise.
13271 (arm_emit_vector_const): Likewise.
13272 (arm_preferred_simd_mode): Likewise.
13273 (arm_output_iwmmxt_tinsr): Likewise.
13274 (thumb1_output_casesi): Likewise.
13275 (thumb2_output_casesi): Likewise.
13276 (arm_emit_load_exclusive): Likewise.
13277 (arm_emit_store_exclusive): Likewise.
13278 (arm_expand_compare_and_swap): Likewise.
13279 (arm_evpc_neon_vuzp): Likewise.
13280 (arm_evpc_neon_vzip): Likewise.
13281 (arm_evpc_neon_vrev): Likewise.
13282 (arm_evpc_neon_vtrn): Likewise.
13283 (arm_evpc_neon_vext): Likewise.
13284 (arm_validize_comparison): Likewise.
13285 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13286 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13287 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13288 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13289 (c6x_preferred_simd_mode): Likewise.
13290 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13291 (epiphany_rtx_costs): Likewise.
13292 * config/epiphany/predicates.md (proper_comparison_operator):
13294 * config/frv/frv.c (condexec_memory_operand): Likewise.
13295 (frv_emit_move): Likewise.
13296 (output_move_single): Likewise.
13297 (output_condmove_single): Likewise.
13298 (frv_hard_regno_mode_ok): Likewise.
13299 (frv_matching_accg_mode): Likewise.
13300 * config/h8300/h8300.c (split_adds_subs): Likewise.
13301 (h8300_rtx_costs): Likewise.
13302 (h8300_print_operand): Likewise.
13303 (compute_mov_length): Likewise.
13304 (output_logical_op): Likewise.
13305 (compute_logical_op_length): Likewise.
13306 (compute_logical_op_cc): Likewise.
13307 (h8300_shift_needs_scratch_p): Likewise.
13308 (output_a_shift): Likewise.
13309 (compute_a_shift_length): Likewise.
13310 (compute_a_shift_cc): Likewise.
13311 (expand_a_rotate): Likewise.
13312 (output_a_rotate): Likewise.
13313 * config/i386/i386.c (classify_argument): Likewise.
13314 (function_arg_advance_32): Likewise.
13315 (function_arg_32): Likewise.
13316 (function_arg_64): Likewise.
13317 (function_value_64): Likewise.
13318 (ix86_gimplify_va_arg): Likewise.
13319 (ix86_legitimate_constant_p): Likewise.
13320 (put_condition_code): Likewise.
13321 (split_double_mode): Likewise.
13322 (ix86_avx256_split_vector_move_misalign): Likewise.
13323 (ix86_expand_vector_logical_operator): Likewise.
13324 (ix86_split_idivmod): Likewise.
13325 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13326 (ix86_build_const_vector): Likewise.
13327 (ix86_build_signbit_mask): Likewise.
13328 (ix86_match_ccmode): Likewise.
13329 (ix86_cc_modes_compatible): Likewise.
13330 (ix86_expand_branch): Likewise.
13331 (ix86_expand_sse_cmp): Likewise.
13332 (ix86_expand_sse_movcc): Likewise.
13333 (ix86_expand_int_sse_cmp): Likewise.
13334 (ix86_expand_vec_perm_vpermi2): Likewise.
13335 (ix86_expand_vec_perm): Likewise.
13336 (ix86_expand_sse_unpack): Likewise.
13337 (ix86_expand_int_addcc): Likewise.
13338 (ix86_split_to_parts): Likewise.
13339 (ix86_vectorize_builtin_gather): Likewise.
13340 (ix86_vectorize_builtin_scatter): Likewise.
13341 (avx_vpermilp_parallel): Likewise.
13342 (inline_memory_move_cost): Likewise.
13343 (ix86_tieable_integer_mode_p): Likewise.
13344 (x86_maybe_negate_const_int): Likewise.
13345 (ix86_expand_vector_init_duplicate): Likewise.
13346 (ix86_expand_vector_init_one_nonzero): Likewise.
13347 (ix86_expand_vector_init_one_var): Likewise.
13348 (ix86_expand_vector_init_concat): Likewise.
13349 (ix86_expand_vector_init_interleave): Likewise.
13350 (ix86_expand_vector_init_general): Likewise.
13351 (ix86_expand_vector_set): Likewise.
13352 (ix86_expand_vector_extract): Likewise.
13353 (emit_reduc_half): Likewise.
13354 (ix86_emit_i387_round): Likewise.
13355 (ix86_mangle_type): Likewise.
13356 (ix86_expand_round_sse4): Likewise.
13357 (expand_vec_perm_blend): Likewise.
13358 (canonicalize_vector_int_perm): Likewise.
13359 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13360 (expand_vec_perm_1): Likewise.
13361 (expand_vec_perm_interleave3): Likewise.
13362 (expand_vec_perm_even_odd_pack): Likewise.
13363 (expand_vec_perm_even_odd_1): Likewise.
13364 (expand_vec_perm_broadcast_1): Likewise.
13365 (ix86_vectorize_vec_perm_const_ok): Likewise.
13366 (ix86_expand_vecop_qihi): Likewise.
13367 (ix86_expand_mul_widen_hilo): Likewise.
13368 (ix86_expand_sse2_abs): Likewise.
13369 (ix86_expand_pextr): Likewise.
13370 (ix86_expand_pinsr): Likewise.
13371 (ix86_preferred_simd_mode): Likewise.
13372 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13373 * config/i386/sse.md (*andnot<mode>3): Likewise.
13374 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13375 (*<code><mode>3): Likewise.
13376 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13377 (ia64_expand_atomic_op): Likewise.
13378 (ia64_arg_type): Likewise.
13379 (ia64_mode_to_int): Likewise.
13380 (ia64_scalar_mode_supported_p): Likewise.
13381 (ia64_vector_mode_supported_p): Likewise.
13382 (expand_vec_perm_broadcast): Likewise.
13383 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13384 (iq2000_function_arg_advance): Likewise.
13385 (iq2000_function_arg): Likewise.
13386 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13387 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13388 (m68k_libcall_value): Likewise.
13389 (m68k_function_value): Likewise.
13390 (sched_attr_op_type): Likewise.
13391 * config/mcore/mcore.c (mcore_output_move): Likewise.
13392 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13394 (microblaze_function_arg): Likewise.
13395 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13396 (mips_print_operand): Likewise.
13397 (mips_mode_ok_for_mov_fmt_p): Likewise.
13398 (mips_vector_mode_supported_p): Likewise.
13399 (mips_preferred_simd_mode): Likewise.
13400 (mips_expand_vpc_loongson_even_odd): Likewise.
13401 (mips_expand_vec_unpack): Likewise.
13402 (mips_expand_vi_broadcast): Likewise.
13403 (mips_expand_vector_init): Likewise.
13404 (mips_expand_vec_reduc): Likewise.
13405 (mips_expand_msa_cmp): Likewise.
13406 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13407 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13408 (cc_flags_for_mode): Likewise.
13409 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13410 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13411 (nds32_output_casesi_pc_relative): Likewise.
13412 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13413 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13414 (nvptx_gen_unpack): Likewise.
13415 (nvptx_gen_pack): Likewise.
13416 (nvptx_gen_shuffle): Likewise.
13417 (nvptx_gen_wcast): Likewise.
13418 (nvptx_preferred_simd_mode): Likewise.
13419 * config/pa/pa.c (pa_secondary_reload): Likewise.
13420 * config/pa/predicates.md (base14_operand): Likewise.
13421 * config/powerpcspe/powerpcspe-c.c
13422 (altivec_resolve_overloaded_builtin): Likewise.
13423 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13425 (rs6000_preferred_simd_mode): Likewise.
13426 (output_vec_const_move): Likewise.
13427 (rs6000_expand_vector_extract): Likewise.
13428 (rs6000_split_vec_extract_var): Likewise.
13429 (reg_offset_addressing_ok_p): Likewise.
13430 (rs6000_legitimate_offset_address_p): Likewise.
13431 (rs6000_legitimize_address): Likewise.
13432 (rs6000_emit_set_const): Likewise.
13433 (rs6000_const_vec): Likewise.
13434 (rs6000_emit_move): Likewise.
13435 (spe_build_register_parallel): Likewise.
13436 (rs6000_darwin64_record_arg_recurse): Likewise.
13437 (swap_selector_for_mode): Likewise.
13438 (spe_init_builtins): Likewise.
13439 (paired_init_builtins): Likewise.
13440 (altivec_init_builtins): Likewise.
13441 (do_load_for_compare): Likewise.
13442 (rs6000_generate_compare): Likewise.
13443 (rs6000_expand_float128_convert): Likewise.
13444 (emit_load_locked): Likewise.
13445 (emit_store_conditional): Likewise.
13446 (rs6000_output_function_epilogue): Likewise.
13447 (rs6000_handle_altivec_attribute): Likewise.
13448 (rs6000_function_value): Likewise.
13449 (emit_fusion_gpr_load): Likewise.
13450 (emit_fusion_p9_load): Likewise.
13451 (emit_fusion_p9_store): Likewise.
13452 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13453 (fusion_gpr_mem_load): Likewise.
13454 (fusion_addis_mem_combo_load): Likewise.
13455 (fusion_addis_mem_combo_store): Likewise.
13456 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13457 (fusion_gpr_mem_load): Likewise.
13458 (fusion_addis_mem_combo_load): Likewise.
13459 (fusion_addis_mem_combo_store): Likewise.
13460 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13462 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13463 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13464 (rs6000_preferred_simd_mode): Likewise.
13465 (output_vec_const_move): Likewise.
13466 (rs6000_expand_vector_extract): Likewise.
13467 (rs6000_split_vec_extract_var): Likewise.
13468 (reg_offset_addressing_ok_p): Likewise.
13469 (rs6000_legitimate_offset_address_p): Likewise.
13470 (rs6000_legitimize_address): Likewise.
13471 (rs6000_emit_set_const): Likewise.
13472 (rs6000_const_vec): Likewise.
13473 (rs6000_emit_move): Likewise.
13474 (rs6000_darwin64_record_arg_recurse): Likewise.
13475 (swap_selector_for_mode): Likewise.
13476 (paired_init_builtins): Likewise.
13477 (altivec_init_builtins): Likewise.
13478 (rs6000_expand_float128_convert): Likewise.
13479 (emit_load_locked): Likewise.
13480 (emit_store_conditional): Likewise.
13481 (rs6000_output_function_epilogue): Likewise.
13482 (rs6000_handle_altivec_attribute): Likewise.
13483 (rs6000_function_value): Likewise.
13484 (emit_fusion_gpr_load): Likewise.
13485 (emit_fusion_p9_load): Likewise.
13486 (emit_fusion_p9_store): Likewise.
13487 * config/rx/rx.c (rx_gen_move_template): Likewise.
13488 (flags_from_mode): Likewise.
13489 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13490 (s390_slb_comparison): Likewise.
13491 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13492 (s390_vector_mode_supported_p): Likewise.
13493 (s390_cc_modes_compatible): Likewise.
13494 (s390_match_ccmode_set): Likewise.
13495 (s390_canonicalize_comparison): Likewise.
13496 (s390_emit_compare_and_swap): Likewise.
13497 (s390_branch_condition_mask): Likewise.
13498 (s390_rtx_costs): Likewise.
13499 (s390_secondary_reload): Likewise.
13500 (__SECONDARY_RELOAD_CASE): Likewise.
13501 (s390_expand_cs): Likewise.
13502 (s390_preferred_simd_mode): Likewise.
13503 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13504 * config/sh/sh.c (sh_print_operand): Likewise.
13505 (dump_table): Likewise.
13506 (sh_secondary_reload): Likewise.
13507 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13508 * config/sh/sh.md (casesi_worker_1): Likewise.
13509 (casesi_worker_2): Likewise.
13510 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13511 (fcc_comparison_operator): Likewise.
13512 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13513 (emit_soft_tfmode_cvt): Likewise.
13514 (sparc_preferred_simd_mode): Likewise.
13515 (output_cbranch): Likewise.
13516 (sparc_print_operand): Likewise.
13517 (sparc_expand_vec_perm_bmask): Likewise.
13518 (vector_init_bshuffle): Likewise.
13519 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13520 (spu_vector_mode_supported_p): Likewise.
13521 (spu_expand_insv): Likewise.
13522 (spu_emit_branch_or_set): Likewise.
13523 (spu_handle_vector_attribute): Likewise.
13524 (spu_builtin_splats): Likewise.
13525 (spu_builtin_extract): Likewise.
13526 (spu_builtin_promote): Likewise.
13527 (spu_expand_sign_extend): Likewise.
13528 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13529 (tilegx_simd_int): Likewise.
13530 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13531 (tilepro_simd_int): Likewise.
13532 * config/v850/v850.c (const_double_split): Likewise.
13533 (v850_print_operand): Likewise.
13534 (ep_memory_offset): Likewise.
13535 * config/vax/vax.c (vax_rtx_costs): Likewise.
13536 (vax_output_int_move): Likewise.
13537 (vax_output_int_add): Likewise.
13538 (vax_output_int_subtract): Likewise.
13539 * config/visium/predicates.md (visium_branch_operator): Likewise.
13540 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13541 (visium_print_operand_address): Likewise.
13542 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13543 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13544 (xtensa_expand_conditional_branch): Likewise.
13545 (xtensa_copy_incoming_a7): Likewise.
13546 (xtensa_output_literal): Likewise.
13547 * dfp.c (decimal_real_maxval): Likewise.
13548 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13550 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13551 Alan Hayward <alan.hayward@arm.com>
13552 David Sherwood <david.sherwood@arm.com>
13554 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13555 (mode_nunits_inline): Likewise.
13556 (mode_inner_inline): Likewise.
13557 (mode_unit_size_inline): Likewise.
13558 (mode_unit_precision_inline): Likewise.
13559 (emit_insn_modes_h): Likewise. Also emit a #define of the
13561 (emit_mode_wider): Add an E_ prefix to mode names.
13562 (emit_mode_complex): Likewise.
13563 (emit_mode_inner): Likewise.
13564 (emit_mode_adjustments): Likewise.
13565 (emit_mode_int_n): Likewise.
13566 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13567 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13568 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13569 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13570 (CRC32_BUILTIN, ENTRY): Likewise.
13571 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13572 (aarch64_pop_regs): Likewise.
13573 (aarch64_process_components): Likewise.
13574 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13575 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13576 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13577 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13578 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13579 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13581 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13582 (pcmpistr): Likewise.
13583 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13584 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13585 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13587 * config/rl78/rl78.c (mduc_regs): Likewise.
13588 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13589 (htm_expand_builtin): Likewise.
13590 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13591 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13592 * config/xtensa/xtensa.c (print_operand): Likewise.
13593 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13594 (NUM_MODE_VECTOR_INT): Likewise.
13595 * genoutput.c (null_operand): Likewise.
13596 (output_operand_data): Likewise.
13597 * genrecog.c (print_parameter_value): Likewise.
13598 * lra.c (debug_operand_data): Likewise.
13600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13602 * dwarf2out.c (scompare_loc_descriptor_wide)
13603 (scompare_loc_descriptor_narrow): New functions, split out from...
13604 (scompare_loc_descriptor): ...here.
13605 * expmed.c (emit_store_flag_int): New function, split out from...
13606 (emit_store_flag): ...here.
13608 2017-08-30 Richard Biener <rguenther@suse.de>
13610 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13611 (dwarf2out_early_finish): Move setting of AT_pubnames from
13612 early debug output to early finish.
13614 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13616 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13617 and -mdata-region to the assembler.
13619 2017-08-30 Richard Biener <rguenther@suse.de>
13621 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13623 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13624 (dwarf2out_early_global_decl): Only generate a DIE for the
13625 abstract origin if it doesn't already exist or is a declaration DIE.
13626 (resolve_addr): Do not add the linkage name twice when
13627 generating a stub DIE for the DW_TAG_GNU_call_site target.
13629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13631 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13632 Use machine_mode rather than int for arg1_mode.
13634 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13637 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13638 that the second argument of the built-in functions to unpack
13639 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13640 switch statement instead a lot of if statements.
13641 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13642 Allow 64-bit values to be in Altivec registers as well as
13643 traditional floating point registers.
13644 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13646 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13648 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13649 MAX_REGS_PER_ADDRESS == 1.
13651 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13653 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13654 * config/i386/i386.c (ix86_option_override_internal): Simplify
13655 setting of opts->x_flag_entry.
13657 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13658 Jakub Jelinek <jakub@redhat.com>
13659 Richard Biener <rguenther@suse.de>
13661 PR tree-optimization/81503
13662 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13663 folded constant fits in the target type; reorder tests for clarity.
13665 2017-08-29 Martin Liska <mliska@suse.cz>
13667 * passes.def: Include pass_lower_switch.
13668 * stmt.c (dump_case_nodes): Remove and move to
13669 tree-switch-conversion.
13670 (case_values_threshold): Likewise.
13671 (expand_switch_as_decision_tree_p): Likewise.
13672 (emit_case_decision_tree): Likewise.
13673 (expand_case): Likewise.
13674 (balance_case_nodes): Likewise.
13675 (node_has_low_bound): Likewise.
13676 (node_has_high_bound): Likewise.
13677 (node_is_bounded): Likewise.
13678 (emit_case_nodes): Likewise.
13679 (struct simple_case_node): New struct.
13680 (add_case_node): Remove.
13681 (emit_case_dispatch_table): Use vector instead of case_list.
13682 (reset_out_edges_aux): Remove.
13683 (compute_cases_per_edge): Likewise.
13684 (expand_case): Build list of simple_case_node.
13685 (expand_sjlj_dispatch_table): Use it.
13686 * tree-switch-conversion.c (struct case_node): Moved from
13687 stmt.c and adjusted.
13688 (emit_case_nodes): Likewise.
13689 (node_has_low_bound): Likewise.
13690 (node_has_high_bound): Likewise.
13691 (node_is_bounded): Likewise.
13692 (case_values_threshold): Likewise.
13693 (reset_out_edges_aux): Likewise.
13694 (compute_cases_per_edge): Likewise.
13695 (add_case_node): Likewise.
13696 (dump_case_nodes): Likewise.
13697 (balance_case_nodes): Likewise.
13698 (expand_switch_as_decision_tree_p): Likewise.
13699 (emit_jump): Likewise.
13700 (emit_case_decision_tree): Likewise.
13701 (try_switch_expansion): Likewise.
13702 (do_jump_if_equal): Likewise.
13703 (emit_cmp_and_jump_insns): Likewise.
13704 (fix_phi_operands_for_edge): New function.
13705 (record_phi_operand_mapping): Likewise.
13706 (class pass_lower_switch): New pass.
13707 (pass_lower_switch::execute): New function.
13708 (make_pass_lower_switch): Likewise.
13709 (conditional_probability):
13710 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13711 * tree-pass.h: Add make_pass_lower_switch.
13713 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13716 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13719 2017-08-29 Richard Biener <rguenther@suse.de>
13721 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13722 we do not add a DW_AT_inline attribute twice.
13723 (gen_subprogram_die): Remove code setting DW_AT_inline on
13724 DECL_ABSTRACT_P nodes.
13726 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13728 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13729 calls to internal functions.
13730 (gimplify_modify_expr): Likewise.
13731 * tree-call-cdce.c (use_internal_fn): Likewise.
13732 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13733 (convert_to_divmod): Set the nothrow flag.
13734 * tree-if-conv.c (predicate_mem_writes): Likewise.
13735 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13736 (vectorizable_call): Likewise.
13737 (vectorizable_store): Likewise.
13738 (vectorizable_load): Likewise.
13739 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13740 (vect_recog_mask_conversion_pattern): Likewise.
13742 2017-08-29 Martin Liska <mliska@suse.cz>
13745 * gcc.c (driver_handle_option): Add new argument.
13746 * opts-common.c (handle_option): Pass
13747 target_option_override_hook.
13748 * opts-global.c (lang_handle_option): Add new option.
13749 (set_default_handlers): Add new argument.
13750 (decode_options): Likewise.
13751 * opts.c (target_handle_option): Likewise.
13752 (common_handle_option): Call target_option_override_hook.
13753 * opts.h (struct cl_option_handler_func): Add hook for
13754 target option override.
13755 (struct cl_option_handlers): Likewise.
13756 (set_default_handlers): Add new argument.
13757 (decode_options): Likewise.
13758 (common_handle_option): Likewise.
13759 (target_handle_option): Likewise.
13760 * toplev.c (toplev::main): Pass targetm.target_option.override
13763 2017-08-29 Richard Biener <rguenther@suse.de>
13764 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13766 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13767 life to the active subtree.
13769 2017-08-28 Jeff Law <law@redhat.com>
13771 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13772 derive_equivalences.
13773 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13774 Code moved into....
13775 (edge_info::derive_equivalences): New private member function
13777 * tree-ssa-dom.c (class edge_info): Changed from a struct
13778 to a class. Add ctor/dtor, methods and data members.
13779 (edge_info::edge_info): Renamed from allocate_edge_info.
13780 Initialize additional members.
13781 (edge_info::~edge_info): New.
13782 (free_dom_edge_info): Delete the edge info.
13783 (record_edge_info): Use new class & associated member functions.
13784 Tighten forms for testing for edge equivalences.
13785 (record_temporary_equivalences): Iterate over the simple
13786 equivalences rather than assuming there's only one per edge.
13787 (cprop_into_successor_phis): Iterate over the simple
13788 equivalences rather than assuming there's only one per edge.
13789 (optimize_stmt): Use operand_equal_p rather than pointer
13790 equality for mini-DSE code.
13792 2017-08-28 Nathan Sidwell <nathan@acm.org>
13794 * gcc.c (execute): Fold SIGPIPE handling into switch
13795 statement. Adjust internal error message.
13797 2017-08-28 Richard Biener <rguenther@suse.de>
13800 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13801 Do nothing for removed DIEs.
13803 2017-08-28 Richard Biener <rguenther@suse.de>
13805 PR tree-optimization/81977
13806 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13809 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13812 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13813 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13814 using targetm.gen_mem_thread_fence.
13816 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13819 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13820 predicate to register_operand. Reorder operands.
13821 (*btr<mode>): Ditto.
13822 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13823 (*btr<mode>_mask): Ditto.
13825 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13827 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13828 * config/rs6000/xmmintrin.h: New file.
13829 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13831 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13834 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13835 parameter and_insn and return it.
13836 (recombine_lvx_pattern): Insert a copy to ensure availability of
13837 the base register of the copied masking operation at the point of
13838 the instruction replacement.
13839 (recombine_stvx_pattern): Likewise.
13841 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13843 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13844 undocumented switches.
13845 (-mpower9-dform-vector): Likewise.
13846 (-mpower9-dform): Likewise.
13847 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13848 comments to delete references to -mpower9-dform* switches.
13849 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13850 Delete reference to -mpower9-dform* switches, test for
13851 -mpower9-vector instead.
13852 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13853 (OTHER_P9_VECTOR_MASKS): Likewise.
13854 (POWERPC_MASKS): Likewise.
13855 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13856 tests against -mpower9-dform* to -mpower9-vector. Delete code
13857 that checked for -mpower9-dform* consistancy with other options.
13858 Add test for -mpower9-misc to enable other power9 switches.
13859 (rs6000_init_hard_regno_mode_ok): Likewise.
13860 (rs6000_option_override_internal): Likewise.
13861 (rs6000_emit_prologue): Likewise.
13862 (rs6000_emit_epilogue): Likewise.
13863 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13864 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13865 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13867 (emit_fusion_p9_store): Likewise.
13868 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13869 resetting these macros if the assembler does not support ISA 3.0
13871 (TARGET_P9_DFORM_VECTOR): Likewise.
13872 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13873 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13875 2017-08-25 Alan Modra <amodra@gmail.com>
13877 PR rtl-optimization/81747
13878 * cse.c (cse_extended_basic_block): Don't attempt to record
13879 equivalences for degenerate conditional branches that branch
13880 to their fall-through.
13882 2017-08-24 Martin Sebor <msebor@redhat.com>
13884 PR middle-end/81908
13885 * gimple-fold.c (size_must_be_zero_p): New function.
13886 (gimple_fold_builtin_memory_op): Call it.
13888 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13890 * config/rs6000/mm_malloc.h: New file.
13892 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13894 PR tree-optimization/81913
13895 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13896 analysis when either IVs in condition can wrap.
13898 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13900 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13901 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13903 2017-08-24 Richard Biener <rguenther@suse.de>
13906 * targhooks.c (default_target_can_inline_p): Properly
13907 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13908 is present and always compare.
13909 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13910 infer -mfpmath=sse from TARGET_SSE_P.
13911 (ix86_can_inline_p): Properly use target_option_default_node when
13912 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13914 2017-08-24 Richard Biener <rguenther@suse.de>
13917 * dwarf2out.c (output_die): Handle flag_generate_offload like
13919 (output_comp_unit): Likewise.
13920 (gen_array_type_die): Likewise.
13921 (dwarf2out_early_finish): Likewise.
13922 (note_variable_value_in_expr): Likewise.
13923 (dwarf2out_finish): Likewise. Adjust assert.
13924 * cgraphunit.c (symbol_table::compile): Move setting of
13925 flag_generate_offload earlier ...
13926 (symbol_table::finalize_compilation_unit): ... here, before
13927 early debug finalization.
13929 2017-08-24 Richard Biener <rguenther@suse.de>
13931 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13932 and ipa-fnsummary.h.
13933 (ix86_can_inline_p): When ix86_fpmath flags do not match
13934 check whether the callee uses FP math at all.
13936 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13938 PR middle-end/81931
13939 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13940 instead of TYPE_PRECISION.
13942 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13943 Alan Hayward <alan.hayward@arm.com>
13944 David Sherwood <david.sherwood@arm.com>
13946 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13948 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13950 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13952 (ix86_frame::stack_realign_allocate): New field.
13953 (struct machine_frame_state): Modify comments.
13954 (machine_frame_state::sp_realigned_fp_end): New field.
13955 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13956 layout calculation.
13957 (sp_valid_at): Add assertion to assure no attempt to access invalid
13958 offset of a realigned stack.
13959 (fp_valid_at): Likewise.
13960 (choose_baseaddr): Modify comments.
13961 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13962 ix86_expand_prologue.
13963 (ix86_expand_prologue): Modify stack realignment and allocation.
13964 (ix86_expand_epilogue): Modify comments.
13965 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13966 avx2_runtime, avx512f, and avx512f_runtime.
13968 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13970 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13971 (mstackrealign): Do not init to -1.
13972 * config/i386/i386.c (ix86_option_override_internal):
13973 Check opts_set, not opts when setting default value of
13974 opts->x_ix86_force_align_arg_pointer.
13976 2017-08-23 Richard Biener <rguenther@suse.de>
13978 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13979 lang_hooks.decl_printable_name.
13980 * print-rtl-function.c (print_rtx_function): Likewise.
13981 * tree-pretty-print.c (dump_function_header): Likewise.
13983 2017-08-23 Richard Biener <rguenther@suse.de>
13986 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13987 -g0 at compile-time.
13989 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13991 PR middle-end/19706
13992 * doc/sourcebuild.texi (Other hardware attributes):
13995 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13997 PR middle-end/19706
13998 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13999 Fix single-use check.
14001 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14003 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14005 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14007 * doc/install.texi: Modify to add more details on running selected
14010 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14012 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14013 is combined with -mabi=ms.
14014 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14017 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14019 PR tree-optimization/81488
14020 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14021 and cached_basis fields.
14022 (MAX_SPREAD): New constant.
14023 (alloc_cand_and_find_basis): Initialize new fields.
14024 (clear_visited): New function.
14025 (create_phi_basis_1): Rename from create_phi_basis, set visited
14026 and cached_basis fields.
14027 (create_phi_basis): New wrapper function.
14028 (phi_add_costs_1): Rename from phi_add_costs, add spread
14029 parameter, set visited field, short-circuit when limits reached.
14030 (phi_add_costs): New wrapper function.
14031 (record_phi_increments_1): Rename from record_phi_increments, set
14033 (record_phi_increments): New wrapper function.
14034 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14035 (phi_incr_cost): New wrapper function.
14036 (all_phi_incrs_profitable_1): Rename from
14037 all_phi_incrs_profitable, set visited field.
14038 (all_phi_incrs_profitable): New wrapper function.
14040 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14041 Alan Hayward <alan.hayward@arm.com>
14042 David Sherwood <david.sherwood@arm.com>
14044 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14045 that takes the outer and inner modes.
14046 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14047 comparison as the canonical test for a paradoxical subreg.
14048 * combine.c (simplify_set): Use paradoxical_subreg_p.
14049 (make_extraction): Likewise.
14050 (force_to_mode): Likewise.
14051 (rtx_equal_for_field_assignment_p): Likewise.
14052 (gen_lowpart_for_combine): Likewise.
14053 (simplify_comparison): Likewise.
14054 * cse.c (equiv_constant): Likewise.
14055 * expmed.c (store_bit_field_1): Likewise.
14056 * final.c (alter_subreg): Likewise.
14057 * fwprop.c (propagate_rtx): Likewise.
14058 (forward_propagate_subreg): Likewise.
14059 * ira-conflicts.c (ira_build_conflicts): Likewise.
14060 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14061 * lra-constraints.c (curr_insn_transform): Likewise.
14062 (split_reg): Likewise.
14063 * lra-eliminations.c (move_plus_up): Likewise.
14064 (lra_eliminate_regs_1): Likewise.
14065 * recog.c (general_operand): Likewise.
14066 * ree.c (combine_reaching_defs): Likewise.
14067 * reload.c (push_reload): Likewise.
14068 (find_reloads): Likewise.
14069 * reload1.c (elimination_effects): Likewise.
14070 (compute_reload_subreg_offset): Likewise.
14071 (choose_reload_regs): Likewise.
14072 * rtlanal.c (subreg_lsb_1): Likewise.
14073 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14074 (simplify_subreg): Likewise.
14075 * var-tracking.c (track_loc_p): Likewise.
14076 * emit-rtl.c (byte_lowpart_offset): Likewise.
14077 (paradoxical_subreg_p): Delete out-of-line definition.
14079 2017-08-22 Jeff Law <law@redhat.com>
14081 PR tree-optimization/81741
14082 PR tree-optimization/71947
14083 * tree-ssa-dom.c: Include tree-inline.h.
14084 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14085 equivalences if one is more expensive to compute than the other.
14086 * tree-ssa-scopedtables.h (class const_or_copies): Make
14087 record_const_or_copy_raw method private.
14088 (class avail_exprs_stack): New method simplify_binary_operation.
14089 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14090 avail_exprs_stack::simplify_binary_operation as needed.
14091 (avail_exprs_stack::simplify_binary_operation): New function.
14093 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14095 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14096 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14097 (DOT_SYMBOLS): Likewise.
14098 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14099 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14100 (RS6000_ABI_NAME): Likewise.
14101 (TARGET_CMODEL): Likewise.
14102 (TOC_SECTION_ASM_OP): Likewise.
14103 (SET_CMODEL): New macro.
14104 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14106 2017-08-22 Richard Biener <rguenther@suse.de>
14108 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14109 to allow for free-lang-data replacements similar to verify_type_variant.
14111 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14112 Alan Hayward <alan.hayward@arm.com>
14113 David Sherwood <david.sherwood@arm.com>
14115 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14116 VOIDmode for the LABEL_REF.
14118 2017-08-22 Richard Biener <rguenther@suse.de>
14120 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14122 2017-08-22 Richard Biener <rguenther@suse.de>
14124 * common.opt (feliminate-dwarf2-dups): Ignore.
14125 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14126 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14127 same_die_p_wrap, compute_section_prefix,
14128 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14129 (comdat_symbol_id, comdat_symbol_number): Likewise.
14130 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14132 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14133 (output_die): Mark unreachable path unreachable.
14134 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14135 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14136 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14137 (dwarf2out_early_finish): Likewise.
14139 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14141 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14143 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14146 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14147 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14148 (avr_attribute_table) <io, io_low, address>: Initialize
14149 .decl_required with true.
14151 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14153 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14154 undocumented debugging options.
14155 (-mvsx-scalar-double): Likewise.
14156 (-mallow-df-permute): Likewise.
14157 (-mvectorize-builtins): Likewise.
14158 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14159 (rs6000_builtin_vectorized_function): Likewise.
14160 (rs6000_builtin_md_vectorized_function): Likewise.
14161 (rs6000_opt_vars): Likewise.
14163 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14166 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14167 (*btrq_imm): Rename from *btrq.
14168 (*btcq_imm): Rename from *btcq.
14169 (btsc): New code attribute.
14170 (*<btsc><mode>): New insn pattern.
14171 (*btr<mode>): Ditto.
14172 (*<btsc><mode>_mask): New insn_and_split pattern.
14173 (*btr<mode>_mask): Ditto.
14175 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14176 Alan Hayward <alan.hayward@arm.com>
14177 David Sherwood <david.sherwood@arm.com>
14179 * function.c (pad_below): Simplify padding calculation.
14181 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14182 Alan Hayward <alan.hayward@arm.com>
14183 David Sherwood <david.sherwood@arm.com>
14185 * target.def (function_prologue): Remove frame size argument.
14186 (function_epilogue): Likewise.
14187 * doc/tm.texi: Regenerate.
14188 * final.c (final_start_function): Update call to function_prologue.
14189 (final_end_function): Update call to function_epilogue.
14190 (default_function_pro_epilogue): Remove frame size argument.
14191 * output.h (default_function_pro_epilogue): Likewise.
14192 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14193 (arm_output_function_prologue): Likewise.
14194 * config/frv/frv.c (frv_function_prologue): Likewise.
14195 (frv_function_epilogue): Likewise.
14196 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14197 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14198 (ia64_output_function_epilogue): Likewise.
14199 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14200 (m32r_output_function_epilogue): Likewise.
14201 * config/microblaze/microblaze.c (microblaze_function_prologue)
14202 (microblaze_function_epilogue): Likewise.
14203 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14204 (mips_output_function_epilogue): Likewise.
14205 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14206 (mmix_target_asm_function_epilogue): Likewise.
14207 * config/msp430/msp430.c (msp430_start_function): Likewise.
14208 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14209 (nds32_asm_function_epilogue): Likewise.
14210 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14211 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14212 (pa_output_function_epilogue): Likewise.
14213 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14214 (rs6000_output_function_epilogue): Likewise.
14215 * config/rl78/rl78.c (rl78_start_function): Likewise.
14216 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14217 (rs6000_output_function_epilogue): Likewise.
14218 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14219 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14220 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14221 (sparc_asm_function_epilogue): Likewise.
14223 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14225 * tree.h (type_has_mode_precision_p): New function.
14226 * convert.c (convert_to_integer_1): Use it.
14227 * expr.c (expand_expr_real_2): Likewise.
14228 (expand_expr_real_1): Likewise.
14229 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14230 * match.pd: Likewise.
14231 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14232 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14233 * tree-tailcall.c (process_assignment): Likewise.
14234 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14235 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14236 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14237 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14238 (vectorizable_assignment): Likewise.
14239 (vectorizable_shift): Likewise.
14240 (vectorizable_operation): Likewise.
14241 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14243 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14245 * match.pd: Add pow (C, x) simplification.
14247 2017-08-21 Richard Biener <rguenther@suse.de>
14249 PR tree-optimization/81900
14250 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14251 for blocks with abnormal predecessors.
14252 (compute_antic): Do not set visited flag prematurely.
14254 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14257 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14259 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14261 * stor-layout.h (vector_type_mode): Move to...
14262 * tree.h (vector_type_mode): ...here.
14263 * stor-layout.c (vector_type_mode): Move to...
14264 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14266 2017-08-21 Richard Biener <rguenther@suse.de>
14268 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14269 register_external_die hooks.
14270 (debug_false_tree_charstarstar_uhwistar): Declare.
14271 (debug_nothing_tree_charstar_uhwi): Likewise.
14272 * debug.c (do_nothing_debug_hooks): Adjust.
14273 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14274 (debug_nothing_tree_charstar_uhwi): Likewise.
14275 * dbxout.c (dbx_debug_hooks): Adjust.
14276 (xcoff_debug_hooks): Likewise.
14277 * sdbout.c (sdb_debug_hooks): Likewise.
14278 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14279 * dwarf2out.c (macinfo_label_base): New global.
14280 (dwarf2out_register_external_die): New function for the
14281 register_external_die hook.
14282 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14283 (dwarf2_debug_hooks): Use them.
14284 (dwarf2_lineno_debug_hooks): Adjust.
14285 (struct die_struct): Add with_offset flag.
14286 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14287 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14288 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14289 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14290 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14291 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14292 defining section names for the early LTO debug variants.
14293 (reset_indirect_string): New helper.
14294 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14295 (print_dw_val): Add support for offsetted symbol references.
14296 (get_ultimate_context): Split out from is_cxx.
14297 (is_cxx): Use get_ultimate_context.
14298 (is_fortran): Add decl overload.
14299 (compute_comp_unit_symbol): Split out worker from
14300 compute_section_prefix.
14301 (compute_section_prefix): Call compute_comp_unit_symbol and
14302 set comdat_type_p here.
14303 (output_die): Skip DIE symbol output for the LTO added one.
14304 Handle DIE symbol references with offset.
14305 (output_comp_unit): Guard section name mangling properly.
14306 For LTO debug sections emit a symbol at the section beginning
14307 which we use to refer to its DIEs.
14308 (add_abstract_origin_attribute): For DIEs registered via
14309 dwarf2out_register_external_die directly refer to the early
14310 DIE rather than indirectly through the shadow one we created.
14311 Remove obsolete call to dwarf2out_abstract_function for
14312 non-function/block origins.
14313 (gen_array_type_die): When generating early LTO debug do
14314 not emit DW_AT_string_length.
14315 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14316 late when in LTO. As suggested place a gcc_unreachable for
14317 the DECL_ABSTRACT_P case.
14318 (gen_subprogram_die): Avoid another specification DIE
14319 for early built declarations/definitions for the late LTO case.
14320 (gen_variable_die): Add type references for late duplicated VLA dies
14322 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14323 we have the abstract instance already.
14324 (process_scope_var): Adjust decl DIE contexts in LTO which
14325 first puts them in limbo.
14326 (gen_decl_die): Do not generate type DIEs late apart from
14327 types for VLAs or for decls we do not yet have a DIE. Do not
14328 call dwarf2out_abstract_function late.
14329 (dwarf2out_early_global_decl): Make sure to create DIEs
14330 for abstract instances of a decl first.
14331 (dwarf2out_late_global_decl): Adjust comment.
14332 (output_macinfo_op): With multiple macro sections use
14333 macinfo_label_base to distinguish labels.
14334 (output_macinfo): Likewise. Update macinfo_label_base.
14335 Pass in the line info label.
14336 (note_variable_value_in_expr): When generating LTO resolve
14337 all variable values here by generating DIEs as needed.
14338 (init_sections_and_labels): Add early LTO debug flag parameter
14339 and generate different sections and names if set. Add generation
14340 counter for the labels so we can have multiple of them.
14341 (reset_dies): Helper to allow DIEs to be output multiple times.
14342 (dwarf2out_finish): When outputting DIEs to the fat part of an
14343 LTO object first reset DIEs.
14344 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14345 (modified_type_die): Check for decl_ultimate_origin being self
14347 (gen_type_die_with_usage): Likewise.
14348 (gen_typedef_die): Allow decl_ultimate_origin being self.
14349 (set_decl_abstract_flags): Remove.
14350 (set_block_abstract_flags): Likewise.
14351 (dwarf2out_abstract_function): Treat the early generated DIEs
14352 as the abstract copy and only add DW_AT_inline and
14353 DW_AT_artificial here and call set_decl_origin_self.
14354 If the DIE has an abstract origin don't do anything.
14355 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14356 if we have none yet (Go fails to build one, PR78628).
14357 (variably_modified_type_p): Prevent endless recursion for Ada
14358 cyclic pointer types.
14359 * lto-streamer-in.c: Include debug.h.
14360 (dref_queue): New global.
14361 (lto_read_tree_1): Stream in DIE references.
14362 (lto_input_tree): Register DIE references.
14363 (input_function): Stream DECL_DEBUG_ARGS.
14364 * lto-streamer-out.c: Include debug.h.
14365 (lto_write_tree_1): Output DIE references.
14366 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14367 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14368 (output_function): Stream DECL_DEBUG_ARGS.
14369 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14370 Stream DECL_ABSTRACT_ORIGIN.
14371 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14372 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14373 DECL_CONTEXT for file-scope decls.
14374 * lto-streamer.h (struct dref_entry): Declare.
14375 (dref_queue): Likewise.
14376 * cfgexpand.c (pass_expand::execute): Do not call the
14377 outlining_inline_function hook here.
14378 * lto-wrapper.c (debug_obj): New global.
14379 (tool_cleanup): Unlink it if required.
14380 (debug_objcopy): New function.
14381 (run_gcc): Handle early debug sections in the IL files by
14382 extracting them to separate files, partially linkin them and
14383 feeding the result back as result to the linker.
14384 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14385 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14386 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14387 sections into a separate segment.
14388 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14390 (darwin_asm_dwarf_section): Likewise.
14391 (darwin_asm_output_dwarf_offset): Likewise.
14392 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14394 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14395 Alan Hayward <alan.hayward@arm.com>
14396 David Sherwood <david.sherwood@arm.com>
14398 * read-md.h (md_reader::record_potential_iterator_use): Replace
14399 pointer argument with an rtx and an index.
14400 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14401 (apply_mode_iterator): Likewise.
14402 (apply_code_iterator): Likewise.
14403 (apply_int_iterator): Likewise.
14404 (apply_subst_iterator): Likewise.
14405 (record_iterator_use): Likewise.
14406 (record_attribute_use): Likewise.
14407 (md_reader::record_potential_iterator_use): Likewise. Update calls
14408 to record_iterator_use and apply_iterator.
14409 (iterator_use): Replace ptr with x and index.
14410 (attribute_use): Likewise.
14411 (apply_attribute_uses): Update calls to apply_iterator.
14412 (apply_iterators): Likewise. Update initialization of iterator_use.
14413 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14414 and record_potential_iterator_use.
14415 (rtx_reader::read_rtx_operand): Likewise.
14417 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14418 Alan Hayward <alan.hayward@arm.com>
14419 David Sherwood <david.sherwood@arm.com>
14421 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14424 2017-08-21 Richard Biener <rguenther@suse.de>
14426 PR middle-end/81884
14427 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14428 at struct end conservatively when comparing common bases.
14430 2017-08-21 Richard Biener <rguenther@suse.de>
14432 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14433 (mem_ref_in_stmt): Remove.
14434 (determine_max_movement): Use ref index to get at the reference.
14435 (invariantness_dom_walker::before_dom_children): Deal with
14436 lim data already initialized.
14437 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14439 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14441 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14442 (TARGET_ISA_ROUND): Ditto.
14443 (TARGET_ROUND): Ditto.
14444 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14445 * config/i386/i386.md: Ditto.
14446 * config/i386/sse.md: Ditto.
14447 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14448 with OPTION_MASK_ISA_SSE4_1.
14450 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14453 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14454 __builtin_ia32_lzcnt_u16.
14456 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14459 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14460 (rs6000_set_current_function): Rewrite function to use it.
14462 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14465 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14466 and TYPE_WARN_IF_NOT_ALIGN.
14467 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14468 (handle_warn_if_not_align): New.
14469 (place_union_field): Call handle_warn_if_not_align.
14470 (place_field): Call handle_warn_if_not_align.
14471 Copy TYPE_WARN_IF_NOT_ALIGN.
14472 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14473 (layout_type): Likewise.
14474 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14476 (tree_decl_common): Add warn_if_not_align.
14477 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14478 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14479 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14480 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14481 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14482 * doc/extend.texi: Document warn_if_not_aligned attribute.
14483 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14485 2017-08-17 Martin Liska <mliska@suse.cz>
14488 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14489 (get_data_dependence): Use it as pointer type.
14490 (distribute_loop): Likewise.
14492 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14494 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14495 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14496 (p8_vmrgow_<mode>_direct): New define_insn.
14497 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14498 handle endianness for vmrgew and vmrgow permute patterns.
14500 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14502 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14503 * config/rs6000/rs6000-cpus.def: Remove comment.
14504 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14505 (POWERPC_MASKS): Likewise.
14506 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14507 use of TARGET_VSX_TIMODE.
14508 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14509 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14510 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14511 (rs6000_option_override_internal): Remove dead code.
14512 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14513 (rs6000_legitimize_reload_address): Likewise.
14514 (rs6000_legitimate_address_p): Likewise.
14515 (rs6000_opt_masks): Delete "vsx-timode".
14516 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14517 from function comment.
14518 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14519 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14520 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14522 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14523 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14524 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14525 (VSX_TI): Likewise.
14527 (define_peephole2): Likewise.
14529 2017-08-17 Martin Sebor <msebor@redhat.com>
14532 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14533 past the end of an array.
14534 (test_pp_format): Add test cases.
14536 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14538 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14539 missing ECF_NOTHROW flags.
14541 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14544 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14545 operands residing in integer registers.
14546 (*vsx_le_perm_load_<mode>): Likewise.
14547 (*vsx_le_perm_store_<mode>): Likewise.
14548 (define_peephole2): Add peepholes to optimize the above.
14550 2017-08-17 Marek Polacek <polacek@redhat.com>
14552 PR middle-end/81814
14553 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14554 to mimic what shorten_compare did. Change the return type to bool.
14555 (fold_cond_expr_with_comparison): Update call to
14556 operand_equal_for_comparison_p.
14557 (fold_ternary_loc): Likewise.
14559 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14561 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14563 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14565 2017-08-17 Richard Biener <rguenther@suse.de>
14567 * tree-ssa-structalias.c (solve_graph): When propagating
14568 to successors update the graphs succ edges and avoid duplicate work.
14570 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14573 * config/i386/i386.c (ix86_option_override_internal): Save target
14574 specific options after ix86_stack_protector_guard_reg was changed.
14576 2017-08-17 Richard Biener <rguenther@suse.de>
14578 PR tree-optimization/81827
14579 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14580 (new_var_info): Initialize it conservatively.
14581 (get_call_vi): Mark register vars.
14582 (new_scalar_tmp_constraint_exp): Likewise.
14583 (handle_rhs_call): Likewise.
14584 (handle_const_call): Likewise.
14585 (create_function_info_for): Likewise.
14586 (solve_constraints): Sort varinfos to separate register from
14587 non-register vars to pack points-to solution bitmaps during
14590 2017-08-17 Marek Polacek <polacek@redhat.com>
14592 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14594 2017-08-17 Richard Biener <rguenther@suse.de>
14596 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14597 to true when overflow is undefined and we saturated the result.
14599 2017-08-17 Alan Modra <amodra@gmail.com>
14602 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14603 Don't use store multiple if only one reg needs saving.
14604 (interesting_frame_related_regno): New function.
14605 (rs6000_frame_related): Don't emit frame info for regs that
14607 (rs6000_emit_epilogue): Likewise.
14609 2017-08-16 Nathan Sidwell <nathan@acm.org>
14611 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14612 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14613 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14614 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14615 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14616 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14617 (verify_type): Adjust for TYPE_BINFO move.
14618 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14619 process TYPE_BINFO directly.
14620 (hash_tree): Likewise.
14621 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14623 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14626 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14628 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14630 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14633 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14634 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14635 Add ix86_binary_operator_ok to insn constraint.
14636 (*iordi_1_bts): Ditto.
14637 (*xordi_1_btc): Ditto.
14638 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14639 Update corresponding peephole2 pattern.
14643 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14645 PR tree-optimization/81832
14646 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14647 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14649 2017-08-16 Marek Polacek <polacek@redhat.com>
14652 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14653 regarding size_zero_node.
14655 2017-08-16 Martin Liska <mliska@suse.cz>
14658 * config.gcc: Respect previously set extra_objs in case
14661 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14663 PR tree-optimization/81835
14664 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14665 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14666 not depend on the phi.
14668 2017-08-16 Alan Modra <amodra@gmail.com>
14670 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14673 2017-08-16 Alan Modra <amodra@gmail.com>
14675 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14676 (save_reg_p): ..into this. Update all callers.
14677 (first_reg_to_save): Simplify.
14679 2017-08-16 Alan Modra <amodra@gmail.com>
14681 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14684 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14688 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14689 constant count if that count is less than 32.
14691 2017-08-15 Nathan Sidwell <nathan@acm.org>
14693 * gcc.c (execute): Emit friendlier message if inferior is killed
14694 by an external cause.
14696 2017-08-15 Richard Biener <rguenther@suse.de>
14698 PR tree-optimization/81790
14699 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14700 CONSTRUCTORs from simplifying and VN.
14702 2017-08-14 Martin Sebor <msebor@redhat.com>
14704 * builtin-attrs.def: Add comments.
14706 2017-08-14 Martin Sebor <msebor@redhat.com>
14709 * doc/extend.texi (attribute nonstring): Document new attribute.
14711 2017-08-14 Martin Sebor <msebor@redhat.com>
14714 * tree-diagnostic.c (default_tree_printer): Handle %G.
14715 * gimple-pretty-print.h (percent_G_format): Declare new function.
14716 * gimple-pretty-print.c (percent_G_format): Define.
14717 * tree-pretty-print.c (percent_K_format): Add argument.
14719 2017-08-14 Martin Sebor <msebor@redhat.com>
14721 PR translation/79998
14722 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14723 Remove a stray space.
14725 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14728 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14729 (*iordi_1_bts): Ditto.
14730 (*xordi_1_btc): Ditto.
14732 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14735 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14736 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14738 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14739 quoted strings, and make more translator-friendly.
14740 (darwin_rs6000_override_options): Likewise.
14741 (rs6000_option_override_internal): Likewise.
14742 (rs6000_return_in_memory): Fix overlong line.
14743 (init_cmulative_args): Use quoted strings, and make more
14744 translator-friendly.
14745 (rs6000_pass_by_reference): Fix overlong line.
14746 (def_builtin): Use quoted strings.
14747 (altivec_expand_predicate_builtin): Use quoted strings, and make
14748 more translator-friendly.
14749 (htm_expand_builtin): Use quoted strings.
14750 (cpu_expand_builtin): Use quoted strings, and make more
14751 translator-friendly.
14752 (altivec_expand_builtin): Likewise.
14753 (paired_expand_predicate_builtin): Likewise.
14754 (rs6000_invalid_builtin): Likewise.
14755 (builtin_function_type): Use quoted strings.
14756 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14757 more translator-friendly.
14758 (rs6000_trampoline_init): Likewise.
14759 (rs6000_handle_altivec_attribute): Likewise.
14760 (rs6000_inner_target_options): Use quoted strings.
14761 (rs6000_disable_incompatible_switches): Likewise.
14762 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14763 strings, and make more translator-friendly.
14764 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14766 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14768 PR tree-optimization/81799
14769 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14770 cond_expr to simple gimple operand.
14772 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14774 PR middle-end/46932
14775 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14777 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14781 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14782 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14783 TARGET_GASISR_PROLOGUES.
14784 * config/avr/avr.c (avr_option_override): Same.
14785 (avr_pass_pre_proep::execute): Same.
14787 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14790 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14791 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14793 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14795 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14796 define_insn_and_split. Split to a memory load from 0 in
14797 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14798 using PTR mode iterator.
14799 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14800 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14801 (*add_tp_<mode>): Redefine as define_insn_and_split.
14802 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14803 address space. Merge with *add_tp_x32 using PTR mode iterator.
14804 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14805 Split to an add with a memory load from 0 in
14806 DEFAULT_TLS_SEG_REG address space.
14808 2017-08-12 Andrew Pinski <apinski@cavium.com>
14810 * config/aarch64/aarch64-option-extensions.def (rdma):
14811 Fix feature string to what Linux prints out in /proc/cpuinfo.
14813 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14816 * dwarf2out.c (modified_type_die): For C typedef types that have
14817 an ultimate origin, process the ultimate origin instead of the
14819 (gen_typedef_die): Assert that input DECLs have no ultimate
14821 (gen_type_die_with_usage): For typedef variants that have an
14822 ultimate origin, just call gen_decl_die on the original DECL.
14823 (process_scope_var): Avoid creating DIEs for local typedefs and
14824 concrete static variables.
14826 2017-08-12 Alan Modra <amodra@gmail.com>
14830 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14831 match gnu-user.h startfile.
14832 (ENDFILE_LINUX_SPEC): Similarly.
14834 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14837 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14839 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14841 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14842 * config/aarch64/aarch64.md (mov<mode>): Change.
14843 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14844 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14845 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14847 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14849 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14850 for storage order barriers.
14852 2017-08-11 Martin Liska <mliska@suse.cz>
14855 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14856 variables of void type.
14858 2017-08-11 Martin Liska <mliska@suse.cz>
14860 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14861 TARGET_SUPPORTS_ALIASES.
14862 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14863 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14864 (optimize_weakref): Likewise.
14865 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14866 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14867 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14869 2017-08-11 Martin Liska <mliska@suse.cz>
14872 * config/i386/i386.c (make_resolver_func): Do complete
14873 refactoring of the function.
14875 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14878 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14879 * config/i386/i386.c (ix86_stack_protect_guard): Use
14880 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14881 * doc/invoke.texi (x86 Options): Document
14882 -mstack-protector-guard-symbol= option.
14884 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14886 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14887 * config/i386/i386.c (ix86_split_stack_guard): New function.
14888 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14889 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14890 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14891 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14892 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14893 (split_stack_space_check): Call ix86_split_stack_guard.
14895 2017-08-10 Martin Sebor <msebor@redhat.com>
14897 * print-tree.c (print_node): Print location using the established
14899 Replace spaces with colons.
14900 (debug_raw, debug): Ditto.
14902 2017-08-10 Martin Sebor <msebor@redhat.com>
14905 * pretty-print.c (pp_format): Correct the handling of %s precision.
14907 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14910 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14912 (ix86_finalize_stack_frame_flags): This. Also clear
14913 frame_pointer_needed if -fno-omit-frame-pointer is used without
14915 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14916 with ix86_finalize_stack_frame_flags.
14917 (ix86_expand_epilogue): Likewise.
14918 (ix86_expand_split_stack_prologue): Likewise.
14919 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14921 2017-08-10 Martin Liska <mliska@suse.cz>
14924 * c-attribs.c (handle_target_attribute):
14925 Report warning for an empty string argument of target attribute.
14927 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14930 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14932 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14933 or DECL_NONLOCAL labels.
14934 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14935 or DECL_NONLOCAL labels here.
14937 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14939 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14940 to indicate when early gimple folding has been disabled.
14941 (rs6000_gimple_fold_builtin): Add debug content.
14942 (rs6000_invalid_builtin): Fix whitespace.
14943 (rs6000_expand_builtin): Fix whitespace.
14944 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14946 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14949 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14950 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14952 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14954 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14955 (qdf24xx): Likewise.
14956 * config/aarch64/aarch64.md: Include falkor.md.
14957 * config/aarch64/falkor.md: New.
14959 2017-08-09 Marek Polacek <polacek@redhat.com>
14962 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14963 * diagnostic.c (emit_diagnostic): Add a comment.
14964 (emit_diagnostic_valist): New function.
14966 2017-08-09 Marek Polacek <polacek@redhat.com>
14969 * input.c (make_location): New overload.
14970 * input.h (make_location): Declare.
14972 2017-08-08 Alan Modra <amodra@gmail.com>
14973 H.J. Lu <hongjiu.lu@intel.com>
14976 * gcc.c (NO_PIE_SPEC): Delete.
14977 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14979 (LINK_PIE_SPEC): ..to here.
14980 (LINK_COMMAND_SPEC): Support -no-pie.
14981 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14982 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14984 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14985 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14986 (ENDFILE_CRTEND_SPEC): Similarly.
14988 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14991 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14992 (mstack-protector-guard-offset=): Ditto.
14993 * config/i386/i386.c (ix86_option_override): Handle
14994 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14996 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14997 ix86_stack_protect_guard_offset variables.
14998 (TARGET_STACK_PROTECT_GUARD): Always define.
14999 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15000 and -mstack-protector-guard-offset= options.
15002 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15004 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15005 boundary case for the last candidate.
15007 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15009 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15010 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15012 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15014 PR middle-end/19706
15015 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15016 * config/aarch64/aarch64-builtins.c
15017 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15018 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15019 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15021 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15022 Andrew Pinski <pinskia@gmail.com>
15024 PR middle-end/19706
15025 * internal-fn.def (XORSIGN): New.
15026 * optabs.def (xorsign_optab): New.
15027 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15028 (convert_expand_mult_copysign): New.
15029 (pass_optimize_widening_mul::execute): Call
15030 convert_expand_mult_copysign.
15032 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15034 PR tree-optimization/81354
15035 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15036 Insert on edges rather than explicitly creating landing pads.
15037 (analyze_candidates_and_replace): Commit edge inserts.
15039 2017-08-08 Richard Biener <rguenther@suse.de>
15041 PR middle-end/81719
15042 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15043 (expand_simple_operations): Also look through ADDR_EXPRs with
15044 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15046 2017-08-08 Richard Biener <rguenther@suse.de>
15048 PR tree-optimization/81723
15049 * tree-vect-slp.c (struct bst_traits): New hash traits.
15050 (bst_fail): New global.
15051 (vect_build_slp_tree_2): New worker, split out from ...
15052 (vect_build_slp_tree): ... this now wrapping it with using
15053 bst_fail set to cache SLP tree build fails. Properly handle
15055 (vect_analyze_slp_instance): Allocate and free bst_fail.
15057 2017-08-08 Martin Liska <mliska@suse.cz>
15060 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15061 LABEL_DECLs that can be from a different function.
15063 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15065 PR tree-optimization/81744
15066 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15067 loop's number of iterations.
15069 2017-08-08 Martin Liska <mliska@suse.cz>
15071 * asan.c: Include header files.
15072 * attribs.c (build_decl_attribute_variant): New function moved
15074 (build_type_attribute_qual_variant): Likewise.
15075 (cmp_attrib_identifiers): Likewise.
15076 (simple_cst_list_equal): Likewise.
15077 (omp_declare_simd_clauses_equal): Likewise.
15078 (attribute_value_equal): Likewise.
15079 (comp_type_attributes): Likewise.
15080 (build_type_attribute_variant): Likewise.
15081 (lookup_ident_attribute): Likewise.
15082 (remove_attribute): Likewise.
15083 (merge_attributes): Likewise.
15084 (merge_type_attributes): Likewise.
15085 (merge_decl_attributes): Likewise.
15086 (merge_dllimport_decl_attributes): Likewise.
15087 (handle_dll_attribute): Likewise.
15088 (attribute_list_equal): Likewise.
15089 (attribute_list_contained): Likewise.
15090 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15091 (lookup_attribute_by_prefix): Likewise.
15092 * bb-reorder.c: Include header files.
15093 * builtins.c: Likewise.
15094 * calls.c: Likewise.
15095 * cfgexpand.c: Likewise.
15096 * cgraph.c: Likewise.
15097 * cgraphunit.c: Likewise.
15098 * convert.c: Likewise.
15099 * dwarf2out.c: Likewise.
15100 * final.c: Likewise.
15101 * fold-const.c: Likewise.
15102 * function.c: Likewise.
15103 * gimple-expr.c: Likewise.
15104 * gimple-fold.c: Likewise.
15105 * gimple-pretty-print.c: Likewise.
15106 * gimple.c: Likewise.
15107 * gimplify.c: Likewise.
15108 * hsa-common.c: Likewise.
15109 * hsa-gen.c: Likewise.
15110 * internal-fn.c: Likewise.
15111 * ipa-chkp.c: Likewise.
15112 * ipa-cp.c: Likewise.
15113 * ipa-devirt.c: Likewise.
15114 * ipa-fnsummary.c: Likewise.
15115 * ipa-inline.c: Likewise.
15116 * ipa-visibility.c: Likewise.
15118 * lto-cgraph.c: Likewise.
15119 * omp-expand.c: Likewise.
15120 * omp-general.c: Likewise.
15121 * omp-low.c: Likewise.
15122 * omp-offload.c: Likewise.
15123 * omp-simd-clone.c: Likewise.
15124 * opts-global.c: Likewise.
15125 * passes.c: Likewise.
15126 * predict.c: Likewise.
15127 * sancov.c: Likewise.
15128 * sanopt.c: Likewise.
15129 * symtab.c: Likewise.
15130 * toplev.c: Likewise.
15131 * trans-mem.c: Likewise.
15132 * tree-chkp.c: Likewise.
15133 * tree-eh.c: Likewise.
15134 * tree-into-ssa.c: Likewise.
15135 * tree-object-size.c: Likewise.
15136 * tree-parloops.c: Likewise.
15137 * tree-profile.c: Likewise.
15138 * tree-ssa-ccp.c: Likewise.
15139 * tree-ssa-live.c: Likewise.
15140 * tree-ssa-loop.c: Likewise.
15141 * tree-ssa-sccvn.c: Likewise.
15142 * tree-ssa-structalias.c: Likewise.
15143 * tree-ssa.c: Likewise.
15144 * tree-streamer-in.c: Likewise.
15145 * tree-vectorizer.c: Likewise.
15146 * tree-vrp.c: Likewise.
15147 * tsan.c: Likewise.
15148 * ubsan.c: Likewise.
15149 * varasm.c: Likewise.
15150 * varpool.c: Likewise.
15151 * tree.c: Remove functions moved to attribs.[ch].
15152 * tree.h: Likewise.
15153 * config/aarch64/aarch64.c: Add attrs.h header file.
15154 * config/alpha/alpha.c: Likewise.
15155 * config/arc/arc.c: Likewise.
15156 * config/arm/arm.c: Likewise.
15157 * config/avr/avr.c: Likewise.
15158 * config/bfin/bfin.c: Likewise.
15159 * config/c6x/c6x.c: Likewise.
15160 * config/cr16/cr16.c: Likewise.
15161 * config/cris/cris.c: Likewise.
15162 * config/darwin.c: Likewise.
15163 * config/epiphany/epiphany.c: Likewise.
15164 * config/fr30/fr30.c: Likewise.
15165 * config/frv/frv.c: Likewise.
15166 * config/ft32/ft32.c: Likewise.
15167 * config/h8300/h8300.c: Likewise.
15168 * config/i386/winnt.c: Likewise.
15169 * config/ia64/ia64.c: Likewise.
15170 * config/iq2000/iq2000.c: Likewise.
15171 * config/lm32/lm32.c: Likewise.
15172 * config/m32c/m32c.c: Likewise.
15173 * config/m32r/m32r.c: Likewise.
15174 * config/m68k/m68k.c: Likewise.
15175 * config/mcore/mcore.c: Likewise.
15176 * config/microblaze/microblaze.c: Likewise.
15177 * config/mips/mips.c: Likewise.
15178 * config/mmix/mmix.c: Likewise.
15179 * config/mn10300/mn10300.c: Likewise.
15180 * config/moxie/moxie.c: Likewise.
15181 * config/msp430/msp430.c: Likewise.
15182 * config/nds32/nds32-isr.c: Likewise.
15183 * config/nds32/nds32.c: Likewise.
15184 * config/nios2/nios2.c: Likewise.
15185 * config/nvptx/nvptx.c: Likewise.
15186 * config/pa/pa.c: Likewise.
15187 * config/pdp11/pdp11.c: Likewise.
15188 * config/powerpcspe/powerpcspe.c: Likewise.
15189 * config/riscv/riscv.c: Likewise.
15190 * config/rl78/rl78.c: Likewise.
15191 * config/rx/rx.c: Likewise.
15192 * config/s390/s390.c: Likewise.
15193 * config/sh/sh.c: Likewise.
15194 * config/sol2.c: Likewise.
15195 * config/sparc/sparc.c: Likewise.
15196 * config/spu/spu.c: Likewise.
15197 * config/stormy16/stormy16.c: Likewise.
15198 * config/tilegx/tilegx.c: Likewise.
15199 * config/tilepro/tilepro.c: Likewise.
15200 * config/v850/v850.c: Likewise.
15201 * config/vax/vax.c: Likewise.
15202 * config/visium/visium.c: Likewise.
15203 * config/xtensa/xtensa.c: Likewise.
15205 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15208 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15209 constraints since the -mupper-regs-* switches have been
15211 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15212 into a vector from a double word element that was extracted from
15213 another vector, and eliminate extra XXPERMDI instructions.
15214 (vsx_concat_<mode>_2): Likewise.
15215 (vsx_concat_<mode>_3): Likewise.
15216 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15217 concat to allow optimizing inserts from previous extracts.
15219 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15221 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15222 memory reference to a SSP offset in TLS address space.
15223 (ix86_print_operand) <case '@'>: Remove.
15224 (ix86_print_operand_punct_valid_p): Remove '@' code.
15225 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15226 UNSPEC_SP_TLS_TEST.
15227 (stack_tls_protect_set_<mode>): Remove.
15228 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15229 (stack_tls_protect_test_<mode>): Remove.
15230 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15232 2017-08-07 Olivier Hainque <hainque@adacore.com>
15235 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15237 2017-08-07 Douglas Rupp <rupp@adacore.com>
15239 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15240 variable was referenced as multidir in command.
15242 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15245 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15248 2017-08-07 Martin Liska <mliska@suse.cz>
15250 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15251 * config/rl78/rl78.c: Add include of attribs.h.
15252 * config/sh/sh.c: Likewise.
15253 * config/v850/v850.c: Likewise.
15255 2017-08-07 Tom de Vries <tom@codesourcery.com>
15257 PR middle-end/78266
15258 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15260 2017-08-07 Martin Liska <mliska@suse.cz>
15262 * config/mips/mips.c: Include attribs.h.
15264 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15267 * doc/invoke.texi: Document change in behvaior for -Ofast for
15270 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15272 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15274 (aarch64_pop_regs): Likewise.
15275 (aarch64_gen_load_pair): Likewise.
15276 (aarch64_save_callee_saves): Likewise.
15277 (aarch64_restore_callee_saves): Likewise.
15279 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15281 * config/i386/i386.c: Revert the last change.
15283 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15286 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15288 (ix86_finalize_stack_frame_flags): This. Also clear
15289 frame_pointer_needed if -fno-omit-frame-pointer is used without
15291 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15292 with ix86_finalize_stack_frame_flags.
15293 (ix86_expand_epilogue): Likewise.
15294 (ix86_expand_split_stack_prologue): Likewise.
15296 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15299 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15300 to P_AES for Westmere.
15302 2017-08-07 Jonathan Yong <10walls@gmail.com>
15304 * config/i386/mingw.opt (fset-stack-executable): Removed.
15305 * config/i386/cygming.opt (fset-stack-executable): Moved
15307 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15309 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15311 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15313 2017-08-07 Marek Polacek <polacek@redhat.com>
15315 PR middle-end/81737
15316 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15318 2017-08-07 Martin Liska <mliska@suse.cz>
15320 * attribs.h (canonicalize_attr_name): New function.
15321 (cmp_attribs): Move from c-format.c and adjusted.
15322 (is_attribute_p): Moved from tree.h.
15323 * tree-inline.c: Add new includes.
15324 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15325 (private_is_attribute_p): Remove.
15326 (private_lookup_attribute): Likewise.
15327 (private_lookup_attribute_by_prefix): Simplify.
15328 (remove_attribute): Use is_attribute_p.
15329 * tree.h: Remove removed declarations.
15331 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15333 PR middle-end/81698
15334 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15335 instead of computing it in the function. Formatting fix.
15336 (expand_case): Don't rely on default_edge being the first edge,
15337 clear it if removing it, pass default_edge to
15338 emit_case_dispatch_table.
15339 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15342 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15344 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15345 insn in the function, emit NOP after the insn.
15347 2017-08-06 Tom de Vries <tom@codesourcery.com>
15349 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15352 2017-08-06 Tom de Vries <tom@codesourcery.com>
15354 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15357 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15359 PR tree-optimization/57371
15360 * match.pd: New pattern.
15362 2017-08-04 Marek Polacek <polacek@redhat.com>
15364 PR middle-end/81695
15365 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15366 perform the computation in offset_int.
15368 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15370 PR tree-optimization/81136
15371 * tree-vectorizer.h: Include tree-hash-traits.h.
15372 (vec_base_alignments): New typedef.
15373 (vec_info): Add a base_alignments field.
15374 (vect_record_base_alignments): Declare.
15375 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15377 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15378 (create_data_ref): Add an is_conditional_in_stmt argument.
15379 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15380 the is_conditional_in_stmt field.
15381 (data_ref_loc): Add an is_conditional_in_stmt field.
15382 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15383 (find_data_references_in_stmt): Update call to create_data_ref.
15384 (graphite_find_data_references_in_stmt): Likewise.
15385 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15386 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15387 (vect_record_base_alignment): New function.
15388 (vect_record_base_alignments): Likewise.
15389 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15390 for nested statements even if we fail to compute a misalignment.
15391 Use pooled base alignments for unconditional references.
15392 (vect_find_same_alignment_drs): Compare base addresses instead
15394 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15395 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15397 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15399 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15400 Add an explicit name for the enum. Use auto_vec for slp_instances
15401 and grouped_stores.
15402 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15404 (_bb_vec_info): Add a constructor and destructor.
15405 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15406 (destroy_loop_vec_info): Delete.
15407 (vect_destroy_datarefs): Likewise.
15408 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15409 (vec_info::vec_info): New function.
15410 (vec_info::~vec_info): Likewise.
15411 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15412 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15413 destroy_loop_vec_info.
15414 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15415 (_loop_vec_info::_loop_vec_info): ...this.
15416 (destroy_loop_vec_info): Replace with...
15417 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15418 the stmt_vec_infos. Leave handling of vec_info information to its
15419 destructor. Remove explicit vector releases.
15420 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15421 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15422 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15423 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15424 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15425 (destroy_bb_vec_info): Replace with...
15426 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15427 information to its destructor.
15428 (vect_slp_analyze_bb_1): Use new and delete instead of
15429 new_bb_vec_info and destroy_bb_vec_info.
15430 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15433 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15435 * tree-data-ref.h (subscript): Add access_fn field.
15436 (data_dependence_relation): Add could_be_independent_p.
15437 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15438 (same_access_functions): Move to tree-data-ref.c.
15439 * tree-data-ref.c (ref_contains_union_access_p): New function.
15440 (access_fn_component_p): Likewise.
15441 (access_fn_components_comparable_p): Likewise.
15442 (dr_analyze_indices): Add a reference to access_fn_component_p.
15443 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15445 (constant_access_functions): Likewise.
15446 (add_other_self_distances): Likewise.
15447 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15448 (initialize_data_dependence_relation): Use XCNEW and remove
15449 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15450 of access functions that have the same type. Allow the
15451 subsequence to end with different bases in some circumstances.
15452 Record the chosen access functions in SUB_ACCESS_FN.
15453 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15454 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15455 (subscript_dependence_tester_1): Likewise dra and drb.
15456 (build_classic_dist_vector): Update calls accordingly.
15457 (subscript_dependence_tester): Likewise.
15458 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15459 DDR_COULD_BE_INDEPENDENT_P.
15460 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15461 comp_alias_ddrs instead of may_alias_ddrs.
15462 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15464 (vect_analyze_data_ref_dependence): Use it if
15465 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15466 distance vectors if that fails.
15467 (dependence_distance_ge_vf): New function.
15468 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15469 LOOP_VINFO_MAY_ALIAS_DDRS.
15471 2017-08-04 Richard Biener <rguenther@suse.de>
15473 PR middle-end/81705
15474 * fold-const.c (fold_binary_loc): Properly restrict
15475 minus_var0 && minus_var1 case when associating undefined overflow
15478 2017-08-04 Tom de Vries <tom@codesourcery.com>
15480 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15482 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15484 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15485 Don't start diagnostic messages with a capital letter.
15486 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15488 (rs6000_invalid_builtin): Likewise.
15489 (rs6000_trampoline_init): Likewise.
15491 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15494 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15495 after setting changeable df flags.
15497 2017-08-03 Richard Biener <rguenther@suse.de>
15499 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15500 up if the use is in USE - X.
15502 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15504 * toplev.c (dumpfile.h): New include.
15505 (internal_error_reentered): New static function. Use it...
15506 (internal_error_function): ...here to handle reentered internal_error.
15508 2017-08-03 Richard Biener <rguenther@suse.de>
15510 PR middle-end/81148
15511 * fold-const.c (split_tree): Add minus_var and minus_con
15512 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15513 here but always use minus_*.
15514 (associate_trees): Assert we never associate with MINUS_EXPR
15515 and NULL first operand. Do not recurse for PLUS_EXPR operands
15516 when associating as MINUS_EXPR either.
15517 (fold_binary_loc): Track minus_var and minus_con.
15519 2017-08-03 Tom de Vries <tom@codesourcery.com>
15522 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15523 ACCEL_COMPILER, apply finish_options on
15524 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15526 2017-08-03 Tom de Vries <tom@codesourcery.com>
15529 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15530 function_entry_patch_area_size > 0.
15532 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15535 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15536 instead of 10??LU, perform unit multiplication in wide_int,
15537 don't change alloc_object_size_limit if the limit is larger
15540 PR tree-optimization/81655
15541 PR tree-optimization/81588
15542 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15543 the case when ranges[i].low and high are 1 for unsigned type with
15546 PR middle-end/81052
15547 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15548 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15550 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15552 * tree-vrp.h: Add include guard.
15554 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15557 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15558 (ud2): New insn pattern.
15559 * config/i386/i386.c (ix86_expand_epilogue):
15560 For naked functions, generate ud2 instead of trap insn.
15562 2017-08-02 Marek Polacek <polacek@redhat.com>
15565 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15567 2017-08-02 Tom de Vries <tom@codesourcery.com>
15568 Cesar Philippidis <cesar@codesourcery.com>
15570 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15571 Add missing edge probabilities.
15573 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15575 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15576 Correct endianness.
15578 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15580 PR middle-end/79499
15581 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15582 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15583 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15584 prologue_seq sequences - if any.
15586 2017-08-02 Richard Biener <rguenther@suse.de>
15588 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15589 via vectors if supported, integer extracts via punning if supported
15590 or otherwise vector extracts.
15592 2017-08-02 Richard Biener <rguenther@suse.de>
15594 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15596 (bitmap_insert_into_set): ... this.
15598 2017-08-02 Richard Biener <rguenther@suse.de>
15600 PR tree-optimization/81633
15602 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15604 PR tree-optimization/71752
15605 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15607 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15609 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15610 (machine_function::call_ms2sysv_pad_out): Remove field.
15611 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15612 (ix86_compute_frame_layout): Likewise.
15614 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15617 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15618 attribute with interrupt attribute.
15620 2017-08-01 Andrew Pinski <apinski@cavium.com>
15622 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15623 BIT_INSERT_EXPR's operand 1
15624 to see if the types precision matches.
15626 2017-08-01 Martin Liska <mliska@suse.cz>
15628 PR middle-end/70140
15629 * builtins.c (expand_builtin_memcpy_args): Remove.
15630 (expand_builtin_memcpy): Call newly added function
15631 expand_builtin_memory_copy_args.
15632 (expand_builtin_memcpy_with_bounds): Likewise.
15633 (expand_builtin_mempcpy): Remove last argument.
15634 (expand_builtin_mempcpy_with_bounds): Likewise.
15635 (expand_builtin_memory_copy_args): New function created from
15636 expand_builtin_mempcpy_args with small modifications.
15637 (expand_builtin_mempcpy_args): Remove.
15638 (expand_builtin_stpcpy): Remove unused argument.
15639 (expand_builtin): Likewise.
15640 (expand_builtin_with_bounds): Likewise.
15642 2017-08-01 Martin Liska <mliska@suse.cz>
15645 Make mempcpy more optimal (PR middle-end/70140).
15647 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15650 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15651 __builtin_vec_cmpne verify both arguments are compatible vectors
15652 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15653 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15654 move computation of aligned to after checking the argument types.
15658 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15659 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15662 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15663 Doug Rupp <rupp@adacore.com>
15664 Olivier Hainque <hainque@adacore.com>
15666 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15667 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15669 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15670 for TARGET_OS_CPP_BUILTIN.
15671 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15672 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15674 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15675 passing required abi options to the assembler for EABI configurations.
15676 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15677 of .text.hot and .text.unlikely sections for kernel modules when
15678 using ARM style exceptions.
15679 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15680 options. Add EXTRA_CC1_SPEC.
15681 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15683 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15685 (ARM_TARGET2_DWARF_FORMAT): Define.
15686 * config/arm/t-vxworks: Adjust multilib control to removal of the
15687 Diab command line options.
15689 2017-08-01 Martin Liska <mliska@suse.cz>
15691 PR gcov-profile/81561
15692 * gcov.c (unblock): Make unblocking safe as we need to preserve
15693 index correspondence of blocks and block_lists.
15695 2017-08-01 Richard Biener <rguenther@suse.de>
15697 PR tree-optimization/81181
15698 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15699 (compute_antic): ... end of iteration here.
15701 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15703 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15704 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15705 (ftree-slp-vectorize): Likewise.
15706 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15707 can no longer be set independent of flag_tree_loop_vectorize.
15708 * omp-general.c (emp_max_vf): Likewise.
15709 * opts.c (enable_fdo_optimizations): Remove references to
15710 flag_tree_vectorize, these are now implicit.
15711 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15712 and leave it for the options machinery.
15714 2017-08-01 Martin Liska <mliska@suse.cz>
15716 PR middle-end/70140
15717 * builtins.c (expand_builtin_memcpy_args): Remove.
15718 (expand_builtin_memcpy): Call newly added function
15719 expand_builtin_memory_copy_args.
15720 (expand_builtin_memcpy_with_bounds): Likewise.
15721 (expand_builtin_mempcpy): Remove last argument.
15722 (expand_builtin_mempcpy_with_bounds): Likewise.
15723 (expand_builtin_memory_copy_args): New function created from
15724 expand_builtin_mempcpy_args with small modifications.
15725 (expand_builtin_mempcpy_args): Remove.
15726 (expand_builtin_stpcpy): Remove unused argument.
15727 (expand_builtin): Likewise.
15728 (expand_builtin_with_bounds): Likewise.
15730 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15733 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15734 print "ds:" only for immediates in generic address space.
15736 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15739 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15740 (ix86_function_ok_for_sibcall): Return false for naked functions.
15742 2017-08-01 Richard Biener <rguenther@suse.de>
15744 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15745 (compute_antic): Seed worklist with exit block predecessors.
15746 * cfganal.c (dfs_find_deadend): For a cycle return the source
15747 of the edge closing it.
15749 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15751 * config/aarch64/aarch64.c
15752 (aarch64_can_const_movi_rtx_p): Move 0 check.
15754 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15756 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15757 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15759 * match.pd: Ditto in address comparison pattern.
15761 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15763 PR tree-optimization/81627
15764 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15765 closed ssa form for store-store chain.
15767 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15769 PR tree-optimization/81620
15770 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15771 for store-store chain.
15773 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15775 PR tree-optimization/81588
15776 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15777 ranges[i].in_p, invert comparison code ccode. For >/>=,
15778 swap rhs1 and rhs2 and comparison code unconditionally,
15779 for </<= don't do that. Don't swap rhs1/rhs2 again if
15780 ranges[i].in_p, instead invert comparison code ccode if
15781 opcode or oe->rank is BIT_IOR_EXPR.
15784 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15785 a direct optab to conversion optab.
15786 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15787 with GET_MODE_INNER as last argument instead of optab_handler.
15788 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15789 vector extraction if possible and optab is available.
15790 * expr.c (store_constructor): Use convert_optab_handler instead
15791 of optab_handler. Use vector initialization from smaller
15792 vectors if possible and optab is available.
15793 * tree-vect-stmts.c (vectorizable_load): Likewise.
15794 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15795 now have two modes.
15796 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15797 of vec_init from half-sized vectors with the same element mode.
15798 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15799 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15800 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15801 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15802 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15803 after mode in gen_vec_extract* calls.
15804 (vec_extract<mode>): Renamed to ...
15805 (vec_extract<mode><ssescalarmodelower>): ... this.
15806 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15807 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15808 element mode after mode in gen_vec_init* calls.
15809 (VEC_INIT_HALF_MODE): New mode iterator.
15810 (vec_init<mode>): Renamed to ...
15811 (vec_init<mode><ssescalarmodelower>): ... this.
15812 (vec_init<mode><ssehalfvecmodelower>): New expander.
15813 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15814 (vec_extractv2sfsf): ... this.
15815 (vec_initv2sf): Renamed to ...
15816 (vec_initv2sfsf): ... this.
15817 (vec_extractv2si): Renamed to ...
15818 (vec_extractv2sisi): ... this.
15819 (vec_initv2si): Renamed to ...
15820 (vec_initv2sisi): ... this.
15821 (vec_extractv4hi): Renamed to ...
15822 (vec_extractv4hihi): ... this.
15823 (vec_initv4hi): Renamed to ...
15824 (vec_initv4hihi): ... this.
15825 (vec_extractv8qi): Renamed to ...
15826 (vec_extractv8qiqi): ... this.
15827 (vec_initv8qi): Renamed to ...
15828 (vec_initv8qiqi): ... this.
15829 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15830 (vec_init<mode>): Renamed to ...
15831 (vec_init<mode><VEC_base_l>): ... this.
15832 (vec_extract<mode>): Renamed to ...
15833 (vec_extract<mode><VEC_base_l>): ... this.
15834 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15835 (vec_initv2sfsf): ... this.
15836 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15837 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15838 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15839 element mode after mode in gen_vec_init* calls.
15840 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15841 (vec_init<mode><Vel>): ... this.
15842 (vec_extract<mode>): Renamed to ...
15843 (vec_extract<mode><Vel>): ... this.
15844 * config/aarch64/iterators.md (Vel): New mode attribute.
15845 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15846 Add element mode after mode in gen_vec_extract* calls.
15847 * config/s390/vector.md (non_vec_l): New mode attribute.
15848 (vec_extract<mode>): Renamed to ...
15849 (vec_extract<mode><non_vec_l>): ... this.
15850 (vec_init<mode>): Renamed to ...
15851 (vec_init<mode><non_vec_l>): ... this.
15852 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15853 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15855 * config/arm/iterators.md (V_elem_l): New mode attribute.
15856 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15857 (vec_extract<mode><V_elem_l>): ... this.
15858 (vec_extractv2di): Renamed to ...
15859 (vec_extractv2didi): ... this.
15860 (vec_init<mode>): Renamed to ...
15861 (vec_init<mode><V_elem_l>): ... this.
15862 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15863 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15864 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15865 Add element mode after gen_vec_extract* calls.
15866 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15867 (vec_init<mode><unitmode>): ... this.
15868 (vec_extract<mode>): Renamed to ...
15869 (vec_extract<mode><unitmode>): ... this.
15870 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15871 (vec_init<mode><unitmode>): ... this.
15872 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15873 (vec_initv2sfsf): ... this.
15874 (vec_extractv2sf): Renamed to ...
15875 (vec_extractv2sfsf): ... this.
15876 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15877 Add element mode after gen_vec_extract* calls.
15878 * config/mips/mips.md (unitmode): New mode iterator.
15879 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15880 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15881 * config/spu/spu.md (inner_l): New mode attribute.
15882 (vec_init<mode>): Renamed to ...
15883 (vec_init<mode><inner_l>): ... this.
15884 (vec_extract<mode>): Renamed to ...
15885 (vec_extract<mode><inner_l>): ... this.
15886 * config/sparc/sparc.md (veltmode): New mode iterator.
15887 (vec_init<VMALL:mode>): Renamed to ...
15888 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15889 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15890 (vec_initv2sisi): ... this.
15891 (vec_initv2sf): Renamed to ...
15892 (vec_initv2sfsf): ... this.
15893 (vec_extractv2sf): Renamed to ...
15894 (vec_extractv2sfsf): ... this.
15895 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15896 (vec_init<mode>): Renamed to ...
15897 (vec_init<mode><VEC_base_l>): ... this.
15898 (vec_extract<mode>): Renamed to ...
15899 (vec_extract<mode><VEC_base_l>): ... this.
15900 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15901 (vec_initv2sfsf): ... this.
15902 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15903 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15904 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15905 gen_vec_init* calls.
15907 2017-08-01 Richard Biener <rguenther@suse.de>
15909 PR tree-optimization/81297
15910 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15911 TREE_OVERFLOW from INTEGER_CSTs.
15913 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15915 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15917 2017-07-31 Carl Love <cel@us.ibm.com>
15919 * config/rs6000/rs6000-c: Add support for built-in functions
15920 vector signed char vec_xl_be (signed long long, signed char *);
15921 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15922 vector signed int vec_xl_be (signed long long, signed int *);
15923 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15924 vector signed long long vec_xl_be (signed long long, signed long long *);
15925 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15926 vector signed short vec_xl_be (signed long long, signed short *);
15927 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15928 vector double vec_xl_be (signed long long, double *);
15929 vector float vec_xl_be (signed long long, float *);
15930 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15931 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15932 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15934 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15935 (altivec_expand_builtin): Add switch statement to call
15936 altivec_expand_xl_be for each builtin.
15937 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15938 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15939 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15940 __builtin_vsx_le_be_v16qi.
15941 * doc/extend.texi: Update the built-in documentation file for the
15942 new built-in functions.
15944 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15947 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15949 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15951 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15953 * config.gcc: Add z14.
15954 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15955 CPU model numbers for z13s and z14.
15956 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15958 * config/s390/s390-opts.h (enum processor_type): Rename
15959 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15960 * config/s390/s390.c (processor_table): Add field for CPU name to
15961 be passed to Binutils.
15962 (s390_asm_output_machine_for_arch): Use the new field in
15963 processor_table for Binutils.
15964 (s390_expand_builtin): Replace arch12 with z14.
15965 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15966 (s390_get_sched_attrmask): Likewise.
15967 (s390_get_unit_mask): Likewise.
15968 * config/s390/s390.opt: Add z14 to processor_type enum.
15970 2017-07-31 Martin Jambor <mjambor@suse.cz>
15973 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15974 regardless of optimization level.
15976 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15977 Martin Liska <mliska@suse.cz>
15979 * predict.def: Remove old comment and adjust probability.
15980 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15981 PREDICT statements.
15983 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15986 * config/i386/i386.c (ix86_function_naked): New function.
15987 (ix86_can_use_return_insn_p): Return false for naked functions.
15988 (ix86_expand_prologue): Skip prologue for naked functions.
15989 (ix86_expand_epilogue): Skip epilogue for naked functions
15990 and emit trap instruction.
15991 (ix86_warn_func_return): New function.
15992 (ix86_attribute_table): Add "naked" attribute specification.
15993 (TARGET_WARN_FUNC_RETURN): Define.
15994 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15996 2017-07-31 Martin Liska <mliska@suse.cz>
15998 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15999 (dump_gimple_bb_header): Always dump BB info.
16000 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16002 2017-07-31 Martin Liska <mliska@suse.cz>
16005 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16006 also with current_function_decl non-null equality.
16008 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16011 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16012 change type to the element type, instead add eltype variable and
16013 use it where we are interested in the element type.
16015 PR tree-optimization/81603
16016 * ipa-polymorphic-call.c
16017 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16018 offset arithmetic in offset_int, bail out if the resulting bit offset
16019 doesn't fit into shwi.
16021 2017-07-31 Martin Liska <mliska@suse.cz>
16023 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16024 (gimplify_save_expr): Fix comment.
16026 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16029 * config/i386/i386.c (ix86_function_arg): Update arguments for
16031 (ix86_compute_frame_layout): Set the initial stack offset to
16032 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16033 INCOMING_FRAME_SP_OFFSET.
16034 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16035 stack before exception handler returns.
16036 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16037 the 'ERROR_CODE' for exception handler.
16039 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16041 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16042 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16043 (ASM_OUTPUT_REG_POP): Ditto.
16044 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16045 instead of asm_fprintf to output pure string.
16047 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16049 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16050 to imported_module_or_decl hook.
16051 (debug_nothing_tree_tree_tree_bool): Remove.
16052 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16053 * debug.c (do_nothing_debug_hooks): Use
16054 debug_nothing_tree_tree_tree_bool_bool instead of
16055 debug_nothing_tree_tree_tree_bool.
16056 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16057 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16058 * sdbout.c (sdb_debug_hooks): Likewise.
16059 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16060 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16062 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16063 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16064 attribute, don't add anything.
16066 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16068 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16069 (fold_build2_stat_loc): Likewise.
16070 (fold_build3_stat_loc): Likewise.
16071 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16072 (fold_build1_loc): Remove macro.
16073 (fold_build2_loc): Likewise.
16074 (fold_build3_loc): Likewise.
16076 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16078 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16079 (gimple_build_debug_bind_source_stat): Likewise.
16080 * gimple.h (gimple_build_debug_bind): Remove macro.
16081 (gimple_build_debug_bind_source): Likewise.
16083 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16085 * bitmap.c (bitmap_alloc): Adjust.
16086 (bitmap_gc_alloc): Likewise.
16087 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16089 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16091 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16092 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16093 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16094 (bitmap_gc_alloc_stat): Likewise.
16095 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16097 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16099 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16100 * rtl.h (shallow_copy_rtx): Remove macro.
16102 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16104 * emit-rtl.c (gen_raw_REG): Adjust.
16105 * gengenrtl.c (gendef): Likewise.
16106 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16107 * rtl.h (rtx_alloc): Remove macro.
16109 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16111 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16112 (build_tree_list_stat): Likewise.
16113 * tree.h (build_tree_list): Remove macro.
16114 (build_tree_list_vec): Likewise.
16116 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16118 * tree.c (make_vector_stat): Remove _stat from name.
16119 (build_vector_stat): Likewise.
16120 * tree.h (make_vector_stat): Remove macro.
16121 (build_vector_stat): Likewise.
16123 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16125 * tree.h (build_var_debug_value): Remove prototype.
16127 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16129 * tree.c (tree_cons_stat): Remove _stat from name.
16130 * tree.h (tree_cons): Remove macro.
16132 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16134 * tree.c (build_vl_exp_stat): Remove _stat from name.
16135 * tree.h (build_vl_exp): Remove macro.
16137 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16139 * tree.c (build_decl_stat): Remove _stat from name.
16140 * tree.h (build_decl): Remove macro.
16142 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16144 * gimple.c (gimple_build_with_ops_stat): Adjust.
16145 (gimple_alloc_stat): Remove _stat from name.
16146 * gimple.h (gimple_alloc): Remove macro.
16148 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16150 * tree.c (make_tree_vec_stat): Remove _stat from name.
16151 (grow_tree_vec_stat): Likewise.
16152 * tree.h (make_tree_vec_stat): Adjust prototype.
16153 (grow_tree_vec_stat): Likewise.
16154 (make_tree_vec): Remove macro.
16155 (grow_tree_vec): Likewise.
16157 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16159 * fold-const.c (fold_build1_stat_loc): Adjust.
16160 (fold_build2_stat_loc): Likewise.
16161 (fold_build3_stat_loc): Likewise.
16162 * tree.c (build0_stat): Remove _stat from name.
16163 (build1_stat): Likewise.
16164 (build2_stat): Likewise.
16165 (build3_stat): Likewise.
16166 (build4_stat): Likewise.
16167 (build5_stat): Likewise.
16168 * tree.h (build1_loc): Remove macro, and rename _stat function
16170 (build2_loc): Likewise.
16171 (build3_loc): Likewise.
16172 (build4_loc): Likewise.
16173 (build5_loc): Likewise.
16175 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16177 * tree.c (make_int_cst_stat): Remove _stat from name.
16178 * tree.h (make_int_cst_stat): Adjust prototype.
16179 (make_int_cst): Remove macro.
16181 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16183 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16184 * tree.h (make_tree_binfo_stat): Adjust prototype.
16185 (make_tree_binfo): Remove.
16187 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16189 * tree.c (copy_node_stat): Rename to copy_node.
16190 (build_distinct_type_copy): Adjust.
16191 * tree.h (copy_node_stat): Adjust prototype.
16192 (copy_node): Remove macro.
16194 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16196 * tree.c (make_node_stat): rename to make_node.
16197 (build_tree_list_stat): Adjust.
16198 (build0_stat): Likewise.
16199 (build2_stat): Likewise.
16200 (build3_stat): Likewise.
16201 (build4_stat): Likewise.
16202 (build5_stat): Likewise.
16203 (build_decl_stat): Likewise.
16204 * tree.h (make_node_stat): Adjust prototype.
16205 (make_node): remove macro.
16207 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16209 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16210 (PPC_FEATURE2_SCV): Likewise.
16211 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16213 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16215 * config/aarch64/aarch64.c
16216 (aarch64_internal_mov_immediate): Add new special pattern.
16217 * config/aarch64/aarch64.md (*movdi_aarch64):
16218 Add reg/32bit const mov case.
16220 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16221 Richard Sandiford <richard.sandiford@linaro.org>
16223 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16224 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16225 Add integer and movi cases.
16226 (movi-split-hf-df-sf split, fp16): New.
16227 (enabled): Added TARGET_FP_F16INST.
16228 * config/aarch64/iterators.md (GPF_HF): New.
16229 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16231 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16233 * config/aarch64/aarch64.c
16234 (aarch64_simd_container_mode): Add prototype.
16235 (aarch64_expand_mov_immediate): Add HI support.
16236 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16237 (aarch64_can_const_movi_rtx_p): New.
16238 (aarch64_preferred_reload_class):
16239 Remove restrictions of using FP registers for certain SIMD operations.
16240 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16241 (aarch64_valid_floating_const): Add integer move validation.
16242 (aarch64_simd_imm_scalar_p): Remove.
16243 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16244 (aarch64_legitimate_constant_p): Expand list of supported cases.
16245 * config/aarch64/aarch64-protos.h
16246 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16247 (aarch64_reinterpret_float_as_int): New.
16248 (aarch64_simd_imm_scalar_p): Remove.
16249 * config/aarch64/constraints.md (Uvi): New.
16250 (Dd): Split into Ds and new Dd.
16251 * config/aarch64/aarch64.md (*movsi_aarch64):
16253 (*movdi_aarch64): Add SIMD mov case.
16255 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16257 * tree-predcom.c: (struct chain): Handle store-store chain in which
16258 stores for elimination only store loop invariant values.
16259 (execute_pred_commoning_chain): Ditto.
16260 (prepare_initializers_chain_store_elim): Ditto.
16261 (prepare_finalizers): Ditto.
16262 (is_inv_store_elimination_chain): New function.
16263 (initialize_root_vars_store_elim_1): New function.
16265 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16267 * tree-predcom.c: Revise general description of the pass.
16268 (enum chain_type): New enum type for store elimination.
16269 (struct chain): New field supporting store elimination.
16270 (struct component): Ditto.
16271 (dump_chain): Dump store-stores chain.
16272 (release_chain): Release resources.
16273 (split_data_refs_to_components): Compute and create component
16274 contains only stores for elimination.
16275 (get_chain_last_ref_at): New function.
16276 (make_invariant_chain): Initialization.
16277 (make_rooted_chain): Specify chain type in parameter and record it.
16278 (add_looparound_copies): Skip for store-stores chain.
16279 (determine_roots_comp): Compute type of chain and pass it to
16281 (initialize_root_vars_store_elim_2): New function.
16282 (finalize_eliminated_stores): New function.
16283 (remove_stmt): Handle store for elimination.
16284 (execute_pred_commoning_chain): Execute predictive commoning on
16285 store-store chains.
16286 (determine_unroll_factor): Skip unroll for store-stores chain.
16287 (prepare_initializers_chain_store_elim): New function.
16288 (prepare_initializers_chain): Hanlde store-store chain.
16289 (prepare_finalizers_chain, prepare_finalizers): New function.
16290 (tree_predictive_commoning_loop): Return integer value indicating
16291 if loop is unrolled or lcssa form is corrupted.
16292 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16294 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16296 * tree-predcom.c (initialize_root): Delete.
16297 (execute_pred_commoning_chain): Initialize root vars and replace
16298 reference of non-combined chain directly, rather than call above
16301 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16303 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16304 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16306 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16308 * tree-predcom.c (struct chain): New field init_seq.
16309 (release_chain): Release init_seq.
16310 (prepare_initializers_chain): Record intialization stmts in above
16312 (insert_init_seqs): New function.
16313 (tree_predictive_commoning_loop): Call insert_init_seqs.
16315 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16317 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16319 2017-07-28 Richard Biener <rguenther@suse.de>
16321 * match.pd: Remove superfluous :c.
16322 * genmatch.c (simplify::id): Add member.
16323 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16325 (current_id): New global.
16326 (dt_node::parent): Move from ...
16327 (dt_operand::parent): ... here. Add for_id member.
16328 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16329 (decision_tree::find_node): Relax order requirement when
16330 merging DT_TRUE nodes to ones inbetween the current simplify
16331 and the one we try to merge with. Add diagnostic whenever
16332 we need to enforce pattern order by not merging.
16333 (decision_tree::insert): Set current_id.
16334 (decision_tree::print_node): Dump parent node and for_id.
16335 (parser::last_id): Add member.
16336 (parser::push_simplify): Assign unique id.
16337 (parser::parser): Initialize last_id.
16339 2017-07-28 Martin Liska <mliska@suse.cz>
16342 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16343 gimple_build_debug_bind.
16345 2017-07-28 Richard Biener <rguenther@suse.de>
16347 PR tree-optimization/81502
16348 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16350 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16352 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16353 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16354 for BIT_FIELD_REF args.
16355 * fold-const.c (make_bit_field_ref): Likewise.
16356 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16358 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16361 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16362 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16363 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16364 Or it into SANITIZER_UNDEFINED.
16365 * ubsan.c: Include gimple-fold.h and varasm.h.
16366 (ubsan_expand_ptr_ifn): New function.
16367 (instrument_pointer_overflow): New function.
16368 (maybe_instrument_pointer_overflow): New function.
16369 (instrument_object_size): Formatting fix.
16370 (pass_ubsan::execute): Call instrument_pointer_overflow
16371 and maybe_instrument_pointer_overflow.
16372 * internal-fn.c (expand_UBSAN_PTR): New function.
16373 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16374 * sanitizer.def (__ubsan_handle_pointer_overflow,
16375 __ubsan_handle_pointer_overflow_abort): New builtins.
16376 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16377 * internal-fn.def (UBSAN_PTR): New internal function.
16378 * opts.c (sanitizer_opts): Add pointer-overflow.
16379 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16380 * fold-const.c (build_range_check): Compute pointer range check in
16381 integral type if pointer arithmetics would be needed. Formatting
16384 2017-07-28 Martin Liska <mliska@suse.cz>
16387 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16388 parameters that are of a variable-length.
16390 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16392 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16394 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16395 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16396 (CRT_CALL_STATIC_FUNCTION): Likewise.
16397 (ASM_DEFAULT_SPEC): New define.
16398 (ASM_SPEC32): Likewise.
16399 (ASM_SPEC64): Likewise.
16400 (ASM_SPEC_COMMON): Likewise.
16401 (ASM_SPEC): Likewise.
16402 (INVALID_64BIT): Likewise.
16403 (LINK_OS_DEFAULT_SPEC): Likewise.
16404 (LINK_OS_SPEC32): Likewise.
16405 (LINK_OS_SPEC64): Likewise.
16406 (POWERPC_LINUX): Likewise.
16407 (PTRDIFF_TYPE): Likewise.
16408 (RESTORE_FP_PREFIX): Likewise.
16409 (RESTORE_FP_SUFFIX): Likewise.
16410 (SAVE_FP_PREFIX): Likewise.
16411 (SAVE_FP_SUFFIX): Likewise.
16412 (SIZE_TYPE): Likewise.
16413 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16414 (TARGET_64BIT): Likewise.
16415 (TARGET_64BIT): Likewise.
16416 (TARGET_AIX): Likewise.
16417 (WCHAR_TYPE_SIZE): Likewise.
16418 (WCHAR_TYPE): Undefine.
16419 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16420 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16421 (CPP_OS_RTEMS_SPEC): Delete.
16422 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16423 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16425 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16427 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16429 PR tree-optimization/81578
16430 * tree-parloops.c (build_new_reduction): Bail out if
16431 reduction_code isn't one of the standard OpenMP reductions.
16432 Move the details printing after that decision.
16434 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16436 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16437 related to reload_in_progress.
16438 (splat_input_operand): Likewise.
16439 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16441 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16443 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16444 (TARGET_INSTANTIATE_DECLS): Likewise.
16445 (legitimate_indexed_address_p): Delete reload_in_progress code.
16446 (rs6000_debug_legitimate_address_p): Likewise.
16447 (rs6000_eliminate_indexed_memrefs): Likewise.
16448 (rs6000_emit_le_vsx_store): Likewise.
16449 (rs6000_emit_move_si_sf_subreg): Likewise.
16450 (rs6000_emit_move): Likewise.
16451 (register_to_reg_type): Likewise.
16452 (rs6000_pre_atomic_barrier): Likewise.
16453 (rs6000_machopic_legitimize_pic_address): Likewise.
16454 (rs6000_allocate_stack_temp): Likewise.
16455 (rs6000_address_for_fpconvert): Likewise.
16456 (rs6000_address_for_altivec): Likewise.
16457 (rs6000_secondary_memory_needed_rtx): Delete function.
16458 (rs6000_check_sdmode): Likewise.
16459 (rs6000_alloc_sdmode_stack_slot): Likewise.
16460 (rs6000_instantiate_decls): Likewise.
16461 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16462 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16463 Delete reload_in_progress.
16464 (*vec_reload_and_plus_<mptrsize>): Likewise.
16465 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16466 (vsx_div_v2di): Likewise.
16467 (vsx_udiv_v2di): Likewise.
16469 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16471 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16472 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16473 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16474 (rs6000_debug_reg_global): Delete print of LRA status.
16475 (rs6000_option_override_internal): Delete dead LRA related code.
16476 (rs6000_lra_p): Delete function.
16477 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16479 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16481 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16482 * config/riscv/rtems.h: New file.
16484 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16485 Sudakshina Das <sudi.das@arm.com>
16487 * config/aarch64/aarch64.md
16488 (define_split for and<mode>3nr_compare): Move
16489 non aarch64_logical_operand to a register.
16490 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16491 register immediate operand to a register.
16492 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16494 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16496 PR middle-end/81564
16497 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16499 2017-07-27 Richard Biener <rguenther@suse.de>
16501 PR tree-optimization/81573
16502 PR tree-optimization/81494
16503 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16504 multi defuse cycle case.
16506 2017-07-27 Richard Biener <rguenther@suse.de>
16508 PR tree-optimization/81571
16509 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16512 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16514 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16515 earlier and only if MASK_FPU is set. Adjust formatting.
16517 2017-07-27 Martin Liska <mliska@suse.cz>
16519 * opt-functions.awk: Add validation of value of Init.
16520 * optc-gen.awk: Pass new argument.
16522 2017-07-27 Martin Liska <mliska@suse.cz>
16524 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16525 Fix wrong condition.
16527 2017-07-27 Martin Liska <mliska@suse.cz>
16529 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16530 BBs and edges seen by autoFDO.
16532 2017-07-27 Richard Biener <rguenther@suse.de>
16534 PR tree-optimization/81502
16535 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16536 with incompatible but same sized type.
16537 (execute_update_addresses_taken): Likewise.
16539 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16541 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16542 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16544 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16547 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16548 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16549 Change s_operand to memory_operand.
16551 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16553 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16554 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16555 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16556 Emit instructions rather than returning an expression. Handle TFmode
16557 and KFmode by casting to TImode.
16558 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16559 (rs6000_emit_le_vsx_store): Likewise.
16560 * config/rs6000/vsx.md (VSX_TI): New iterator.
16561 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16562 (*vsx_le_undo_permute_<mode>): Likewise.
16563 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16564 emit the split sequence.
16565 (*vsx_le_perm_store_<mode>): Likewise.
16567 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16569 PR tree-optimization/81555
16570 PR tree-optimization/81556
16571 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16572 if true, force CHANGED for the recursive invocation.
16573 (reassociate_bb): Remember original length of ops array, pass
16574 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16576 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16577 attributes for noipa attribute. For naked attribute use
16578 lookup_attribute first before lookup_attribute_spec.
16579 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16581 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16582 for functions with noipa attribute.
16583 (cgraph_externally_visible_p): Return true for functions with noipa
16585 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16586 for functions with noipa attribute.
16587 * doc/extend.texi: Document noipa function attribute.
16588 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16589 also for functions with noipa attribute.
16590 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16592 2017-07-26 Andrew Pinski <apinski@cavium.com>
16594 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16595 vec_unalign_load_cost and vec_unalign_store_cost.
16597 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16599 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16600 -mvsx-small-integer option.
16601 (ISA_3_0_MASKS_IEEE): Likewise.
16602 (OTHER_VSX_VECTOR_MASKS): Likewise.
16603 (POWERPC_MASKS): Likewise.
16604 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16605 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16606 code, only testing for DImode being allowed in non-VSX floating
16608 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16609 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16611 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16612 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16613 TARGET_P8_VECTOR test.
16614 (rs6000_secondary_reload_simple_move): Likewise.
16615 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16616 since TARGET_P9_VECTOR was already tested.
16617 (rs6000_opt_masks): Remove -mvsx-small-integer.
16618 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16619 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16621 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16622 test for TARGET_VEXTRACTUB was used, and that uses
16624 (p9 extract splitter): Likewise.
16625 (vsx_extract_<mode>_di_p9): Likewise.
16626 (vsx_extract_<mode>_store_p9): Likewise.
16627 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16628 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16629 the elimination of TARGET_VSX_SMALL_INTEGER.
16630 (vsx_extract_<mode>_p8): Likewise.
16631 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16632 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16633 (vsx_set_<mode>_p9): Likewise.
16634 (vsx_set_v4sf_p9): Likewise.
16635 (vsx_set_v4sf_p9_zero): Likewise.
16636 (vsx_insert_extract_v4sf_p9): Likewise.
16637 (vsx_insert_extract_v4sf_p9_2): Likewise.
16638 * config/rs6000/rs6000.md (sign extend splitter): Change
16639 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16640 (floatsi<mode>2_lfiwax_mem): Likewise.
16641 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16642 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16643 since a test for TARGET_P9_VECTOR was used.
16644 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16645 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16646 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16647 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16648 TARGET_P8_VECTOR test.
16649 (fix_trunc<mode>si2_stfiwx): Likewise.
16650 (fix_trunc<mode>si2_internal): Likewise.
16651 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16652 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16654 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16655 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16656 TARGET_P8_VECTOR test.
16657 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16658 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16659 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16661 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16662 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16663 since a test for TARGET_P9_VECTOR was used.
16664 (splitter for loading small constants): Likewise.
16666 2017-07-26 Andrew Pinski <apinski@cavium.com>
16668 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16671 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16674 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16675 (fp_valid_at): Likewise.
16677 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16679 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16680 (qdf24xx_addrcost_table): Likewise.
16681 (cortexa57_tunings): Update to use generic_branch_cost.
16682 (cortexa72_tunings): Likewise.
16683 (cortexa73_tunings): Likewise.
16684 (qdf24xx_tunings): Likewise.
16686 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16688 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16689 (thunderx2t99_branch_cost): Likewise.
16690 (cortexa35_tunings): Update to use generic_branch_cost.
16691 (cortexa53_tunings): Likewise.
16692 (cortexa57_tunings): Likewise.
16693 (cortexa72_tunings): Likewise.
16694 (cortexa73_tunings): Likewise.
16695 (thunderx2t99_tunings): Likewise.
16697 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16699 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16700 (sparc_option_override): Honour MASK_FSMULD.
16701 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16702 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16703 * config/sparc/sparc.opt (mfsmuld): New option.
16704 * doc/invoke.texi (mfsmuld): Document option.
16706 2017-07-26 Marek Polacek <polacek@redhat.com>
16708 PR middle-end/70992
16709 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16711 2017-07-26 Richard Biener <rguenther@suse.de>
16713 * gimple-match-head.c (do_valueize): Return OP if valueize
16715 (get_def): New helper to get at the def stmt of a SSA name
16716 if valueize allows.
16717 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16718 do_valueize to get at the def stmt.
16719 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16721 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16723 PR middle-end/46932
16724 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16726 2017-07-26 Martin Liska <mliska@suse.cz>
16729 * function.c (expand_function_start): Make expansion of
16730 nonlocal_goto_save_area after parm_birth_insn.
16732 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16734 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16735 from all CPU target flags enable members.
16737 2017-07-26 Richard Biener <rguenther@suse.de>
16739 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16740 (decision_tree::gen): Make 'type' const.
16741 (write_predicate): Likewise.
16743 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16745 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16746 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16747 (rs6000_option_override_internal): Likewise.
16748 (rs6000_expand_vector_set): Likewise.
16749 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16750 (TARGET_UPPER_REGS_SF): Likewise.
16751 (TARGET_UPPER_REGS_DI): Likewise.
16752 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16753 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16754 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16755 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16756 (Splitters for DI constants in Altivec registers): Likewise.
16757 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16758 (vsx_set_v4sf_p9): Likewise.
16759 (vsx_set_v4sf_p9_zero): Likewise.
16760 (vsx_insert_extract_v4sf_p9): Likewise.
16761 (vsx_insert_extract_v4sf_p9_2): Likewise.
16763 2017-07-25 Carl Love <cel@us.ibm.com>
16765 * doc/extend.texi: Update the built-in documentation file for the
16766 existing built-in functions
16767 vector signed char vec_cnttz (vector signed char);
16768 vector unsigned char vec_cnttz (vector unsigned char);
16769 vector signed short vec_cnttz (vector signed short);
16770 vector unsigned short vec_cnttz (vector unsigned short);
16771 vector signed int vec_cnttz (vector signed int);
16772 vector unsigned int vec_cnttz (vector unsigned int);
16773 vector signed long long vec_cnttz (vector signed long long);
16774 vector unsigned long long vec_cnttz (vector unsigned long long);
16776 2017-07-25 Andrew Pinski <apinski@cavium.com>
16778 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16779 accesses where the use is for the first operand of a BIT_INSERT.
16781 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16784 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16785 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16787 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16789 * config/i386/gstabs.h: Delete.
16790 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16792 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16794 * config/i386/i386.c (ix86_decompose_address): Do not check for
16795 register RTX when looking at index_reg or base_reg.
16796 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16798 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16800 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16801 to update EH info here.
16803 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16805 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16807 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16809 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16811 2017-07-25 Torsten Duwe <duwe@suse.de>
16813 * common.opt: Introduce -fpatchable-function-entry
16814 command line option, and its variables function_entry_patch_area_size
16815 and function_entry_patch_area_start.
16816 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16817 including a two-value parser.
16818 * target.def (print_patchable_function_entry): New target hook.
16819 * targhooks.h (default_print_patchable_function_entry): New function.
16820 * targhooks.c (default_print_patchable_function_entry): Likewise.
16821 * toplev.c (process_options): Switch off IPA-RA if
16822 patchable function entries are being generated.
16823 * varasm.c (assemble_start_function): Look at the
16824 patchable-function-entry command line switch and current
16825 function attributes and maybe generate NOP instructions by
16826 calling the print_patchable_function_entry hook.
16827 * doc/extend.texi: Document patchable_function_entry attribute.
16828 * doc/invoke.texi: Document -fpatchable_function_entry
16829 command line option.
16830 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16832 * doc/tm.texi: Re-generate.
16834 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16837 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16838 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16840 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16842 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16844 2017-07-25 Richard Biener <rguenther@suse.de>
16846 PR tree-optimization/81455
16847 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16848 not walk in cycles when looking for guards.
16850 2017-07-25 Richard Biener <rguenther@suse.de>
16852 PR tree-optimization/81529
16853 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16854 when optimizing backedge uses.
16856 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16858 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16860 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16861 to dl_section_ref. On AIX, append an expression to subtract
16862 the size of the section length to dl_section_ref.
16864 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16866 * configure.ac: If any of the config.* scripts fail, exit 1.
16867 * configure: Regenerate.
16869 2017-07-25 Richard Biener <rguenther@suse.de>
16871 PR middle-end/81546
16872 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16873 of immediate uses, be more verbose on errors.
16875 2017-07-25 Richard Biener <rguenther@suse.de>
16877 PR tree-optimization/81510
16878 * tree-vect-loop.c (vect_is_simple_reduction): When the
16879 reduction stmt is not inside the loop bail out.
16881 2017-07-25 Richard Biener <rguenther@suse.de>
16883 PR tree-optimization/81303
16884 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16885 profitability check against LOOP_VINFO_NITERSM1.
16887 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16889 * domwalk.c (cmp_bb_postorder): Simplify.
16890 (sort_bbs_postorder): New function. Use it...
16891 (dom_walker::walk): ...here to optimize common cases.
16893 2017-07-25 Martin Liska <mliska@suse.cz>
16896 * ipa-visibility.c (function_and_variable_visibility): Make the
16897 redirection just on target that supports aliasing.
16898 Fix GNU coding style.
16900 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16903 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16905 * config/arm/rtems.h: Likewise.
16906 * config/bfin/rtems.h: Likewise.
16907 * config/i386/rtemself.h: Likewise.
16908 * config/lm32/rtems.h: Likewise.
16909 * config/m32c/rtems.h: Likewise.
16910 * config/m68k/rtemself.h: Likewise.
16911 * config/microblaze/rtems.h: Likewise.
16912 * config/mips/rtems.h: Likewise.
16913 * config/moxie/rtems.h: Likewise.
16914 * config/nios2/rtems.h: Likewise.
16915 * config/powerpcspe/rtems.h: Likewise.
16916 * config/rs6000/rtems.h: Likewise.
16917 * config/rtems.h: Likewise.
16918 * config/sh/rtems.h: Likewise.
16919 * config/sh/rtemself.h: Likewise.
16920 * config/sparc/rtemself.h: Likewise.
16922 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16925 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16926 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16927 * tree-ssa-structalias.c (alias_get_name): Same.
16929 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16932 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16933 instructions if no du chain is found.
16935 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16937 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16939 2017-07-25 Richard Biener <rguenther@suse.de>
16941 PR middle-end/81505
16942 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16945 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16947 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16948 upper-regs options.
16949 (ISA_2_7_MASKS_SERVER): Likewise.
16950 (ISA_3_0_MASKS_IEEE): Likewise.
16951 (OTHER_P8_VECTOR_MASKS): Likewise.
16952 (OTHER_VSX_VECTOR_MASKS): Likewise.
16953 (POWERPC_MASKS): Likewise.
16954 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16955 duplicate list of options.
16956 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16957 explicit -mupper-regs options.
16958 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16959 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16960 alias for -mupper-regs-df.
16961 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16962 (rs6000_init_hard_regno_mode_ok): Likewise.
16963 (rs6000_option_override_internal): Likewise.
16964 (rs6000_opt_masks): Likewise.
16965 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16966 options in terms of whether -mvsx or -mpower8-vector was used.
16967 (TARGET_UPPER_REGS_DI): Likewise.
16968 (TARGET_UPPER_REGS_SF): Likewise.
16969 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16970 -mupper-regs-* options.
16972 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16974 * passes.c (emergency_dump_function): Print some empty lines and a
16975 header before the RTL dump.
16977 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16979 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16981 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16984 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16985 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16987 2017-07-24 Carl Love <cel@us.ibm.com>
16989 * config/rs6000/rs6000-c.c: Add support for built-in functions
16990 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16991 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16992 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16993 vec_extract_fp_from_shortl): Add defines for the two builtins.
16994 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16995 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16997 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16998 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16999 * doc/extend.texi: Update the built-in documentation file for the
17000 new built-in function.
17002 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17005 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17007 * doc/generic.texi: Likewise.
17008 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17009 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17011 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17013 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17014 (aarch64_mls_elt_merge<mode>): Likewise.
17016 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17018 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17019 having __cxa_atexit.
17021 2017-07-23 Michael Collison <michael.collison@arm.com>
17023 * config/arm/arm.c (arm_option_override): Deprecate
17024 use of -mstructure-size-boundary.
17025 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17026 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17028 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17031 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17032 Reduce cost estimate for direct moves.
17034 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17037 * config/i386/i386.c (ix86_option_override_internal): Disable
17038 BMI, BMI2 and TBM instructions for -m16.
17040 2017-07-21 Carl Love <cel@us.ibm.com>
17042 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17043 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17044 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17045 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17046 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17047 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17048 VMULOSW): New enum "unspec" values.
17049 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17050 altivec_vmulosw): New patterns.
17051 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17052 VMULOSW): Add definitions.
17054 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17056 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17057 (qdf24xx): Likewise.
17058 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17059 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17060 (AARCH64_FL_V8_1): Renumber.
17061 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17062 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17063 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17064 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17065 rdma to feature modifiers list.
17067 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17069 PR middle-end/56727
17070 * ipa-visibility (function_and_variable_visibility): Convert
17071 recursive PLT call to direct call if appropriate.
17073 2017-07-21 Andrew Pinski <apinski@cavium.com>
17075 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17076 operand 1 to see if the types precision matches.
17077 * fold-const.c (operand_equal_p): Likewise.
17079 2017-07-21 Richard Biener <rguenther@suse.de>
17081 PR tree-optimization/81303
17082 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17083 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17084 (vect_peeling_hash_get_lowest_cost): Adjust.
17085 (vect_enhance_data_refs_alignment): Likewise. Use
17086 vect_get_peeling_costs_all_drs to compute the penalty for no
17087 peeling to match up costs.
17089 2017-07-21 Richard Biener <rguenther@suse.de>
17091 PR tree-optimization/81500
17092 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17093 we didn't identify a reduction path.
17095 2017-07-21 Tom de Vries <tom@codesourcery.com>
17096 Cesar Philippidis <cesar@codesourcery.com>
17098 PR gcov-profile/81442
17099 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17102 2017-07-21 Tom de Vries <tom@codesourcery.com>
17105 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17107 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17108 nvptx_override_options_after_change.
17110 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17112 * dwarf2out.c (output_file_names): Avoid double testing for
17113 dwarf_version >= 5.
17115 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17117 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17119 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17121 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17123 (try_crossjump_to_edge): Do not punt on partitioned functions.
17125 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17127 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17128 Put all BBs reachable only via paths crossing cold region to cold
17130 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17132 2016-07-21 Richard Biener <rguenther@suse.de>
17134 PR tree-optimization/81303
17135 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17136 into account prologue and epilogue iterations when raising
17137 min_profitable_iters to sth at least covering one vector iteration.
17139 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17141 * config/arm/arm.c (arm_test_cpu_arch_dat):
17144 2017-07-20 Nathan Sidwell <nathan@acm.org>
17146 Remove TYPE_METHODS.
17147 * tree.h (TYPE_METHODS): Delete.
17148 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17149 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17150 (dbxout_type_methods): Scan TYPE_FIELDS.
17151 (dbxout_type): Don't check TYPE_METHODS here.
17152 * function.c (use_register_for_decl): Always ignore register for
17153 class types when not optimizing.
17154 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17155 * tree.c (free_lang_data_in_type): Stitch out member functions and
17156 templates from TYPE_FIELDS.
17157 (build_distinct_type_copy, verify_type_variant,
17158 verify_type): Member fns are on TYPE_FIELDS.
17159 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17160 * tree-pretty-print.c (dump_generic_node): Likewise.
17162 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17165 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17166 V2TImode and V4TImode.
17167 (ix86_expand_vector_extract): Likewise.
17168 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17169 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17170 (ssescalarmode): Handle V4TImode and V2TImode.
17171 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17172 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17173 (VEXTRACTI128_MODE): New mode iterator.
17174 (splitter for *vec_extractv?ti first element): New.
17175 (VEC_INIT_MODE): New mode iterator.
17176 (vec_init<mode>): Consolidate 3 expanders into one using
17177 VEC_INIT_MODE mode iterator.
17179 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17181 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17182 non_spilled_static_chain_regno_p.
17184 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17186 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17188 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17190 * bb-reorder.c (connect_traces): Allow copying of blocks within
17193 2017-07-20 Richard Biener <rguenther@suse.de>
17195 * gimple.h (gimple_phi_result): Add gphi * overload.
17196 (gimple_phi_result_ptr): Likewise.
17197 (gimple_phi_arg): Likewise. Adjust index assert to only
17198 allow actual argument accesses rather than all slots available
17200 (gimple_phi_arg_def): Add gphi * overload.
17201 * tree-phinodes.c (make_phi_node): Initialize only actual
17203 (resize_phi_node): Clear memory not covered by old node,
17204 do not initialize excess argument slots.
17205 (reserve_phi_args_for_new_edge): Initialize new argument slot
17208 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17210 PR tree-optimization/81388
17212 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17214 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17215 by removing computation of may_be_zero.
17217 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17218 Tom de Vries <tom@codesourcery.com>
17220 PR middle-end/81030
17221 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17222 when gimple level profile disagrees with what RTL expander did.
17224 2017-07-20 Richard Biener <rguenther@suse.de>
17226 PR tree-optimization/61171
17227 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17228 (vect_analyze_stmt): Add slp instance parameter.
17229 (vectorizable_reduction): Likewise.
17230 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17231 (vect_is_simple_reduction): Deal with chains not detected
17232 as SLP reduction chain, specifically not properly associated
17233 chains containing a mix of plus/minus.
17234 (get_reduction_op): Remove.
17235 (get_initial_defs_for_reduction): Simplify, pass in whether
17236 this is a reduction chain, pass in the SLP node for the PHIs.
17237 (vect_create_epilog_for_reduction): Get the SLP instance as
17239 (vectorizable_reduction): Get the SLP instance as arg.
17240 During analysis remember the SLP node with the PHIs in the
17241 instance. Simplify getting at the vectorized reduction PHIs.
17242 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17243 through SLP instance.
17244 (vect_slp_analyze_operations): Likewise.
17245 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17246 (vect_transform_stmt): Likewise.
17248 2017-07-20 Tom de Vries <tom@codesourcery.com>
17250 PR tree-optimization/81489
17251 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17252 read of phi arg location to before loop that modifies phi.
17254 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17256 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17259 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17261 PR middle-end/81331
17262 * except.c (execute): Fix ordering issue.
17264 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17266 PR rtl-optimization/81423
17267 * combine.c (make_compound_operation_int): Don't try to optimize
17268 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17270 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17272 PR rtl-optimization/81423
17273 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17274 with a constant that is -1 in the truncated to mode.
17276 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17278 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17279 (determine_unlikely_bbs): ... here.
17280 * predict.h (propagate_unlikely_bbs_forward): Declare.
17281 * cfgexpand.c (pass_expand::execute): Use it.
17282 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17284 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17285 propagate_unlikely_bbs_forward.
17287 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17289 PR middle-end/81331
17290 * except.c (maybe_add_nop_after_section_switch): New function.
17293 2017-07-19 Tom de Vries <tom@codesourcery.com>
17295 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17297 2017-07-19 Tom de Vries <tom@codesourcery.com>
17299 * gimple.h (gimple_phi_arg): Make assert more strict.
17301 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17303 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17304 * config/rs6000/mmintrin.h: New file.
17305 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17307 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17309 PR tree-optimization/81346
17310 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17312 2017-07-19 Tom de Vries <tom@codesourcery.com>
17314 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17316 2017-07-19 Tom de Vries <tom@codesourcery.com>
17318 * config/nvptx/nvptx-modes.def: Add V2DImode.
17319 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17320 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17321 (nvptx_output_mov_insn): Handle lack of mov.b128.
17322 (nvptx_print_operand): Handle 'H' and 'L' codes.
17323 (nvptx_vector_mode_supported): Allow V2DImode.
17324 (nvptx_preferred_simd_mode): New function.
17325 (nvptx_data_alignment): New function.
17326 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17327 nvptx_preferred_simd_mode.
17328 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17330 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17332 2017-07-19 Tom de Vries <tom@codesourcery.com>
17334 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17335 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17336 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17337 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17338 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17339 (mov<VECIM>_insn): New define_insn.
17340 (define_expand "mov<VECIM>): New define_expand.
17342 2017-07-19 Tom de Vries <tom@codesourcery.com>
17344 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17346 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17348 PR tree-optimization/81346
17349 * fold-const.h (fold_div_compare, range_check_type): Declare.
17350 * fold-const.c (range_check_type): New function.
17351 (build_range_check): Use range_check_type.
17352 (fold_div_compare): No longer static, rewritten into
17353 a match.pd helper function.
17354 (fold_comparison): Don't call fold_div_compare here.
17355 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17356 as helper function.
17358 2017-07-19 Nathan Sidwell <nathan@acm.org>
17360 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17361 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17362 * tree.c (find_decls_types_r, verify_type): Use
17363 TYPE_{MIN,MAX}_VALUE_RAW.
17364 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17365 (hash_tree): Likewise.
17366 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17368 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17371 2017-07-18 Tom de Vries <tom@codesourcery.com>
17373 PR middle-end/81464
17374 * omp-expand.c (expand_omp_for_static_chunk): Handle
17375 equal-argument loop exit phi.
17377 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17380 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17381 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17382 operand 2 predicate.
17383 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17384 operand 2 predicate.
17385 (ror,rol -> rorx splitters): Use const_int_operand as
17386 operand 2 predicate.
17388 2017-06-18 Richard Biener <rguenther@suse.de>
17390 PR tree-optimization/81410
17391 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17392 the gap in the ! slp_perm SLP case after each group.
17394 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17396 PR middle-end/81463
17397 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17400 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17402 PR middle-end/81462
17403 * predict.c (set_even_probabilities): Cleanup; do not affect
17404 probabilities that are already known.
17405 (combine_predictions_for_bb): Call even when count is set.
17407 2017-07-18 Nathan Sidwell <nathan@acm.org>
17409 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17412 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17415 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17416 optimization for loop niter analysis.
17418 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17421 * config/avr/avr.c (avr_optimize_casesi): Don't use
17422 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17424 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17426 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17427 body_cost_vec from _vect_peel_extended_info.
17428 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17429 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17432 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17434 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17436 2017-07-18 Richard Biener <rguenther@suse.de>
17438 PR tree-optimization/80620
17439 PR tree-optimization/81403
17440 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17441 info when re-using a VN table entry.
17443 2017-07-18 Richard Biener <rguenther@suse.de>
17445 PR tree-optimization/81418
17446 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17447 vectype_in. Verify that with lane-reducing reduction operations
17448 we have a single def-use cycle.
17450 2017-07-17 Carl Love <cel@us.ibm.com>
17452 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17454 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17455 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17456 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17457 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17458 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17459 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17460 VMULOSW): New enum "unspec" values.
17461 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17462 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17463 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17464 altivec_vmulosw): New patterns.
17465 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17466 VMULOSW): Add definitions.
17467 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17469 * config/alpha/alpha.c: Include predict.h.
17471 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17473 * tree-vrp.c (compare_assert_loc): Fix comparison function
17474 to return predictable results.
17476 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17478 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17480 (subdi3): Likewise.
17481 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17482 * doc/invoke.texi (mexpand-adddi): Update text.
17484 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17486 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17487 that also clobbers the CC register. The old expand code is moved
17489 (*arc_clzsi2): ... here.
17490 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17491 the CC register. The old expand code is moved to ...
17492 (arc_ctzsi2): ... here.
17494 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17496 * config/arc/arc.opt (mindexed-loads): Use initial value
17497 TARGET_INDEXED_LOADS_DEFAULT.
17498 (mauto-modify-reg): Use initial value
17499 TARGET_AUTO_MODIFY_REG_DEFAULT.
17500 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17501 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17502 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17503 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17505 2017-07-17 Martin Liska <mliska@suse.cz>
17508 * opts.c (finish_options): Do not allow -fgnu-tm
17509 w/ -fsanitize={kernel-,}address. Say sorry.
17511 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17514 * tree-loop-distribution.c (classify_partition): Only assert on
17515 numer of iterations.
17516 (merge_dep_scc_partitions): Delete prameter. Update function call.
17517 (distribute_loop): Remove code handling loop with unknown niters.
17518 (pass_loop_distribution::execute): Skip loop with unknown niters.
17520 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17523 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17524 function sort_partitions_by_post_order.
17526 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17528 PR tree-optimization/81374
17529 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17530 the max index of basic blocks, rather than number of basic blocks.
17532 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17534 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17536 (arc_legitimate_pic_operand_p): Likewise.
17537 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17539 (arc_needs_pcl_p): Likewise.
17540 (arc_legitimate_pc_offset_p): Likewise.
17541 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17542 function is also used in constrains.md.
17543 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17544 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17545 PLUS. Only return true/false in known cases, otherwise assert.
17546 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17547 is already called in arc_legitimate_constant_p.
17548 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17550 (LEGITIMATE_PIC_OPERAND_P): Use
17551 arc_raw_symbolic_reference_mentioned_p function.
17552 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17557 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17558 Andrew Burgess <andrew.burgess@embecosm.com>
17560 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17561 (arc_return_address_register): New function.
17562 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17563 (arc_handle_fndecl_attribute): Add naked attribute.
17564 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17565 (TARGET_WARN_FUNC_RETURN): Likewise.
17566 (arc_allocate_stack_slots_for_args): New function.
17567 (arc_warn_func_return): Likewise.
17568 (machine_function): Change type fn_type.
17569 (arc_compute_function_type): Consider new naked function type,
17570 change function return type.
17571 (arc_must_save_register): Adapt to handle new
17572 arc_compute_function_type's return type.
17573 (arc_expand_prologue): Likewise.
17574 (arc_expand_epilogue): Likewise.
17575 (arc_return_address_regs): Delete.
17576 (arc_return_address_register): New function.
17577 (arc_epilogue_uses): Use above function.
17578 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17579 (arc_function_type): Change encoding, add naked type.
17580 (ARC_INTERRUPT_P): Change to handle the new encoding.
17581 (ARC_FAST_INTERRUPT_P): Likewise.
17582 (ARC_NORMAL_P): Define.
17583 (ARC_NAKED_P): Likewise.
17584 (arc_compute_function_type): Delete prototype.
17585 * config/arc/arc.md (in_ret_delay_slot): Use
17586 arc_return_address_register function.
17587 (simple_return): Likewise.
17588 (p_return_i): Likewise.
17590 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17592 PR tree-optimization/81428
17593 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17594 can't be built for those types.
17596 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17598 Remove stuff dead since r239246.
17600 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17601 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17602 (avr_inform_devices): Remove dead stuff.
17604 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17606 * config/arm/arm_neon.h: Fix softp typo.
17608 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17610 PR tree-optimization/81365
17611 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17612 aggregate moves onto bb predecessor edges, make sure there are no
17613 loads that could alias the lhs in between the start of bb and the
17616 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17619 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17620 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17621 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17623 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17625 PR tree-optimization/81396
17626 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17627 (init_symbolic_number): Initialize it to 1.
17628 (perform_symbolic_merge): Add n_ops from both operands into the new
17630 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17631 without base_addr as useless if they need more than one operation.
17632 (bswap_replace): Handle !bswap case for NULL base_addr.
17634 2017-07-17 Tom de Vries <tom@codesourcery.com>
17637 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17640 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17642 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17643 conditional builtin define __FIX_LEON3FT_B2BST.
17645 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17647 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17648 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17651 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17653 PR rtl-optimization/81424
17654 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17655 to remove potential trapping from operands if -fnon-call-exceptions.
17657 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17659 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17660 profile_proability for scalling.
17661 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17663 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17665 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17667 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17669 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17670 fixpoint arithmetics.
17672 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17674 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17675 fixpoint arithmetics.
17677 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17679 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17680 fixpoint arithmetics.
17682 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17684 * profile-count.h (profile_probability::from_reg_br_prob_note,
17685 profile_probability::to_reg_br_prob_note): New functions.
17686 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17687 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17688 * predict.c (probability_reliable_p): Update.
17689 (edge_probability_reliable_p): Update.
17690 (br_prob_note_reliable_p): Update.
17691 (invert_br_probabilities): Update.
17692 (add_reg_br_prob_note): New function.
17693 (combine_predictions_for_insn): Update.
17694 * asan.c (asan_clear_shadow): Update.
17695 * cfgbuild.c (compute_outgoing_frequencies): Update.
17696 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17697 (update_br_prob_note): Update.
17698 (rtl_verify_edges): Update.
17699 (purge_dead_edges): Update.
17700 (fixup_reorder_chain): Update.
17701 * emit-rtl.c (try_split): Update.
17702 * ifcvt.c (cond_exec_process_insns): Update.
17703 (cond_exec_process_if_block): Update.
17704 (dead_or_predicable): Update.
17705 * internal-fn.c (expand_addsub_overflow): Update.
17706 (expand_neg_overflow): Update.
17707 (expand_mul_overflow): Update.
17708 * loop-doloop.c (doloop_modify): Update.
17709 * loop-unroll.c (compare_and_jump_seq): Update.
17710 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17711 * predict.h: Update.
17712 * reorg.c (mostly_true_jump): Update.
17714 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17715 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17716 * config/arc/arc.c: (emit_unlikely_jump): Update.
17717 * config/arm/arm.c: (emit_unlikely_jump): Update.
17718 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17719 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17720 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17721 (ix86_print_operand): Update.
17722 (ix86_split_fp_branch): Update.
17723 (predict_jump): Update.
17724 * config/ia64/ia64.c (ia64_print_operand): Update.
17725 * config/mmix/mmix.c (mmix_print_operand): Update.
17726 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17727 (rs6000_expand_split_stack_prologue): Update.
17728 * config/rs6000/rs6000.c: Update.
17729 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17730 (s390_expand_vec_movstr): Update.
17731 (s390_expand_cs_tdsi): Update.
17732 (s390_expand_split_stack_prologue): Update.
17733 * config/sh/sh.c (sh_print_operand): Update.
17734 (expand_cbranchsi4): Update.
17735 (expand_cbranchdi4): Update.
17736 * config/sparc/sparc.c (output_v9branch): Update.
17737 * config/spu/spu.c (get_branch_target): Update.
17738 (ea_load_store_inline): Update.
17739 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17740 * config/tilepro/tilepro.c: Update.
17742 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17744 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17745 (gimplify_save_expr): Likewise.
17747 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17749 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17751 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17753 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17755 * timevar.def (TV_IPA_FNSUMMARY): Define.
17757 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17759 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17760 to back store errata sensitive sequence from being generated.
17761 (sqrtdf2_fix): Likewise.
17763 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17765 * tree-ssa-threadupdate.c (compute_path_counts,
17766 update_joiner_offpath_counts): Use profile_probability.
17768 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17771 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17773 * config/arm/arm-c.c (arm_cpu_builtins): Define
17774 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17776 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17778 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17779 array entries to represent __ieee128 versions of the
17780 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17781 scalar_extract_sig, and scalar_insert_exp built-in functions.
17782 (altivec_resolve_overloaded_builtin): Add special case handling
17783 for the __builtin_scalar_insert_exp function, as represented by
17784 the P9V_BUILTIN_VEC_VSIEDP constant.
17785 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17786 exponent support for __ieee128 argument.
17787 (VSESQP): Add scalar extract signature support for __ieee128
17789 (VSTDCNQP): Add scalar test negative support for __ieee128
17791 (VSIEQP): Add scalar insert exponent support for __int128 argument
17792 with __ieee128 result.
17793 (VSIEQPF): Add scalar insert exponent support for __ieee128
17794 argument with __ieee128 result.
17795 (VSTDCQP): Add scalar test data class support for __ieee128
17797 (VSTDCNQP): Add overload support for scalar test negative with
17798 __ieee128 argument.
17799 (VSTDCQP): Add overload support for scalar test data class
17800 __ieee128 argument.
17801 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17802 UNSPEC_VSX_SXSIGDP.
17803 (UNSPEC_VSX_SIEXPQP): New constant.
17804 (xsxexpqp): New insn for VSX scalar extract exponent quad
17806 (xsxsigqp): New insn for VSX scalar extract significand quad
17808 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17809 precision with floating point argument.
17810 (xststdcqp): New expand for VSX scalar test data class quad
17812 (xststdcnegqp): New expand for VSX scalar test negative quad
17814 (xststdcqp): New insn to match expansions for VSX scalar test data
17815 class quad precision and VSX scalar test negative quad precision.
17816 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17817 special case operand checking to enforce that second operand of
17818 VSX scalar test data class with quad precision argument is a 7-bit
17820 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17821 prototypes and descriptions of __ieee128 versions of
17822 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17823 scalar_test_data_class, and scalar_test_neg built-in functions.
17825 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17827 PR tree-optimization/81162
17828 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17829 replace a negate with an add.
17831 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17833 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17835 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17837 * config/arm/arm-c.c (arm_cpu_builtins): Define
17838 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17840 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17842 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17843 (armv8-r): Set ARM Cortex-R52 as default CPU.
17844 * config/arm/arm-tables.opt: Regenerate.
17845 * config/arm/arm-tune.md: Regenerate.
17846 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17848 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17849 extension for -mcpu=cortex-r52.
17851 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17853 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17854 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17855 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17856 (fp-armv8): Define it as FP_ARMv8 only.
17857 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17858 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17860 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17861 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17862 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17863 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17864 than TARGET_FPU_ARMV8.
17865 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17866 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17867 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17869 * config/arm/neon.md (neon_vrint): Likewise.
17870 (neon_vcvt): Likewise.
17871 (neon_<fmaxmin_op><mode>): Likewise.
17872 (<fmaxmin><mode>3): Likewise.
17873 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17874 * config/arm/predicates.md (arm_cond_move_operator): Check against
17875 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17877 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17879 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17880 to top of function.
17882 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17884 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17885 loop in comment with memset.
17887 2017-07-14 Martin Liska <mliska@suse.cz>
17889 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17890 * dwarf2out.c (is_java): Remove the function.
17891 (output_pubname): Remove usage of the function.
17892 (lower_bound_default): Remove usage of DW_LANG_Java.
17893 (gen_compile_unit_die): Likewise.
17894 * gcc.c: Remove compiler defaults for .java and .zip files.
17895 * gimple-expr.c (remove_suffix): Change as there's no longer
17896 extension than 4-letter one.
17897 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17898 (gimplify_save_expr): Likewise.
17899 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17900 as it's possible even for other languages than Java.
17901 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17902 * lto-opts.c (lto_write_options): Remove reference to Java.
17903 * opts.c (strip_off_ending): Update file extension handling.
17904 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17905 * tree-eh.c (lower_resx): Likewise.
17906 * tree.c (free_lang_data_in_type): Remove dead code.
17907 (find_decls_types_r): Likewise.
17908 (build_common_builtin_nodes): Remove Java from a comment.
17909 (verify_type): Remove dead code.
17910 * varasm.c (assemble_external): Remove Java from a comment.
17912 2017-07-14 Martin Liska <mliska@suse.cz>
17914 * opts.c (finish_options): Add quotes.
17915 (common_handle_option): Likewise.
17917 2017-07-14 Martin Liska <mliska@suse.cz>
17919 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17920 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17921 Remove N_SO_PASCAL.
17922 * dwarf2out.c (lower_bound_default): Do not handle
17924 (gen_compile_unit_die): Likewise.
17925 * gcc.c: Remove default extension binding for GNU Pascal.
17926 * stmt.c: Remove Pascal language from a comment.
17927 * xcoffout.c: Likewise.
17929 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17932 * diagnostic-show-locus.c (fixit_cmp): New function.
17933 (layout::layout): Sort m_fixit_hints.
17934 (column_range::column_range): Assert that the values are valid.
17935 (struct char_span): New struct.
17936 (correction::overwrite): New method.
17937 (struct source_line): New struct.
17938 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17939 calls in terms of classes source_line and char_span, and
17940 correction::overwrite.
17941 (selftest::test_overlapped_fixit_printing_2): New function.
17942 (selftest::diagnostic_show_locus_c_tests): Call it.
17944 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17946 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17947 early if there is no lhs.
17949 2017-07-13 Martin Liska <mliska@suse.cz>
17951 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17952 (gen_reference_type_die): Likewise.
17953 * stor-layout.c: Remove Pascal-related comment.
17955 2017-07-13 Martin Liska <mliska@suse.cz>
17957 * opts.c (finish_options): Add quotes to error messages.
17958 (parse_sanitizer_options): Likewise.
17960 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17962 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17964 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17966 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17968 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17970 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17972 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17974 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17977 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17978 provides the hardware capability bits, define the macro
17979 __BUILTIN_CPU_SUPPORTS__.
17980 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17981 if GLIBC does not provide the hardware capability bits. Add a
17982 gcc_unreachable call if the built-in cpu function is neither
17983 __builtin_cpu_is nor __builtin_cpu_supports.
17984 (rs6000_get_function_versions_dispatcher): Change the warning
17985 that an old GLIBC is used which does not export the capability
17986 bits to be an error.
17987 * doc/extend.texi (target_clones attribute): Document the
17988 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17989 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17990 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17991 the macros defined by GCC if the newer GLIBC is available.
17993 2017-07-12 Jeff Law <law@redhat.com>
17995 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17996 remaining includes slightly.
17997 * config/riscv/riscv-builtins.c: Include profile-count.h.
17999 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18002 * config/avr/avr.c (avr_set_current_function): In diagnostic
18003 messages: Quote keywords and (parts of) identifiers.
18004 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18007 2017-07-12 Carl Love <cel@us.ibm.com>
18009 * config/rs6000/rs6000-c.c: Add support for built-in functions
18010 vector bool char vec_revb (vector bool char);
18011 vector bool short vec_revb (vector short char);
18012 vector bool int vec_revb (vector bool int);
18013 vector bool long long vec_revb (vector bool long long);
18014 * doc/extend.texi: Update the built-in documentation file for the
18015 new built-in functions.
18017 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18019 * config/s390/s390.md: Remove movcc splitter.
18021 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18023 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18024 load/store on condition.
18026 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18029 * config/avr/avr.c (avr_encode_section_info)
18030 [progmem && !TREE_READONLY]: Error if progmem object needs
18033 2017-07-11 Michael Collison <michael.collison@arm.com>
18035 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18038 2017-07-11 Carl Love <cel@us.ibm.com>
18040 * config/rs6000/rs6000-c.c: Add support for builtins
18041 vector unsigned int vec_parity_lsbb (vector signed int);
18042 vector unsigned int vec_parity_lsbb (vector unsigned int);
18043 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18044 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18045 vector unsigned long long vec_parity_lsbb (vector signed long long);
18046 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18047 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18048 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18049 * doc/extend.texi: Update the built-in documentation file for the
18050 new built-in functions.
18052 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18054 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18055 (layout::m_primary_loc): New field.
18056 (layout::layout): Initialize new field. Move location filtering
18057 logic from here to...
18058 (layout::maybe_add_location_range): ...this new method. Add
18059 support for filtering to just the lines already specified by other
18061 (layout::will_show_line_p): New method.
18062 (gcc_rich_location::add_location_if_nearby): New method.
18063 (selftest::test_add_location_if_nearby): New test function.
18064 (selftest::diagnostic_show_locus_c_tests): Call it.
18065 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18068 2017-07-11 Tom de Vries <tom@codesourcery.com>
18070 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18071 (bb_first_real_insn): New function.
18072 (nvptx_single): Add extra initialization of broadcasted condition
18075 2017-07-11 Nathan Sidwell <nathan@acm.org>
18077 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18079 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18081 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18082 Binutils doc as TEXI will mess them up.
18083 * doc/invoke.texi (AVR Options): Same here.
18085 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18087 * config/sparc/sparc.opt (mfix-ut700): New option.
18088 (mfix-gr712rc): Likewise.
18089 (sparc_fix_b2bst): New variable.
18090 * doc/invoke.texi (SPARC options): Document them.
18091 (ARM options): Fix warnings.
18092 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18093 instructions to prevent sequences that can trigger the store-store
18094 errata for certain LEON3FT processors.
18095 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18096 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18097 * config/sparc/sparc.md (fix_b2bst): New attribute.
18098 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18100 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18103 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18105 (*rsqrtsf2_sse): Ditto.
18107 (div<mode>3): Macroize insn from divdf3 and divsf3
18108 using MODEF mode iterator.
18110 2017-07-10 Martin Sebor <msebor@redhat.com>
18112 PR tree-optimization/80397
18113 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18114 instead of testing for equality to INTEGER_TYPE.
18116 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18118 * config.gcc: Remove uclibc from arc target spec.
18120 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18122 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18124 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18127 * lto-wrapper.c (remove_option): New function.
18128 (merge_and_complain): Merge PIC/PIE options more realistically.
18130 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18132 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18136 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18137 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18138 * config.in: Regenerate.
18139 * configure: Regenerate.
18140 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18141 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18142 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18143 (TARGET_GASISR_PROLOGUES): ...target mask.
18144 * common/config/avr/avr-common.c
18145 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18146 Set -mgas-isr-prologues.
18147 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18148 INSERT_PASS_BEFORE for it.
18149 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18150 * config/avr/avr.c (avr_option_override)
18151 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18152 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18153 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18154 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18155 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18157 (make_avr_pass_pre_proep): New function.
18158 (emit_push_sfr) <treg>: Add argument to function and use it
18159 instead of TMP_REG.
18160 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18161 and set machine->gasisr.yes.
18162 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18163 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18164 __gcc_isr.n_pushed to .L__stack_usage.
18165 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18166 (avr_asm_final_postscan_insn): ...this new static function.
18167 * config/avr/avr.h (machine_function)
18168 <is_no_gccisr, use_L__stack_usage>: New fields.
18169 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18170 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18171 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18172 (gasisr, *gasisr): New expander and insn.
18173 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18174 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18175 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18177 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18179 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18182 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18184 Move jump-tables out of .text again.
18187 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18188 (ASM_OUTPUT_ADDR_VEC): New function.
18189 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18190 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18191 INSN_ADDRESSes as asm comment.
18192 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18193 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18194 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18195 * config/avr/avr.md (*tablejump): Adjust comment.
18196 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18197 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18199 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18200 (avr_output_addr_vec): New proto.
18201 (avr_log_t) <insn_addresses>: New field.
18203 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18206 * config/i386/i386.c (ix86_function_arg_advance): Set
18207 outgoing_args_on_stack to true if there are outgoing arguments
18209 (ix86_function_arg): Likewise.
18210 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18211 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18212 * config/i386/i386.h (machine_function): Add
18213 outgoing_args_on_stack.
18215 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18217 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18218 supporting pthreds.
18219 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18221 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18223 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18224 (REAL_H): Remove $(MACHMODE_H).
18225 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18227 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18228 $(MACHMODE_H) and double-int.h.
18229 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18231 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18234 2017-07-07 Andrew Pinski <apinski@cavium.com>
18236 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18237 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18239 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18241 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18242 Add warning if GCC was not configured to link against a GLIBC that
18243 exports the hardware capability bits.
18244 (make_resolver_func): Make resolver function private and not a
18245 COMDAT function. Create the name with clone_function_name instead
18246 of make_unique_name.
18249 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18250 correct operand in doing the split.
18252 2017-07-07 Carl Love <cel@us.ibm.com>
18254 * config/rs6000/rs6000-c: Add support for built-in function
18255 vector unsigned short vec_pack_to_short_fp32 (vector float,
18257 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18258 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18259 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18260 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18261 (convert_4f32_8i16): Add define_expand.
18262 * doc/extend.texi: Update the built-in documentation file for the
18263 new built-in function.
18265 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18267 * config/sparc/m8.md: New file.
18268 * config/sparc/sparc.md: Include m8.md.
18270 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18272 * config/sparc/sparc.opt: New option -mvis4b.
18273 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18274 (sparc_option_override): Handle VIS4B.
18275 (enum sparc_builtins): Define
18276 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18277 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18278 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18279 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18280 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18281 (check_constant_argument): New function.
18282 (sparc_vis_init_builtins): Define builtins
18283 __builtin_vis_dictunpack{8,16,32},
18284 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18285 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18286 __builtin_vis_fpcmpde{8,16,32}shl and
18287 __builtin_vis_fpcmpur{8,16,32}shl.
18288 (sparc_expand_builtin): Check that the constant operands to
18289 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18290 constant and in range.
18291 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18293 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18294 (SPARC_IMM5_P): Likewise.
18295 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18296 (enabled): Handle vis4b.
18297 (UNSPEC_DICTUNPACK): New unspec.
18298 (UNSPEC_FPCMPSHL): Likewise.
18299 (UNSPEC_FPUCMPSHL): Likewise.
18300 (UNSPEC_FPCMPDESHL): Likewise.
18301 (UNSPEC_FPCMPURSHL): Likewise.
18302 (cpu_feature): New CPU feature `vis4b'.
18303 (dictunpack{8,16,32}): New insns.
18304 (FPCSMODE): New mode iterator.
18305 (fpcscond): New code iterator.
18306 (fpcsucond): Likewise.
18307 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18308 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18309 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18310 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18311 * config/sparc/constraints.md: Define constraints `q' for unsigned
18312 2-bit integer constants and `t' for unsigned 5-bit integer
18314 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18316 (imm5_operand_dictunpack16): Likewise.
18317 (imm5_operand_dictunpack32): Likewise.
18318 (imm2_operand): Likewise.
18319 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18320 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18321 ditunpack* and fpcmp*shl builtins.
18323 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18325 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18326 * config.in: Add HAVE_AS_SPARC6 define.
18327 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18329 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18331 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18332 (CPP_CPU_SPEC): Handle m8.
18333 (ASM_CPU_SPEC): Likewise.
18334 * config/sparc/sparc-opts.h (enum processor_type): Add
18336 * config/sparc/sparc.c (m8_costs): New struct.
18337 (sparc_option_override): Handle TARGET_CPU_m8.
18338 (sparc32_initialize_trampoline): Likewise.
18339 (sparc64_initialize_trampoline): Likewise.
18340 (sparc_issue_rate): Likewise.
18341 (sparc_register_move_cost): Likewise.
18342 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18343 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18344 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18345 (CPP_CPU_SPEC): Handle M8.
18346 (ASM_CPU_SPEC): Likewise.
18347 (AS_M8_FLAG): Define.
18348 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18349 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18350 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18352 * configure: Regenerate.
18353 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18356 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18358 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18360 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18361 ("*movdi_insn_sp32"): Do not set v3pipe.
18362 ("*movsi_insn"): Likewise.
18363 ("*movdi_insn_sp64"): Likewise.
18364 ("*movsf_insn"): Likewise.
18365 ("*movdf_insn_sp32"): Likewise.
18366 ("*movdf_insn_sp64"): Likewise.
18367 ("*zero_extendsidi2_insn_sp64"): Likewise.
18368 ("*sign_extendsidi2_insn"): Likewise.
18369 ("*mov<VM32:mode>_insn"): Likewise.
18370 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18371 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18372 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18373 ("<vlop:code><VL:mode>3"): Likewise.
18374 ("*not_<vlop:code><VL:mode>3"): Likewise.
18375 ("*nand<VL:mode>_vis"): Likewise.
18376 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18377 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18378 ("one_cmpl<VL:mode>2"): Likewise.
18379 ("faligndata<VM64:mode>_vis"): Likewise.
18380 ("alignaddrsi_vis"): Likewise.
18381 ("alignaddrdi_vis"): Likweise.
18382 ("alignaddrlsi_vis"): Likewise.
18383 ("alignaddrldi_vis"): Likewise.
18384 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18385 ("bmaskdi_vis"): Likewise.
18386 ("bmasksi_vis"): Likewise.
18387 ("bshuffle<VM64:mode>_vis"): Likewise.
18388 ("cmask8<P:mode>_vis"): Likewise.
18389 ("cmask16<P:mode>_vis"): Likewise.
18390 ("cmask32<P:mode>_vis"): Likewise.
18391 ("pdistn<P:mode>_vis"): Likewise.
18392 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18394 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18396 * config/sparc/sparc.md ("subtype"): New insn attribute.
18397 ("*wrgsr_sp64"): Set insn subtype.
18398 ("*rdgsr_sp64"): Likewise.
18399 ("alignaddrsi_vis"): Likewise.
18400 ("alignaddrdi_vis"): Likewise.
18401 ("alignaddrlsi_vis"): Likewise.
18402 ("alignaddrldi_vis"): Likewise.
18403 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18404 ("fexpand_vis"): Likewise.
18405 ("fpmerge_vis"): Likewise.
18406 ("faligndata<VM64:mode>_vis"): Likewise.
18407 ("bshuffle<VM64:mode>_vis"): Likewise.
18408 ("cmask8<P:mode>_vis"): Likewise.
18409 ("cmask16<P:mode>_vis"): Likewise.
18410 ("cmask32<P:mode>_vis"): Likewise.
18411 ("fchksm16_vis"): Likewise.
18412 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18413 ("fmean16_vis"): Likewise.
18414 ("fp<plusminus_insn>64_vis"): Likewise.
18415 ("<plusminus_insn>v8qi3"): Likewise.
18416 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18417 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18418 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18419 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18420 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18421 ("*movqi_insn"): Likewise.
18422 ("*movhi_insn"): Likewise.
18423 ("*movsi_insn"): Likewise.
18424 ("movsi_pic_gotdata_op"): Likewise.
18425 ("*movdi_insn_sp32"): Likewise.
18426 ("*movdi_insn_sp64"): Likewise.
18427 ("movdi_pic_gotdata_op"): Likewise.
18428 ("*movsf_insn"): Likewise.
18429 ("*movdf_insn_sp32"): Likewise.
18430 ("*movdf_insn_sp64"): Likewise.
18431 ("*zero_extendhisi2_insn"): Likewise.
18432 ("*zero_extendqihi2_insn"): Likewise.
18433 ("*zero_extendqisi2_insn"): Likewise.
18434 ("*zero_extendqidi2_insn"): Likewise.
18435 ("*zero_extendhidi2_insn"): Likewise.
18436 ("*zero_extendsidi2_insn_sp64"): Likewise.
18437 ("ldfsr"): Likewise.
18438 ("prefetch_64"): Likewise.
18439 ("prefetch_32"): Likewise.
18440 ("tie_ld32"): Likewise.
18441 ("tie_ld64"): Likewise.
18442 ("*tldo_ldub_sp32"): Likewise.
18443 ("*tldo_ldub1_sp32"): Likewise.
18444 ("*tldo_ldub2_sp32"): Likewise.
18445 ("*tldo_ldub_sp64"): Likewise.
18446 ("*tldo_ldub1_sp64"): Likewise.
18447 ("*tldo_ldub2_sp64"): Likewise.
18448 ("*tldo_ldub3_sp64"): Likewise.
18449 ("*tldo_lduh_sp32"): Likewise.
18450 ("*tldo_lduh1_sp32"): Likewise.
18451 ("*tldo_lduh_sp64"): Likewise.
18452 ("*tldo_lduh1_sp64"): Likewise.
18453 ("*tldo_lduh2_sp64"): Likewise.
18454 ("*tldo_lduw_sp32"): Likewise.
18455 ("*tldo_lduw_sp64"): Likewise.
18456 ("*tldo_lduw1_sp64"): Likewise.
18457 ("*tldo_ldx_sp64"): Likewise.
18458 ("*mov<VM32:mode>_insn"): Likewise.
18459 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18460 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18462 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18464 * config/sparc/sparc.md ("type"): New insn type viscmp.
18465 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18467 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18468 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18469 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18470 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18472 ("n7_vis_logical_11cycle"): Likewise.
18473 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18474 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18475 * config/sparc/niagara.md ("niag_vis"): Likewise.
18476 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18477 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18479 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18481 * config/sparc/sparc.md: New instruction type `bmask'.
18482 (bmaskdi_vis): Use the `bmask' type.
18483 (bmasksi_vis): Likewise.
18484 * config/sparc/ultra3.md (us3_array): Likewise.
18485 * config/sparc/niagara7.md (n7_array): Likewise.
18486 * config/sparc/niagara4.md (n4_array): Likewise.
18487 * config/sparc/niagara2.md (niag2_vis): Likewise.
18488 (niag3_vis): Likewise.
18489 * config/sparc/niagara.md (niag_vis): Likewise.
18491 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18493 * ipa-comdats.c: Remove optimize check from gate.
18494 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18495 for functions not optimized.
18496 (ipa_fn_summary_read): Skip optimize check.
18497 (ipa_fn_summary_write): Likewise.
18498 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18500 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18502 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18503 (check_callers): Give up on caller which is not optimized.
18504 (inline_small_functions): Likewise.
18505 (ipa_inline): Do not give up when not optimizing.
18506 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18507 away unoptimizes cdtors.
18508 (whole_program_function_and_variable_visibility): Do
18509 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18510 * ipa.c (process_references): Do not check optimize.
18511 (symbol_table::remove_unreachable_nodes): Update optimize check.
18512 (set_writeonly_bit): Update optimize check.
18513 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18514 (pass_ipa_single_use::gate): Remove.
18516 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18518 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18519 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18520 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18521 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18522 permute_load, permute_store, adjust_extract, adjust_splat,
18523 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18524 replace_swap_with_copy, dump_swap_insn_table,
18525 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18526 recombine_lvx_pattern, recombine_stvx_pattern,
18527 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18528 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18529 to file rs6000-p8swap.c.
18530 * config/rs6000/rs6000-p8swap.c: New file.
18531 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18532 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18533 and rs6000*-*-* targets.
18535 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18537 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18539 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18541 * lto-wrapper.c (merge_and_complain): Do not merge
18542 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18543 fsigned_zeros, ftrapping_math, fwrapv.
18544 (append_compiler_options): Do not track these options.
18545 (append_linker_options): Likewie
18547 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18549 * cgraphunit.c (cgraph_node::finalize_function): When
18550 !flag_toplevel_reorde set no_reorder flag.
18551 (varpool_node::finalize_decl): Likewise.
18552 (symbol_table::compile): Drop no toplevel reorder path.
18554 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18556 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18557 edges; zero probability is not better than uninitialized.
18559 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18561 * asan.h (asan_sanitize_allocas_p): Declare.
18562 * asan.c (asan_sanitize_allocas_p): New function.
18563 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18564 (handle_builtin_alloca): Likewise.
18565 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18566 if !asan_sanitize_allocas_p.
18567 * params.def (asan-instrument-allocas): Add new option.
18568 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18569 * opts.c (common_handle_option): Disable allocas sanitization for
18572 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18574 * asan.c: Include gimple-fold.h.
18575 (get_last_alloca_addr): New function.
18576 (handle_builtin_stackrestore): Likewise.
18577 (handle_builtin_alloca): Likewise.
18578 (asan_emit_allocas_unpoison): Likewise.
18579 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18580 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18581 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18582 (instrument_builtin_call): Pass gimple iterator to
18583 get_mem_refs_of_builtin_call.
18584 (last_alloca_addr): New global.
18585 * asan.h (asan_emit_allocas_unpoison): Declare.
18586 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18587 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18588 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18589 if function calls alloca.
18590 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18591 * gimple-fold.h (replace_call_with_value): Declare.
18592 * internal-fn.c: Include asan.h.
18593 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18594 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18596 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18598 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18599 (C_SELFTEST_FLAGS): New.
18600 (CPP_SELFTEST_FLAGS): New.
18601 (SELFTEST_DEPS): New, from deps of s-selftest.
18602 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18603 (CPP_SELFTEST_DEPS): New.
18604 (selftest): Add dependency on s-selftest-c++.
18605 (s-selftest): Rename to...
18606 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18607 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18608 than SELFTEST_FLAGS.
18609 (selftest-gdb): Rename to...
18610 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18612 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18613 (selftest-valgrind): Rename to...
18614 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18616 (selftest-valgrind): Reintroduce as an alias for
18617 selftest-c-valgrind.
18618 (s-selftest-c++): New.
18619 (selftest-c++-gdb): New.
18620 (selftest-c++-valgrind): New.
18622 2017-07-06 Olivier Hainque <hainque@adacore.com>
18624 * gcc.c (process_command): When deciding if undefined variables
18625 should be ignored when processing specs, accept "gcc -v" as well.
18627 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18629 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18630 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18632 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18634 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18635 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18636 * config/arm/arm-tables.opt: Regenerate.
18637 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18639 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18641 2017-07-06 Carl Love <cel@us.ibm.com>
18643 * ChangeLog: Clean up from mid air collision
18645 2017-07-06 Carl Love <cel@us.ibm.com>
18647 * config/rs6000/rs6000-c.c: Add support for built-in functions
18648 vector signed int vec_subc (vector signed int, vector signed int);
18649 vector signed __int128 vec_subc (vector signed __int128,
18650 vector signed __int128);
18651 vector unsigned __int128 vec_subc (vector unsigned __int128,
18652 vector unsigned __int128);
18653 vector signed int vec_sube (vector signed int, vector signed int,
18654 vector signed int);
18655 vector unsigned int vec_sube (vector unsigned int,
18656 vector unsigned int,
18657 vector unsigned int);
18658 vector signed __int128 vec_sube (vector signed __int128,
18659 vector signed __int128,
18660 vector signed__int128);
18661 vector unsigned __int128 vec_sube (vector unsigned __int128,
18662 vector unsigned __int128,
18663 vector unsigned __int128);
18664 vector signed int vec_subec (vector signed int, vector signed int,
18665 vector signed int);
18666 vector unsigned int vec_subec (vector unsigned int,
18667 vector unsigned int,
18668 vector unsigned int);
18669 vector signed __int128 vec_subec (vector signed __int128,
18670 vector signed __int128,
18671 vector signed__int128);
18672 vector unsigned __int128 vec_subec (vector unsigned __int128,
18673 vector unsigned __int128,
18674 vector unsigned __int128);
18675 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18676 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18677 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18678 BU_ALTIVEC_OVERLOAD_X definitions.
18679 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18680 * doc/extend.texi: Update the built-in documentation file for the new
18681 built-in functions.
18683 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18686 * diagnostic-show-locus.c (layout::layout): Use start and finish
18687 spelling location for the start and finish of each range.
18688 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18689 Add unused aspect param.
18690 * input.c (expand_location_1): Add "aspect" param, and use it
18691 to access the correct part of the location.
18692 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18694 (expand_location_to_spelling_point): Likewise.
18695 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18696 param, and pass it to expand_location_1.
18698 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18700 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18701 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18702 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18703 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18704 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18705 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18706 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18707 _mm_maskz_getmant_ss): New intrinsics.
18708 (__builtin_ia32_getexpss128_mask): Changed to ...
18709 __builtin_ia32_getexpss128_round ... this.
18710 (__builtin_ia32_getexpsd128_mask): Changed to ...
18711 __builtin_ia32_getexpsd128_round ... this.
18712 * config/i386/i386-builtin-types.def
18713 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18714 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18715 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18716 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18717 __builtin_ia32_getmantss_mask_round): New builtins.
18718 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18719 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18720 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18721 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18722 * config/i386/sse.md
18723 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18724 avx512f_sgetexp<mode><mask_scalar_name>
18725 <round_saeonly_scalar_name> ... this.
18726 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18727 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18728 vgetexp<ssescalarmodesuffix>
18729 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18730 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18731 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18732 avx512f_vgetmant<mode><mask_scalar_name>
18733 <round_saeonly_scalar_name> ... this.
18734 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18735 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18736 vgetmant<ssescalarmodesuffix>
18737 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18738 %0<mask_scalar_operand4>, %1, %2
18739 <round_saeonly_scalar_mask_op4>, %3} ... this.
18740 * config/i386/subst.md (mask_scalar_operand4,
18741 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18742 round_saeonly_scalar_nimm_predicate): New subst attributes.
18744 2017-07-06 Julia Koval <julia.koval@intel.com>
18746 * config/i386/i386.c (ix86_erase_embedded_rounding):
18747 Remove code for old rounding pattern.
18749 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18751 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18753 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18755 * doc/sourcebuild.texi (Test Directives, Variants of
18756 dg-require-support): Add documentation for dg-require-stack-check.
18758 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18760 * config/i386/subst.md (mask_scalar, round_scalar,
18761 round_saeonly_scalar): New meta-templates.
18762 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18763 round_scalar_mask_operand3, round_scalar_mask_op3,
18764 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18765 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18766 round_saeonly_scalar_constraint,
18767 round_saeonly_scalar_prefix): New subst attribute.
18768 * config/i386/sse.md
18769 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18770 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18771 <round_scalar_name> ... this.
18772 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18773 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18774 <round_scalar_name> ... this.
18775 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18776 <sse>_vm<code><mode>3<mask_scalar_name>
18777 <round_saeonly_scalar_name> ... this.
18778 (v<plusminus_mnemonic><ssescalarmodesuffix>
18779 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18780 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18781 v<plusminus_mnemonic><ssescalarmodesuffix>
18782 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18783 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18784 (v<multdiv_mnemonic><ssescalarmodesuffix>
18785 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18786 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18787 v<multdiv_mnemonic><ssescalarmodesuffix>
18788 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18789 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18790 (v<maxmin_float><ssescalarmodesuffix>
18791 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18792 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18793 v<maxmin_float><ssescalarmodesuffix>
18794 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18795 %0<mask_scalar_operand3>, %1, %<iptr>2
18796 <round_saeonly_scalar_mask_op3>} ... this.
18798 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18800 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18801 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18803 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18804 Alan Hayward <alan.hayward@arm.com>
18805 David Sherwood <david.sherwood@arm.com>
18807 * combine.c (simplify_if_then_else): Remove "enum" before
18809 * compare-elim.c (can_eliminate_compare): Likewise.
18810 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18812 (aarch64_lookup_simd_builtin_type): Likewise.
18813 (aarch64_simd_builtin_type): Likewise.
18814 (aarch64_init_simd_builtin_types): Likewise.
18815 (aarch64_simd_expand_args): Likewise.
18816 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18818 (aarch64_reverse_mask): Likewise.
18819 (aarch64_simd_emit_reg_reg_move): Likewise.
18820 (aarch64_gen_adjusted_ldpstp): Likewise.
18821 (aarch64_ccmp_mode_to_code): Likewise.
18822 (aarch64_operands_ok_for_ldpstp): Likewise.
18823 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18824 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18826 (aarch64_min_divisions_for_recip_mul): Likewise.
18827 (aarch64_reassociation_width): Likewise.
18828 (aarch64_get_condition_code_1): Likewise.
18829 (aarch64_simd_emit_reg_reg_move): Likewise.
18830 (aarch64_simd_attr_length_rglist): Likewise.
18831 (aarch64_reverse_mask): Likewise.
18832 (aarch64_operands_ok_for_ldpstp): Likewise.
18833 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18834 (aarch64_gen_adjusted_ldpstp): Likewise.
18835 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18837 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18838 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18839 (arm_lookup_simd_builtin_type): Likewise.
18840 (arm_simd_builtin_type): Likewise.
18841 (arm_init_simd_builtin_types): Likewise.
18842 (arm_expand_builtin_args): Likewise.
18843 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18844 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18845 (ft32_setup_incoming_varargs): Likewise.
18846 (ft32_function_arg): Likewise.
18847 (ft32_function_arg_advance): Likewise.
18848 (ft32_pass_by_reference): Likewise.
18849 (ft32_arg_partial_bytes): Likewise.
18850 (ft32_valid_pointer_mode): Likewise.
18851 (ft32_addr_space_pointer_mode): Likewise.
18852 (ft32_addr_space_legitimate_address_p): Likewise.
18853 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18855 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18856 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18857 (iamcu_alignment): Likewise.
18858 (canonicalize_vector_int_perm): Likewise.
18859 (ix86_noce_conversion_profitable_p): Likewise.
18860 (ix86_mpx_bound_mode): Likewise.
18861 (ix86_operands_ok_for_move_multiple): Likewise.
18862 * config/microblaze/microblaze-protos.h
18863 (microblaze_expand_conditional_branch_reg): Likewise.
18864 * config/microblaze/microblaze.c
18865 (microblaze_expand_conditional_branch_reg): Likewise.
18866 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18868 (rs6000_reassociation_width): Likewise.
18869 (rs6000_invalid_binary_op): Likewise.
18870 (fusion_p9_p): Likewise.
18871 (emit_fusion_p9_load): Likewise.
18872 (emit_fusion_p9_store): Likewise.
18873 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18875 (riscv_hard_regno_mode_ok_p): Likewise.
18876 (riscv_address_insns): Likewise.
18877 (riscv_split_symbol): Likewise.
18878 (riscv_legitimize_move): Likewise.
18879 (riscv_function_value): Likewise.
18880 (riscv_hard_regno_nregs): Likewise.
18881 (riscv_expand_builtin): Likewise.
18882 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18883 (riscv_build_integer): Likewise.
18884 (riscv_split_integer): Likewise.
18885 (riscv_legitimate_constant_p): Likewise.
18886 (riscv_cannot_force_const_mem): Likewise.
18887 (riscv_regno_mode_ok_for_base_p): Likewise.
18888 (riscv_valid_base_register_p): Likewise.
18889 (riscv_valid_offset_p): Likewise.
18890 (riscv_valid_lo_sum_p): Likewise.
18891 (riscv_classify_address): Likewise.
18892 (riscv_legitimate_address_p): Likewise.
18893 (riscv_address_insns): Likewise.
18894 (riscv_load_store_insns): Likewise.
18895 (riscv_force_binary): Likewise.
18896 (riscv_split_symbol): Likewise.
18897 (riscv_force_address): Likewise.
18898 (riscv_legitimize_address): Likewise.
18899 (riscv_move_integer): Likewise.
18900 (riscv_legitimize_const_move): Likewise.
18901 (riscv_legitimize_move): Likewise.
18902 (riscv_address_cost): Likewise.
18903 (riscv_subword): Likewise.
18904 (riscv_output_move): Likewise.
18905 (riscv_canonicalize_int_order_test): Likewise.
18906 (riscv_emit_int_order_test): Likewise.
18907 (riscv_function_arg_boundary): Likewise.
18908 (riscv_pass_mode_in_fpr_p): Likewise.
18909 (riscv_pass_fpr_single): Likewise.
18910 (riscv_pass_fpr_pair): Likewise.
18911 (riscv_get_arg_info): Likewise.
18912 (riscv_function_arg): Likewise.
18913 (riscv_function_arg_advance): Likewise.
18914 (riscv_arg_partial_bytes): Likewise.
18915 (riscv_function_value): Likewise.
18916 (riscv_pass_by_reference): Likewise.
18917 (riscv_setup_incoming_varargs): Likewise.
18918 (riscv_print_operand): Likewise.
18919 (riscv_elf_select_rtx_section): Likewise.
18920 (riscv_save_restore_reg): Likewise.
18921 (riscv_for_each_saved_reg): Likewise.
18922 (riscv_register_move_cost): Likewise.
18923 (riscv_hard_regno_mode_ok_p): Likewise.
18924 (riscv_hard_regno_nregs): Likewise.
18925 (riscv_class_max_nregs): Likewise.
18926 (riscv_memory_move_cost): Likewise.
18927 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18928 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18929 (rl78_addr_space_address_mode): Likewise.
18930 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18932 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18933 (rs6000_reassociation_width): Likewise.
18934 (rs6000_invalid_binary_op): Likewise.
18935 (fusion_p9_p): Likewise.
18936 (emit_fusion_p9_load): Likewise.
18937 (emit_fusion_p9_store): Likewise.
18938 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18939 (ok_for_simple_move_operands): Likewise.
18940 (ok_for_simple_move_strict_operands): Likewise.
18941 (ok_for_simple_arith_logic_operands): Likewise.
18942 (visium_legitimize_reload_address): Likewise.
18943 (visium_select_cc_mode): Likewise.
18944 (output_cbranch): Likewise.
18945 (visium_split_double_move): Likewise.
18946 (visium_expand_copysign): Likewise.
18947 (visium_expand_int_cstore): Likewise.
18948 (visium_expand_fp_cstore): Likewise.
18949 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18950 (visium_function_arg): Likewise.
18951 (visium_function_arg_advance): Likewise.
18952 (visium_libcall_value): Likewise.
18953 (visium_setup_incoming_varargs): Likewise.
18954 (visium_legitimate_constant_p): Likewise.
18955 (visium_legitimate_address_p): Likewise.
18956 (visium_legitimize_address): Likewise.
18957 (visium_secondary_reload): Likewise.
18958 (visium_register_move_cost): Likewise.
18959 (visium_memory_move_cost): Likewise.
18960 (prepare_move_operands): Likewise.
18961 (ok_for_simple_move_operands): Likewise.
18962 (ok_for_simple_move_strict_operands): Likewise.
18963 (ok_for_simple_arith_logic_operands): Likewise.
18964 (visium_function_value_1): Likewise.
18965 (rtx_ok_for_offset_p): Likewise.
18966 (visium_legitimize_reload_address): Likewise.
18967 (visium_split_double_move): Likewise.
18968 (visium_expand_copysign): Likewise.
18969 (visium_expand_int_cstore): Likewise.
18970 (visium_expand_fp_cstore): Likewise.
18971 (visium_split_cstore): Likewise.
18972 (visium_select_cc_mode): Likewise.
18973 (visium_split_cbranch): Likewise.
18974 (output_cbranch): Likewise.
18975 (visium_print_operand_address): Likewise.
18976 * expmed.c (flip_storage_order): Likewise.
18977 * expmed.h (emit_cstore): Likewise.
18978 (flip_storage_order): Likewise.
18979 * genrecog.c (validate_pattern): Likewise.
18980 * hsa-gen.c (gen_hsa_addr): Likewise.
18981 * internal-fn.c (expand_arith_overflow): Likewise.
18982 * ira-color.c (allocno_copy_cost_saving): Likewise.
18983 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18984 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18985 (process_invariant_for_inheritance): Likewise.
18986 * lra-eliminations.c (move_plus_up): Likewise.
18987 * omp-low.c (lower_oacc_reductions): Likewise.
18988 * simplify-rtx.c (simplify_subreg): Likewise.
18989 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18990 (TARGET_CHKP_BOUND_MODE): Likewise..
18991 * targhooks.c (default_chkp_bound_mode): Likewise.
18992 (default_setup_incoming_vararg_bounds): Likewise.
18993 * targhooks.h (default_chkp_bound_mode): Likewise.
18994 (default_setup_incoming_vararg_bounds): Likewise.
18995 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18996 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18997 (have_whole_vector_shift): Likewise.
18998 * tree-vect-stmts.c (vectorizable_load): Likewise.
18999 * doc/tm.texi: Regenerate.
19001 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19003 Graceful degrade if Binutils PR21472 is not available.
19006 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19007 .rodata in flash test fails.
19008 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19009 * confgure: Regenerate.
19010 * config.in: Regenerate.
19011 * config/avr/avr.c (avr_asm_named_section)
19012 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19013 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19014 (avr_asm_init_sections): Same.
19016 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19018 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19019 (fma<VH:mode>4_intrinsic): Likewise.
19020 (*fmsub<VCVTF:mode>4): Likewise.
19021 (*fmsub<VH:mode>4_intrinsic): Likewise.
19023 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19026 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19027 Don't depend on "optimize > 0".
19028 (out_movhi_r_mr, out_movqi_mr_r): Same.
19029 (out_movhi_mr_r, out_movqi_r_mr): Same.
19030 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19031 io_address_operand on "optimize > 0".
19033 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19035 * tree-loop-distribution.c: Add general explanantion on the pass.
19036 (generate_loops_for_partition): Mark distributed loop.
19037 (pg_add_dependence_edges): New parameter. Handle alias data
19038 dependence specially and record it in the parameter if asked.
19039 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19040 (init_partition_graph_vertices, add_partition_graph_edge): New.
19041 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19042 (free_partition_graph_vdata, build_partition_graph): New.
19043 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19044 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19045 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19046 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19047 (version_for_distribution_p, finalize_partitions): New.
19048 (distribute_loop): Handle alias data dependence specially. Factor
19049 out loop fusion code as functions and call these functions.
19051 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19053 * tree-loop-distribution.c (classify_partition): New parameter and
19054 better handle reduction statement.
19055 (rdg_build_partitions): Revise comment.
19056 (distribute_loop): Compute statements in all partitions and pass it
19057 to classify_partition.
19059 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19061 * tree-loop-distribution.c (enum partition_type): New.
19062 (struct partition): New field type.
19063 (partition_merge_into): Add parameter. Update partition type.
19064 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19065 (build_rdg_partition_for_vertex): Compute partition type.
19066 (rdg_build_partitions): Dump partition type.
19067 (distribute_loop): Update calls to partition_merge_into.
19069 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19071 * tree-loop-distribution.c (struct ddr_hasher): New.
19072 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19074 (classify_partition): Call get_data_dependence.
19075 (pg_add_dependence_edges): Ditto.
19076 (distribute_loop): Release data dependence hash table.
19078 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19080 * tree-loop-distribution.c (ref_base_address): Delete.
19081 (similar_memory_accesses): Rename ...
19082 (share_memory_accesses): ... to this. Check if partitions access
19083 the same memory reference.
19084 (distribute_loop): Call share_memory_accesses.
19086 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19088 * tree-loop-distribution.c (struct partition): New field recording
19089 its data reference.
19090 (partition_alloc, partition_free): Init and release data refs.
19091 (partition_merge_into): Merge data refs.
19092 (build_rdg_partition_for_vertex): Collect data refs for partition.
19093 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19095 (distribute_loop): Remve data refs from vertice data of partition
19098 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19100 * tree-loop-distribution.c (params.h): Include header file.
19101 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19102 (datarefs_vec): New global var.
19103 (create_rdg_vertices): Use datarefs_vec directly.
19104 (free_rdg): Don't free data references.
19105 (build_rdg): Update use. Don't free data references.
19106 (distribute_loop): Compute global variable for data references.
19107 Bail out if there are too many data references.
19109 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19111 * tree-loop-distribution.c (loop_nest): New global var.
19112 (build_rdg): Use loop directly, rather than loop nest.
19113 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19115 (distribute_loop): Compute global variable loop nest. Update use.
19117 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19119 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19120 (partition_merge_into): New parameter. Dump reason for fusion.
19121 (distribute_loop): Update use of partition_merge_into.
19123 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19125 * tree-loop-distribution.c (bb_top_order_index): New.
19126 (bb_top_order_index_size, bb_top_order_cmp): New.
19127 (stmts_from_loop): Use topological order.
19128 (pass_loop_distribution::execute): Compute and release topological
19129 order for basic blocks.
19131 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19133 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19136 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19138 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19139 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19140 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19141 * internal-fn.def (LOOP_DIST_ALIAS): New.
19142 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19143 (fold_loop_internal_call): ... this.
19144 (vect_loop_dist_alias_call): New function.
19145 (set_uid_loop_bbs): Call fold_loop_internal_call.
19146 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19149 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19152 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19153 Require dead FLAGS_REG at the beginning of a peephole.
19155 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19158 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19159 arguments in the call to __builtin_ia32_sbb_u32.
19160 (_subborrow_u64): Swap _X and _Y arguments in the call to
19161 __builtin_ia32_sbb_u64.
19163 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19166 * tree-vrp.c (compare_assert_loc): Turn into a function template
19167 with stable template parameter. Only test if a->e is NULL,
19168 !a->e == !b->e has been verified already. Use e == NULL or
19169 e != NULL instead of e or ! e tests. If stable is true, don't use
19170 iterative_hash_expr, on the other side allow a or b or both NULL
19171 and sort the NULLs last.
19172 (process_assert_insertions): Sort using compare_assert_loc<false>
19173 instead of compare_assert_loc, later sort using
19174 compare_assert_loc<true> before calling process_assert_insertions_for
19175 in a loop. Use break instead of continue once seen NULL pointer.
19177 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19179 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19180 Cortex-R7 and Cortex-R8 processors.
19182 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19184 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19185 uninitialized while src is not.
19187 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19189 * common/config/arm/arm-common.c: Adjust include path for
19191 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19192 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19193 (arm-cpu-data.h): Likewise.
19194 (arm-cpu-cdata.h): Likewise.
19195 * config/arm/arm-cpu.h: Delete.
19196 * config/arm/arm-cpu-cdata.h: Delete.
19197 * config/arm/arm-cpu-data.h: Delete.
19199 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19201 * config/arm/arm-cpus.in (cortex-a55): New.
19202 (cortex-a75): Likewise.
19203 (cortex-a75.cortex-a55): Likewise.
19204 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19206 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19207 * config/arm/arm-cpu-cdata.h: Regenerate.
19208 * config/arm/arm-cpu-data.h: Regenerate.
19209 * config/arm/arm-cpu.h: Regenerate.
19210 * config/arm/arm-tables.opt: Regenerate.
19211 * config/arm/arm-tune.md: Regenerate.
19213 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19215 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19217 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19219 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19222 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19224 PR tree-optimization/81292
19225 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19226 full_string_p, also call adjust_related_strinfos if the adjustment
19227 is simple, otherwise invalidate related strinfos.
19229 2017-07-04 Martin Liska <mliska@suse.cz>
19232 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19233 newly created variable as DECL_IGNORED_P.
19235 2017-07-04 Martin Liska <mliska@suse.cz>
19238 * ipa-inline.c (inline_small_functions):
19239 Use xstrdup_for_dump.
19241 2017-07-04 Tom de Vries <tom@codesourcery.com>
19243 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19245 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19248 * config/darwin.c (darwin_function_switched_text_sections):
19251 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19253 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19255 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19257 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19259 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19261 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19262 min_profitable_iters, and th as inclusive lower bounds.
19263 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19264 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19265 for min_profitable_iters and min_profitable_estimate.
19266 (vect_transform_loop): Treat th as an inclusive lower bound.
19267 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19269 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19272 * config/darwin.c (darwin_function_switched_text_sections):
19273 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19274 in two pieces, and suppress the use of buf.
19276 2017-07-03 Nathan Sidwell <nathan@acm.org>
19278 * hash-table.h (hash_table_mod1): Fix indentation.
19280 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19282 PR middle-end/81290
19283 * predict.c (force_edge_cold): Be more careful about propagation
19285 * profile-count.h (profile_probability::guessed,
19286 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19288 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19290 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19292 * doc/invoke.texi (rcpc architecture extension): Document it.
19294 2017-07-03 Richard Biener <rguenther@suse.de>
19296 PR tree-optimization/60510
19297 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19298 the scalar reduction PHI and use it.
19299 (vectorizable_reduction): Properly guard the single_defuse_cycle
19300 path for non-SLP reduction chains where we cannot use it.
19301 Rework reduc_def/index and vector type deduction. Rework
19302 vector operand gathering during reduction op code-gen.
19303 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19304 chains dissolve the chain and leave it to non-SLP reduction
19307 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19309 * tree-data-ref.h (dr_alignment): Declare.
19310 * tree-data-ref.c (dr_alignment): New function.
19311 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19312 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19314 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19316 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19318 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19319 and base_misalignment fields.
19320 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19321 * tree-data-ref.c: Include builtins.h.
19322 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19323 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19324 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19325 * tree-vect-data-refs.c: Include tree-cfg.h.
19326 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19327 fields instead of calculating an alignment here.
19328 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19329 innermost_loop_behavior fields.
19331 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19333 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19335 (DR_STEP_ALIGNMENT): New macro.
19336 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19337 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19338 (create_data_ref): Print it.
19339 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19340 to tell whether the step preserves vector (mis)alignment.
19341 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19342 Move the check for an integer step and generalise to all INTEGER_CST.
19343 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19344 Print the outer step alignment.
19346 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19348 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19349 with offset_alignment.
19350 (DR_ALIGNED_TO): Delete.
19351 (DR_OFFSET_ALIGNMENT): New macro.
19352 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19353 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19354 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19355 (create_data_ref): Likewise.
19356 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19357 (vect_analyze_data_refs): Likewise.
19358 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19359 creating dummy innermost behavior.
19361 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19363 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19364 with a "innermost_loop_behavior *" and refeence tree.
19365 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19366 (create_data_ref): Update call accordingly.
19367 * tree-predcom.c (find_looparound_phi): Likewise.
19369 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19371 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19372 fields with dr_wrt_vec_loop.
19373 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19374 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19375 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19376 (vect_dr_behavior): New function.
19377 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19378 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19379 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19380 track whether the step preserves the misalignment.
19381 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19382 Use vect_dr_behavior.
19383 (vect_setup_realignment): Update call accordingly.
19384 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19385 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19386 call to vect_create_addr_base_for_vector_ref.
19387 (vect_create_cond_for_align_checks): Likewise.
19388 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19389 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19390 (vect_recog_mask_conversion_pattern): Likewise.
19391 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19392 (new_stmt_vec_info): Remove redundant zeroing.
19394 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19396 * common/config/arm/arm-common.c (arm_be8_option): New function.
19397 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19398 (ISA_ARMv6): Add isa_bit_be8.
19399 * config/arm/arm.h (arm_be8_option): Add prototype.
19400 (BE8_SPEC_FUNCTION): New define.
19401 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19402 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19403 (mlittle-endian): Similarly.
19404 (mbe8, mbe32): New options.
19405 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19406 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19408 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19410 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19412 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19414 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19415 (cleanup_tree_cfg_bb): Use it.
19416 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19418 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19420 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19423 * loop-doloop.c (add_test): Update profile.
19425 2017-07-03 Martin Liska <mliska@suse.cz>
19428 * sanopt.c (rewrite_usage_of_param): New function.
19429 (sanitize_rewrite_addressable_params): Likewise.
19430 (pass_sanopt::execute): Call rewrite_usage_of_param.
19432 2017-07-03 Richard Biener <rguenther@suse.de>
19434 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19435 back to using VIEW_CONVERT_EXPR.
19437 2017-07-03 Martin Liska <mliska@suse.cz>
19440 * doc/extend.texi: Document when a resolver function is
19441 generated for target_clones.
19443 2017-07-03 Martin Liska <mliska@suse.cz>
19445 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19446 and shadow memory of auto variables which are subject of
19447 use-after-scope sanitization.
19448 (asan_expand_mark_ifn): Add do set only when is_poison.
19450 2016-07-03 Richard Biener <rguenther@suse.de>
19452 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19454 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19455 (vectorizable_reduction): Perform reduction PHI creation when
19456 visiting a reduction PHI and adjust and simplify code generation
19457 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19458 (vect_transform_loop): Visit reduction PHIs.
19459 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19460 defs into the SLP tree.
19461 (vect_build_slp_tree): Reduction defs terminate the recursion.
19462 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19464 (vect_get_vec_defs_for_stmt_copy): Export.
19465 (vect_get_vec_defs): Likewise.
19466 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19468 (vect_get_vec_defs_for_stmt_copy): Declare.
19469 (vect_get_vec_defs): Likewise.
19471 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19473 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19474 parameter with a "loop" parameter and use it instead of the
19475 loop containing DR_STMT. Don't check simple_iv when doing
19476 BB analysis. Describe the two analysis modes in the comment.
19478 2017-07-03 Tom de Vries <tom@codesourcery.com>
19480 PR tree-optimization/69468
19481 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19482 (find_same_succ_bb): Handle ignore_edge_flags.
19484 2017-07-03 Tom de Vries <tom@codesourcery.com>
19486 PR tree-optimization/81192
19487 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19489 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19491 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19493 2017-07-03 Tom de Vries <tom@codesourcery.com>
19495 PR tree-optimization/81192
19496 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19497 BB_SAME_SUCC (bb) == NULL.
19499 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19501 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19504 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19506 * dumpfile.c: Include profile-count.h
19507 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19509 (insert_cond_bb): Update profile.
19510 * tree-cfg.h (insert_cond_bb): Update prototype.
19511 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19512 * tree-dump.c: Do not include tree-cfg.
19514 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19516 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19518 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19520 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19523 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19525 * tree-complex.c (expand_complex_div_wide): update profile.
19527 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19528 Alan Hayward <alan.hayward@arm.com>
19529 David Sherwood <david.sherwood@arm.com>
19531 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19532 (CORETYPES_H): New define.
19533 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19534 (insn-modes-inline.h, s-modes-inline-h): New rules.
19535 (generated_files): Add insn-modes-inline.h.
19536 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19537 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19538 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19539 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19540 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19541 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19542 (build/gencodes.o, build/genconditions.o): Likewise.
19543 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19544 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19545 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19546 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19547 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19548 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19549 * coretypes.h: Include everything up to real.h for generators.
19550 Include insn-modes.h first. Include wide-int-print.h after
19551 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19552 * machmode.h: Don't include insn-modes.h here.
19553 * function-tests.c: Remove includes of signop.h, machmode.h,
19554 double-int.h and wide-int.h.
19556 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19558 * optc-save-gen.awk: Likewise.
19559 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19560 * godump.c: Remove include of wide-int-print.h.
19561 * pretty-print.h: Likewise.
19562 * wide-int-print.cc: Likewise.
19563 * wide-int.cc: Likewise.
19564 * hash-map-tests.c: Remove include of signop.h.
19565 * hash-set-tests.c: Likewise.
19566 * rtl-tests.c: Likewise.
19567 * mkconfig.sh: Remove include of machmode.h.
19568 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19570 (emit_insn_modes_inline_h): ...this new function. Emit the code
19571 into an insn-modes-inline.h header file, adding appropriate
19572 include guards and end comments.
19573 (emit_insn_modes_c_header): Remove include of machmode.h.
19574 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19576 (main): Handle -i flag and call emit_insn_modes_inline_h when
19579 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19581 * tree-ssa-strlen.c (strinfo): Rename the length field to
19582 nonzero_chars. Add a full_string_p field.
19583 (compare_nonzero_chars, zero_length_string_p): New functions.
19584 (get_addr_stridx): Add an offset_out parameter.
19585 Use compare_nonzero_chars.
19586 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19587 (new_strinfo): Update after above changes to strinfo.
19588 (set_endptr_and_length): Set full_string_p.
19589 (get_string_length): Update after above changes to strinfo.
19590 (unshare_strinfo): Update call to new_strinfo.
19591 (maybe_invalidate): Likewise.
19592 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19593 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19594 as a uhwi instead of an shwi. Update after above changes to
19595 strinfo and new_strinfo.
19596 (zero_length_string): Assert that chainsi contains full strings.
19597 Use zero_length_string_p. Update call to new_strinfo.
19598 (adjust_related_strinfos): Update after above changes to strinfo.
19599 Copy full_string_p from origsi.
19600 (adjust_last_stmt): Use zero_length_string_p.
19601 (handle_builtin_strlen): Update after above changes to strinfo and
19602 new_strinfo. Install the lhs as the string length if the previous
19603 entry didn't describe a full string.
19604 (handle_builtin_strchr): Update after above changes to strinfo
19606 (handle_builtin_strcpy): Likewise.
19607 (handle_builtin_strcat): Likewise.
19608 (handle_builtin_malloc): Likewise.
19609 (handle_pointer_plus): Likewise.
19610 (handle_builtin_memcpy): Likewise. Track nonzero characters
19611 that aren't necessarily followed by a nul terminator.
19612 (handle_char_store): Likewise.
19614 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19616 PR tree-optimization/80769
19617 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19618 for malloc and calloc. Document the new invariant that all related
19619 strinfos have delayed lengths or none do.
19620 (verify_related_strinfos): Move earlier in file.
19621 (set_endptr_and_length): New function, split out from...
19622 (get_string_length): ...here. Also set the lengths of related
19624 (zero_length_string): Assert that chainsi has known (rather than
19626 (adjust_related_strinfos): Likewise.
19628 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19630 PR tree-optimization/81136
19631 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19632 assert that two references with the same misalignment have the same
19633 compile-time misalignment if those compile-time misalignments
19636 2017-07-01 Andi Kleen <ak@linux.intel.com>
19638 * print-tree.c (print_node): Print all attributes.
19640 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19642 * cfg.c (scale_bbs_frequencies): New function.
19643 * cfg.h (scale_bbs_frequencies): Declare it.
19644 * cfgloopanal.c (single_likely_exit): Cleanup.
19645 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19647 (scale_loop_profile): Likewise.
19648 (loop_version): Likewise.
19649 (create_empty_loop_on_edge): Update.
19650 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19651 scale_loop_frequencies, scale_loop_profile, loopify,
19652 loop_version): Update prototypes.
19653 * modulo-sched.c (sms_schedule): Update.
19654 * predict.c (unlikely_executed_edge_p): Also check probability.
19655 (probably_never_executed_edge_p): Fix typo.
19656 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19657 * tree-parloops.c (gen_parallel_loop): Update.
19658 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19659 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19660 * tree-ssa-loop-split.c (split_loop): Update.
19661 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19662 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19663 (vect_loop_versioning): Update.
19664 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19666 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19668 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19670 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19672 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19673 to keep profile consistent.
19675 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19677 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19678 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19679 * profile-count.h (max_safe_multiplier): Make unsigned.
19680 (profile_count::guessed_zero): New.
19682 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19684 * bb-reorder.c (fix_up_crossing_landing_pad,
19685 fix_crossing_conditional_branches): Use make_single_succ_edge
19686 to keep profile consistent.
19688 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19690 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19693 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19696 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19697 the right scopes, make sure cond_jump isn't preserved between multiple
19698 iterations. Search for fallthru edge whenever there are 3+ edges and
19699 use find_fallthru_edge for it.
19701 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19703 Patch by Alexander Monakov <amonakov@ispras.ru>
19704 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19705 probabilities consistently.
19707 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19709 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19710 emit_cmp_and_jump_insns.
19712 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19715 * tree-inline.c (expand_call_inline): Combine profile statuses.
19717 2017-06-30 Andrew Pinski <apinski@cavium.com>
19719 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19720 fold_stmt returned true.
19722 2017-06-30 Nathan Sidwell <nathan@acm.org>
19724 * ggc.h (empty_string): Delete.
19725 * cfgexpand.c (expand_asm_stmt): Use plain "".
19726 * optabs.c (expand_asm_memory_barrier): Likewise.
19727 * stringpool.c (empty_string): Delete.
19728 (digit_vector, digit_string): Delete.
19729 (ggc_alloc_string): Use plain "", don't optimize single digit
19730 strings. Use ggc_alloc_atomic.
19732 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19734 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19735 comparison set and one other set, use the cost of the non-comparison
19738 2017-06-30 Nathan Sidwell <nathan@acm.org>
19740 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19743 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19745 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19746 loops. Remove now unneeded calls to gimple_switch_set_label() that
19747 just set removed labels to NULL_TREE.
19749 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19751 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19752 (set_range_info): ...here. Only call set_range_info_raw if domain
19754 (set_nonzero_bits): Call set_range_info_raw.
19755 * tree-ssanames.h (set_range_info_raw): New.
19757 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19760 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19761 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19762 of nonimmediate_operand and <store_mask_constraint> instead of m
19763 for the input operand. For V8FI iterator, always split if input
19764 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19765 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19766 <store_mask_predicate> instead of register_operand and
19767 <store_mask_constraint> instead of v for the input operand. Make
19768 sure both operands aren't MEMs for if not <mask_applied>.
19770 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19772 * lto-wrapper.c (copy_file) Close both file descriptors before
19775 2017-06-30 Martin Liska <mliska@suse.cz>
19778 * multiple_target.c (create_dispatcher_calls): Make ifunc
19779 also for function that don't have calls or are not referenced.
19781 2017-06-30 Richard Biener <rguenther@suse.de>
19783 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19784 analyze the first scalar stmt. Move vector type computation
19785 for the BB case here from ...
19786 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19787 live operation processing in the SLP case properly.
19789 2017-06-30 Richard Biener <rguenther@suse.de>
19791 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19793 2017-06-30 Martin Liska <mliska@suse.cz>
19796 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19797 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19799 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19801 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19804 2017-06-30 Richard Biener <rguenther@suse.de>
19806 PR tree-optimization/81249
19807 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19808 condition reduction result to original scalar type.
19810 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19812 * profile-count.h (enum profile_quality): Fix typos and whitespace
19815 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19817 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19818 type for branch probabilities.
19820 2017-06-29 Julian Brown <julian@codesourcery.com>
19821 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19823 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19824 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19825 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19826 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19828 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19830 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19831 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19832 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19833 CC usage from generic code to here.
19834 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19835 CC usage into the target macros.
19837 2017-06-29 Maya Rashish <coypu@sdf.org>
19839 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19842 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19844 * arm/arm-builtins.c: Include profile-count.h
19845 * except.c (sjlj_emit_function_enter): Use
19846 profile_probability::unlikely.
19848 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19850 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19851 and tocrel_offset be pointer args rather than implicitly using
19853 (legitimate_constant_pool_address_p, rs6000_emit_move,
19854 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19855 tocrel_offset and use in toc_relative_expr_p call.
19856 (print_operand, print_operand_address): Use static tocrel_base_oac
19857 and tocrel_offset_oac.
19858 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19861 2017-06-29 Maya Rashish <coypu@sdf.org>
19863 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19865 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19867 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19868 objects, take into account only the alignment of 'op0' and 'mode1' if
19871 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19873 * ccmp.c (ccmp_tree_comparison_p): New function.
19874 (ccmp_candidate_p): Update to use above function.
19875 (get_compare_parts): New function.
19876 (expand_ccmp_next): Update to use new functions.
19877 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19879 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19880 take mode as argument.
19881 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19882 * expr.c (expand_expr_real_1): Pass mode as argument.
19884 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19886 * combine.c (combine_instructions): Print insns to dump_file, together
19889 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19891 * asan.c (asan_emit_stack_protection): Update.
19892 (create_cond_insert_point): Update.
19893 * auto-profile.c (afdo_propagate_circuit): Update.
19894 * basic-block.h (struct edge_def): Turn probability to
19895 profile_probability.
19896 (EDGE_FREQUENCY): Update.
19897 * bb-reorder.c (find_traces_1_round): Update.
19898 (better_edge_p): Update.
19899 (sanitize_hot_paths): Update.
19900 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19901 (make_single_succ_edge): Update.
19902 (check_bb_profile): Update.
19903 (dump_edge_info): Update.
19904 (update_bb_profile_for_threading): Update.
19905 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19906 probabilitycount to 0.
19907 * cfgbuild.c (compute_outgoing_frequencies): Update.
19908 * cfgcleanup.c (try_forward_edges): Update.
19909 (outgoing_edges_match): Update.
19910 (try_crossjump_to_edge): Update.
19911 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19912 (expand_gimple_tailcall): Update.
19913 (construct_init_block): Use make_single_succ_edge.
19914 (construct_exit_block): Use make_single_succ_edge.
19915 * cfghooks.c (verify_flow_info): Update.
19916 (redirect_edge_succ_nodup): Update.
19917 (split_edge): Update.
19918 (account_profile_record): Update.
19919 * cfgloopanal.c (single_likely_exit): Update.
19920 * cfgloopmanip.c (scale_loop_profile): Update.
19921 (set_zero_probability): Remove.
19922 (duplicate_loop_to_header_edge): Update.
19923 * cfgloopmanip.h (loop_version): Update prototype.
19924 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19925 (force_nonfallthru_and_redirect): Update.
19926 (update_br_prob_note): Update.
19927 (rtl_verify_edges): Update.
19928 (purge_dead_edges): Update.
19929 (rtl_lv_add_condition_to_bb): Update.
19930 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19931 * cgraphunit.c (init_lowered_empty_function): Update.
19932 (cgraph_node::expand_thunk): Update.
19933 * cilk-common.c: Include profile-count.h
19934 * dojump.c (inv): Remove.
19935 (jumpifnot): Update.
19936 (jumpifnot_1): Update.
19937 (do_jump_1): Update.
19939 (do_jump_by_parts_greater_rtx): Update.
19940 (do_compare_rtx_and_jump): Update.
19941 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19942 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19943 * dwarf2cfi.c: Include profile-count.h
19944 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19945 (sjlj_emit_dispatch_table): Likewise.
19946 * explow.c: Include profile-count.h
19947 * expmed.c (emit_store_flag_force): Update.
19948 (do_cmp_and_jump): Update.
19949 * expr.c (compare_by_pieces_d::generate): Update.
19950 (compare_by_pieces_d::finish_mode): Update.
19951 (emit_block_move_via_loop): Update.
19952 (store_expr_with_bounds): Update.
19953 (store_constructor): Update.
19954 (expand_expr_real_2): Update.
19955 (expand_expr_real_1): Update.
19956 * expr.h (try_casesi, try_tablejump): Update prototypes.
19957 * gimple-pretty-print.c (dump_probability): Update.
19958 (dump_profile): New.
19959 (dump_gimple_label): Update.
19960 (dump_gimple_bb_header): Update.
19961 * graph.c (draw_cfg_node_succ_edges): Update.
19962 * hsa-gen.c (convert_switch_statements): Update.
19963 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19964 (find_if_case_1): Update.
19965 (find_if_case_2): Update.
19966 * internal-fn.c (expand_arith_overflow_result_store): Update.
19967 (expand_addsub_overflow): Update.
19968 (expand_neg_overflow): Update.
19969 (expand_mul_overflow): Update.
19970 (expand_vector_ubsan_overflow): Update.
19971 * ipa-cp.c (good_cloning_opportunity_p): Update.
19972 * ipa-split.c (split_function): Use make_single_succ_edge.
19973 * ipa-utils.c (ipa_merge_profiles): Update.
19974 * loop-doloop.c (add_test): Update.
19975 (doloop_modify): Update.
19976 * loop-unroll.c (compare_and_jump_seq): Update.
19977 (unroll_loop_runtime_iterations): Update.
19978 * lra-constraints.c (lra_inheritance): Update.
19979 * lto-streamer-in.c (input_cfg): Update.
19980 * lto-streamer-out.c (output_cfg): Update.
19981 * mcf.c (adjust_cfg_counts): Update.
19982 * modulo-sched.c (sms_schedule): Update.
19983 * omp-expand.c (expand_omp_for_init_counts): Update.
19984 (extract_omp_for_update_vars): Update.
19985 (expand_omp_ordered_sink): Update.
19986 (expand_omp_for_ordered_loops): Update.
19987 (expand_omp_for_generic): Update.
19988 (expand_omp_for_static_nochunk): Update.
19989 (expand_omp_for_static_chunk): Update.
19990 (expand_cilk_for): Update.
19991 (expand_omp_simd): Update.
19992 (expand_omp_taskloop_for_outer): Update.
19993 (expand_omp_taskloop_for_inner): Update.
19994 * omp-simd-clone.c (simd_clone_adjust): Update.
19995 * optabs.c (expand_doubleword_shift): Update.
19996 (expand_abs): Update.
19997 (emit_cmp_and_jump_insn_1): Update.
19998 (expand_compare_and_swap_loop): Update.
19999 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20000 * predict.c (predictable_edge_p): Update.
20001 (edge_probability_reliable_p): Update.
20002 (set_even_probabilities): Update.
20003 (combine_predictions_for_insn): Update.
20004 (combine_predictions_for_bb): Update.
20005 (propagate_freq): Update.
20006 (estimate_bb_frequencies): Update.
20007 (force_edge_cold): Update.
20008 * profile-count.c (profile_count::dump): Add missing space into dump.
20009 (profile_count::debug): Add newline.
20010 (profile_count::differs_from_p): Explicitly convert to unsigned.
20011 (profile_count::stream_in): Update.
20012 (profile_probability::dump): New member function.
20013 (profile_probability::debug): New member function.
20014 (profile_probability::differs_from_p): New member function.
20015 (profile_probability::differs_lot_from_p): New member function.
20016 (profile_probability::stream_in): New member function.
20017 (profile_probability::stream_out): New member function.
20018 * profile-count.h (profile_count_quality): Rename to ...
20019 (profile_quality): ... this one.
20020 (profile_probability): New.
20021 (profile_count): Update.
20022 * profile.c (compute_branch_probabilities): Update.
20023 * recog.c (peep2_attempt): Update.
20024 * sched-ebb.c (schedule_ebbs): Update.
20025 * sched-rgn.c (find_single_block_region): Update.
20026 (compute_dom_prob_ps): Update.
20027 (schedule_region): Update.
20028 * sel-sched-ir.c (compute_succs_info): Update.
20029 * stmt.c (struct case_node): Update.
20030 (do_jump_if_equal): Update.
20031 (get_outgoing_edge_probs): Update.
20032 (conditional_probability): Update.
20033 (emit_case_dispatch_table): Update.
20034 (expand_case): Update.
20035 (expand_sjlj_dispatch_table): Update.
20036 (emit_case_nodes): Update.
20037 * targhooks.c: Update.
20038 * tracer.c (better_p): Update.
20039 (find_best_successor): Update.
20040 * trans-mem.c (expand_transaction): Update.
20041 * tree-call-cdce.c: Update.
20042 * tree-cfg.c (gimple_split_edge): Upate.
20043 (move_sese_region_to_fn): Upate.
20044 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20045 * tree-eh.c (lower_resx): Upate.
20046 (cleanup_empty_eh_move_lp): Upate.
20047 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20048 * tree-inline.c (copy_edges_for_bb): Update.
20049 (copy_cfg_body): Update.
20050 * tree-parloops.c (gen_parallel_loop): Update.
20051 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20052 (gimple_gen_time_profiler): Update.
20053 * tree-ssa-dce.c (remove_dead_stmt): Update.
20054 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20055 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20056 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20057 (unloop_loops): Update.
20058 (try_peel_loop): Update.
20059 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20060 * tree-ssa-loop-split.c (connect_loops): Update.
20061 (split_loop): Update.
20062 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20063 (hoist_guard): Update.
20064 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20065 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20066 (value_replacement): Update.
20067 * tree-ssa-reassoc.c (branch_fixup): Update.
20068 * tree-ssa-tail-merge.c (replace_block_by): Update.
20069 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20070 (create_edge_and_update_destination_phis): Update.
20071 (compute_path_counts): Update.
20072 (recompute_probabilities): Update.
20073 (update_joiner_offpath_counts): Update.
20074 (freqs_to_counts_path): Update.
20075 (duplicate_thread_path): Update.
20076 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20077 (struct switch_conv_info): Update.
20078 (gen_inbound_check): Update.
20079 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20080 (vect_do_peeling): Update.
20081 (vect_loop_versioning): Update.
20082 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20083 (optimize_mask_stores): Update.
20084 * ubsan.c (ubsan_expand_null_ifn): Update.
20085 * value-prof.c (gimple_divmod_fixed_value): Update.
20086 (gimple_divmod_fixed_value_transform): Update.
20087 (gimple_mod_pow2): Update.
20088 (gimple_mod_pow2_value_transform): Update.
20089 (gimple_mod_subtract): Update.
20090 (gimple_mod_subtract_transform): Update.
20091 (gimple_ic): Update.
20092 (gimple_stringop_fixed_value): Update.
20093 (gimple_stringops_transform): Update.
20094 * value-prof.h: Update.
20096 2017-06-29 Carl Love <cel@us.ibm.com>
20098 * config/rs6000/rs6000-c.c: Add support for built-in functions
20099 vector signed int vec_signed (vector float);
20100 vector signed long long vec_signed (vector double);
20101 vector signed int vec_signed2 (vector double, vector double);
20102 vector signed int vec_signede (vector double);
20103 vector signed int vec_signedo (vector double);
20104 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20105 instruction generator.
20106 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20107 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20108 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20110 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20111 vunsignede_v2df): Add define_expands.
20112 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20113 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20114 VEC_UNSIGNEDO): Add definitions.
20115 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20116 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20117 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20118 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20119 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20120 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20121 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20122 vec_unsignede, vec_unsignedo): Add builtin defines.
20123 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20125 * doc/extend.texi: Update the built-in documentation file for the
20126 new built-in functions.
20128 2017-06-29 Richard Biener <rguenther@suse.de>
20130 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20131 reduction chains to LOOP_VINFO_REDUCTIONS.
20132 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20133 SLP reductions after processing reduction chains.
20135 2017-06-29 Nathan Sidwell <nathan@acm.org>
20137 * builtins.c (fold_builtin_FUNCTION): Use
20138 lang_hooks.decl_printable_name.
20140 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20142 PR middle-end/81194
20143 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20144 with only one label.
20145 * stmt.c (expand_case): Assert NCASES is greater than one.
20147 2017-06-29 Richard Biener <rguenther@suse.de>
20149 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20151 (group_case_labels): Likewise.
20152 (find_taken_edge): Push sanity checking on val to workers...
20153 (find_taken_edge_cond_expr): ... here
20154 (find_taken_edge_switch_expr): ... and here, handle cases
20155 with just a default label.
20156 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20157 (group_case_labels): Likewise.
20158 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20159 group_case_labels does anything cleanup the CFG again.
20161 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20163 PR tree-optimization/81196
20164 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20165 exit condition comparing two IVs.
20167 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20169 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20170 profile to the dummy entry at the end of the list of architectures.
20171 * config/arm/arm-cpu-cdata.h: Regenerated.
20173 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20174 Michael Collison <michael.collison@arm.com>
20177 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20179 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20180 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20181 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20182 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20183 of shift when the shift amount is masked with constant equal to
20184 the size of the mode.
20185 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20188 2017-06-29 Martin Liska <mliska@suse.cz>
20190 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20192 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20195 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20198 (edge_growth_cache_entry::edge_growth_cache_entry): New
20200 (reset_edge_growth_cache): Update to use constructor.
20202 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20204 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20205 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20206 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20208 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20210 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20211 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20213 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20215 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20216 (*-linux-uclibc*): Add t-uclibc tmake_file.
20217 * config/t-musl: New.
20218 * config/t-uclibc: New.
20220 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20222 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20224 (gen_comm_data): Emit architectural setting of arch_prof.
20225 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20227 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20228 (armv8-m.base, armv8-m.main): Likewise.
20229 * arm-protos.h (arm_build_target): Add profile field.
20230 (arch_option): Likewise.
20231 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20233 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20234 arm_active_target.profile.
20236 2017-06-28 Richard Biener <rguenther@suse.de>
20238 PR middle-end/81227
20239 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20240 TYPE_OVERFLOW_WRAPS.
20241 * match.pd (negate_expr_p): Likewise.
20242 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20243 fold_build2, not fold_binary.
20245 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20247 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20248 Convert memory address to Pmode.
20249 (aarch64_print_operand): Assert MEM operands are always Pmode.
20251 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20254 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20255 Remove redundant if.
20256 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20257 * config/arm/aarch-common-protos.h
20258 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20259 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20261 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20264 * multiple_target.c (create_dispatcher_calls): Set the default
20265 clone to be static, not public.
20267 2017-06-28 Richard Biener <rguenther@suse.de>
20269 * tree-vect-loop.c (vectorizable_reduction): Move special
20270 cond reduction IV var creation ...
20271 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20272 parameter. Use STMT_VINFO_VECTYPE.
20273 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20276 2017-06-28 Martin Liska <mliska@suse.cz>
20279 * ipa-visibility.c (non_local_p): Handle visibility.
20281 2017-06-28 Martin Liska <mliska@suse.cz>
20284 * common.opt: Add IntegerRange to various options.
20285 * opt-functions.awk (integer_range_info): New function.
20286 * optc-gen.awk: Add integer_range_info to cl_options struct.
20287 * opts-common.c (decode_cmdline_option): Handle
20288 CL_ERR_INT_RANGE_ARG.
20289 (cmdline_handle_error): Likewise.
20290 * opts.c (print_filtered_help): Show valid interval in
20291 when --help is provided.
20292 * opts.h (struct cl_option): Add range_min and range_max fields.
20293 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20295 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20297 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20298 (x * C EQ/NE y * C): New transformation.
20300 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20302 * genmultilib (combination_space): Accept '+' in option names.
20304 2017-06-28 Martin Liska <mliska@suse.cz>
20307 * asan.c (instrument_derefs): Bail out inner references
20308 that are hard register variables.
20310 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20313 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20314 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20316 2017-06-28 Richard Biener <rguenther@suse.de>
20318 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20319 (vect_get_slp_defs): Adjust.
20320 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20322 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20324 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20325 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20326 (vectorizable_reduction): Adjust.
20327 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20329 (vect_get_slp_defs): Likewise.
20330 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20331 (vectorizable_bswap): Adjust.
20332 (vectorizable_call): Likewise.
20333 (vectorizable_conversion): Likewise.
20334 (vectorizable_assignment): Likewise.
20335 (vectorizable_shift): Likewise.
20336 (vectorizable_operation): Likewise.
20337 (vectorizable_store): Likewise.
20338 (vectorizable_condition): Likewise.
20339 (vectorizable_comparison): Likewise.
20341 2017-06-28 Michael Collison <michael.collison@arm.com>
20344 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20346 (arm_gen_movmemqi): Removed unused variable 'i'.
20347 Convert 'for' loop into 'while' loop.
20348 (arm_expand_prologue): Remove last unnecessary set of insn.
20349 (thumb_pop): Remove unused variable 'pushed_words'.
20350 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20352 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20354 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20355 * config/s390/s390-protos.h: Add prototype of
20356 s390_rel_address_ok_p.
20357 * config/s390/s390.c (s390_got_symbol): New function.
20358 (s390_rel_address_ok_p): New function.
20359 (legitimize_pic_address): Use s390_rel_address_ok_p.
20360 (s390_load_got): Use s390_got_symbol.
20361 (s390_option_override): Issue error if
20362 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20363 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20365 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20367 2017-06-27 Andrew Pinski <apinski@cavium.com>
20369 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20370 (X * copysign (1.0, X)): New pattern.
20371 (X * copysign (1.0, -X)): New pattern.
20372 (copysign (-1.0, CST)): New pattern.
20374 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20376 * genmultilib (combination_space): Remove variable.
20377 Validate reuse rules against regular expression for any sequence
20378 of multilib options in any order.
20380 2017-06-27 Michael Collison <michael.collison@arm.com>
20382 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20383 call aarch64_split_simd_combine.
20384 * (aarch64_combine_internal<mode>): Delete pattern.
20385 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20386 Allow register and subreg operands.
20388 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20390 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20391 specific need, just fallback on defaults.
20392 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20394 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20395 Olivier Hainque <hainque@adacore.com>
20397 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20399 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20400 targets. Pick a default if no particular attempt applied.
20401 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20404 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20406 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20407 (x86_64-wrs-vxworks7): Likewise.
20409 2017-06-27 Marek Polacek <polacek@redhat.com>
20412 * ubsan.c (instrument_null): Check get_base_address's result for null.
20414 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20416 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20418 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20420 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20421 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20422 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20423 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20424 New function types.
20425 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20426 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20427 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20428 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20429 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20430 BUILT_IN_FEUPDATEENV): New builtins.
20431 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20432 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20433 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20434 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20436 (builtin_structptr_types): Adjust size.
20437 * tree.c (builtin_structptr_types): Add four entries.
20439 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20440 Olivier Hainque <hainque@adacore.com>
20442 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20443 (TLS_SYM): New local macro, forcing reference to __tls__ on
20444 link command lines for VxWorks 7 RTPs, triggering initialization
20446 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20447 OS features TLS support, true for RTPs on VxWorks 7.
20448 * config/vxworks.c (vxworks_override_options): Setup emutls
20451 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20453 * predict.c (test_prediction_value_range): Use -1U instead of -1
20454 to avoid narrowing conversion warning.
20455 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20456 to avoid narrowing conversion warning.
20457 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20459 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20461 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20463 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20464 64bit configurations.
20465 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20466 (SIZE_TYPE): Likewise.
20467 * config/vxworks.c (vxworks_emutls_var_fields): Use
20468 long_unsigned_type_node instead of unsigned_type_node as the offset
20469 field type, which is "pointer" mode in emutls.c.
20471 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20474 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20476 PR middle-end/81207
20477 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20478 gimple_vuse copying separately from gimple_vdef copying.
20480 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20482 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20485 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20486 Olivier Hainque <hainque@adacore.com>
20488 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20489 for all vxworks7 targets.
20490 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20491 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20492 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20493 variations for VX6/VX7 and 32/64bits later on in ...
20494 (VXWORKS_LIB_SPEC): Leverage new macros.
20495 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20496 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20498 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20500 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20501 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20503 2017-06-26 Carl Love <cel@us.ibm.com>
20505 * config/rs6000/rs6000-c.c: Add support for built-in functions
20506 vector bool char vec_reve (vector bool char);
20507 vector signed char vec_reve (vector signed char);
20508 vector unsigned char vec_reve (vector unsigned char);
20509 vector bool int vec_reve (vector bool int);
20510 vector signed int vec_reve (vector signed int);
20511 vector unsigned int vec_reve (vector unsigned int);
20512 vector bool long long vec_reve (vector bool long long);
20513 vector signed long long vec_reve (vector signed long long);
20514 vector unsigned long long vec_reve (vector unsigned long long);
20515 vector bool short vec_reve (vector bool short);
20516 vector signed short vec_reve (vector signed short);
20517 vector double vec_reve (vector double);
20518 vector float vec_reve (vector float);
20519 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20520 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20521 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20522 (altivec_vreve): New pattern.
20523 * config/rs6000/altivec.h (vec_reve): New define.
20524 * doc/extend.texi (vec_rev): Update the built-in documentation file
20525 for the new built-in functions.
20527 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20529 PR tree-optimization/71815
20530 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20532 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20534 (slsr_process_phi): Likewise.
20535 (replace_uncond_cands_and_profitable_phis): Don't replace a
20536 multiply candidate with a stride of 1 (copy or cast).
20537 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20539 (lowest_cost_path): Likewise.
20540 (total_savings): Likewise.
20542 2017-06-26 Richard Biener <rguenther@suse.de>
20545 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20546 Use def_builtin_pure for all gather builtins.
20548 2017-06-26 Richard Biener <rguenther@suse.de>
20550 PR tree-optimization/81203
20551 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20552 non-dominating BBs.
20554 2017-06-26 Marek Polacek <polacek@redhat.com>
20557 * doc/invoke.texi: Document -Wmultistatement-macros.
20559 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20561 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20562 arm_neon_ok_no_float_abi effective target.
20564 2017-06-26 Richard Biener <rguenther@suse.de>
20566 PR tree-optimization/80928
20567 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20568 (copy_bbs): Set BB_DUPLICATED flag early.
20569 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20571 (execute_on_shrinking_pred): Likewise.
20572 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20573 BB_DUPLICATED blocks.
20574 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20575 iterate over all PHIs considering removal of *gsi.
20577 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20579 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20582 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20584 * config/rs6000/rs6000-string.c: (expand_block_clear,
20585 do_load_for_compare, select_block_compare_mode,
20586 compute_current_alignment, expand_block_compare,
20587 expand_strncmp_align_check, expand_strn_compare,
20588 expand_block_move, rs6000_output_load_multiple)
20589 Move functions related to string/block move/compare
20590 to a separate file.
20591 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20592 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20593 for this function which is now used in two files.
20594 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20595 * config.gcc: Add rs6000-string.o to extra_objs for
20596 targets powerpc*-*-* and rs6000*-*-*.
20598 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20601 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20602 32-bit, since indexed is not valid for DImode.
20603 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20604 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20605 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20606 (define_peephole2 for Altivec d-form store): Likewise.
20609 * multiple_target.c (create_dispatcher_calls): Only create the
20610 dispatcher call if the function is the default clone of a
20611 versioned function.
20613 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20615 PR middle-end/80902
20616 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20617 a call, force the call to not be a tail call.
20619 2017-06-23 Jeff Law <law@redhat.com>
20621 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20624 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20626 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20627 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20628 handling for early expansion of vector shifts (sl,sr,sra,rl).
20629 (builtin_function_type): Add vector shift right instructions
20630 to the unsigned argument list.
20632 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20634 rtl-optimizatoin/79286
20635 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20636 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20637 trap. PIC register plus a const unspec without offset can never trap.
20639 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20641 * tree.h (builtin_structptr_type): New type.
20642 (builtin_structptr_types): Declare new array.
20643 * tree.c (builtin_structptr_types): New array.
20644 (free_lang_data, build_common_tree_nodes): Use it.
20646 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20649 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20652 2017-06-22 Matt Turner <mattst88@gmail.com>
20654 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20655 Lake models to skylake case. Assume skylake for unknown
20656 models with clflushopt.
20658 2017-06-22 Jeff Law <law@redhat.com>
20660 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20661 frame sizes that do not satisfy aarch64_uimm12_shift.
20663 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20665 * profile-count.h (apply_probability,
20666 apply_scale, probability_in): Fix checks for zero.
20668 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20670 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20671 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20673 2016-06-22 Richard Biener <rguenther@suse.de>
20675 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20676 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20677 REDUC_MAX_EXPR support.
20678 (vectorizable_reduction): Likewise.
20679 (vect_create_epilog_for_reduction): Likewise.
20681 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20683 * match.pd (A / (1 << B) -> A >> B): New.
20684 * generic-match-head.c: Include optabs-tree.h.
20685 * gimple-match-head.c: Likewise.
20686 * optabs-tree.h (target_supports_op_p): New.
20687 * optabs-tree.c (target_supports_op_p): New.
20689 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20691 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20692 $gcc_cv_ld --help output.
20693 (gcc_cv_ld_demangle): Likewise.
20694 (gcc_cv_ld_eh_frame_hdr): Likewise.
20695 (gcc_cv_ld_pie): Likewise.
20696 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20697 (gcc_cv_ld_buildid): Likewise.
20698 (gcc_cv_ld_sysroot): Likewise.
20699 (ld_bndplt_support): Likewise.
20700 (ld_pushpopstate_support): Likewise.
20701 * configure: Regenerate.
20702 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20704 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20707 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20708 operands indexes to avoid gap between operands and match_dups.
20710 2017-06-21 Andrew Pinski <apinski@cavium.com>
20712 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20713 Increment Arith_shift and Arith_shift_reg by 1.
20714 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20716 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20717 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20718 (aarch64_strip_extend): Add new argument and test for it.
20719 (aarch64_cheap_mult_shift_p): New function.
20720 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20721 add a cost if it is true.
20722 Update calls to aarch64_strip_extend.
20723 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20725 2017-06-21 Andrew Pinski <apinski@cavium.com>
20727 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20729 (thunderxt88): Likewise.
20730 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20731 (thunderx_prefetch_tune): New variable.
20732 (thunderx2t99_prefetch_tune): Update for the correct values.
20733 (thunderxt88_tunings): New variable.
20734 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20735 generic_prefetch_tune.
20736 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20738 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20740 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20741 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20742 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20743 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20744 (aarch64_atomic_cas<mode>, GPI): Likewise.
20746 2017-06-21 Martin Liska <mliska@suse.cz>
20748 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20749 statements on cold and hot labels.
20750 * predict.c (tree_estimate_probability_bb): Remove the
20751 prediction from this place.
20753 2017-06-21 Martin Liska <mliska@suse.cz>
20755 PR tree-optimization/79489
20756 * gimplify.c (maybe_add_early_return_predict_stmt): New
20758 (gimplify_return_expr): Call the function.
20759 * predict.c (tree_estimate_probability_bb): Remove handling
20761 * predict.def: Update comment about early return predictor.
20762 * gimple-predict.h (is_gimple_predict): New function.
20763 * predict.def: Change default value of early return to 66.
20764 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20766 * passes.def: Put pass_strip_predict_hints to the beginning of
20769 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20771 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20772 FUNCTION_DECL declarations.
20773 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20775 (dwaf2out_decl): Likewise.
20776 * godump.c (go_early_global_decl): Skip call to the real debug hook
20777 for FUNCTION_DECL declarations.
20778 * passes.c (rest_of_decl_compilation): Skip call to the
20779 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20780 -fdump-go-spec is passed.
20782 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20784 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20785 Reorder for compactness.
20786 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20787 (def_builtin_pure, def_builtin_pure2): New functions.
20788 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20790 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20792 * match.pd (nop_convert): New predicate.
20793 ((A +- CST1) +- CST2): Allow some NOP conversions.
20795 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20798 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20799 with ctors/dtors if GOVD_SHARED is set.
20801 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20803 * config/aarch64/aarch64.md (movti_aarch64):
20804 Emit mov rather than orr.
20805 (movtf_aarch64): Likewise.
20806 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20807 Emit mov rather than orr.
20809 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20811 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20812 Swap alternatives, make integer dup more expensive.
20814 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20816 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20817 Return true for non-tls symbols.
20819 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20821 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20822 (cortex-a75): Likewise.
20823 (cortex-a75.cortex-a55): Likewise.
20824 * config/aarch64/aarch64-tune.md: Regenerate.
20825 * doc/invoke.texi (-mtune): Document new values for -mtune.
20827 2017-06-21 Tom de Vries <tom@codesourcery.com>
20829 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20830 stack_size feature.
20831 (Effective-Target Keywords, Other attributes): Suggest using
20832 dg-add-options stack_size feature to get stack limit in stack_size
20833 effective target documentation.
20835 2017-06-21 Julian Brown <julian@codesourcery.com>
20836 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20838 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20839 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20840 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20842 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20843 attribute type list for neon_multiply.
20844 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20845 attribute type list for neon_multiply.
20846 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20847 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20848 attribute type list for neon_multiply.
20849 * config/arm/types.md (crypto_pmull): Add.
20850 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20851 attribute type list.
20853 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20855 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20858 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20860 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20861 to make sure not to dereference a NULL cost_classes_ptr pointer.
20863 2017-06-20 Carl Love <cel@us.ibm.com>
20865 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20866 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20867 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20868 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20869 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20870 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20871 VMULOSW): New enum "unspec" values.
20872 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20873 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20874 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20875 altivec_vmulosw): New patterns.
20876 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20877 VMULOSW): Add definitions.
20879 2017-06-20 Julia Koval <julia.koval@intel.com>
20881 * config/i386/i386.c: Fix rounding expand for new pattern.
20882 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20884 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20886 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20887 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20889 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20891 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20894 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20896 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20897 architecture, then by implementer ID.
20898 * config/aarch64/aarch64-tune.md: Regenerate.
20900 2017-06-20 Richard Biener <rguenther@suse.de>
20902 PR middle-end/81097
20903 * fold-const.c (split_tree): Fold to type before negating.
20905 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20907 * diagnostic-show-locus.c
20908 (selftest::test_fixit_deletion_affecting_newline): New function.
20909 (selftest::diagnostic_show_locus_c_tests): Call it.
20911 2017-06-20 Andreas Schwab <schwab@suse.de>
20914 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20917 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20919 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20920 __ARM_FEATURE_COPROC according to support.
20922 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20924 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20925 Rewritten to avoid overflow for > 32-bit pointers.
20928 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20929 by removing enum keyword.
20930 (ubsan_type_descriptor): Likewise. Formatting fix.
20933 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20934 splitter): Require TARGET_SSE2 in the condition.
20936 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20939 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20940 for doing vector set of SFmode on ISA 3.0.
20941 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20942 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20944 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20945 SFmode value into a V4SF variable that was extracted from another
20946 V4SF variable without converting the element to double precision
20947 and back to single precision vector format.
20948 (vsx_insert_extract_v4sf_p9_2): Likewise.
20950 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20952 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20953 in UWHI to avoid undefined overflow.
20956 * ubsan.h (enum ubsan_encode_value_phase): New.
20957 (ubsan_encode_value): Change second argument to
20958 enum ubsan_encode_value_phase with default value of
20959 UBSAN_ENCODE_VALUE_GENERIC.
20960 * ubsan.c (ubsan_encode_value): Change second argument to
20961 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20962 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20963 create_tmp_var_raw instead of create_tmp_var and use a
20965 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20966 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20967 ubsan_encode_value callers.
20970 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20971 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20972 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20974 2017-06-19 Richard Biener <rguenther@suse.de>
20976 PR middle-end/81118
20977 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20978 estimates if we changed anything.
20980 2017-06-19 Richard Biener <rguenther@suse.de>
20982 PR tree-optimization/80887
20983 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20984 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20985 simplified lookups, then reset mprts_hook.
20986 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20988 (try_to_simplify): Likewise.
20990 2017-06-19 Martin Liska <mliska@suse.cz>
20993 * gimplify.c (gimplify_switch_expr):
20994 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20996 2017-06-19 Martin Liska <mliska@suse.cz>
20998 * doc/install.texi: Document that PGO runs in 4 stages.
21000 2017-06-19 Martin Liska <mliska@suse.cz>
21003 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21004 to dispatcher function name.
21005 * multiple_target.c (replace_function_decl): New function.
21006 (create_dispatcher_calls): Redirect both edges and references.
21008 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21010 * profile-count.c (profile_count::dump): Dump quality.
21011 (profile_count::differs_from_p): Update for unsigned val.
21012 * profile-count.h (profile_count_quality): New enum.
21013 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21015 2017-06-19 Richard Biener <rguenther@suse.de>
21017 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21018 struct function as arg.
21019 (estimate_numbers_of_iterations): Export overload with loop arg.
21020 (free_numbers_of_iterations_estimates_loop): Use an overload of
21021 free_numbers_of_iterations_estimates instead.
21022 * tree-cfg.c (remove_bb): Adjust.
21023 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21024 * tree-parloops.c (gen_parallel_loop): Likewise.
21025 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21027 (tree_unroll_loops_completely): Likewise.
21028 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21029 Use an overload instead and export.
21030 (estimated_loop_iterations): Adjust.
21031 (max_loop_iterations): Likewise.
21032 (likely_max_loop_iterations): Likewise.
21033 (estimate_numbers_of_iterations): Take struct function as arg
21035 (loop_exits_before_overflow): Adjust.
21036 (free_numbers_of_iterations_estimates_loop): Use an overload.
21037 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21038 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21040 2017-06-19 Richard Biener <rguenther@suse.de>
21043 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21044 RANGE_EXPR conservatively.
21046 2017-06-16 Carl Love <cel@us.ibm.com>
21048 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21049 definitions for vec_float, vec_float2, vec_floato,
21050 vec_floate built-ins.
21051 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21052 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21054 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21055 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21056 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21057 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21058 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21059 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21060 vec_floato): Add builtin defines.
21061 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21062 Update the built-in documentation file for the new built-in
21065 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21067 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21068 (mthumb): Mark as the negative of -marm.
21070 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21072 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21074 (ARM Options, -mtune): Document that this accepts the same
21075 extension options as -mcpu.
21076 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21078 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21080 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21081 permitted extensions.
21083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21085 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21086 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21087 (armv8-m.main): Add option +nodsp.
21088 * config/arm/arm-cpu-cdata.h: Regenerated.
21090 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21092 * config/arm/t-fuchsia: New file.
21093 * config.gcc (arm*-*-fuchsia*): Use it.
21095 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21097 * config/arm/t-symbian: Rewrite for new option infrastructure.
21099 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21101 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21102 (MULTILIB_REQUIRED): Likewise.
21104 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21106 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21107 (MULTILIB_RESUE): Likewise.
21108 (MULTILIB_MATCHES): Likewise.
21109 (MULTLIB_REQUIRED): Likewise.
21111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21113 * config/arm/t-rtems: Rewrite for new option framework.
21115 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21117 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21118 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21119 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21120 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21121 * config/arm/t-multilib: ... here.
21122 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21123 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21124 armv7-a and armv8*-a when A-profile libraries have not been built.
21125 * config/arm/t-rmprofile: Rewrite.
21127 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21129 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21130 with a backslash. Remove the backslash after substituting unescaped
21132 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21134 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21136 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21137 the driver to pass through the default setting of -mfloat-abi.
21138 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21140 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21141 (all_feat_combs): New rule.
21142 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21144 * config/arm/t-aprofile: Rewrite.
21146 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21148 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21149 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21150 fpu is not specified by the user/command-line.
21151 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21152 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21153 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21154 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21155 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21156 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21158 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21160 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21161 * config/arm/t-arm-elf: Rewritten.
21163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21165 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21166 have some floating-point instructions.
21167 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21168 (TARGET_MAYBE_HARD_FLOAT): New macro.
21169 * config/arm/arm-builtins.c (arm_init_builtins): Use
21170 TARGET_MAYBE_HARD_FLOAT.
21171 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21175 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21176 (configargs.h): Include it.
21177 (arm_print_hint_for_fpu_option): New function.
21178 (arm_parse_fpu_option): New function.
21179 (candidate_extension): New class.
21180 (arm_canon_for_multilib): New function.
21181 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21182 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21183 (ARCH_CANONICAL_SPECS): New macro.
21184 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21186 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21188 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21189 are set after handling multilib fragments. Set target_cpu_default2
21192 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21194 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21196 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21197 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21199 (check_arch): Likewise.
21200 * config/arm/arm.c (arm_configure_build_target): Handle
21201 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21202 options in the default.
21204 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21206 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21207 when an option is an alias of another.
21208 * config/arm/parsecpu.awk (optalias): New parser token.
21209 (gen_comm_data): Mark non-alias options as such. Emit entries
21210 for extension aliases.
21211 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21212 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21213 (armv6kz, armv6zk, armv6t2): Likewise.
21214 (armv7): Make vfpv3-d16 an alias.
21215 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21217 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21218 Sort in canonical order.
21219 (armv8-a): Sort in canonical order.
21220 (armv8.1-a, armv8.2-a): Likewise.
21221 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21223 (cortex-a9): Sort in canonical order.
21224 * config/arm/arm.c (selftests.h): Include it.
21225 (arm_test_cpu_arch_data): New function.
21226 (arm_run_self_tests): New function.
21227 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21228 (targetm): Move declaration to the end of the file.
21229 * arm-cpu-cdata.h: Regenerated.
21231 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21233 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21234 call to target_mode_check describing the type of option passed.
21235 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21236 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21237 arm_parse_cpu_option_name to match parameters against list of
21239 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21240 arm_arch_core_flags data structure.
21241 * config/arm/arm-cpu_cdata.h: Regenerated.
21243 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21245 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21247 (arm_print_hint_for_cpu_option): Likewise.
21248 (arm_print_hint_for_arch_option): Likewise.
21249 (arm_parse_cpu_option_name): Likewise.
21250 (arm_parse_arch_option_name): Likewise.
21251 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21252 of entries in the all_fpus list.
21253 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21254 (arm_parse_cpu_option_name): Declare.
21255 (arm_parse_arch_option_name): Declare.
21256 (arm_parse_option_features): Declare.
21257 (arm_intialize_isa): Declare.
21258 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21260 (gen_comm_data): ... here. Make definitions non-static.
21261 * config/arm/arm-cpu-data.h: Regenerated.
21262 * config/arm/arm-cpu-cdata.h: Regenerated.
21264 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21266 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21267 (cpu_arch_extension): New structure.
21268 (cpu_arch_option, arch_option, cpu_option): New structures.
21269 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21270 architecture types.
21271 (gen_data): Generate new format data tables.
21272 * config/arm/arm.c (cpu_tune): New structure.
21273 (cpu_option, processors): Delete.
21274 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21275 (arm_print_hint_for_cpu_option): ... this and ...
21276 (arm_print_hint_for_arch_option): ... this.
21277 (arm_parse_arch_cpu_name): Delete. Replace with ...
21278 (arm_parse_cpu_option_name): ... this and ...
21279 (arm_parse_arch_option_name): ... this.
21280 (arm_unrecognized_feature): Change type of target parameter to
21282 (arm_parse_arch_cpu_features): Delete. Replace with ...
21283 (arm_parse_option_features): ... this.
21284 (arm_configure_build_target): Rework to use new configuration data
21286 (arm_print_tune_info): Rework for new configuration data tables.
21287 * config/arm/arm-cpu-data.h: Regenerated.
21288 * config/arm/arm-cpu.h: Regenerated.
21290 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21292 * Makefile.in (OBJS): Move sbitmap.o from here ...
21293 (OBJS-libcommon): ... to here.
21295 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21297 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21298 (ISA_ALL_CRYPTO): New macro.
21299 (ISA_ALL_SIMD): New macro
21300 (ISA_ALL_FP): New macro.
21301 * config/arm/arm.c (fpu_bitlist): Update initializer.
21302 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21304 (arm9e): Add fpu. Add option for nofp
21305 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21306 (arm926ej-s, arm1026ej-s): Likewise.
21307 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21308 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21309 neon-fp16, neon-vfpv4, nofp and nosimd.
21310 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21311 (cortex-a8): Add fpu. Add option for nofp.
21312 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21313 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21314 (cortex-r4f): Add fpu.
21315 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21316 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21318 (cortex-r8): Likewise.
21319 (cortex-m4): Add fpu. Add option for nofp.
21320 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21321 (cortex-a17.cortex-a7): Likewise.
21322 (cortex-a32): Add fpu. Add options for crypto and nofp.
21323 (cortex-a35, cortex-a53): Likewise.
21324 (cortex-a57): Add fpu. Add option for crypto.
21325 (cortex-a72, cortex-a73): Likewise.
21326 (exynos-m1): Likewise.
21327 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21328 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21329 (cortex-m33): Add fpu. Add option for nofp.
21330 * config/arm/arm-cpu-cdata.h: Regenerated
21331 * config/arm/arm-cpu-data.h: Regenerated.
21333 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21335 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21336 (armv5te, armv5tej): Likewise.
21337 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21338 (armv7): Add options fp and vfpv3-d16.
21339 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21340 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21342 (armv7ve): Likewise.
21343 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21344 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21345 (armv8-a): Add nocrypto option.
21346 (armv8.1-a, armv8.2-a): Likewise.
21347 (armv8-m.main): add options fp, fp.dp and nofp.
21349 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21351 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21353 (armv8-a+crc): Delete.
21354 (armv8.1-a): Add options simd, crypto and nofp.
21355 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21356 (armv8.2-a+fp16): Delete.
21357 (armv8-m.main): Add option dsp.
21358 (armv8-m.main+dsp): Delete.
21359 (cortex-a8): Add fpu. Add nofp option.
21360 (cortex-a9): Add fpu. Add nofp and nosimd options.
21361 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21362 link to main cpu and architecture data structures.
21363 (gen_comm_data): Only put isa attributes from the main architecture
21365 (option): New statement for architecture and CPU entries.
21366 * arm.c (struct cpu_option): New structure.
21367 (struct processors): Add entry for options.
21368 (arm_unrecognized_feature): New function.
21369 (arm_parse_arch_cpu_name): Ignore any characters after the first
21371 (arm_parse_arch_cpu_feature): New function.
21372 (arm_configure_build_target): Separate out any CPU and architecture
21373 features and parse separately. Don't error out if -mfpu=auto is
21374 used with only an architecture string.
21375 (arm_print_asm_arch_directives): New function.
21376 (arm_file_start): Call it.
21377 * config/arm/arm-cpu-cdata.h: Regenerated.
21378 * config/arm/arm-cpu-data.h: Likewise.
21379 * config/arm/arm-tables.opt: Likewise.
21381 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21383 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21384 assembler when it is not -mfpu=auto.
21386 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21388 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21389 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21390 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21391 (ASM_CPU_SPEC): Rewrite.
21392 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21393 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21394 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21395 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21396 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21397 copied string is NUL-terminated. Also strip any characters prefixed
21399 (arm_rewrite_selected_arch): New function.
21400 (arm_rewrite_march): New function.
21402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21404 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21405 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21406 (march, mcpu, mtune): Convert to string-based options.
21407 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21408 (arm_parse_arch_cpu_name): New function.
21409 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21410 identify selected architecture or CPU.
21411 (arm_option_save): New function.
21412 (TARGET_OPTION_SAVE): Redefine.
21413 (arm_option_restore): Restore string options.
21414 (arm_option_print): Print string options.
21416 2017-06-16 Martin Sebor <msebor@redhat.com>
21418 PR tree-optimization/80933
21419 PR tree-optimization/80934
21420 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21421 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21422 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21423 (gimple_fold_builtin): Call them.
21425 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21427 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21428 as unlikely; update profile.
21430 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21432 * predict.c (force_edge_cold): Handle declaring edges impossible
21435 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21437 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21439 (try_unroll_loop_completely): Fix reporting.
21441 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21443 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21445 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21448 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21449 if given a non-constant argument for an intrinsic which requires a
21452 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21454 * profile.c (compare_freqs): New function.
21455 (branch_prob): Sort edge list.
21456 (find_spanning_tree): Assume that the list is priority sorted.
21458 2017-06-16 Richard Biener <rguenther@suse.de>
21460 PR tree-optimization/81090
21461 * passes.def (pass_record_bounds): Remove.
21462 * tree-pass.h (make_pass_record_bounds): Likewise.
21463 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21464 make_pass_record_bounds): Likewise.
21465 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21466 not free niter estimates at the beginning but at the end.
21467 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21469 2017-06-16 Richard Biener <rguenther@suse.de>
21471 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21472 initializer to workaround ICE in host GCC 4.8.
21474 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21476 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21478 (clone_inlined_nodes): Update.
21480 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21482 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21483 prefetch settings, and enable prefetching by default at -O3.
21485 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21487 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21488 Set flag_prefetch_loop_arrays according to tuning data.
21490 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21492 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21493 New tune structure.
21494 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21495 [Unrelated to main purpose of the patch] Place the pointer field last
21496 to enable type checking errors when tune structure are wrongly merged.
21497 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21498 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21499 (thunderx2t99_prefetch_tune): New tune constants.
21500 (tune_params *_tunings): Update all tunings (no functional change).
21501 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21502 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21503 from tunings structures.
21505 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21508 * ubsan.c (instrument_null): Add T argument, use it instead
21509 of computing it based on IS_LHS.
21510 (instrument_object_size): Likewise.
21511 (pass_ubsan::execute): Adjust instrument_null and
21512 instrument_object_size callers to pass gimple_get_lhs or
21513 gimple_assign_rhs1 result to it. Use instrument_null instead of
21514 calling get_base_address and instrument_mem_ref. Handle
21515 aggregate call arguments for object-size sanitization.
21517 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21519 PR tree-optimization/81089
21520 * tree-vrp.c (is_masked_range_test): Validate operands of
21523 2017-06-15 Martin Sebor <msebor@redhat.com>
21526 * dumpfile.c (dump_register): Avoid calling memset to initialize
21527 a class with a default ctor.
21528 * gcc.c (struct compiler): Remove const qualification.
21529 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21530 * hash-table.h: Ditto.
21531 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21533 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21534 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21536 * params.h (struct param_info): Make struct members non-const.
21537 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21538 with default initialization.
21539 * vec.h (vec_copy_construct, vec_default_construct): New helper
21541 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21542 with vec_copy_construct.
21543 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21544 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21545 * doc/invoke.texi (-Wclass-memaccess): Document.
21547 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21549 * emit-rtl.h (is_leaf): Update comment about local
21550 register allocator.
21552 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21555 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21556 for a variable to have a section before checking if the section has a
21558 Set section to.persistent if persistent attribute is set.
21559 Warn if .persistent attribute is used on an automatic variable.
21561 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21563 PR rtl-optimization/80474
21564 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21566 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21568 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21571 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21573 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21574 (casesi): Emit bounds checking as RTL.
21575 (casesi_internal_mips16_<mode>): Remove bounds checking.
21577 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21579 * config/xtensa/xtensa.c (xtensa_option_override): Append
21580 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21581 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21582 xtensa_doloop_hooks): Define unconditionally.
21583 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21585 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21586 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21587 for it in xtensa_option_override.
21588 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21589 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21591 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21593 * doc/cppopts.texi: Document '-' special value to -MF.
21595 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21597 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21598 (cortex_a53_fconst): Likewise.
21599 (cortex_a53_fpmul): Likewise.
21600 (cortex_a53_f_load_64): Likewise.
21601 (cortex_a53_f_load_many): Likewise.
21602 (cortex_a53_advsimd_alu): Likewise.
21603 (cortex_a53_advsimd_alu_q): Likewise.
21604 (cortex_a53_advsimd_mul): Likewise.
21605 (cortex_a53_advsimd_mul_q): Likewise.
21606 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21607 Add missing fmul, r2f_cvt and fconst cases.
21609 2017-06-14 Richard Biener <rguenther@suse.de>
21611 PR middle-end/81088
21612 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21614 (fold_binary_loc): When associating do not treat pre-existing
21615 TREE_OVERFLOW on literal constants as a reason to allow
21616 TREE_OVERFLOW on associated literal constants.
21618 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21620 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21621 (MASK_FEATURES): New macro.
21622 * config/sparc/sparc.c (sparc_option_override): Remove the special
21623 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21625 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21627 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21628 a division of 0 if non-call exceptions are enabled.
21630 2017-06-14 Andrew Pinski <apinski@cavium.com>
21631 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21634 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21635 Improve vector initialization code gen for only variable case.
21637 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21639 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21641 2017-06-14 Richard Biener <rguenther@suse.de>
21643 PR tree-optimization/81083
21644 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21647 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21649 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21650 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21651 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21652 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21653 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21654 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21656 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21658 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21659 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21661 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21663 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21665 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21667 * config/rs6000/t-rtems: Don't handle SPE.
21669 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21671 * config/rs6000/t-linux: Don't handle SPE.
21673 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21675 * config/rs6000/eabispe.h: Delete file.
21677 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21679 * config/rs6000/t-spe: Delete file.
21681 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21683 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21684 (rs6000_legitimate_offset_address_p): Return false for anything in
21685 V2SImode or V2SFmode.
21687 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21689 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21690 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21691 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21693 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21694 (rs6000_legitimate_offset_address_p): Ditto.
21695 (rs6000_emit_move): Ditto.
21696 (rs6000_init_builtins): Remove V4HI_type_node.
21698 2017-06-13 Martin Liska <mliska@suse.cz>
21701 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21702 (gate_asan): Likewise.
21703 * asan.h (asan_no_sanitize_address_p): Remove the function.
21704 (sanitize_flags_p): New function.
21705 * builtins.def: Fix coding style.
21706 * common.opt: Use renamed enum value.
21707 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21708 * doc/extend.texi: Document no_sanitize attribute.
21709 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21710 to SANITIZE_UNDEFINED_NONDEFAULT.
21711 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21712 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21713 Use sanitize_flags_p.
21714 * gimplify.c (gimplify_function_tree): Likewise.
21715 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21716 * opts.c (parse_no_sanitize_attribute): New function.
21717 (common_handle_option): Use renamed enum value.
21718 * opts.h (parse_no_sanitize_attribute): Declare.
21719 * tree.c (sanitize_flags_p): New function.
21720 * tree.h: Declared here.
21721 * tsan.c: Use sanitize_flags_p.
21722 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21723 (instrument_mem_ref): Likewise.
21724 (instrument_bool_enum_load): Likewise.
21725 (do_ubsan_in_current_function): Remove the function.
21726 (pass_ubsan::execute): Use sanitize_flags_p.
21727 * ubsan.h: Remove do_ubsan_in_current_function
21728 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21729 (dump_function_to_file): Use it here.
21731 2017-06-13 Martin Jambor <mjambor@suse.cz>
21733 PR tree-optimization/80803
21734 PR tree-optimization/81063
21735 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21736 (propagate_subaccesses_across_link): Enqueue subtree whenever
21737 necessary instead of relying on the caller.
21739 2017-06-13 Martin Jambor <mjambor@suse.cz>
21741 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21742 that have a first_link.
21743 (sort_and_splice_var_accesses): Do not check first_link before
21745 (subtree_mark_written_and_enqueue): Likewise.
21746 (propagate_all_subaccesses): Likewise and do not stop at first
21747 parent with a first_link.
21749 2017-06-13 Martin Jambor <mjambor@suse.cz>
21751 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21754 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21756 * match.pd: New pattern.
21758 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21760 * tree-vrp.c (is_masked_range_test): New function.
21761 (register_edge_assert_for): Determine ranges for
21764 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21766 PR tree-optimization/67328
21767 * fold-const.c (maskable_range_p): New function.
21768 (build_range_check): Generate bittests if possible.
21770 2017-06-13 Martin Liska <mliska@suse.cz>
21772 * gimple-pretty-print.c (dump_probability): Add new argument.
21773 (dump_edge_probability): Dump both probability and count.
21774 (dump_gimple_label): Likewise.
21775 (dump_gimple_bb_header): Likewise.
21777 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21780 * config/avr/avr-devices.c: Fix indentation.
21781 * config/avr/gen-avr-mmcu-specs.c: Dito.
21783 2017-06-13 Richard Biener <rguenther@suse.de>
21785 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21786 instead get vector type from stmt_info.
21787 (vectorizable_reduction): Adjust. Remove dead code.
21789 2017-06-13 Richard Biener <rguenther@suse.de>
21791 PR middle-end/81065
21792 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21793 case of C * (x * C2 + C3).
21794 (fold_addr_of_array_ref_difference): Properly fold index difference.
21796 2017-06-12 David S. Miller <davem@davemloft.net>
21799 * config/sparc/sparc.md (return expander): Emit frame blockage if
21800 function uses alloca.
21802 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21804 * combine.c (make_field_assignment): Check len rather than the mode
21805 precision when calling force_to_mode.
21807 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21809 Support multilibs and devices that see flash in RAM address range.
21812 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21813 (avr_mcu_t) <flash_pm_offset>: New field.
21814 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21815 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21816 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21817 (AVR_TINY_PM_OFFSET): Remove macro.
21818 * config/avr/avr.opt (-mshort-calls): New option.
21819 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21820 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21821 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21822 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21823 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21824 instead of avr_arch->have_jmp_call.
21825 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21826 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21827 avr_arch->flash_pm_offset to define.
21828 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21829 new field flash_pm_offset. Add entry for avrxmega3.
21830 (avr_texinfo): Add entry for avrxmega3.
21831 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21832 attiny212, attiny214,
21833 attiny412, attiny414, attiny416, attiny417,
21834 attiny814, attiny816, attiny817,
21835 attiny1614, attiny1616, attiny1617,
21836 attiny3214, attiny3216, attiny3217.
21837 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21838 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21839 (avr_print_operand_address) [AVR_TINY]: Same.
21840 (avr_asm_init_sections) <readonly_data_section>: Only patch
21841 callback if avr_arch->flash_pm_offset = 0.
21842 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21843 for rodata if avr_arch->flash_pm_offset != 0.
21844 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21845 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21846 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21847 (m_options): Append opt_rcall.
21848 (m_dirnames): Append dir_rcall.
21849 * config/avr/t-multilib: Regenerate.
21851 * configure.ac [target=avr]: Check whether avrxmega3 default
21852 linker description file works as needed.
21853 * configure: Regenerate.
21854 * doc/avr-mmcu.texi: Regenerate.
21855 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21856 <__AVR_ARCH__>: Document avrxmega3 and 103.
21857 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21858 <__AVR_SHORT_CALLS__>: Document it.
21859 <__AVR_PM_BASE_ADDRESS__>: Document it.
21860 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21861 (AVR Variable Attributes) <progmem>: Document this is
21862 not needed for avrxmega3.
21863 (AVR Named Address Spaces) <__flash>: Dito.
21865 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21867 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21869 2017-06-12 Doug Rupp <rupp@adacore.com>
21871 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21872 Append vxworks-stdint.h to the tm_file list.
21873 * config/vxworks-stdint.h: New file.
21875 2017-06-12 Martin Liska <mliska@suse.cz>
21877 PR tree-optimization/81041
21878 * tree-profile.c (gimple_gen_ic_func_profiler):
21879 Create an extra BB in profile-generate
21880 (gimple_gen_time_profiler): Likewise.
21882 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21884 PR tree-optimization/81003
21885 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21886 (update_range_test): Use it instead of force_gimple_operand_gsi.
21888 2017-06-12 Richard Biener <rguenther@suse.de>
21890 PR tree-optimization/81053
21891 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21892 with backedge value not defined in loop. Simplify def stmt
21895 2017-06-11 Tom de Vries <tom@codesourcery.com>
21898 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21900 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21901 nvptx_cannot_force_const_mem.
21903 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21905 * opts.c (finish_options): Move test for flag_split_stack after
21906 it has been initialized.
21908 2017-06-11 Jason Merrill <jason@redhat.com>
21910 * tree.h (id_equal): New.
21911 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21912 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21913 instead of strcmp of IDENTIFIER_POINTER.
21915 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21917 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21918 (mark_all_inlined_calls_cdtor): Fix formating.
21919 (inline_transform): Rescale profile before inlining.
21921 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21923 * cgraph.h (cgraph_edge::clone): Update prototype.
21924 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21925 (cgraph_node::create_clone): Update.
21926 (cgraph_node::create_version_clone): Update.
21927 * tree-inline.c (copy_bb): Update.
21928 (expand_call_inline): Update.
21930 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21932 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21933 factored out from ...
21934 (rs6000_emit_prologue): ... here.
21936 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21938 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21939 factored out from ...
21940 (rs6000_emit_prologue): ... here.
21942 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21944 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21946 (handle_missing_profiles): Fix computation of tp_first_run.
21947 (counts_to_freqs): Do not touch freqs when count is 0.
21949 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21951 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21954 2017-06-10 Tom de Vries <tom@codesourcery.com>
21956 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21957 attributes): Document signal effective target.
21959 2017-06-10 Tom de Vries <tom@codesourcery.com>
21961 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21962 Document effective target stack_size.
21964 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21966 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21967 to the edit_context if they can be auto-applied.
21969 2017-06-9 Ian Lance Taylor <iant@golang.org>
21971 * opts.c (finish_options): If -fsplit-stack, disable implicit
21972 -forder-blocks-and-partition.
21973 * doc/invoke.texi (Optimize Options): Document that when using
21974 -fsplit-stack -forder-blocks-and-partition is not implicitly
21977 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21979 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21980 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21981 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21982 * builtins.def (abort, trap, unreachable): Declare cold.
21983 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21984 * tree-core.h (ECF_COLD): New.
21985 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21986 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21988 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21990 * predict.c (unlikely_executed_stmt_p): Cleanup.
21992 2017-06-09 Richard Biener <rguenther@suse.de>
21994 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21995 model if the ref is always written to.
21997 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21999 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22001 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22003 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22006 2017-06-09 Tom de Vries <tom@codesourcery.com>
22009 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22010 "target cannot support label values" when encountering LABEL_REF.
22012 2017-06-09 Martin Liska <mliska@suse.cz>
22014 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22015 (gimple_gen_ic_func_profiler): Emit direct comparison
22016 of __gcov_indirect_call_callee with NULL.
22017 (gimple_gen_time_profiler): Change probability from
22018 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22020 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22022 * profile.c (edge_gcov_counts): Turn to pointer.
22023 (compute_branch_probabilities, compute_branch_probabilities): Update.
22024 (branch_prob): Do not clear edge_gcov_count.
22025 * profile.h (edge_gcov_counts): Turn to pointer.
22026 (edge_gcov_count): Update.
22028 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22030 * gimple.h (gimple_check_failed): Mark cold.
22032 2017-06-09 Richard Biener <rguenther@suse.de>
22034 PR tree-optimization/66623
22035 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22036 refactor check_reduction into two parts, properly computing
22037 whether we have to check reduction validity for outer loop
22040 2017-06-09 Richard Biener <rguenther@suse.de>
22042 PR tree-optimization/79483
22043 * graphite-scop-detection.c (order): New global.
22044 (get_order): Compute bb to order mapping that satisfies code
22045 generation constraints.
22046 (cmp_pbbs): New helper.
22047 (build_scops): Start domwalk at entry block, sort generated
22050 2017-06-09 Richard Biener <rguenther@suse.de>
22052 PR middle-end/81007
22053 * ipa-polymorphic-call.c
22054 (ipa_polymorphic_call_context::restrict_to_inner_class):
22055 Skip FIELD_DECLs with error_mark_node type.
22056 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22059 2017-06-09 Martin Liska <mliska@suse.cz>
22061 * predict.c (struct branch_predictor): New struct.
22062 (test_prediction_value_range): New test.
22063 (predict_c_tests): New function.
22064 * selftest-run-tests.c (selftest::run_tests): Run the function.
22065 * selftest.h: Declare new tests.
22067 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22070 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22071 gen_add3_insn did not fail.
22072 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22073 r0, construct that number in a temporary reg and add that reg to r0.
22074 If asked to put the result in r0 as well, fail.
22076 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22078 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22079 for early expansion of vec_eqv.
22081 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22083 PR middle-end/81005
22084 * ubsan.c (instrument_null): Avoid pointless code temporary.
22085 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22087 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22091 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22094 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22097 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22099 * predict.c (unlikely_executed_edge_p): Move ahead.
22100 (probably_never_executed_edge_p): Use it.
22102 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22104 PR middle-end/79988
22105 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22106 gimple_call_builtin_p call.
22108 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22110 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22111 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22112 rtl_check_failed_type2, rtl_check_failed_code1,
22113 rtl_check_failed_code2, rtl_check_failed_code_mode,
22114 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22115 rtvec_check_failed_bounds, rtl_check_failed_flag,
22116 _fatal_insn_not_found, _fatal_insn): Likewise.
22117 * tree.h (tree_contains_struct_check_failed,
22118 tree_check_failed, tree_not_check_failed,
22119 tree_class_check_failed, tree_range_check_failed,
22120 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22121 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22122 tree_operand_check_failed, omp_clause_check_failed,
22123 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22126 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22128 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22129 flag_branch_probabilities.
22130 * ipa-inline.c (edge_badness): Likewise.
22131 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22132 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22133 * predict.c (maybe_hot_frequency_p): Likewise.
22134 (probably_never_executed): Likewise.
22135 * sched-ebb.c (schedule_ebbs): Likewise.
22136 * sched-rgn.c (find_single_block_region): Likewise.
22137 * tracer.c (tail_duplicate): Likewise.
22139 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22141 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22142 longer requires x_flag_profile_use.
22144 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22146 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22147 instead of flag_reorder_blocks_and_partition.
22148 * dbxout.c (dbxout_function_end): Likewise.
22149 * dwarf2out.c (gen_subprogram_die): Likewise.
22150 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22151 * hw-doloop.c (reorg_loops): Likewise.
22152 * varasm.c (assemble_start_function,
22153 assemble_end_function): Likewise.
22154 (decide_function_section): Do not check for
22155 flag_reorder_blocks_and_partition.
22157 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22159 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22161 (chkp_get_hard_register_fake_addr_expr): Ditto.
22162 (chkp_build_addr_expr): Add check for hard reg case.
22163 (chkp_parse_array_and_component_ref): Ditto.
22164 (chkp_find_bounds_1): Ditto.
22165 (chkp_process_stmt): Don't generate bounds store for
22168 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22170 * predict.c (maybe_hot_bb_p): Do not check profile status.
22171 (maybe_hot_edge_p): Likewise.
22172 (probably_never_executed): Check for zero counts even if profile
22174 (unlikely_executed_edge_p): New function.
22175 (unlikely_executed_stmt_p): New function.
22176 (unlikely_executed_bb_p): New function.
22177 (set_even_probabilities): Use unlikely predicates.
22178 (combine_predictions_for_bb): Likewise.
22179 (predict_paths_for_bb): Likewise.
22180 (predict_paths_leading_to_edge): Likewise.
22181 (determine_unlikely_bbs): New function.
22182 (estimate_bb_frequencies): Use it.
22183 (compute_function_frequency): Use zero counts even if profile is
22185 * profile-count.h: Fix typo.
22187 2017-08-08 Julia Koval <julia.koval@intel.com>
22189 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22190 _mm512_mask_cvtsepi16_storeu_epi8,
22191 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22192 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22193 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22194 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22195 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22196 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22197 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22198 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22199 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22200 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22201 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22202 __builtin_ia32_pmovuswb256mem_mask,
22203 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22204 __builtin_ia32_pmovwb512mem_mask): New builtins.
22206 2017-08-08 Julia Koval <julia.koval@intel.com>
22208 PR target/73350,80862
22209 * config/i386/subst.md (round): Fix round pattern.
22210 * config/i386/i386.c (ix86_erase_embedded_rounding):
22211 Fix erasing rounding for the fixed pattern.
22213 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22215 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22217 2017-06-08 Martin Liska <mliska@suse.cz>
22219 PR gcov-profile/80911
22220 * gcov.c (block_info::block_info): New constructor.
22222 2017-06-07 Carl Love <cel@us.ibm.com>
22224 * config/rs6000/rs6000-c: The return type of the following
22225 built-in functions was implemented as int not long long. Fix sign
22226 of return value for the unsigned version of vec_mulo and vec_mule.
22227 vector unsigned long long vec_bperm (vector unsigned long long,
22228 vector unsigned char)
22229 vector signed long long vec_mule (vector signed int,
22231 vector unsigned long long vec_mule (vector unsigned int,
22232 vector unsigned int)
22233 vector signed long long vec_mulo (vector signed int,
22235 vector unsigned long long vec_mulo (vector unsigned int,
22236 vector unsigned int)
22237 * doc/extend.texi: Fix the documentation for the built-in
22240 2017-06-07 Carl Love <cel@us.ibm.com>
22243 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22246 2017-06-07 Carl Love <cel@us.ibm.com>
22248 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22249 support, Generate doublehv for signed int/float for BE case only.
22251 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22253 * doc/invoke.texi (mcx16): Rewrite.
22255 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22257 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22258 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22259 *mov<mode>_softfloat, and an anonymous splitter): Use
22260 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22262 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22264 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22266 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22267 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22268 (rs6000_debug_reg_global): Adjust.
22269 (rs6000_init_hard_regno_mode_ok): Adjust.
22270 (rs6000_dbx_register_number): Adjust.
22271 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22272 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22273 Remove SPE_ACC and SPEFSCR.
22274 (REG_ALLOC_ORDER): Ditto.
22275 (FRAME_POINTER_REGNUM): Change to 111.
22276 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22277 (REG_CLASS_NAMES): Ditto.
22278 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22279 (REGISTER_NAMES): Ditto.
22280 (ADDITIONAL_REG_NAMES): Ditto.
22281 (rs6000_reg_names): Ditto.
22282 * config/rs6000/rs6000.md: Renumber some register number
22285 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22287 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22289 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22290 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22292 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22293 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22294 Delete the SPE high registers.
22295 (REG_ALLOC_ORDER): Ditto.
22296 (enum reg_class): Remove SPE_HIGH_REGS.
22297 (REG_CLASS_NAMES): Ditto.
22298 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22299 (REGISTER_NAMES): Ditto.
22300 (rs6000_reg_names): Ditto.
22301 * doc/tm.texi.in: Remove SPE as example.
22302 * doc/tm.texi: Regenerate.
22304 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22306 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22307 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22308 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22309 * config/rs6000/rs6000.md (type): Remove "brinc".
22311 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22313 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22314 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22315 * config/rs6000/linuxspe.h: Delete file.
22316 * config/rs6000/rs6000.md: Don't include spe.md.
22317 * config/rs6000/spe.h: Delete file.
22318 * config/rs6000/spe.md: Delete file.
22319 * config/rs6000/t-rs6000: Remove spe.md.
22321 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22323 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22324 (reg_or_none500mem_operand): Delete.
22325 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22326 instead of reg_or_none500mem_operand.
22328 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22330 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22331 handling of SPE flags.
22332 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22334 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22336 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22338 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22339 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22340 paired_divv2sf3): Similar.
22341 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22342 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22343 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22345 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22346 Rename the paired_* instruction patterns.
22347 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22349 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22350 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22351 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22352 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22353 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22354 PAIRED_VECTOR_MODE.
22355 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22356 (spe_func_has_64bit_regs_p): Delete.
22357 (spe_expand_predicate_builtin): Delete.
22358 (spe_expand_evsel_builtin): Delete.
22359 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22360 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22361 (invalid_e500_subreg): Delete.
22362 (rs6000_legitimize_address): Always force_reg op2 as well, for
22363 paired single memory accesses.
22364 (rs6000_member_type_forces_blk): Delete.
22365 (rs6000_spe_function_arg): Delete.
22366 (rs6000_expand_unop_builtin): Delete SPE handling.
22367 (rs6000_expand_binop_builtin): Ditto.
22368 (spe_expand_stv_builtin): Delete.
22369 (bdesc_2arg_spe): Delete.
22370 (spe_expand_builtin): Delete.
22371 (spe_expand_predicate_builtin): Delete.
22372 (spe_expand_evsel_builtin): Delete.
22373 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22374 (spe_init_builtins): Delete.
22375 (spe_func_has_64bit_regs_p): Delete.
22376 (savres_routine_name): Delete "info" parameter. Adjust callers.
22377 (rs6000_emit_stack_reset): Ditto.
22378 (rs6000_dwarf_register_span): Delete.
22379 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22380 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22381 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22383 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22385 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22386 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22387 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22388 mulv2sf3, divv2sf3): Delete expanders.
22390 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22392 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22394 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22396 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22397 * config/rs6000/rs6000.c: Ditto.
22399 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22401 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22402 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22403 comparison_operator.
22405 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22407 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22408 * config/rs6000/rs6000.opt: Ditto.
22409 * config/rs6000/t-rtems: Ditto.
22411 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22413 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22414 TARGET_E500_SINGLE by 0, simplify.
22415 * config/rs6000/rs6000.c: Ditto.
22416 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22417 (spe_build_register_parallel): Delete.
22418 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22419 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22420 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22421 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22422 (E500_CONVERT): Delete.
22423 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22425 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22427 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22428 * config/rs6000/dfp.md: Ditto.
22429 (negdd2, *negdd2_fpr): Merge.
22430 (absdd2, *absdd2_fpr): Merge.
22431 (negtd2, *negtd2_fpr): Merge.
22432 (abstd2, *abstd2_fpr): Merge.
22433 * config/rs6000/e500.h: Delete file.
22434 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22435 TARGET_FPRS by 1 and simplify.
22436 * config/rs6000/rs6000-c.c: Ditto.
22437 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22438 TARGET_DF_SPE by 0.
22439 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22441 * config/rs6000/rs6000.md: Ditto.
22442 (floatdidf2, *floatdidf2_fpr): Merge.
22443 (move_from_CR_gt_bit): Delete.
22444 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22445 (E500_CR_IOR_COMPARE): Delete.
22446 (All patterns that require !TARGET_FPRS): Delete.
22447 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22449 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22451 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22453 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22455 * graphds.c (add_edge): Intitialize edge's attached data.
22456 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22457 pointer parameter. Call pointed function on each edge during
22458 graph traversing. Skip traversing the edge when the function
22460 (graphds_dfs, graphds_scc): Ditto.
22461 (for_each_edge): New parameter. Pass the new parameter to callback
22463 * graphds.h (skip_edge_callback): New function pointer type.
22464 (graphds_dfs, graphds_scc): New function pointer parameter.
22465 (graphds_edge_callback, for_each_edge): New parameter.
22467 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22469 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22470 out code checking if runtime alias check is possible to below ...
22471 Call the new function.
22472 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22473 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22475 2017-06-07 Marek Polacek <polacek@redhat.com>
22478 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22479 TYPE_OVERFLOW_WRAPS check.
22481 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22483 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22484 if versioning is required.
22485 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22486 peeling with the check for versioning.
22488 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22490 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22491 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22492 Set true to new parameter if new ssa variable is defined.
22493 (vect_gen_vector_loop_niters): Refactor. Set range information
22494 for the new vector loop bound variable.
22495 (vect_do_peeling): Ditto.
22497 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22499 * tree-affine.c (ssa.h): Include header file.
22500 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22501 has wrapping overflow behavior.
22503 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22505 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22507 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22509 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22510 (tree_to_aff_combination): ... here.
22512 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22514 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22515 reg_pressure model function.
22516 (ivopts_global_cost_for_size): Delete.
22517 (determine_set_costs, iv_ca_recount_cost): Call new model function
22518 ivopts_estimate_reg_pressure.
22520 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22522 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22523 expensive than udiv. Remove floating point cases from mod.
22525 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22527 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22528 Increase idiv cost.
22530 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22532 * config/aarch64/aarch64.md
22533 (copysignsf3): Fix mask generation.
22535 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22537 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22539 (class dump_manager): Add register_dumps method.
22540 * dumpfile.c: Include langhooks.h.
22541 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22542 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22543 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22544 (dump_manager::dump_register): Start with 512 entries instead of 32.
22545 (dump_manager::register_dumps): New method.
22546 * toplev.c (general_init): Instead of invoking register_dumps
22547 langhook, invoke register_dumps method on the dump manager.
22548 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22551 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22553 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22554 Say that # requires an associated define_split to exist, and that
22555 the define_split must be suitable for use after register allocation.
22557 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22559 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22560 (compute_outgoing_frequencies): Also initialize zero counts.
22561 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22562 around loops; preserve more of profile when nothing changes.
22564 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22566 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22568 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22569 * config/arm/arm-cpu-cdata.h: Regenerate.
22570 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22571 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22572 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22573 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22574 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22576 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22578 * config/arm/t-rmprofile: Likewise.
22579 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22581 2017-06-06 David S. Miller <davem@davemloft.net>
22584 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22585 blockage if function uses alloca.
22587 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22589 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22590 New "uid" fields to hold pretty-print IDs of group and ref.
22591 Memory references are now identified as <group_id>:<ref_id>
22592 instead of using [random] addresses.
22593 (dump_mem_details): Simplify, no functional change.
22594 (dump_mem_ref): Simplify and make output more concise.
22595 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22596 (find_or_create_group): Initialize group uid.
22597 (record_ref): Initialize ref uid. Improve debug output.
22598 (prune_group_by_reuse, should_issue_prefetch_p,)
22599 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22600 (mark_nontemporal_store, determine_loop_nest_reuse):
22601 Improve debug output.
22603 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22605 * dbgcnt.def (prefetch): New debug counter.
22606 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22607 (schedule_prefetches): Stop issueing prefetches if debug counter
22610 2017-06-06 Tom de Vries <tom@codesourcery.com>
22612 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22613 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22615 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22617 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22618 Use aarch64_reg_or_zero predicate for operand 4.
22619 (aarch64_compare_and_swap<mode> define_insn_and_split):
22620 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22621 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22623 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22625 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22626 (arm_compute_save_core_reg_mask): This.
22627 (thumb1_compute_save_reg_mask): Rename into ...
22628 (thumb1_compute_save_core_reg_mask): This.
22629 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22630 (arm_compute_frame_layout): Likewise.
22632 2017-06-06 Richard Biener <rguenther@suse.de>
22634 PR tree-optimization/80974
22635 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22636 keep or clear leaders SSA info.
22638 2017-06-06 Tom de Vries <tom@codesourcery.com>
22640 * config/nvptx/nvptx.c (split_mode_p): New function.
22641 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22643 2017-06-06 Tom de Vries <tom@codesourcery.com>
22645 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22647 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22650 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22653 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22655 * shrink-wrap.c (handle_simple_exit): Update profile.
22656 (try_shrink_wrapping): Upate profile.
22658 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22660 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22661 (tree_guess_outgoing_edge_probabilities): New.
22662 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22663 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22665 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22667 * ipa-split.c (split_function): Initialize return bb profile.
22669 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22671 * profile.c (compute_branch_probabilities): Also initialize
22672 EXIT_BLOCK profile.
22674 2017-06-06 Richard Biener <rguenther@suse.de>
22676 PR tree-optimization/80928
22677 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22678 (vect_analyze_loop_operations): Properly guard analysis for
22680 (vect_transform_loop): Likewise.
22681 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22682 (vect_model_induction_cost): Do not cost for pure SLP.
22683 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22684 of induction in inner loop vectorization.
22685 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22686 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22687 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22689 (vect_analyze_slp_cost_1): Cost induction.
22690 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22691 (vect_get_slp_vect_defs): Likewise.
22692 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22693 (vect_transform_stmt): Handle SLP reductions.
22694 * tree-vectorizer.h (vectorizable_induction): Adjust.
22696 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22698 * config/rs6000/rs6000.c (make_resolver_func): Update
22699 init_lowered_empty_function call.
22701 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22703 * doc/invoke.texi: Document the -fprofile-abs-path option.
22704 * common.opt (fprofile-abs-path): New option.
22705 * gcov-io.h (gcov_write_filename): Declare.
22706 * gcov-io.c (gcov_write_filename): New function.
22707 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22708 * profile.c (output_location): Likewise.
22710 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22712 * shring-wrap.c: Revert accidental commit.
22714 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22716 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22718 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22720 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22722 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22723 profile in callgraph edge.
22724 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22725 (apply_scale): Likewise.
22726 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22727 Also scale profile when inlining function with zero profile.
22728 (initialize_cfun): Update exit block profile even when it is zero.
22729 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22730 when profile is read.
22732 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22734 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22735 (CLONE_*): New constants to define the processors we can generate
22736 code for with the target_clone attribute.
22737 (rs6000_clone_map): New array to identify which clone processors
22738 the current program is running on.
22739 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22740 target_clone attribute.
22741 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22742 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22743 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22744 (cpu_expand_builtin): Add support for target_clone attribute.
22745 (rs6000_valid_attribute_p): Allow "default" attribute.
22746 (get_decl_name): New debug function to simplify printing the
22747 current function name in debugging statements.
22748 (rs6000_clone_priority): New functions to support the target_clone
22749 attribute, and be able to generate code to switch between ISA 2.05
22750 through ISA 3.0 (power6 through power9).
22751 (rs6000_compare_version_priority): Likewise.
22752 (rs6000_get_function_versions_dispatcher): Likewise.
22753 (make_resolver_func): Likewise.
22754 (add_condition_to_bb): Likewise.
22755 (dispatch_function_versions): Likewise.
22756 (rs6000_generate_version_dispatcher_body): Likewise.
22757 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22758 (fusion_gpr_load_p): Fix a spacing issue.
22759 * doc/extend.texi (Common Function Attributes): Document that the
22760 PowerPC supports the target_clone attribute.
22762 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22764 * config/arm/arm.h: explain F symbol found in description of ARM
22765 register allocation in its legend.
22767 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22769 * config/mips/frame-header-opt.c: Include profile-count.h.
22770 * config/riscv/riscv.c: Include profile-count.h
22772 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22774 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22776 (sm_set_flag_if_changed): Add bbs field.
22777 (execute_sm_if_changed_flag_set): Pass BBS.
22778 (execute_sm): Update.
22780 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22782 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22785 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22787 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22788 (peephole2): New peephole2 to emit the above.
22789 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22791 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22793 * config/aarch64/aarch64.c (define_peephole2 above
22794 *sub_<shift>_<mode>): New peephole.
22796 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22798 * config/i386/i386.c (make_resolver_func): Update.
22799 * Makefile.in: Add profile-count.h and profile-count.o
22800 * auto-profile.c (afdo_indirect_call): Update to new API.
22801 (afdo_set_bb_count): Update.
22802 (afdo_propagate_edge): Update.
22803 (afdo_propagate_circuit): Update.
22804 (afdo_calculate_branch_prob): Update.
22805 (afdo_annotate_cfg): Update.
22806 * basic-block.h: Include profile-count.h
22807 (struct edge_def): Turn count to profile_count.
22808 (struct basic_block_def): Likewie.
22809 (REG_BR_PROB_BASE): Move to profile-count.h
22810 (RDIV): Move to profile-count.h
22811 * bb-reorder.c (max_entry_count): Turn to profile_count.
22812 (find_traces): Update.
22813 (rotate_loop):Update.
22814 (connect_traces):Update.
22815 (sanitize_hot_paths):Update.
22816 * bt-load.c (migrate_btr_defs): Update.
22817 * cfg.c (RDIV): Remove.
22818 (init_flow): Use alloc_block.
22819 (alloc_block): Uninitialize count.
22820 (unchecked_make_edge): Uninitialize count.
22821 (check_bb_profile): Update.
22822 (dump_edge_info): Update.
22823 (dump_bb_info): Update.
22824 (update_bb_profile_for_threading): Update.
22825 (scale_bbs_frequencies_int): Update.
22826 (scale_bbs_frequencies_gcov_type): Update.
22827 (scale_bbs_frequencies_profile_count): New.
22828 * cfg.h (update_bb_profile_for_threading): Update.
22829 (scale_bbs_frequencies_profile_count): Declare.
22830 * cfgbuild.c (compute_outgoing_frequencies): Update.
22831 (find_many_sub_basic_blocks): Update.
22832 * cfgcleanup.c (try_forward_edges): Update.
22833 (try_crossjump_to_edge): Update.
22834 * cfgexpand.c (expand_gimple_tailcall): Update.
22835 (construct_exit_block): Update.
22836 * cfghooks.c (verify_flow_info): Update.
22837 (dump_bb_for_graph): Update.
22838 (split_edge): Update.
22839 (make_forwarder_block): Update.
22840 (duplicate_block): Update.
22841 (account_profile_record): Update.
22842 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22843 (get_estimated_loop_iterations): Update.
22844 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22845 (single_likely_exit): Update.
22846 * cfgloopmanip.c (scale_loop_profile): Update.
22848 (set_zero_probability): Update.
22849 (lv_adjust_loop_entry_edge): Update.
22850 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22851 (purge_dead_edges): Update.
22852 (rtl_account_profile_record): Update.
22853 * cgraph.c (cgraph_node::create): Uninitialize count.
22854 (symbol_table::create_edge): Uninitialize count.
22855 (cgraph_update_edges_for_call_stmt_node): Update.
22856 (cgraph_edge::dump_edge_flags): Update.
22857 (cgraph_node::dump): Update.
22858 (cgraph_edge::maybe_hot_p): Update.
22859 * cgraph.h: Include profile-count.h
22860 (create_clone), create_edge, create_indirect_edge): Update.
22861 (cgraph_node): Turn count to profile_count.
22862 (cgraph_edge0: Likewise.
22863 (make_speculative, clone): Update.
22864 (create_edge): Update.
22865 (init_lowered_empty_function): Update.
22866 * cgraphclones.c (cgraph_edge::clone): Update.
22867 (duplicate_thunk_for_node): Update.
22868 (cgraph_node::create_clone): Update.
22869 * cgraphunit.c (cgraph_node::analyze): Update.
22870 (cgraph_node::expand_thunk): Update.
22871 * final.c (dump_basic_block_info): Update.
22872 * gimple-streamer-in.c (input_bb): Update.
22873 * gimple-streamer-out.c (output_bb): Update.
22874 * graphite.c (print_global_statistics): Update.
22875 (print_graphite_scop_statistics): Update.
22876 * hsa-brig.c: Include basic-block.h.
22877 * hsa-dump.c: Include basic-block.h.
22878 * hsa-gen.c (T sum_slice): Update.
22879 (convert_switch_statements):Update.
22880 * hsa-regalloc.c: Include basic-block.h.
22881 * ipa-chkp.c (chkp_produce_thunks): Update.
22882 * ipa-cp.c (struct caller_statistics): Update.
22883 (init_caller_stats): Update.
22884 (gather_caller_stats): Update.
22885 (ipcp_cloning_candidate_p): Update.
22886 (good_cloning_opportunity_p): Update.
22887 (get_info_about_necessary_edges): Update.
22888 (dump_profile_updates): Update.
22889 (update_profiling_info): Update.
22890 (update_specialized_profile): Update.
22891 (perhaps_add_new_callers): Update.
22892 (decide_about_value): Update.
22893 (ipa_cp_c_finalize): Update.
22894 * ipa-devirt.c (struct odr_type_warn_count): Update.
22895 (struct decl_warn_count): Update.
22896 (struct final_warning_record): Update.
22897 (possible_polymorphic_call_targets): Update.
22898 (ipa_devirt): Update.
22899 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22900 * ipa-icf.c (sem_function::merge): Update.
22901 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22902 * ipa-inline.c (compute_uninlined_call_time): Update.
22903 (compute_inlined_call_time): Update.
22904 (want_inline_small_function_p): Update.
22905 (want_inline_self_recursive_call_p): Update.
22906 (edge_badness): Update.
22907 (lookup_recursive_calls): Update.
22908 (recursive_inlining): Update.
22909 (inline_small_functions): Update.
22910 (dump_overall_stats): Update.
22911 (dump_inline_stats): Update.
22912 * ipa-profile.c (ipa_profile_generate_summary): Update.
22913 (ipa_propagate_frequency): Update.
22914 (ipa_profile): Update.
22915 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22916 * ipa-utils.c (ipa_merge_profiles): Update.
22917 * loop-doloop.c (doloop_modify): Update.
22918 * loop-unroll.c (report_unroll): Update.
22919 (unroll_loop_runtime_iterations): Update.
22920 * lto-cgraph.c (lto_output_edge): Update.
22921 (lto_output_node): Update.
22922 (input_node): Update.
22923 (input_edge): Update.
22924 (merge_profile_summaries): Update.
22925 * lto-streamer-in.c (input_cfg): Update.
22926 * lto-streamer-out.c (output_cfg): Update.
22927 * mcf.c (create_fixup_graph): Update.
22928 (adjust_cfg_counts): Update.
22929 (sum_edge_counts): Update.
22930 * modulo-sched.c (sms_schedule): Update.
22931 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22932 * predict.c (maybe_hot_count_p): Update.
22933 (probably_never_executed): Update.
22934 (dump_prediction): Update.
22935 (combine_predictions_for_bb): Update.
22936 (propagate_freq): Update.
22937 (handle_missing_profiles): Update.
22938 (counts_to_freqs): Update.
22939 (rebuild_frequencies): Update.
22940 (force_edge_cold): Update.
22941 * predict.h: Include profile-count.h
22942 (maybe_hot_count_p, counts_to_freqs): UPdate.
22943 * print-rtl-function.c: Do not include cfg.h
22944 * print-rtl.c: Include basic-block.h
22945 * profile-count.c: New file.
22946 * profile-count.h: New file.
22947 * profile.c (is_edge_inconsistent): Update.
22948 (correct_negative_edge_counts): Update.
22949 (is_inconsistent): Update.
22950 (set_bb_counts): Update.
22951 (read_profile_edge_counts): Update.
22952 (compute_frequency_overlap): Update.
22953 (compute_branch_probabilities): Update; Initialize and deinitialize
22955 (branch_prob): Update.
22956 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22957 (edge_gcov_count): New.
22958 (bb_gcov_count): New.
22959 * shrink-wrap.c (try_shrink_wrapping): Update.
22960 * tracer.c (better_p): Update.
22961 * trans-mem.c (expand_transaction): Update.
22962 (ipa_tm_insert_irr_call): Update.
22963 (ipa_tm_insert_gettmclone_call): Update.
22964 * tree-call-cdce.c: Update.
22965 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22966 (gimple_duplicate_sese_tail): Update.
22967 (gimple_account_profile_record): Update.
22968 (execute_fixup_cfg): Update.
22969 * tree-inline.c (copy_bb): Update.
22970 (copy_edges_for_bb): Update.
22971 (initialize_cfun): Update.
22972 (freqs_to_counts): Update.
22973 (copy_cfg_body): Update.
22974 (expand_call_inline): Update.
22975 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22976 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22977 (try_unroll_loop_completely): Update.
22978 (try_peel_loop): Update.
22979 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22980 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22981 * tree-ssa-loop-split.c (connect_loops): Update.
22982 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22983 * tree-ssa-reassoc.c (branch_fixup): Update.
22984 * tree-ssa-tail-merge.c (replace_block_by): Update.
22985 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22986 (compute_path_counts): Update.
22987 (update_profile): Update.
22988 (recompute_probabilities): Update.
22989 (update_joiner_offpath_counts): Update.
22990 (estimated_freqs_path): Update.
22991 (freqs_to_counts_path): Update.
22992 (clear_counts_path): Update.
22993 (ssa_fix_duplicate_block_edges): Update.
22994 (duplicate_thread_path): Update.
22995 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22996 (struct switch_conv_info): Update.
22997 * tree-tailcall.c (decrease_profile): Update.
22998 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22999 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23000 * value-prof.c (check_counter): Update.
23001 (gimple_divmod_fixed_value): Update.
23002 (gimple_mod_pow2): Update.
23003 (gimple_mod_subtract): Update.
23004 (gimple_ic_transform): Update.
23005 (gimple_stringop_fixed_value): Update.
23006 * value-prof.h (gimple_ic): Update.
23008 2017-06-02 Carl Love <cel@us.ibm.com>
23010 * config/rs6000/rs6000-c: Add support for built-in functions
23011 vector double vec_doublee (vector signed int);
23012 vector double vec_doublee (vector unsigned int);
23013 vector double vec_doublee (vector float);
23014 vector double vec_doubleh (vector signed int);
23015 vector double vec_doubleh (vector unsigned int);
23016 vector double vec_doubleh (vector float);
23017 vector double vec_doublel (vector signed int);
23018 vector double vec_doublel (vector unsigned int);
23019 vector double vec_doublel (vector float);
23020 vector double vec_doubleo (vector signed int);
23021 vector double vec_doubleo (vector unsigned int);
23022 vector double vec_doubleo (vector float);.
23023 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23024 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23026 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23027 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23028 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23030 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23031 vec_doublel, vec_doubleh.
23032 * doc/extend.texi: Update the built-in documentation file for the
23033 new built-in functions.
23035 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23038 * ipa-inline-analysis.c (free_growth_caches): Set
23039 edge_removal_hook_holder to NULL after removing it.
23041 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23043 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23044 comparision with zero.
23046 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23047 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23048 for early expansion of vec_min and vec_max builtins.
23049 (builtin_function_type): Add min/max unsigned variants to those
23050 identified as having unsigned arguments.
23052 2017-06-02 Olivier Hainque <hainque@adacore.com>
23054 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23056 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23058 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23059 Use VALL_F16 iterator rather than VALL.
23061 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23063 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23064 Emit CBNZ inside loop when doing a strong exchange and comparing
23065 against zero. Generate the CC flags after the loop.
23067 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23069 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23070 (dl_section_ref): New.
23071 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23072 On AIX, append an expression to subtract the size of the
23073 section length to dl_section_ref.
23075 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23077 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23078 for early expansion of vector absolute builtins.
23080 2017-06-02 Richard Biener <rguenther@suse.de>
23082 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23083 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23085 2017-06-02 Richard Biener <rguenther@suse.de>
23087 PR tree-optimization/80948
23088 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23089 stmt order as well.
23091 2017-06-02 Richard Biener <rguenther@suse.de>
23093 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23095 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23096 for inductions as relevant.
23098 2017-06-02 Richard Biener <rguenther@suse.de>
23100 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23101 (vectorizable_induction): ... this. Remove dead code.
23103 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23105 * builtins. (expand_builtin_alloca): Remove second parameter and
23106 infer its value from the first parameter instead.
23107 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23109 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23111 PR rtl-optimization/80903
23112 * loop-doloop.c (add_test): Unshare sequence.
23114 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23116 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23118 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23120 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23122 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23123 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23124 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23125 (xlogue_layout::get_stub_rtx): Make static.
23126 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23127 (xlogue_layout::compute_stub_managed_regs): Rename to...
23128 (xlogue_layout::count_stub_managed_regs): ...this.
23129 (xlogue_layout::is_stub_managed_reg): New function.
23130 (xlogue_layout::m_stub_names): Rename to...
23131 (xlogue_layout::s_stub_names): ...this, make static.
23132 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23133 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23134 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23135 xlogue_layout::s_stub_names): Instantiate statics.
23136 (stub_managed_regs): Remove.
23137 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23138 (disable_call_ms2sysv_xlogues): Rename to...
23139 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23140 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23142 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23143 change after reload_completed.
23144 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23146 (ix86_expand_prologue): Likewise.
23147 (ix86_expand_epilogue): Likewise.
23148 (ix86_expand_split_stack_prologue): Likewise.
23149 (ix86_compute_frame_layout): Remove frame parameter ...
23150 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23151 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23153 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23154 (ix86_frame): Move from here ...
23155 * config/i386/i386.h (ix86_frame): ... to here.
23156 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23157 complete ix86_frame data structure instead. Remove some_ld_name.
23159 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23161 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23162 symbols that hold a DECL_VALUE_EXPR.
23164 2017-06-01 Martin Jambor <mjambor@suse.cz>
23166 PR tree-optimization/80898
23167 * tree-sra.c (process_subtree_disqualification): Removed.
23168 (disqualify_candidate): Do not acll
23169 process_subtree_disqualification.
23170 (subtree_mark_written_and_enqueue): New function.
23171 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23172 RHS has been disqualified and re-queue LHS if necessary. Apart
23173 from that, ignore disqualified RHS.
23175 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23177 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23178 address fetch for z10 or later.
23180 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23182 * config/arc/arc.md (tst_movb): Add guard when splitting.
23184 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23186 * config/arc/arc.c (arc_can_eliminate): Test against
23187 arc_frame_pointer_needed.
23189 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23191 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23192 to prevent store reordering.
23193 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23194 (type): Add block type.
23195 (stack_tie): Define special instruction to be used in
23198 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23200 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23201 constraint. It is not valid for the pattern.
23202 (noncommutative_binary_comparison): Likewise.
23204 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23206 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23209 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23211 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23212 be used by the reg-alloc.
23214 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23216 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23217 reg-alloc when having mul64 or mul32x16 instructions.
23218 (mulsidi3): Likewise.
23219 (umulsidi3): Likewise.
23220 (mulsi32x16): New pattern.
23221 (mulsi64): Likewise.
23222 (mulsidi64): Likewise.
23223 (umulsidi64): Likewise.
23224 (MUL32x16_REG): Define.
23225 (mul64_600): Use MUL32x16_REG.
23226 (mac64_600): Likewise.
23227 (umul64_600): Likewise.
23228 (umac64_600): Likewise.
23230 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23232 * config/arc/arc.md (mulsi3_700): Make it commutative.
23234 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23236 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23238 (*sign_extendsidi2_insn): Likewise for movstosw.
23240 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23242 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23243 the type of the input discriminant value. Convert the
23244 discriminant value of signedness vary.
23246 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23248 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23249 Add to -Wall section.
23251 2017-06-01 Richard Biener <rguenther@suse.de>
23253 PR middle-end/66313
23254 * fold-const.c (fold_plusminus_mult_expr): If the factored
23255 factor may be zero use a wrapping type for the inner operation.
23256 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23257 and handle moved defs.
23258 (process_assignment): Properly guard the unary op case. Return a
23259 tri-state indicating that moving the stmt before the call may allow
23260 to continue. Pass through to_move.
23261 (find_tail_calls): Handle moving unrelated defs before
23264 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23267 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23268 splitter result in the canonical way.
23270 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23272 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23273 also for 32bit target. Update insn attributes.
23274 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23276 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23278 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23279 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23280 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23281 (_mm_maskz_min_ss): New intrinsics.
23283 2017-05-31 Martin Liska <mliska@suse.cz>
23285 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23286 Change comment style to one we normally use.
23287 (vectorizable_reduction): Likewise.
23288 (vectorizable_induction): Likewise.
23289 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23290 (vectorizable_call): Likewise.
23291 (vectorizable_simd_clone_call): Likewise.
23292 (vectorizable_conversion): Likewise.
23293 (vectorizable_assignment): Likewise.
23294 (vectorizable_shift): Likewise.
23295 (vectorizable_operation): Likewise.
23296 (vectorizable_store): Likewise.
23297 (vectorizable_load): Likewise.
23298 * tree-vectorizer.h: Likewise.
23300 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23302 * passes.c (emergency_dump_function): New.
23303 * tree-pass.h (emergency_dump_function): Declare.
23304 * plugin.c (plugins_internal_error_function): Remove.
23305 * plugin.h (plugins_internal_error_function): Remove declaration.
23306 * toplev.c (internal_error_function): New static function. Use it...
23307 (general_init): ...here.
23309 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23311 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23312 (arc_rtx_costs): Add costs for new patterns.
23313 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23314 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23316 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23318 * tree-ssa-strlen.c (get_next_strinfo): New function.
23319 (get_stridx_plus_constant): Use it.
23320 (zero_length_string): Likewise.
23321 (adjust_related_strinfos): Likewise.
23322 (adjust_last_stmt): Likewise.
23324 2017-05-31 Richard Biener <rguenther@suse.de>
23327 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23328 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23330 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23332 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23333 loop_vinfo argument and use of dependence distance vectors.
23334 Check instead whether the two references differ only in their
23335 initial value and assume that they have the same alignment if the
23336 difference is a multiple of the vector alignment.
23337 (vect_analyze_data_refs_alignment): Update call accordingly.
23339 2017-05-31 Martin Liska <mliska@suse.cz>
23342 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23344 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23346 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23347 (create_intersect_range_checks): Move from ...
23348 * tree-data-ref.c (create_intersect_range_checks_index)
23349 (create_intersect_range_checks): ... to here.
23350 (create_runtime_alias_checks): New function factored from ...
23351 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23352 here. Call above function.
23353 * tree-data-ref.h (create_runtime_alias_checks): New function.
23355 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23357 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23358 segment length for dr_b and compute it in wide_int.
23360 2017-05-31 Richard Biener <rguenther@suse.de>
23362 PR tree-optimization/80906
23363 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23364 and pass through iv_map.
23365 (copy_bb_and_scalar_dependences): Adjust.
23366 (translate_pending_phi_nodes): Likewise.
23367 (copy_loop_close_phi_args): Handle code-generating IVs instead
23370 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23372 * diagnostic-color.c (color_dict): Add "type-diff".
23373 (parse_gcc_colors): Update comment.
23374 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23375 -fdiagnostics-show-template-tree and -fno-elide-type.
23376 (GCC_COLORS): Add type-diff to example.
23378 (-fdiagnostics-show-template-tree): New.
23379 (-fno-elide-type): New.
23380 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23381 the pp_format_decoder callback. Call any m_format_postprocessor's
23383 (pretty_printer::pretty_printer): Initialize
23384 m_format_postprocessor.
23385 (pretty_printer::~pretty_printer): Delete any
23386 m_format_postprocessor.
23387 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23388 (class format_postprocessor): New class.
23389 (struct pretty_printer::format_decoder): Document the new parameters.
23390 (struct pretty_printer::m_format_postprocessor): New field.
23391 * tree-diagnostic.c (default_tree_printer): Update for new
23392 bool and const char ** params.
23393 * tree-diagnostic.h (default_tree_printer): Likewise.
23395 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23397 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23398 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23399 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23400 rs6000_gen_cell_microcode code.
23401 (rs6000_final_prescan_insn): Delete.
23402 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23403 "warn-cell-microcode" entries.
23404 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23405 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23406 throughout. Change cc_reg_not_micro_cr0_operand to
23407 cc_reg_not_cr0_operand throughout.
23408 (*extendhi<mode>2_noload): Delete.
23409 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23410 (mwarn-cell-microcode): Delete.
23411 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23412 -mgen-cell-microcode and -mwarn-cell-microcode.
23414 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23417 * config/i386/constraints.md (Yd): New constraint.
23419 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23420 and (?Yd, r) alternatives. Update insn attributes.
23421 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23422 and (?*Yd, r) alternatives. Update insn attributes.
23423 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23425 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23427 * gimplify.c (gimplify_modify_expr): Don't create a
23428 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23431 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23433 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23435 2017-05-30 Richard Biener <rguenther@suse.de>
23437 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23438 and reduc_def fields.
23439 (STMT_VINFO_REDUC_TYPE): New define.
23440 (STMT_VINFO_REDUC_DEF): Likewise.
23441 (vect_force_simple_reduction): Adjust prototype.
23442 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23443 (vect_is_simple_reduction): Remove check_reduction argument.
23444 (vect_force_simple_reduction): Adjust and set
23445 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23446 (vectorizable_reduction): Do not re-do reduction analysis
23447 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23448 * tree-parloops.c (gather_scalar_reductions): Adjust.
23450 2017-05-30 Richard Biener <rguenther@suse.de>
23452 PR middle-end/80901
23453 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23456 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23458 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23459 Introduce unknown_misalignment parameter and remove vf.
23460 (vect_peeling_hash_get_lowest_cost):
23461 Pass unknown_misalignment parameter.
23462 (vect_enhance_data_refs_alignment):
23463 Fix unsupportable data ref treatment.
23465 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23467 * tree-vect-data-refs.c (vect_get_data_access_cost):
23468 Workaround for SLP handling.
23469 (vect_enhance_data_refs_alignment):
23470 Compute costs for doing no peeling at all, compare to the best
23471 peeling costs so far and avoid peeling if cheaper.
23473 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23475 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23476 Return peeling info and set costs to zero for unlimited cost
23478 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23479 with unknown misalignment. Compute and costs for unknown
23480 misalignment, compare them to the costs for known misalignment
23481 and choose the cheapest for peeling.
23483 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23485 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23486 (vect_get_peeling_costs_all_drs): Create function.
23487 (vect_peeling_hash_get_lowest_cost):
23488 Use vect_get_peeling_costs_all_drs.
23489 (vect_peeling_supportable): Create function.
23490 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23492 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23494 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23495 DR_HAS_NEGATIVE_STEP.
23496 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23497 (vect_enhance_data_refs_alignment): Use.
23498 (vect_duplicate_ssa_name_ptr_info): Use.
23499 * tree-vectorizer.h (dr_misalignment): Use.
23500 (known_alignment_for_access_p): Use.
23502 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23505 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23507 (has_section_name): New function.
23509 2017-05-30 Martin Liska <mliska@suse.cz>
23512 * auto-profile.c (get_function_decl_from_block): Fix
23515 2017-05-30 Richard Biener <rguenther@suse.de>
23517 PR middle-end/80876
23518 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23520 2017-05-30 Martin Liska <mliska@suse.cz>
23522 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23523 * dumpfile.h (struct dump_file_info): Remove ctors.
23525 2017-05-30 Martin Liska <mliska@suse.cz>
23527 * predict.def: Fix GNU coding style.
23529 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23531 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23532 Mark 'to' argument with ATTRIBUTE_UNUSED.
23534 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23536 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23537 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23538 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23541 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23543 * doc/install.texi (Options specification): Restore entry of
23544 --enable-sjlj-exceptions.
23546 2017-05-27 Michael Eager <eager@eagercon.com>
23549 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23551 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23553 * config/microblaze/microblaze.h
23554 (FIXED_REGISTERS): Update in macro.
23555 (CALL_USED_REGISTERS): Update in macro.
23557 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23559 * doc/install.texi: Add links to macOS binary distributions.
23561 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23565 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23567 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23569 2017-05-26 Martin Liska <mliska@suse.cz>
23571 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23573 2017-05-26 Martin Liska <mliska@suse.cz>
23575 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23576 always leading ';; '.
23577 (dump_bb_info): Likewise.
23578 (brief_dump_cfg): Likewise.
23579 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23580 * dumpfile.c: Remove usage of TDF_VERBOSE.
23581 * dumpfile.h (enum dump_kind): Likewise.
23582 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23583 * print-tree.c (debug_verbose): Remove.
23584 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23585 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23586 * tree-diagnostic.c (default_tree_printer): Replace
23587 TDF_DIAGNOSTIC with TDF_SLIM.
23589 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23591 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23592 in parameter loop, rather than loop_vinfo.
23593 (create_intersect_range_checks): Ditto.
23594 (vect_create_cond_for_alias_checks): Update call to above functions.
23596 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23598 PR tree-optimization/80815
23599 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23600 for merging runtime alias checks. Handle negative DR_STEPs.
23602 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23604 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23606 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23607 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23608 out code pruning runtime alias checks.
23609 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23610 factored out from above.
23611 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23613 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23615 (prune_runtime_alias_test_list): New decalaration.
23617 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23619 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23620 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23621 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23622 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23623 (operator==, comp_dr_with_seg_len_pair): Ditto.
23624 (vect_prune_runtime_alias_test_list): Ditto.
23626 2017-05-26 Martin Liska <mliska@suse.cz>
23629 * params.def: Bound partial-inlining-entry-probability param.
23631 2017-05-26 Marek Polacek <polacek@redhat.com>
23634 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23637 2017-05-26 Richard Biener <rguenther@suse.de>
23639 PR tree-optimization/80842
23640 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23643 2017-05-26 Richard Biener <rguenther@suse.de>
23645 PR tree-optimization/80844
23646 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23648 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23650 * doc/md.texi (Machine Constraints): Update x86 family
23651 machine constraints section to match 'config/i386/constraints.md'.
23653 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23655 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23657 2017-05-25 Nathan Sidwell <nathan@acm.org>
23659 * doc/invoke.texi (--enable-languages): Update documentation.
23661 2017-05-25 Martin Liska <mliska@suse.cz>
23663 * dumpfile.c: Add TDF_FOLDING.
23664 * dumpfile.h (enum dump_kind): Likewise.
23665 * genmatch.c (dt_simplify::gen_1): Use it.
23667 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23669 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23671 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23673 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23674 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23676 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23678 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23679 * match.pd (X == C): Rewrite it here.
23680 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23681 with_certain_nonzero_bits2): New predicates.
23682 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23684 2017-05-24 Nathan Sidwell <nathan@acm.org>
23686 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23689 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23692 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23694 * config/powerpcspe: New port. Files are copied from the rs6000
23695 port, with "rs6000" in filenames replaced by "powerpcspe".
23697 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23699 PR rtl-optimization/80754
23700 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23702 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23704 * config/sparc/sparc.md (length): Return the correct value for -mflat
23705 sibcalls to match output_sibcall.
23707 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23711 * config/rs6000/rs6000.c (struct machine_function): Add new field
23713 (rs6000_get_separate_components): Init that field, use it.
23714 (rs6000_components_for_bb): Use the field.
23716 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23718 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23720 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23722 PR middle-end/80823
23723 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23725 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23728 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23729 address against address_operand predicate.
23730 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23732 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23734 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23735 parameters passed indirectly.
23737 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23739 * config/i386/i386.md (*movdi_internal): Remove SSE4
23740 alternative 18 (?r, *v). Update insn attributes.
23741 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23742 Update insn attributes.
23743 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23744 Update insn attributes.
23745 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23746 alternative 1 (r, v). Remove isa attribute.
23747 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23748 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23749 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23751 2017-05-23 Tom de Vries <tom@codesourcery.com>
23753 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23756 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23758 * cgraphunit.c (symbol_table::process_new_functions): Update.
23759 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23760 (inline_generate_summary): Rename to ...
23761 (ipa_fn_summary_generate): ... this one.
23762 (inline_read_summary): Rename to ...
23763 (ipa_fn_summary_read): ... this one.
23764 (inline_write_summary): Rename to ...
23765 (ipa_fn_summary_write): ... this one.
23766 (inline_free_summary): Rename to ...
23767 (ipa_free_fn_summary): ... this one.
23768 (pass_data_local_fn_summary, pass_local_fn_summary,
23769 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23770 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23771 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23772 make_pass_ipa_fn_summary): New.
23773 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23774 inline_write_summary, inline_free_summary): Remove.
23775 (ipa_free_fn_summary) : New.
23776 * ipa-inline.c (ipa_inline): Update.
23777 (pass_ipa_inline): Do not generate summaries.
23778 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23780 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23781 and add pass_ipa_fn_summary.
23782 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23784 (make_pass_inline_parameters): Remove.
23786 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23788 * omp-low.c (struct omp_context): Remove "default_kind" member.
23791 * omp-offload.c (execute_oacc_device_lower): Remove the
23792 parallelism dimensions function attributes for unparallelized
23793 OpenACC kernels constructs.
23795 2017-05-23 Martin Liska <mliska@suse.cz>
23797 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23799 (cgraph_edge::make_speculative): Likewise.
23800 (cgraph_edge::resolve_speculation): Likewise.
23801 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23802 (cgraph_node::dump): Likewise.
23803 * cgraph.h: Likewise.
23804 * cgraphunit.c (analyze_functions): Likewise.
23805 (symbol_table::compile): Likewise.
23806 * ipa-cp.c (print_all_lattices): Likewise.
23807 (determine_versionability): Likewise.
23808 (initialize_node_lattices): Likewise.
23809 (ipcp_verify_propagated_values): Likewise.
23810 (estimate_local_effects): Likewise.
23811 (update_profiling_info): Likewise.
23812 (create_specialized_node): Likewise.
23813 (perhaps_add_new_callers): Likewise.
23814 (decide_about_value): Likewise.
23815 (decide_whether_version_node): Likewise.
23816 (identify_dead_nodes): Likewise.
23817 (ipcp_store_bits_results): Likewise.
23818 * ipa-devirt.c (dump_targets): Likewise.
23819 (ipa_devirt): Likewise.
23820 * ipa-icf.c (sem_item::dump): Likewise.
23821 (sem_function::equals): Likewise.
23822 (sem_variable::equals): Likewise.
23823 (sem_item_optimizer::read_section): Likewise.
23824 (sem_item_optimizer::execute): Likewise.
23825 (congruence_class::dump): Likewise.
23826 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23827 (dump_inline_summary): Likewise.
23828 (estimate_node_size_and_time): Likewise.
23829 (inline_analyze_function): Likewise.
23830 * ipa-inline-transform.c (inline_call): Likewise.
23831 * ipa-inline.c (report_inline_failed_reason): Likewise.
23832 (want_early_inline_function_p): Likewise.
23833 (edge_badness): Likewise.
23834 (update_edge_key): Likewise.
23835 (inline_small_functions): Likewise.
23836 * ipa-profile.c (ipa_profile): Likewise.
23837 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23838 (ipa_make_edge_direct_to_target): Likewise.
23839 (remove_described_reference): Likewise.
23840 (ipa_impossible_devirt_target): Likewise.
23841 (propagate_controlled_uses): Likewise.
23842 (ipa_print_node_params): Likewise.
23843 (ipcp_transform_function): Likewise.
23844 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23845 (propagate_pure_const): Likewise.
23846 * ipa-reference.c (generate_summary): Likewise.
23847 (read_write_all_from_decl): Likewise.
23848 (propagate): Likewise.
23849 (ipa_reference_read_optimization_summary): Likewise.
23850 * ipa-utils.c (ipa_merge_profiles): Likewise.
23851 * ipa.c (walk_polymorphic_call_targets): Likewise.
23852 (symbol_table::remove_unreachable_nodes): Likewise.
23853 (ipa_single_use): Likewise.
23854 * passes.c (execute_todo): Likewise.
23855 * predict.c (drop_profile): Likewise.
23856 * symtab.c (symtab_node::get_dump_name): New function.
23857 (symtab_node::dump_name): Likewise.
23858 (symtab_node::dump_asm_name): Likewise.
23859 (symtab_node::dump_references): Likewise.
23860 (symtab_node::dump_referring): Likewise.
23861 (symtab_node::dump_base): Likewise.
23862 (symtab_node::debug_symtab): Likewise.
23863 * tree-sra.c (convert_callers_for_node): Likewise.
23864 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23865 * value-prof.c (init_node_map): Likewise.
23867 2017-05-23 Martin Liska <mliska@suse.cz>
23869 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23870 and symtab_node::debug_symtab to symbol_table::debug.
23871 * cgraphunit.c (analyze_functions): Use the renamed function.
23872 (symbol_table::compile): Likewise.
23873 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23874 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23875 * passes.c (execute_todo): Likewise.
23876 * symtab.c (symbol_table::dump): New function.
23877 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23879 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23881 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23882 that nonconst implies exec.
23884 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23886 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23887 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23888 (inline_edge_summary_vec): Turn into ...
23889 (ipa_call_summaries): ... this one.
23890 (redirect_to_unreachable, edge_set_predicate,
23891 evaluate_properties_for_edge, inline_summary_alloc,
23892 reset_ipa_call_summary, reset_inline_summary,
23893 inline_summary_t::duplicate): Update.
23894 (inline_edge_duplication_hook): Turn to ...
23895 (ipa_call_summary_t::duplicate): ... this one.
23896 (inline_edge_removal_hook): Turn to ...
23897 (ipa_call_summary_t::remove): ... this one.
23898 (dump_inline_edge_summary): Turn to ...
23899 (dump_ipa_call_summary): ... this one.
23900 (estimate_function_body_sizes): Update.
23901 (inline_update_callee_summaries): Update.
23902 (remap_edge_change_prob): Update.
23903 (remap_edge_summaries): Update.
23904 (inline_merge_summary): Update.
23905 (do_estimate_edge_time): Update.
23906 (inline_generate_summary): Update.
23907 (inline_read_section): Update.
23908 (inline_read_summary): Update.
23909 (inline_free_summary): Update.
23910 * ipa-inline.c (can_inline_edge_p): Update.
23911 (compute_inlined_call_time): Update.
23912 (want_inline_small_function_p): Update.
23913 (edge_badness): Update.
23914 (early_inliner): Update.
23915 * ipa-inline.h (inline_edge_summary): Turn to ...
23916 (ipa_call_summary): ... this one.
23917 (ipa_call_summary_t): New class.
23918 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23919 (ipa_call_summaries): New.
23920 (inline_edge_summary): Remove.
23921 (estimate_edge_growth): Update.
23922 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23923 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23924 * ipa-split.c (execute_split_functions): Update.
23925 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23927 2017-05-23 Tom de Vries <tom@codesourcery.com>
23929 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23930 attributes): Document rdrand effective target.
23932 2017-05-23 Tom de Vries <tom@codesourcery.com>
23934 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23935 attributes): Sort alphabetically.
23937 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23939 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23941 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23944 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23945 V2DF/V2DI splat into two separate patterns, one that handles
23946 registers, and the other that only handles memory. Drop support
23947 for splatting from a GPR on ISA 2.07 and then splitting the
23948 splat into direct move and splat.
23949 (vsx_splat_<mode>_reg): Likewise.
23950 (vsx_splat_<mode>_mem): Likewise.
23952 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23954 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23956 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23958 PR middle-end/80809
23959 * omp-low.c (finish_taskreg_remap): New function.
23960 (finish_taskreg_scan): If unit size of ctx->record_type
23961 is non-constant, unshare the size expression and replace
23962 decls in it with possible outer var refs.
23964 PR middle-end/80809
23965 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23966 GOVD_SHARED rather than GOVD_PRIVATE with it.
23967 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23968 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23970 PR middle-end/80853
23971 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23972 as last argument to build_outer_var_ref for pointer bases of array
23973 section reductions.
23975 2017-05-19 Martin Sebor <msebor@redhat.com>
23977 * print-tree.c (print_node): Print DECL_READ_P flag.
23979 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23981 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23982 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23983 * cgraph.c: Likewise.
23984 * cgraphunit.c: Likewise.
23985 * gengtype.c: Likewise.
23986 * ipa-cp.c: Likewise.
23987 * ipa-devirt.c: Likewise.
23988 * ipa-icf.c: Likewise.
23989 * ipa-predicate.c: Likewise.
23990 * ipa-profile.c: Likewise.
23991 * ipa-prop.c: Likewise.
23992 * ipa-split.c: Likewise.
23994 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23995 edge_predicate_pool, dump_inline_hints,
23996 inline_summary::account_size_time, redirect_to_unreachable,
23997 edge_set_predicate, set_hint_predicate,
23998 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23999 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24000 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24001 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24002 ipa_call_summary_t::remove, initialize_growth_caches,
24003 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24004 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24005 mark_modified, unmodified_parm_1, unmodified_parm,
24006 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24007 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24008 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24009 will_be_nonconstant_predicate, record_modified_bb_info,
24010 get_minimal_bb, record_modified, param_change_prob,
24011 phi_result_unknown_predicate, predicate_for_phi_result,
24012 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24013 estimate_function_body_sizes, compute_inline_parameters,
24014 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24015 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24016 inline_update_callee_summaries, remap_edge_change_prob,
24017 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24018 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24019 inline_analyze_function, inline_summary_t::insert,
24020 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24021 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24022 inline_free_summary): Move to ipa-fnsummary.h
24023 (predicate_t): Remove.
24024 * ipa-fnsummary.c: New file.
24025 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24026 (enum inline_hints_vals, inline_hints, agg_position_info,
24027 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24028 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24029 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24030 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24031 inline_read_summary, inline_write_summary, inline_free_summary,
24032 inline_analyze_function, initialize_inline_failed,
24033 inline_merge_summary, inline_update_overall_summary,
24034 compute_inline_parameters): Move to ipa-fnsummary.h
24035 * ipa-fnsummary.h: New file.
24036 * ipa-inline-transform.h: Include ipa-inline.h.
24037 * ipa-inline.c: LIkewise.
24039 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24041 * ipa-inline.c (edge_badness): Use inlined_time instead of
24042 inline_summaries->get.
24044 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24046 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24048 2017-05-22 Nathan Sidwell <nathan@acm.org>
24050 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24051 (fdump-lang): Document 'raw' option.
24052 * dumpfile.h (TDI_tu): Delete.
24053 * dumpfile.c (dump_files): Remove translation-unit.
24054 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24056 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24058 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24059 command option from $(AWK) call.
24060 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24061 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24062 [FORMAT]: Remove handling of variable.
24063 * config/avr/t-multilib: Regenerate.
24065 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24067 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24069 (dump_inline_summary): Do not print self_time.
24070 (estimate_function_body_sizes): Do not set self_time.
24071 (compute_inline_parameters): Likewise.
24072 (inline_read_section, inline_write_summary): Do not stream self_time.
24073 * ipa-inline.h (inline_summary): Drop self_time.
24075 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24077 * ipa-inline-analysis.c (account_size_time): Rename to ...
24078 (inline_summary::account_size_time): ... this one.
24079 (reset_ipa_call_summary): Turn to ...
24080 (ipa_call_summary::reset): ... this one.
24081 (reset_inline_summary): Turn to ...
24082 (inline_summary::reset): ... this one.
24083 (inline_summary_t::remove): Update.
24084 (inline_summary_t::duplicate): Update.
24085 (ipa_call_summary_t::remove): Update.
24086 (dump_inline_summary): Update.
24087 (estimate_function_body_sizes): Update.
24088 (compute_inline_parameters): Update.
24089 (estimate_node_size_and_time): Update.
24090 (inline_merge_summary): Update.
24091 (inline_update_overall_summary): Update.
24092 (inline_read_section): Update.
24093 (inline_write_summary): Update.
24094 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24095 add account_size_time and reset member functions.
24096 (ipa_call_summary): Add reset function.
24097 * ipa-predicate.h (predicate::operator &): Constify.
24099 2017-05-22 Richard Biener <rguenther@suse.de>
24101 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24103 2017-05-19 Jason Merrill <jason@redhat.com>
24105 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24107 2017-05-19 Marek Polacek <polacek@redhat.com>
24110 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24111 TYPE_OVERFLOW_WRAPS checks.
24113 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24115 * tree-core.h (enum omp_clause_default_kind): Add
24116 "OMP_CLAUSE_DEFAULT_PRESENT".
24117 * tree-pretty-print.c (dump_omp_clause): Handle it.
24118 * gimplify.c (enum gimplify_omp_var_data): Add
24119 "GOVD_MAP_FORCE_PRESENT".
24120 (gimplify_adjust_omp_clauses_1): Map it to
24121 "GOMP_MAP_FORCE_PRESENT".
24122 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24124 * gimplify.c (oacc_default_clause): Clarify.
24126 2017-05-19 Nathan Sidwell <nathan@acm.org>
24128 LANG_HOOK_REGISTER_DUMPS
24129 * toplev.c (general_init): Call register dump lang hook.
24130 * doc/invoke.texi: Document -fdump-lang option family.
24131 * dumpfile.c (dump_files): Remove class dump here.
24132 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24133 * dumpfile.h (tree_dump_index): Remove TDI_class.
24134 * langhooks-def.h (lhd_register_dumps): Declare.
24135 (LANG_HOOKS_REGISTER_DUMPS): Define.
24136 (LANG_HOOKS_INITIALIZER): Add it.
24137 * langhooks.c (lhd_register_dumps): Define.
24138 * langhooks.h (struct lang_hooks): Add register_dumps.
24140 2017-05-19 Nathan Sidwell <nathan@acm.org>
24142 * context.h (context::set_passes): New.
24143 * context.c (context::context): Do not create pass manager.
24144 * toplev.c (general_init): Create pass manager here.
24146 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24148 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24149 use this splitter if two add or or instructions would also work for
24150 the constant we want to generate.
24152 2017-05-19 Richard Biener <rguenther@suse.de>
24155 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24156 predicate evaluation.
24158 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24160 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24162 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24165 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24167 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24168 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24169 (inline_edge_summary_vec): Turn into ...
24170 (ipa_call_summaries): ... this one.
24171 (redirect_to_unreachable, edge_set_predicate,
24172 evaluate_properties_for_edge, inline_summary_alloc,
24173 reset_ipa_call_summary, reset_inline_summary,
24174 inline_summary_t::duplicate): Update.
24175 (inline_edge_duplication_hook): Turn to ...
24176 (ipa_call_summary_t::duplicate): ... this one.
24177 (inline_edge_removal_hook): Turn to ...
24178 (ipa_call_summary_t::remove): ... this one.
24179 (dump_inline_edge_summary): Turn to ...
24180 (dump_ipa_call_summary): ... this one.
24181 (estimate_function_body_sizes): Update.
24182 (inline_update_callee_summaries): Update.
24183 (remap_edge_change_prob): Update.
24184 (remap_edge_summaries): Update.
24185 (inline_merge_summary): Update.
24186 (do_estimate_edge_time): Update.
24187 (inline_generate_summary): Update.
24188 (inline_read_section): Update.
24189 (inline_read_summary): Update.
24190 (inline_free_summary): Update.
24191 * ipa-inline.c (can_inline_edge_p): Update.
24192 (compute_inlined_call_time): Update.
24193 (want_inline_small_function_p): Update.
24194 (edge_badness): Update.
24195 (early_inliner): Update.
24196 * ipa-inline.h (inline_edge_summary): Turn to ...
24197 (ipa_call_summary): ... this one.
24198 (ipa_call_summary_t): New class.
24199 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24200 (ipa_call_summaries): New.
24201 (inline_edge_summary): Remove.
24202 (estimate_edge_growth): Update.
24203 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24204 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24205 * ipa-split.c (execute_split_functions): Update.
24206 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24208 2017-05-19 Richard Biener <rguenther@suse.de>
24210 PR middle-end/80764
24211 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24213 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24215 * config/rs6000/rs6000.c (struct machine_function): Add field
24216 fpr_is_wrapped_separately.
24217 (rs6000_get_separate_components): Use 64 components. Handle the
24218 new FPR components.
24219 (rs6000_components_for_bb): Handle the FPR components.
24220 (rs6000_emit_prologue_components): Handle the FPR components.
24221 (rs6000_emit_epilogue_components): Handle the FPR components.
24222 (rs6000_set_handled_components): Handle the FPR components.
24223 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24224 that are already separately shrink-wrapped.
24225 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24226 that are already separately shrink-wrapped.
24228 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24231 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24234 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24235 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24236 cases where the register allocator uses a move and an offsettable
24237 memory operation to/from a FPR register on ISA 2.06/2.07.
24238 (define_peephole2 for Altivec d-form store): Likewise.
24240 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24243 * config/i386/mmx.md (*mov<mode>_internal): Enable
24244 alternatives 11, 12, 13 and 14 also for 32bit targets.
24245 Remove alternatives 15, 16, 17 and 18.
24246 * config/i386/sse.md (vec_concatv2di): Change
24247 alternative (!x, *y) to (x, ?!*Yn).
24249 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24251 * dumpfile.h (enum dump_kind): Remove stray comma.
24253 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24255 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24256 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24257 predicate::num_conditions
24258 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24259 (CHANGED): turn into predicate::changed.
24260 (agg_position_info): Move to ipa-predicate.h
24261 (add_condition, predicate::add_clause, predicate::operator &=,
24262 predicate::or_with, predicate::evaluate, predicate::probability,
24263 dump_condition, dump_clause, predicate::dump,
24264 predicate::remap_after_duplication, predicate::remap_after_inlining,
24265 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24266 (evaluate_conditions_for_known_args): Update.
24267 (set_cond_stmt_execution_predicate): Update.
24268 * ipa-inline.h: Include ipa-predicate.h
24269 (condition, inline_param_summary, conditions, agg_position_info,
24270 predicate): Move to ipa-predicate.h
24271 * ipa-predicate.c: New file.
24272 * ipa-predicate.h: New file.
24274 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24276 * final.c (leaf_function_p): Check we are not in a sequence.
24278 2017-05-18 Martin Liska <mliska@suse.cz>
24280 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24281 * dumpfile.c (dump_register): Use new enum dump_kind.
24282 (get_dump_file_name): Likewise.
24283 (dump_enable_all): Likewise.
24284 (dump_switch_p_1): Likewise.
24285 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24286 * dumpfile.h (enum dump_kind): New enum type.
24287 (struct dump_file_info): Create constructor and
24288 format fields and comments.
24289 * passes.c (pass_manager::register_one_dump_file):
24291 * statistics.c (statistics_early_init): Likewise.
24292 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24293 TDF_TREE with TDF_SLIM.
24294 (gather_memory_references_ref): Likewise.
24296 2017-05-18 Martin Liska <mliska@suse.cz>
24298 * vec.h (struct vnull): Use it.
24300 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24302 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24303 (true_predicate, false_predicate, true_predicate_p,
24304 false_predicate_p): Remove.
24305 (single_cond_predicate, not_inlined_predicate): Turn to member function
24307 (add_condition): Update.
24308 (add_clause): Turn to...
24309 (predicate::add_clause): ... this one; update; allow passing NULL
24311 (and_predicates): Turn to ...
24312 (predicate::operator &=): ... this one.
24313 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24314 (or_predicates): Turn to ...
24315 (predicate::or_with): ... this one.
24316 (evaluate_predicate): Turn to ...
24317 (predicate::evaluate): ... this one.
24318 (predicate_probability): Turn to ...
24319 (predicate::probability): ... this one.
24320 (dump_condition): Update.
24321 (dump_predicate): Turn to ...
24322 (predicate::dump): ... this one.
24323 (account_size_time): Update.
24324 (edge_set_predicate): Update.
24325 (set_hint_predicate): UPdate.
24326 (evaluate_conditions_for_known_args): Update.
24327 (evaluate_properties_for_edge): Update.
24328 (remap_predicate_after_duplication): Turn to...
24329 (predicate::remap_after_duplication): ... this one.
24330 (remap_hint_predicate_after_duplication): Update.
24331 (inline_summary_t::duplicate): UPdate.
24332 (dump_inline_edge_summary): Update.
24333 (dump_inline_summary): Update.
24334 (set_cond_stmt_execution_predicate): Update.
24335 (set_switch_stmt_execution_predicate): Update.
24336 (compute_bb_predicates): Update.
24337 (will_be_nonconstant_expr_predicate): Update.
24338 (will_be_nonconstant_predicate): Update.
24339 (phi_result_unknown_predicate): Update.
24340 (predicate_for_phi_result): Update.
24341 (array_index_predicate): Update.
24342 (estimate_function_body_sizes): Update.
24343 (estimate_node_size_and_time): Update.
24344 (estimate_ipcp_clone_size_and_time): Update.
24345 (remap_predicate): Rename to ...
24346 (predicate::remap_after_inlining): ... this one.
24347 (remap_hint_predicate): Update.
24348 (inline_merge_summary): Update.
24349 (inline_update_overall_summary): Update.
24350 (estimate_size_after_inlining): Update.
24351 (read_predicate): Rename to ...
24352 (predicate::stream_in): ... this one.
24353 (read_inline_edge_summary): Update.
24354 (write_predicate): Rename to ...
24355 (predicate::stream_out): ... this one.
24356 (write_inline_edge_summary): Update.
24357 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24358 (clause_t): Turn to uint32_t
24359 (predicate): Turn to class; implement constructor and operators
24361 (size_time_entry): Update.
24362 (inline_summary): Update.
24363 (inline_edge_summary): Update.
24365 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24367 * fold-const.c (fold_binary_loc): Move transformation...
24368 * match.pd (C - X CMP X): ... here.
24370 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24372 * config/sparc/sparc.c (sparc_option_override): Set function
24373 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24374 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24376 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24378 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24380 2017-05-18 Marek Polacek <polacek@redhat.com>
24383 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24384 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24385 gimple_assign_load_p.
24387 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24389 PR middle-end/80692
24390 * real.c (do_compare): Give decimal_do_compare preference over
24391 comparing just the signs.
24393 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24395 * doc/md.texi (Canonicalization of Instructions): Describe the
24396 canonical form of instructions that inherently set a condition
24399 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24401 PR middle-end/80775
24402 * tree-cfg.c: Move deletion of unreachable case statements to after
24403 the merging of consecutive case labels.
24405 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24407 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24408 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24409 restoring of callee-saved registers.
24411 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24413 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24414 * config/visium/visium.c (single_set_and_flags): Likewise.
24415 * config/visium/visium.md (Substitutions): Likewise.
24417 2017-05-17 Martin Liska <mliska@suse.cz>
24419 * cfg.c: Introduce dump_flags_t type and
24420 use it instead of int type.
24422 * cfghooks.c: Likewise.
24423 * cfghooks.h (struct cfg_hooks): Likewise.
24424 * cfgrtl.c: Likewise.
24425 * cfgrtl.h: Likewise.
24426 * cgraph.c (cgraph_node::get_body): Likewise.
24427 * coretypes.h: Likewise.
24428 * domwalk.c: Likewise.
24429 * domwalk.h: Likewise.
24430 * dumpfile.c (struct dump_option_value_info): Likewise.
24431 (dump_enable_all): Likewise.
24432 (dump_switch_p_1): Likewise.
24433 (opt_info_switch_p): Likewise.
24434 * dumpfile.h (enum tree_dump_index): Likewise.
24435 (struct dump_file_info): Likewise.
24436 * genemit.c: Likewise.
24437 * generic-match-head.c: Likewise.
24438 * gengtype.c (open_base_files): Likewise.
24439 * gimple-pretty-print.c: Likewise.
24440 * gimple-pretty-print.h: Likewise.
24441 * graph.c (print_graph_cfg): Likewise.
24442 * graphite-scop-detection.c (dot_all_sese): Likewise.
24443 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24444 * loop-unroll.c (report_unroll): Likewise.
24445 * passes.c (pass_manager::register_one_dump_file): Likewise.
24446 * print-tree.c: Likewise.
24447 * statistics.c: Likewise.
24448 * tree-cfg.c: Likewise.
24449 * tree-cfg.h: Likewise.
24450 * tree-dfa.c: Likewise.
24451 * tree-dfa.h: Likewise.
24452 * tree-dump.c (dump_function): Likewise.
24453 * tree-dump.h (struct dump_info): Likewise.
24454 * tree-pretty-print.c: Likewise.
24455 * tree-pretty-print.h: Likewise.
24456 * tree-ssa-live.c: Likewise.
24457 * tree-ssa-live.h: Likewise.
24458 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24459 * tree-vect-loop.c: Likewise.
24460 * tree-vect-slp.c: Likewise.
24462 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24463 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24465 PR tree-optimization/80457
24466 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24467 of all arguments to a statement as scalar_to_vec operations.
24468 (vectorizable_call): Adjust call to vect_model_simple_cost for
24470 (vectorizable_conversion): Likewise.
24471 (vectorizable_assignment): Likewise.
24472 (vectorizable_shift): Likewise.
24473 (vectorizable_operation): Likewise.
24474 (vectorizable_comparison): Likewise.
24475 (vect_is_simple_cond): Record the def types for operands.
24476 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24477 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24478 for statement argument count.
24480 2017-05-16 Carl Love <cel@us.ibm.com>
24482 * config/rs6000/rs6000-c: Add support for built-in functions
24483 vector unsigned long long vec_bperm (vector unsigned long long,
24484 vector unsigned char)
24485 vector signed long long vec_mule (vector signed int,
24487 vector unsigned long long vec_mule (vector unsigned int,
24488 vector unsigned int)
24489 vector signed long long vec_mulo (vector signed int,
24491 vector unsigned long long vec_mulo (vector unsigned int,
24492 vector unsigned int)
24493 vector signed char vec_sldw (vector signed char,
24494 vector signed char,
24496 vector unsigned char vec_sldw (vector unsigned char,
24497 vector unsigned char,
24499 vector signed short vec_sldw (vector signed short,
24500 vector signed short,
24502 vector unsigned short vec_sldw (vector unsigned short,
24503 vector unsigned short,
24505 vector signed int vec_sldw (vector signed int,
24508 vector unsigned int vec_sldw (vector unsigned int,
24509 vector unsigned int,
24511 vector signed long long vec_sldw (vector signed long long,
24512 vector signed long long,
24514 vector unsigned long long vec_sldw (vector unsigned long long,
24515 vector unsigned long long,
24517 * config/rs6000/rs6000-c: Add support for built-in functions
24518 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24519 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24520 * doc/extend.texi: Update the built-in documentation for the
24521 new built-in functions.
24523 2017-05-16 Marek Polacek <polacek@redhat.com>
24527 * tree.c (save_expr): Don't fold the expression.
24529 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24531 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24532 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24533 and (?*y,m). Update insn attributes.
24535 2017-05-16 Martin Liska <mliska@suse.cz>
24537 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24538 flags argument of print_gimple_stmt, print_gimple_expr,
24539 print_generic_stmt and print_generic_expr.
24540 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24541 * coretypes.h: Likewise.
24542 * except.c (dump_eh_tree): Likewise.
24543 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24544 * gimple-pretty-print.h: Likewise.
24545 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24546 (backprop::push_to_worklist): Likewise.
24547 (backprop::pop_from_worklist): Likewise.
24548 (backprop::process_use): Likewise.
24549 (backprop::intersect_uses): Likewise.
24550 (note_replacement): Likewise.
24551 * gimple-ssa-store-merging.c
24552 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24553 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24554 (pass_store_merging::execute): Likewise.
24555 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24556 (ssa_base_cand_dump_callback): Likewise.
24557 (dump_incr_vec): Likewise.
24558 (replace_refs): Likewise.
24559 (replace_mult_candidate): Likewise.
24560 (create_add_on_incoming_edge): Likewise.
24561 (create_phi_basis): Likewise.
24562 (insert_initializers): Likewise.
24563 (all_phi_incrs_profitable): Likewise.
24564 (introduce_cast_before_cand): Likewise.
24565 (replace_one_candidate): Likewise.
24566 * gimplify.c (gimplify_expr): Likewise.
24567 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24568 (set_rename): Likewise.
24569 (rename_uses): Likewise.
24570 (copy_loop_phi_nodes): Likewise.
24571 (add_close_phis_to_merge_points): Likewise.
24572 (copy_loop_close_phi_args): Likewise.
24573 (copy_cond_phi_args): Likewise.
24574 (graphite_copy_stmts_from_block): Likewise.
24575 (translate_pending_phi_nodes): Likewise.
24576 * graphite-poly.c (print_pdr): Likewise.
24577 (dump_gbb_cases): Likewise.
24578 (dump_gbb_conditions): Likewise.
24579 (print_scop_params): Likewise.
24580 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24581 (build_cross_bb_scalars_use): Likewise.
24582 (gather_bbs::before_dom_children): Likewise.
24583 * hsa-dump.c (dump_hsa_immed): Likewise.
24584 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24585 (get_replacement_map): Likewise.
24586 * ipa-inline-analysis.c (dump_condition): Likewise.
24587 (estimate_function_body_sizes): Likewise.
24588 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24589 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24590 * ipa-prop.c (ipa_dump_param): Likewise.
24591 (ipa_print_node_jump_functions_for_edge): Likewise.
24592 (ipa_modify_call_arguments): Likewise.
24593 (ipa_modify_expr): Likewise.
24594 (ipa_dump_param_adjustments): Likewise.
24595 (ipa_dump_agg_replacement_values): Likewise.
24596 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24597 * ipa-pure-const.c (check_stmt): Likewise.
24598 (pass_nothrow::execute): Likewise.
24599 * ipa-split.c (execute_split_functions): Likewise.
24600 * omp-offload.c (dump_oacc_loop_part): Likewise.
24601 (dump_oacc_loop): Likewise.
24602 * trans-mem.c (tm_log_emit): Likewise.
24603 (tm_memopt_accumulate_memops): Likewise.
24604 (dump_tm_memopt_set): Likewise.
24605 (dump_tm_memopt_transform): Likewise.
24606 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24607 (print_loop): Likewise.
24608 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24609 (chkp_gather_checks_info): Likewise.
24610 (chkp_get_check_result): Likewise.
24611 (chkp_remove_check_if_pass): Likewise.
24612 (chkp_use_outer_bounds_if_possible): Likewise.
24613 (chkp_reduce_bounds_lifetime): Likewise.
24614 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24615 (chkp_mark_completed_bounds): Likewise.
24616 (chkp_register_incomplete_bounds): Likewise.
24617 (chkp_mark_invalid_bounds): Likewise.
24618 (chkp_maybe_copy_and_register_bounds): Likewise.
24619 (chkp_build_returned_bound): Likewise.
24620 (chkp_get_bound_for_parm): Likewise.
24621 (chkp_build_bndldx): Likewise.
24622 (chkp_get_bounds_by_definition): Likewise.
24623 (chkp_generate_extern_var_bounds): Likewise.
24624 (chkp_get_bounds_for_decl_addr): Likewise.
24625 * tree-chrec.c (chrec_apply): Likewise.
24626 * tree-data-ref.c (dump_data_reference): Likewise.
24627 (dump_subscript): Likewise.
24628 (dump_data_dependence_relation): Likewise.
24629 (analyze_overlapping_iterations): Likewise.
24630 * tree-inline.c (expand_call_inline): Likewise.
24631 (tree_function_versioning): Likewise.
24632 * tree-into-ssa.c (dump_defs_stack): Likewise.
24633 (dump_currdefs): Likewise.
24634 (dump_names_replaced_by): Likewise.
24635 (dump_update_ssa): Likewise.
24636 (update_ssa): Likewise.
24637 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24638 * tree-parloops.c (build_new_reduction): Likewise.
24639 (try_create_reduction_list): Likewise.
24640 (ref_conflicts_with_region): Likewise.
24641 (oacc_entry_exit_ok_1): Likewise.
24642 (oacc_entry_exit_single_gang): Likewise.
24643 * tree-pretty-print.h: Likewise.
24644 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24645 (get_scalar_evolution): Likewise.
24646 (add_to_evolution): Likewise.
24647 (get_loop_exit_condition): Likewise.
24648 (analyze_evolution_in_loop): Likewise.
24649 (analyze_initial_condition): Likewise.
24650 (analyze_scalar_evolution): Likewise.
24651 (instantiate_scev): Likewise.
24652 (number_of_latch_executions): Likewise.
24653 (gather_chrec_stats): Likewise.
24654 (final_value_replacement_loop): Likewise.
24655 (scev_const_prop): Likewise.
24656 * tree-sra.c (dump_access): Likewise.
24657 (disqualify_candidate): Likewise.
24658 (create_access): Likewise.
24659 (reject): Likewise.
24660 (maybe_add_sra_candidate): Likewise.
24661 (create_access_replacement): Likewise.
24662 (analyze_access_subtree): Likewise.
24663 (analyze_all_variable_accesses): Likewise.
24664 (sra_modify_assign): Likewise.
24665 (initialize_constant_pool_replacements): Likewise.
24666 (find_param_candidates): Likewise.
24667 (decide_one_param_reduction): Likewise.
24668 (replace_removed_params_ssa_names): Likewise.
24669 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24670 * tree-ssa-copy.c (dump_copy_of): Likewise.
24671 (copy_prop_visit_cond_stmt): Likewise.
24672 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24673 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24674 (record_equivalences_from_stmt): Likewise.
24675 * tree-ssa-dse.c (compute_trims): Likewise.
24676 (delete_dead_call): Likewise.
24677 (delete_dead_assignment): Likewise.
24678 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24679 (forward_propagate_into_cond): Likewise.
24680 (pass_forwprop::execute): Likewise.
24681 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24682 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24684 (move_computations_worker): Likewise.
24685 (execute_sm): Likewise.
24686 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24687 (remove_exits_and_undefined_stmts): Likewise.
24688 (remove_redundant_iv_tests): Likewise.
24689 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24690 (adjust_iv_update_pos): Likewise.
24691 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24692 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24693 (value_replacement): Likewise.
24694 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24695 * tree-ssa-pre.c (print_pre_expr): Likewise.
24696 (get_representative_for): Likewise.
24697 (create_expression_by_pieces): Likewise.
24698 (insert_into_preds_of_block): Likewise.
24699 (eliminate_insert): Likewise.
24700 (eliminate_dom_walker::before_dom_children): Likewise.
24701 (eliminate): Likewise.
24702 (remove_dead_inserted_code): Likewise.
24703 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24704 * tree-ssa-reassoc.c (get_rank): Likewise.
24705 (eliminate_duplicate_pair): Likewise.
24706 (eliminate_plus_minus_pair): Likewise.
24707 (eliminate_not_pairs): Likewise.
24708 (undistribute_ops_list): Likewise.
24709 (eliminate_redundant_comparison): Likewise.
24710 (update_range_test): Likewise.
24711 (optimize_range_tests_var_bound): Likewise.
24712 (optimize_vec_cond_expr): Likewise.
24713 (rewrite_expr_tree): Likewise.
24714 (rewrite_expr_tree_parallel): Likewise.
24715 (linearize_expr): Likewise.
24716 (break_up_subtract): Likewise.
24717 (linearize_expr_tree): Likewise.
24718 (attempt_builtin_powi): Likewise.
24719 (attempt_builtin_copysign): Likewise.
24720 (transform_stmt_to_copy): Likewise.
24721 (transform_stmt_to_multiply): Likewise.
24722 (dump_ops_vector): Likewise.
24723 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24724 (print_scc): Likewise.
24725 (set_ssa_val_to): Likewise.
24726 (visit_reference_op_store): Likewise.
24727 (visit_use): Likewise.
24728 (sccvn_dom_walker::before_dom_children): Likewise.
24729 (run_scc_vn): Likewise.
24730 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24732 (expr_hash_elt::print): Likewise.
24733 (const_and_copies::pop_to_marker): Likewise.
24734 (const_and_copies::record_const_or_copy_raw): Likewise.
24735 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24736 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24737 (dump_predicates): Likewise.
24738 (find_uninit_use): Likewise.
24739 (warn_uninitialized_phi): Likewise.
24740 (pass_late_warn_uninitialized::execute): Likewise.
24741 * tree-ssa.c (verify_vssa): Likewise.
24742 (verify_ssa): Likewise.
24743 (maybe_optimize_var): Likewise.
24744 * tree-vrp.c (dump_value_range): Likewise.
24745 (dump_all_value_ranges): Likewise.
24746 (dump_asserts_for): Likewise.
24747 (register_edge_assert_for_2): Likewise.
24748 (vrp_visit_cond_stmt): Likewise.
24749 (vrp_visit_switch_stmt): Likewise.
24750 (vrp_visit_stmt): Likewise.
24751 (vrp_visit_phi_node): Likewise.
24752 (simplify_cond_using_ranges_1): Likewise.
24753 (fold_predicate_in): Likewise.
24754 (evrp_dom_walker::before_dom_children): Likewise.
24755 (evrp_dom_walker::push_value_range): Likewise.
24756 (evrp_dom_walker::pop_value_range): Likewise.
24757 (execute_early_vrp): Likewise.
24759 2017-05-16 Richard Biener <rguenther@suse.de>
24761 * dwarf2out.c (loc_list_from_tree_1): Do not create
24762 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24764 2017-05-16 Richard Biener <rguenther@suse.de>
24766 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24768 (note_variable_value_in_expr): If we resolved the decl ref
24769 do not push to the stack.
24771 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24773 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24774 operations in fast-math mode.
24775 (vaddq_f16): Likewise.
24776 (vmul_f16): Likewise.
24777 (vmulq_f16): Likewise.
24778 (vsub_f16): Likewise.
24779 (vsubq_f16): Likewise.
24780 * config/arm/neon.md (add<mode>3): New.
24782 (fma:<VH:mode>3): New. Also remove outdated comment.
24785 2017-05-16 Martin Liska <mliska@suse.cz>
24789 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24790 (odr_types_equivalent_p): Likewise.
24792 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24794 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24796 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24799 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24800 non-interunit SSE move alternatives with '?'.
24801 (zero-extendsidi peephole2): New peephole to skip intermediate
24802 general register in SSE zero-extend sequence.
24804 2017-05-15 Jeff Law <law@redhat.com>
24806 * reorg.c (relax_delay_slots): Create a new variable to hold
24807 the temporary target rather than clobbering TARGET_LABEL.
24809 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24810 missing argument to extract_bit_field call.
24811 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24813 2017-05-15 Martin Liska <mliska@suse.cz>
24816 * gcc.c (process_command): Do not allow empty argument of -o option.
24818 2017-05-15 Renlin Li <renlin.li@arm.com>
24820 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24821 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24822 * config/aarch64/constraints.md (Usf): Add long call check.
24823 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24824 (call_value): Likewise.
24825 (sibcall): Likewise.
24826 (sibcall_value): Likewise.
24828 (call_value_insn): New.
24829 (sibcall_insn): Update rtx pattern.
24830 (sibcall_value_insn): Likewise.
24831 (call_internal): Remove.
24832 (call_value_internal): Likewise.
24833 (sibcall_internal): Likewise.
24834 (sibcall_value_internal): Likewise.
24835 (call_reg): Likewise.
24836 (call_symbol): Likewise.
24837 (call_value_reg): Likewise.
24838 (call_value_symbol): Likewise.
24840 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24843 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24845 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24847 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24848 compatible with CCGOCmode and with CCZmode.
24850 2017-05-14 Martin Sebor <msebor@redhat.com>
24852 PR middle-end/77671
24853 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24854 (gimple_fold_builtin_snprintf): Same.
24855 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24856 (gimple_fold_builtin_snprintf): Same.
24857 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24858 of character types.
24859 (is_call_safe): New function.
24860 (try_substitute_return_value): Call it.
24861 (try_simplify_call): New function.
24862 (pass_sprintf_length::handle_gimple_call): Call it.
24864 2017-05-14 Martin Sebor <msebor@redhat.com>
24866 PR middle-end/80669
24867 * builtins.c (expand_builtin_stpncpy): Simplify.
24869 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24871 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24872 * config/i386/i386.h
24873 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24874 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24875 (struct machine_function): Add new members call_ms2sysv,
24876 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24877 (struct machine_frame_state): New fields sp_realigned and
24878 sp_realigned_offset.
24879 * config/i386/i386.c
24880 (enum xlogue_stub): New enum.
24881 (enum xlogue_stub_sets): New enum.
24882 (class xlogue_layout): New class.
24883 (struct ix86_frame): New fields stack_realign_allocate_offset,
24884 stack_realign_offset and outlined_save_offset. Modify comments to
24885 detail stack layout when using out-of-line stubs.
24886 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24887 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24888 -mcall-ms2sysv-xlogues.
24889 (stub_managed_regs): New static variable.
24890 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24891 registers managed by out-of-line stub.
24892 (disable_call_ms2sysv_xlogues): New function.
24893 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24894 m->call_ms2sysv when appropriate and compute frame layout for
24896 (sp_valid_at, fp_valid_at): New inline functions.
24897 (choose_basereg): New function.
24898 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24900 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24901 Use align parameter of choose_baseaddr to generated aligned SSE movs
24903 (pro_epilogue_adjust_stack): Modify to track
24904 machine_frame_state::sp_realigned.
24905 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24906 (ix86_nsaved_sseregs): Likewise.
24907 (ix86_emit_save_regs): Likewise.
24908 (ix86_emit_save_regs_using_mov): Likewise.
24909 (ix86_emit_save_sse_regs_using_mov): Likewise.
24910 (get_scratch_register_on_entry): Likewise.
24911 (gen_frame_set): New function.
24912 (gen_frame_load): Likewise.
24913 (gen_frame_store): Likewise.
24914 (emit_outlined_ms2sysv_save): Likewise.
24915 (emit_outlined_ms2sysv_restore): Likewise.
24916 (ix86_expand_prologue): Modify stack re-alignment code and call
24917 emit_outlined_ms2sysv_save when appropriate.
24918 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24919 parameter rtx_insn *insn, which allows the function to be used to only
24920 generate the notes.
24921 (ix86_expand_epilogue): Modify validity checks of frame and stack
24922 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24923 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24924 * config/i386/predicates.md
24925 (save_multiple): New predicate.
24926 (restore_multiple): Likewise.
24927 * config/i386/sse.md
24928 (save_multiple<mode>): New pattern.
24929 (save_multiple_realign<mode>): Likewise.
24930 (restore_multiple<mode>): Likewise.
24931 (restore_multiple_and_return<mode>): Likewise.
24932 (restore_multiple_leave_return<mode>): Likewise.
24933 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24935 2017-05-14 Julia Koval <julia.koval@intel.com>
24937 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24938 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24939 (__builtin_ia32_xsetbv): New builtins.
24940 * config/i386/i386.c (ix86_expand_special_args_builtin):
24942 (ix86_expand_builtin): Special expand for new intrinsics.
24943 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24944 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24945 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24947 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24949 * cfganal.c (inverted_post_order_compute): Change argument type
24951 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24952 * df-core.c (rest_of_handle_df_initialize): Adjust.
24953 (rest_of_handle_df_finish): Likewise.
24954 (df_analyze_1): Likewise.
24955 (df_analyze): Likewise.
24956 (loop_inverted_post_order_compute): Change argument to be a vec *.
24957 (df_analyze_loop): Adjust.
24958 (df_get_n_blocks): Likewise.
24959 (df_get_postorder): Likewise.
24960 * df.h (struct df_d): Change field to be a vec.
24961 * lcm.c (compute_laterin): Adjust.
24962 (compute_available): Likewise.
24963 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24964 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24965 * tree-ssa-pre.c (compute_antic): Likewise.
24967 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24969 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24970 (depth_first_search::depth_first_search): Change structure init
24971 function to this constructor.
24972 (depth_first_search::add_bb): Rename function to this member.
24973 (depth_first_search::execute): Likewise.
24974 (flow_dfs_compute_reverse_finish): Adjust.
24976 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24978 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24979 (longest_simple_path): Likewise.
24980 * shrink-wrap.c (spread_components): Likewise.
24981 (disqualify_problematic_components): Likewise.
24982 (emit_common_heads_for_components): Likewise.
24983 (emit_common_tails_for_components): Likewise.
24984 (insert_prologue_epilogue_for_components): Likewise.
24986 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24988 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24991 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24993 * df-core.c (df_set_blocks): Start using auto_bitmap.
24994 (df_compact_blocks): Likewise.
24995 * df-problems.c (df_rd_confluence_n): Likewise.
24996 * df-scan.c (df_insn_rescan_all): Likewise.
24997 (df_process_deferred_rescans): Likewise.
24998 (df_update_entry_block_defs): Likewise.
24999 (df_update_exit_block_uses): Likewise.
25000 (df_entry_block_bitmap_verify): Likewise.
25001 (df_exit_block_bitmap_verify): Likewise.
25002 (df_scan_verify): Likewise.
25003 * lra-constraints.c (lra_constraints): Likewise.
25004 (undo_optional_reloads): Likewise.
25005 (lra_undo_inheritance): Likewise.
25006 * lra-remat.c (calculate_gen_cands): Likewise.
25007 (do_remat): Likewise.
25008 * lra-spills.c (assign_spill_hard_regs): Likewise.
25009 (spill_pseudos): Likewise.
25010 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25011 (bitmap_set_subtract_values): Likewise.
25013 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25015 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25016 management with auto_bitmap.
25017 (fix_inter_tick): Likewise.
25018 (fix_recovery_deps): Likewise.
25019 * ira.c (add_store_equivs): Likewise.
25020 (find_moveable_pseudos): Likewise.
25021 (split_live_ranges_for_shrink_wrap): Likewise.
25022 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25023 (rtx_reuse_manager::seen_def_p): Likewise.
25024 (rtx_reuse_manager::set_seen_def): Likewise.
25025 * print-rtl.h (class rtx_reuse_manager): Likewise.
25027 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25029 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25031 (migrate_btr_def): Likewise.
25032 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25033 * df-core.c (loop_post_order_compute): Likewise.
25034 (loop_inverted_post_order_compute): Likewise.
25035 * hsa-common.h: Likewise.
25036 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25037 * init-regs.c (initialize_uninitialized_regs): Likewise.
25038 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25039 (inline_small_functions): Likewise.
25040 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25041 * ira.c (combine_and_move_insns): Likewise.
25042 (build_insn_chain): Likewise.
25043 * loop-invariant.c (find_invariants): Likewise.
25044 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25045 * predict.c (tree_predict_by_opcode): Likewise.
25046 (predict_paths_leading_to): Likewise.
25047 (predict_paths_leading_to_edge): Likewise.
25048 (estimate_loops_at_level): Likewise.
25049 (estimate_loops): Likewise.
25050 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25051 (spread_components): Likewise.
25052 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25053 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25054 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25055 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25056 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25057 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25058 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25059 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25060 (mark_threaded_blocks): Likewise.
25061 (thread_through_all_blocks): Likewise.
25062 * tree-ssa.c (verify_ssa): Likewise.
25063 (execute_update_addresses_taken): Likewise.
25064 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25066 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25068 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25070 (post_order_compute): Likewise.
25071 (inverted_post_order_compute): Likewise.
25072 (pre_and_rev_post_order_compute_fn): Likewise.
25074 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25076 * genrecog.c (int_set::int_set): Explicitly construct our
25077 auto_vec base class.
25078 * vec.h (auto_vec::auto_vec): New constructor.
25080 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25082 * bitmap.h (class auto_bitmap): New constructor taking
25083 bitmap_obstack * argument.
25085 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25087 * bitmap.h (class auto_bitmap): Change type of m_bits to
25088 bitmap_head, and adjust ctor / dtor and member operators.
25090 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25092 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25093 when returned register mode doesn't match original mode.
25095 2017-05-12 Jeff Law <law@redhat.com>
25096 Jakub Jelinek <jakub@redhat.com>
25098 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25099 we look for cc setter after the compare-elim changes.
25100 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25101 within the vector to match what compare-elim now expects.
25102 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25103 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25105 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25106 after the compare-elim changes.
25107 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25108 the vector to match what compare-elim now expects.
25109 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25110 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25111 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25112 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25113 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25115 * config/visium/visium.c (single_set_and_flags): Fix where
25116 we look for cc setter after the compare-elim changes.
25117 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25118 with the vector to match what compare-elim now expects.
25119 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25120 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25121 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25122 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25123 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25124 (neg<mode>2_insn_set_overflow): Likewise.
25126 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25128 PR middle-end/79794
25129 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25130 maybe_expand_insn call, set ops[0].target. If still set after call,
25131 set alt_rtl. Add extra arg to recursive calls.
25132 (extract_bit_field): Add alt_rtl argument. Pass to
25134 * expmed.h (extract_bit_field): Fix prototype.
25135 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25136 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25137 to extract_bit_field_calls.
25138 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25139 Pass alt_rtl to extract_bit_field calls.
25140 * calls.c (store_unaligned_arguments_into_psuedos)
25141 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25142 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25144 * optabs.h (struct expand_operand): Add target bitfield.
25146 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25148 * compare-elim.c (try_eliminate_compare): Canonicalize
25149 operation with embedded compare to
25150 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25151 (set (reg) (operation)].
25153 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25155 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25158 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25159 cost of adding a carry flag for ADC instruction.
25160 [case MINUS]: Ignore the cost of subtracting a carry flag
25161 for SBB instruction.
25163 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25165 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25167 * config/rs6000/bmiintrin.h: New file.
25168 * config/rs6000/bmi2intrin.h: New file.
25169 * config/rs6000/x86intrin.h: New file.
25171 2017-05-12 Jeff Law <law@redhat.com>
25173 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25176 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25178 PR middle-end/80707
25179 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25181 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25183 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25184 early expansion of vector divide builtins.
25185 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25186 builtins identified as having unsigned arguments.
25188 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25190 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25191 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25192 expansion of vector logical operations (and, andc, or, xor,
25195 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25197 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25198 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25200 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25202 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25203 early GIMPLE expansion of vector multiplies.
25205 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25207 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25208 TARGET_HAVE_MOVT conditional.
25209 (movt splitter): Likewise.
25211 2017-05-12 Richard Biener <rguenther@suse.de>
25213 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25214 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25215 Fold all stmts not inplace.
25217 2017-05-12 Richard Biener <rguenther@suse.de>
25219 PR tree-optimization/80713
25220 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25221 inserted_exprs bit for not removed stmts.
25223 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25225 PR middle-end/69921
25226 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25227 parallelized" attribute for parallelized OpenACC kernels.
25228 * omp-offload.c (execute_oacc_device_lower): Use it.
25230 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25231 Set "oacc kernels" attribute.
25232 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25233 parameter. Adjust all users.
25234 (oacc_fn_attrib_kernels_p): Remove function.
25235 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25236 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25237 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25238 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25239 assert "oacc kernels" attribute is set.
25241 2017-05-11 Carl Love <cel@us.ibm.com>
25243 * config/rs6000/rs6000-c: Add support for built-in functions
25244 vector unsigned char vec_popcnt (vector signed char)
25245 vector unsigned char vec_popcnt (vector unsigned char)
25246 vector unsigned short vec_popcnt (vector signed short)
25247 vector unsigned short vec_popcnt (vector unsigned short)
25248 vector unsigned int vec_popcnt (vector signed int)
25249 vector unsigned int vec_popcnt (vector unsigned int)
25250 vector unsigned long long vec_popcnt (vector signed long long)
25251 vector unsigned long long vec_popcnt (vector unsigned long long)
25252 vector signed long long vec_slo (vector signed long long,
25253 vector signed char)
25254 vector signed long long vec_slo (vector signed long long,
25255 vector unsigned char)
25256 vector unsigned long long vec_slo (vector unsigned long long,
25257 vector signed char)
25258 vector unsigned long long vec_slo (vector unsigned long long,
25259 vector unsigned char)
25260 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25261 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25262 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25263 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25264 * doc/extend.texi: Update the built-in documentation file for the
25265 new built-in functions.
25267 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25269 * attribs.h (sorted_attr_string): Move machine independent
25270 functions for target clone support from the i386 port to common
25271 code. Rename ix86_function_versions to common_function_versions.
25272 Rename make_name to make_unique_name.
25273 (common_function_versions): Likewise.
25274 (make_unique_name): Likewise.
25275 (make_dispatcher_decl): Likewise.
25276 (is_function_default_version): Likewise.
25277 * attribs.c (attr_strcmp): Likewise.
25278 (sorted_attr_string): Likewise.
25279 (common_function_versions): Likewise.
25280 (make_unique_name): Likewise.
25281 (make_dispatcher_decl): Likewise.
25282 (is_function_default_version): Likewise.
25283 * config/i386/i386.c (attr_strcmp): Likewise.
25284 (sorted_attr_string): Likewise.
25285 (ix86_function_versions): Likewise.
25286 (make_name): Likewise.
25287 (make_dispatcher_decl): Likewise.
25288 (is_function_default_version): Likewise.
25289 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25291 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25294 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25295 Account for direct move costs for vec_construct of integer
25298 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25301 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25302 (UNSPEC_STX_ATOMIC): Ditto.
25303 (loaddi_via_sse): New insn.
25304 (storedi_via_sse): Ditto.
25305 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25306 Update corresponding peephole2 patterns.
25307 (atomic_storedi_fpu): Ditto.
25309 2017-05-11 Julia Koval <julia.koval@intel.com>
25311 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25312 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25314 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25315 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25316 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25318 2017-05-11 Nathan Sidwell <nathan@acm.org>
25320 * graphite-poly.c: Include dumpfile.h.
25322 * dumpfle.h (dump_function): Declare here ...
25323 * tree-dump.h (dump_function): ... not here.
25324 * dumpfile.c: #include tree-cfg.h.
25325 (dump_function): Move here from ...
25326 * tree-dump.c (dump_function): ... here.
25327 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25328 * graphite-poly.c: Don't include tree-dump.h.
25329 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25330 * print-tree.c: Likewise.
25331 * stor-layout.c: Likewise.
25332 * tree-nested.c: Likewise.
25334 * dumpfile.c (dump_start): Use TDF_FLAGS.
25335 (dump_enable_all): Fix TDF_KIND check thinko.
25337 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25339 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25340 array entries to represent two legal parameterizations of the
25341 overloaded __builtin_cmpb function, as represented by the
25342 P6_OV_BUILTIN_CMPB constant.
25343 (altivec_resolve_overloaded_builtin): Add special case handling
25344 for the __builtin_cmpb function, as represented by the
25345 P6_OV_BUILTIN_CMPB constant.
25346 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25347 (BU_P6_64BIT_2): New macro.
25348 (BU_P6_OVERLOAD_2): New macro
25349 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25350 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25351 (CMPB): Add overload support to represent both 32-bit and 64-bit
25352 compare-bytes function.
25353 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25354 support for TARGET_CMPB.
25355 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25356 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25357 documentation of the __builtin_cmpb overloaded built-in function.
25359 2017-05-11 Richard Biener <rguenther@suse.de>
25361 PR tree-optimization/80705
25362 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25363 bases are not vectorizable.
25365 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25367 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25368 when counting register pressure.
25370 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25372 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25373 (iv_ca_more_deps): Renamed to ...
25374 (iv_ca_compare_deps): ... this.
25375 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25377 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25379 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25381 (determine_group_iv_costs): ... here.
25382 (find_inv_vars_cb): Record inv var if it's not recorded before.
25384 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25386 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25387 (get_shiftadd_cost): Ditto.
25389 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25391 * tree-ssa-address.c: Include header file.
25392 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25394 (add_to_parts): Refactor.
25395 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25396 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25399 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25401 PR tree-optimization/53090
25402 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25404 (extract_cond_operands): Detect condition with IV on both sides
25405 and return COMP_IV_EXPR_2.
25406 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25407 (rewrite_use_compare): Simplify by removing call to function
25408 extract_cond_operands.
25410 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25412 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25413 (extract_cond_operands): Detect condition comparing against non-
25414 invariant bound and return appropriate enum value.
25415 (find_interesting_uses_cond): Update use of extract_cond_operands.
25416 Handle its return value accordingly.
25417 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25419 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25421 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25422 nonlinear iv_use computation in loop invariant sensitive way.
25424 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25426 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25427 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25429 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25431 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25432 (dump_cand): Support iv_cand.inv_exprs.
25433 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25435 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25438 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25440 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25442 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25443 as local function. Include necessary header files.
25444 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25446 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25448 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25450 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25452 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25453 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25454 RSHIFT_EXPR and BIT_NOT_EXPR.
25456 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25458 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25459 (adjust_setup_cost): New parameter supporting round up adjustment.
25460 (struct address_cost_data): Delete.
25461 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25462 (split_address_cost, ptr_difference_cost): Delete.
25463 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25464 (struct ainc_cost_data): New struct.
25465 (get_address_cost_ainc): New function.
25466 (get_address_cost, get_computation_cost): Reimplement.
25467 (determine_group_iv_cost_address): Record inv_expr for all uses of
25469 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25470 (iv_ca_has_deps): Reimplemented to ...
25471 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25473 (iv_ca_extend): Call iv_ca_more_deps.
25475 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25477 * tree-ssa-address.c (struct mem_address): Move to header file.
25478 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25479 * tree-ssa-address.h (struct mem_address): Move from C file.
25480 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25482 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25484 * tree-affine.h (aff_combination_type): New interface.
25485 (aff_combination_zero_p): Remove static.
25486 (aff_combination_const_p): New interface.
25487 (aff_combination_singleton_var_p): New interfaces.
25489 2017-05-11 Richard Biener <rguenther@suse.de>
25491 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25492 Skip unreachable blocks and destinations.
25493 (eliminate): Move stmt removal and fixup ...
25494 (fini_eliminate): ... here. Skip inserted exprs.
25495 (pass_pre::execute): Move fini_pre after fini_eliminate.
25496 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25497 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25498 PRE to get rid of dead code that has invalid SSA form and
25499 split critical edges again.
25501 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25503 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25505 2017-05-11 Richard Biener <rguenther@suse.de>
25507 * passes.c (execute_function_todo): Verify loops if they are
25508 said to be up-to-date.
25509 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25510 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25512 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25515 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25516 handle calling assemble_external ourself.
25519 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25520 modes with zero size. Enhance comment.
25522 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25524 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25525 built-ins for vec_xl and vec_xst with short and char pointer
25528 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25530 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25531 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25532 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25533 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25534 (_mm_maskz_min_round_ss): New intrinsics.
25535 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25536 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25537 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25538 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25539 (__builtin_ia32_minss_mask_round): New builtins.
25540 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25541 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25542 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25544 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25545 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25547 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25550 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25552 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25553 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25554 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25555 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25556 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25557 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25558 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25559 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25560 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25561 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25562 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25563 (__builtin_ia32_mulss_mask_round): New builtins.
25564 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25565 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25566 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25568 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25569 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25571 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25574 2017-05-10 Julia Koval <julia.koval@intel.com>
25576 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25577 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25578 (_mm256_setr_m128i): New intrinsics.
25580 2017-05-10 Julia Koval <julia.koval@intel.com>
25582 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25583 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25584 (_mm_maskz_rcp14_ss): New intrinsics.
25585 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25586 (__builtin_ia32_rcp14ss_mask): New builtins.
25587 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25589 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25591 PR tree-optimization/51513
25592 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25593 (assert_unreachable_fallthru_edge_p): Use it.
25594 (group_case_labels_stmt): Likewise.
25595 * tree-cfg.h: Prototype it.
25596 * stmt.c: Include cfghooks.h and tree-cfg.h.
25597 (emit_case_dispatch_table) <gap_label>: New local variable.
25598 Use it to fill dispatch table gaps.
25599 Test for default_label before updating probabilities.
25600 (expand_case) <default_label>: Remove unneeded initialization.
25601 Test for unreachable default case statement and remove its edge.
25602 Set default_label accordingly.
25603 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25605 2017-05-10 Carl Love <cel@us.ibm.com>
25607 * config/rs6000/rs6000-c: Add support for built-in functions
25608 vector signed char vec_neg (vector signed char)
25609 vector signed short int vec_neg (vector short int)
25610 vector signed int vec_neg (vector signed int)
25611 vector signed long long vec_neg (vector signed long long)
25612 vector float vec_neg (vector float)
25613 vector double vec_neg (vector double)
25614 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25616 * config/rs6000/altivec.h: Add define for vec_neg
25617 * doc/extend.texi: Update the built-in documentation for the
25618 new built-in functions.
25620 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25622 PR tree-optimization/77644
25623 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25625 2017-05-10 Nathan Sidwell <nathan@acm.org>
25627 * dumpfile.h (TDI_lang_all): New.
25628 (TDF_KIND): New. Renumber others
25629 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25631 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25633 (get_dump_file_name): Adjust suffix generation.
25634 (dump_enable_all): Use TDF_KIND.
25635 * doc/invoke.texi (-fdump-lang-all): Document.
25637 * dumpfile.h: Tabify.
25639 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25642 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25643 Move member access before delete.
25645 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25647 * tree-inline.c (expand_call_inline): Split block at stmt
25650 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25653 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25654 are now unused after splitting mov{sf,sd}_hardfloat.
25655 (f32_lr2): Likewise.
25656 (f32_lm): Likewise.
25657 (f32_lm2): Likewise.
25658 (f32_li): Likewise.
25659 (f32_li2): Likewise.
25660 (f32_lv): Likewise.
25661 (f32_sr): Likewise.
25662 (f32_sr2): Likewise.
25663 (f32_sm): Likewise.
25664 (f32_sm2): Likewise.
25665 (f32_si): Likewise.
25666 (f32_si2): Likewise.
25667 (f32_sv): Likewise.
25668 (f32_dm): Likewise.
25669 (f32_vsx): Likewise.
25670 (f32_av): Likewise.
25671 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25672 For movsf, order stores so the VSX stores occur before the GPR
25673 store which encourages the register allocator to use a traditional
25674 FPR instead of a GPR. For movsd, order the stores so that the GPR
25675 store comes before the VSX stores to allow the power6 to work.
25676 This is due to the power6 not having a 32-bit integer store
25677 instruction from a FPR.
25678 (movsf_hardfloat): Likewise.
25679 (movsd_hardfloat): Likewise.
25681 2017-05-09 Martin Sebor <msebor@redhat.com>
25683 PR translation/80280
25684 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25687 PR translation/80280
25688 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25689 data member added in r247778.
25690 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25692 2017-05-09 Nathan Sidwell <nathan@acm.org>
25694 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25696 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25699 2017-05-09 Marek Polacek <polacek@redhat.com>
25701 * doc/invoke.texi: Fix typo.
25703 2017-05-09 Richard Biener <rguenther@suse.de>
25705 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25706 (vrp_val_is_min): Likewise.
25707 (set_value_range_to_value): Likewise.
25708 (set_value_range_to_nonnegative): Likewise.
25709 (gimple_assign_nonzero_p): Likewise.
25710 (gimple_stmt_nonzero_p): Likewise.
25711 (vrp_int_const_binop): Likewise. Remove unreachable case.
25712 (adjust_range_with_scev): Adjust comments.
25713 (compare_range_with_value): Likewise.
25714 (extract_range_from_phi_node): Likewise.
25715 (test_for_singularity): Likewise.
25717 2017-05-09 Richard Biener <rguenther@suse.de>
25719 * tree-vrp.c (get_single_symbol): Add assert that we don't
25720 get overflowed constants as invariant part.
25721 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25722 checks. Use wi::cmp instead of recursing for integer constants.
25723 (compare_values): Just ignore whether we assumed undefined
25724 overflow instead of failing the compare.
25725 (extract_range_for_var_from_comparison_expr): Add comment before the
25726 TREE_NO_WARNING sets.
25727 (test_for_singularity): Likewise.
25728 (extract_range_from_comparison): Do not disable optimization
25729 when we assumed undefined overflow.
25730 (extract_range_basic): Remove init of unused var.
25732 2017-05-09 Richard Biener <rguenther@suse.de>
25734 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25735 (extract_range_from_multiplicative_op_1): Adjust.
25736 (extract_range_from_binary_expr_1): Use int_const_binop.
25738 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25741 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25742 rs6000_store_data_bypass_p in seven define_bypass directives and
25743 in several comments.
25744 * config/rs6000/rs6000-protos.h: Add prototype for
25745 rs6000_store_data_bypass_p function.
25746 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25747 function implements slightly different (rs6000-specific) semantics
25748 than store_data_bypass_p, returning false rather than aborting
25749 with assertion error when arguments do not satisfy the
25750 requirements of store data bypass.
25751 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25752 rs6000_store_data_bypass_p.
25754 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25756 * config/xtensa/xtensa-protos.h
25757 (xtensa_initial_elimination_offset): New declaration.
25758 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25759 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25760 macro definition, add case for FRAME_POINTER_REGNUM when
25761 FRAME_GROWS_DOWNWARD.
25762 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25763 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25764 xtensa_initial_elimination_offset.
25766 2017-05-08 Nathan Sidwell <nathan@acm.org>
25768 * doc/invoke.texi: Alphabetize -fdump options.
25770 2017-05-08 Martin Sebor <msebor@redhat.com>
25772 PR translation/80280
25773 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25775 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25777 * target.def (compute_frame_layout): New optional target hook.
25778 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25779 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25780 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25782 * reload1.c (verify_initial_elim_offsets): Likewise.
25783 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25784 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25785 (arm_get_frame_offsets): Split up into this ...
25786 (arm_compute_frame_layout): ... and this function.
25788 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25790 * config/aarch64/constraints.md (Usa): New constraint.
25791 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25793 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25795 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25796 with_multilib_list after it has been checked.
25798 2017-05-08 Richard Biener <rguenther@suse.de>
25800 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25801 (bitmap_set_subtract_values): Likewise.
25803 2017-05-08 Richard Biener <rguenther@suse.de>
25805 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25806 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25808 (gimple_stmt_nonzero_warnv_p): Rename to ...
25809 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25811 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25812 (extract_range_basic): Adjust, do not disable propagation on
25813 strict overflow sensitive simplification.
25814 (vrp_visit_cond_stmt): Likewise.
25816 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25818 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25819 body size unconditionally.
25821 2017-05-07 Jeff Law <law@redhat.com>
25824 2017-05-06 Jeff Law <law@redhat.com>
25825 PR tree-optimization/78496
25826 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25829 PR tree-optimization/78496
25830 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25831 (simplify_stmt_using_ranges): Call it.
25832 (vrp_dom_walker::before_dom_children): Extract equivalences
25833 from an ASSERT_EXPR with an equality comparison against a
25836 2017-05-06 Jeff Law <law@redhat.com>
25838 PR tree-optimization/78496
25839 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25842 PR tree-optimization/78496
25843 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25844 (simplify_stmt_using_ranges): Call it.
25845 (vrp_dom_walker::before_dom_children): Extract equivalences
25846 from an ASSERT_EXPR with an equality comparison against a
25849 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25851 * lra-constraints.c (lra_copy_reg_equiv): New function.
25852 (split_reg): Use it to copy equivalence information from the
25853 original register to the spill register.
25855 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25857 PR rtl-optimization/75964
25858 * simplify-rtx.c (simplify_const_relational_operation): Remove
25859 invalid handling of comparisons of integer ABS.
25861 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25863 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25864 initialize to zero.
25865 (init_regs): Remove declaration.
25866 (function_arg_advance_32): Initialize error_p as boolean variable.
25868 2017-05-05 Nathan Sidwell <nathan@acm.org>
25870 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25871 lines. Use for (;;).
25873 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25875 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25876 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25877 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25878 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25879 VF=2 that require versioning.
25881 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25883 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25886 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25888 * diagnostic.h (diagnostic_override_option_index): Convert from
25889 macro to inline function.
25891 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25893 * diagnostic.c (last_module_changed_p): New function.
25894 (set_last_module): New function.
25895 (diagnostic_report_current_module): Convert macro usage to
25896 the above functions.
25897 * diagnostic.h (diagnostic_context::last_module): Strengthen
25898 from const line_map * to const line_map_ordinary *.
25899 (diagnostic_last_module_changed): Delete macro.
25900 (diagnostic_set_last_module): Delete macro.
25902 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25904 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25905 with diagnostic_report_diagnostic.
25906 (diagnostic_n_impl_richloc): Likewise.
25907 * diagnostic.h (report_diagnostic): Delete macro.
25908 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25909 with diagnostic_report_diagnostic.
25910 * substring-locations.c (format_warning_va): Likewise.
25912 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25914 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25915 save/restor of format_spec. Move option-printing code to...
25916 (print_option_information): ...this new function, and
25917 reimplement by simply printing to the pretty_printer,
25918 rather than appending to the format string.
25920 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25922 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25923 handling logic into...
25924 (update_effective_level_from_pragmas): ...this new function.
25926 2017-05-04 Andrew Waterman <andrew@sifive.com>
25928 * config/riscv/riscv.opt (mstrict-align): New option.
25929 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25930 (SLOW_UNALIGNED_ACCESS): Define.
25931 (riscv_slow_unaligned_access): Declare.
25932 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25934 (riscv_slow_unaligned_access): New variable.
25935 (rocket_tune_info): Set slow_unaligned_access to true.
25936 (optimize_size_tune_info): Set slow_unaligned_access to false.
25937 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25938 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25939 (riscv_option_override): Set riscv_slow_unaligned_access.
25940 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25942 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25944 * config/riscv/riscv.md: Unify indentation.
25946 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25951 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25953 (extendsi<mode>2): Add support for doing sign extension via
25954 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25955 don't have ISA 3.0 instructions.
25956 (extendsi<mode>2 splitter): Likewise.
25957 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25958 generate the normal insns since SImode can now go in vector
25959 registers. Disallow the special UNSPECs needed for previous
25960 machines to hide SImode being used. Add new insns
25961 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25962 (fix_trunc<mode>si2_stfiwx): Likewise.
25963 (fix_trunc<mode>si2_internal): Likewise.
25964 (fixuns_trunc<mode>si2): Likewise.
25965 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25966 (fctiw<u>z_<mode>_smallint): Likewise.
25967 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25968 of floating point to 32-bit integer from doing a direct move to
25969 the GPR registers to do a store.
25970 (fctiwz_<mode>): Break long line.
25972 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25974 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25975 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25976 (addr_list, addr_offset_valid_p): New.
25977 (split_address_groups): Check offset validity with above function.
25978 (gt-tree-ssa-loop-ivopts.h): Include header file.
25980 2017-05-05 Nathan Sidwell <nathan@acm.org>
25982 * config.gcc (arm*-*-*): Add missing 'fi'.
25984 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25986 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25987 included in -fopt-info does not matter.
25988 * doc/optinfo.texi (-fopt-info): Fix description of default
25989 behavour. Explicitly say order of options included in -fopt-info
25992 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25994 * config.gcc: Allow combinations of aprofile and rmprofile values for
25995 --with-multilib-list.
25996 * config/arm/t-multilib: New file.
25997 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25998 variables. Remove setting of ISA and floating-point ABI in
25999 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26000 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26001 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26002 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26004 * config/arm/t-rmprofile: Likewise except for the matches changes.
26005 * doc/install.texi (--with-multilib-list): Document the combination of
26006 aprofile and rmprofile values and warn about pitfalls in doing that.
26008 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26010 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26011 (movdi_aarch64): Likewise.
26013 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26015 PR tree-optimization/80632
26016 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26018 (build_arrays): Initialize it for virtual phis.
26019 (fix_phi_nodes): Use it for virtual phis.
26021 PR tree-optimization/80558
26022 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26023 [x, y] op z into [x op, y op z] for op & or | if conditions
26026 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26027 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26030 * config/arm/arm.md (use_literal_pool): Remove.
26031 (64-bit immediate split): No longer takes cost into consideration
26032 if arm_disable_literal_pool is enabled.
26033 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26034 used when arm_disable_literal_pool is enabled.
26035 (arm_max_const_double_inline_cost): Remove use of
26036 arm_disable_literal_pool.
26037 (push_minipool_fix): Add assert.
26038 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26039 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26040 (no_literal_pool_sf_immediate): New.
26042 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26044 PR tree-optimization/80613
26045 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26046 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26048 2017-05-05 Richard Biener <rguenther@suse.de>
26050 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26052 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26054 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26055 of this flag from insn conditions due to removal from r247495.
26057 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26059 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26061 (arm_early_store_addr_dep_ptr): Likewise.
26062 * config/arm/aarch-common-protos.h
26063 (arm_early_load_addr_dep_ptr): Add prototype.
26064 (arm_early_store_addr_dep_ptr): Likewise.
26065 * config/arm/cortex-a53.md: Add new bypasses.
26067 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26069 * tree.c (next_type_uid): Change type to unsigned.
26070 (type_hash_canon): Decrement back next_type_uid if
26071 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26072 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26075 2017-05-04 Martin Sebor <msebor@redhat.com>
26077 * builtins.c: Fix a trivial typo in a comment.
26079 PR middle-end/79234
26080 * builtins.c (check_sizes): Adjust to handle reading past the end.
26081 Avoid printing excessive upper bound of ranges. Use %E to print
26082 tree nodes instead of converting them to %wu.
26083 (expand_builtin_memchr): New function.
26084 (compute_dest_size): Rename...
26085 (compute_objsize): ...to this.
26086 (expand_builtin_memcpy): Adjust.
26087 (expand_builtin_mempcpy): Adjust.
26088 (expand_builtin_strcat): Adjust.
26089 (expand_builtin_strcpy): Adjust.
26090 (check_strncat_sizes): Adjust.
26091 (expand_builtin_strncat): Adjust.
26092 (expand_builtin_strncpy): Adjust and simplify.
26093 (expand_builtin_memset): Adjust.
26094 (expand_builtin_bzero): Adjust.
26095 (expand_builtin_memcmp): Adjust.
26096 (expand_builtin): Handle memcmp.
26097 (maybe_emit_chk_warning): Check strncat just once.
26099 2017-05-04 Martin Sebor <msebor@redhat.com>
26101 PR preprocessor/79214
26102 PR middle-end/79222
26103 PR middle-end/79223
26104 * builtins.c (check_sizes): Add inlining context and issue
26105 warnings even when -Wno-system-headers is set.
26106 (check_strncat_sizes): Same.
26107 (expand_builtin_strncat): Same.
26108 (expand_builtin_memmove): New function.
26109 (expand_builtin_stpncpy): Same.
26110 (expand_builtin): Handle memmove and stpncpy.
26112 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26114 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26115 which is not used any more.
26117 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26119 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26121 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26123 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26124 (cortexa53_tunings): Likewise.
26125 (cortexa57_tunings): Likewise.
26126 (cortexa72_tunings): Likewise.
26127 (cortexa73_tunings): Likewise.
26129 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26131 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26132 Set loop alignment to 8.
26134 2017-05-04 Martin Sebor <msebor@redhat.com>
26136 PR translation/80280
26137 * builtins.c (expand_builtin_object_size): Add missing quoting to
26138 %D and like directives.
26139 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26140 (hsa_type_for_tree_type): Same.
26141 (verify_function_arguments): Same.
26142 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26143 * varasm.c (get_section): Same.
26146 2017-05-04 Martin Sebor <msebor@redhat.com>
26148 PR translation/80280
26149 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26151 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26153 * config/aarch64/aarch64.c (generic_addrcost_table):
26154 Change HI/TI mode setting.
26156 2017-05-04 Martin Jambor <mjambor@suse.cz>
26158 PR tree-optimization/80622
26159 * tree-sra.c (comes_initialized_p): New function.
26160 (build_accesses_from_assign): Only set write lazily when
26161 comes_initialized_p is false.
26162 (analyze_access_subtree): Use comes_initialized_p.
26163 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26164 instead of testing for PARM_DECL.
26166 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26168 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26169 constraint on operand 0 to allow more general addressing modes.
26170 Adjust output template.
26171 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26173 * config/aarch64/aarch64-protos.h
26174 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26175 * config/aarch64/constraints.md (Dp): New address constraint.
26176 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26179 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26181 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26182 update use of estimate_ipcp_clone_size_and_time.
26183 (estimate_local_effects): Update use of
26184 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26185 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26186 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26187 Return nonspecialized time.
26189 2017-05-04 Richard Biener <rguenther@suse.de>
26191 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26192 for the last VUSE which def dominates the PHI. Directly call
26194 (get_continuation_for_phi_1): Remove.
26196 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26198 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26199 to explain the use of truncating division. Cap the number of
26200 iterations to the maximum given by nb_iterations_upper_bound,
26203 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26205 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26206 * configure: Regenerate.
26207 * config.in: Regenerate.
26208 * config/i386/driver-mingw32.c: new file.
26209 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26210 * config.host: Link driver-mingw32.o on MinGW host.
26211 * doc/install.texi: Document new --enable-mingw-wildcard configure
26214 2017-05-04 Marek Polacek <polacek@redhat.com>
26216 PR tree-optimization/80612
26217 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26219 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26220 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26222 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26223 (movt splitter): Likewise.
26224 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26225 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26226 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26227 block for Thumb-1 with MOVT.
26228 (thumb2_legitimate_address_p): Move code block ...
26229 (can_avoid_literal_pool_for_label_p): ... into this new function.
26230 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26232 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26233 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26234 "M-profile targets with the MOVT instruction".
26236 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26238 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26239 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26240 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26242 2017-05-04 Martin Liska <mliska@suse.cz>
26244 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26245 variable cond_code.
26247 2017-05-04 Richard Biener <rguenther@suse.de>
26249 * tree.c (array_at_struct_end_p): Handle arrays at struct
26250 end with flexarrays more conservatively. Refactor and treat
26251 arrays of arrays or aggregates more strict. Fix
26252 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26253 * tree.c (array_at_struct_end_p): Adjust prototype.
26254 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26255 * gimple-fold.c (get_range_strlen): Likewise.
26256 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26258 2017-05-04 Richard Biener <rguenther@suse.de>
26260 PR tree-optimization/31130
26261 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26263 (supports_overflow_infinity): Likewise.
26264 (is_negative_overflow_infinity): Likewise.
26265 (is_positive_overflow_infinity): Likewise.
26266 (is_overflow_infinity): Likewise.
26267 (stmt_overflow_infinity): Likewise.
26268 (overflow_infinity_range_p): Likewise.
26269 (usable_range_p): Remove as always returning true.
26270 (make_overflow_infinity): Remove.
26271 (negative_overflow_infinity): Likewise.
26272 (positive_overflow_infinity): Likewise.
26273 (avoid_overflow_infinity): Likewise.
26274 (set_value_range): Adjust accordingly.
26275 (set_value_range_to_nonnegative): Likewise, remove now unused
26276 overflow_infinity arg.
26277 (vrp_operand_equal_p): Adjust.
26278 (update_value_range): Likewise.
26279 (range_int_cst_singleton_p): Likewise.
26280 (operand_less_p): Likewise.
26281 (compare_values_warnv): Likewise.
26282 (extract_range_for_var_from_comparison_expr): Likewise.
26283 (vrp_int_const_binop): Likewise.
26284 (zero_nonzero_bits_from_vr): Likewise.
26285 (extract_range_from_multiplicative_op_1): Likewise.
26286 (extract_range_from_binary_expr_1): Likewise.
26287 (extract_range_from_unary_expr): Likewise.
26288 (extract_range_from_comparison): Likewise.
26289 (extract_range_basic): Likewise.
26290 (adjust_range_with_scev): Likewise.
26291 (compare_ranges): Likewise.
26292 (compare_range_with_value): Likewise.
26293 (dump_value_range): Likewise.
26294 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26296 (simplify_cond_using_ranges): Adjust.
26298 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26300 * brig-builtins.def: Added a builtin for class_f64.
26301 * builtin-types.def: Added a builtin type needed by class_f64.
26303 2017-05-03 Jason Merrill <jason@redhat.com>
26305 * timevar.def: Add TV_CONSTEXPR.
26307 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26309 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26311 2017-05-03 Martin Jambor <mjambor@suse.cz>
26313 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26314 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26315 * ipa-cp.c (ipcp_propagate_stage): Do not call
26316 ipa_update_after_lto_read.
26317 * ipa-inline.c (ipa_inline): Likewise.
26319 2017-05-03 Martin Jambor <mjambor@suse.cz>
26321 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26322 tag. Added a default constructor and a destructor.
26323 (ipa_edge_args_sum_t): New class;
26324 (ipa_edge_args_sum): Declare.
26325 (ipa_edge_args_vector): Remove declaration.
26326 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26327 (ipa_free_edge_args_substructures): Remove declaration.
26328 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26329 (ipa_edge_args_info_available_for_edge_p): Likewise.
26330 * ipa-prop.c (ipa_edge_args_vector): Removed.
26331 (edge_removal_hook_holder): Likewise.
26332 (edge_duplication_hook_holder): Likewise.
26333 (ipa_edge_args_sum): New variable.
26334 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26335 ipa_edge_args_vector.
26336 (ipa_free_edge_args_substructures): Likewise.
26337 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26338 ipa_edge_args_vector.
26339 (ipa_edge_removal_hook): Turned into method
26340 ipa_edge_args_sum_t::remove.
26341 (ipa_edge_duplication_hook): Turned into method
26342 ipa_edge_args_sum_t::duplicate.
26343 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26344 registering edge hooks.
26345 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26346 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26347 ipa_edge_args_sum instead of ipa_edge_args_vector.
26348 * ipa-profile.c (ipa_profile): Likewise.
26350 2017-05-03 Martin Jambor <mjambor@suse.cz>
26352 * symbol-summary.h (function_summary): New method exists.
26353 (function_summary::symtab_removal): Deallocate through release.
26354 (call_summary): New class.
26355 (gt_ggc_mx): New overload.
26356 (gt_pch_nx): Likewise.
26357 (gt_pch_nx): Likewise.
26359 2017-05-03 Jeff Law <law@redhat.com>
26361 PR tree-optimization/78496
26362 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26363 from simplify_cond_using_ranges. Split off code to walk
26364 backwards through casts into ...
26365 (simplify_cond_using_ranges_2): New function.
26366 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26367 (execute_vrp): After identifying jump threads, call
26368 simplify_cond_using_ranges_2.
26370 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26373 * ipa-inline.h (inline_summary): Add ctor.
26374 (create_ggc): Do not use ggc_cleared_alloc.
26376 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26377 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26379 * gcc.c (handle_braces): Support escaping in switch matching
26381 * doc/invoke.texi (Spec Files): Document it.
26382 Remove superfluous @code markup in items.
26384 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26386 * diagnostic-show-locus.c (struct column_range): New struct.
26387 (get_affected_columns): New function.
26388 (get_printed_columns): New function.
26389 (struct correction): New struct.
26390 (correction::ensure_capacity): New function.
26391 (correction::ensure_terminated): New function.
26392 (struct line_corrections): New struct.
26393 (line_corrections::~line_corrections): New dtor.
26394 (line_corrections::add_hint): New function.
26395 (layout::print_trailing_fixits): Reimplement in terms of the new
26397 (selftest::test_overlapped_fixit_printing): New function.
26398 (selftest::diagnostic_show_locus_c_tests): Call it.
26400 2017-05-03 Nathan Sidwell <nathan@acm.org>
26402 Canonicalize canonical type hashing
26403 * tree.h (type_hash_canon_hash): Declare.
26404 * tree.c (type_hash_list, attribute_hash_list): Move into
26405 type_hash_canon_hash.
26406 (build_type_attribute_qual_variant): Break out hash code calc into
26407 type_hash_canon_hash.
26408 (type_hash_canon_hash): New. Generic type hash computation.
26409 (build_range_type_1, build_array_type_1, build_function_type,
26410 build_method_type_directly, build_offset_type, build_complex_type,
26411 make_vector_type): Call it.
26413 2017-05-03 Richard Biener <rguenther@suse.de>
26415 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26416 When all DRs have unknown misaligned do not always peel
26417 when there is a store but apply the same costing model as if
26418 there were only loads.
26420 2017-05-03 Richard Biener <rguenther@suse.de>
26423 PR tree-optimization/80492
26424 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26425 compare_base_decls returning dont-know properly.
26427 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26429 * config/arm/iterators.md (CCSI): New mode iterator.
26430 (arch): New mode attribute.
26431 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26432 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26433 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26434 code iterator for success result mode.
26435 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26436 the corresponding new insn generators.
26438 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26441 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26442 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26444 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26446 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26447 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26448 (DDR_A): Wrap DDR argument in brackets.
26449 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26450 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26451 (DDR_REVERSED_P): Likewise.
26453 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26455 PR tree-optimization/79472
26456 * tree-switch-conversion.c (struct switch_conv_info): Add
26457 contiguous_range and default_case_nonstandard fields.
26458 (collect_switch_conv_info): Compute contiguous_range and
26459 default_case_nonstandard fields, don't clear final_bb if
26460 contiguous_range and only the default case doesn't have the required
26462 (check_all_empty_except_final): Set default_case_nonstandard instead
26463 of failing if contiguous_range and the default case doesn't have empty
26465 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26466 and only the default case doesn't have the required constants. Skip
26468 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26469 if default_case_nonstandard.
26470 (build_constructors): Build constant 1 just once. Assert that default
26471 values aren't inserted in between cases if contiguous_range. Skip
26473 (build_arrays): Skip virtual phis.
26474 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26475 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26476 Handle virtual phis.
26477 (gen_inbound_check): Handle default_case_nonstandard case.
26478 (process_switch): Adjust check_final_bb caller. Call
26479 gather_default_values with the first non-default case instead of
26480 default case if default_case_nonstandard.
26482 2017-05-02 Nathan Sidwell <nathan@acm.org>
26484 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26485 check. Fix formatting.
26487 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26489 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26490 errors when comparing specialized and unspecialized times.
26492 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26494 * diagnostic-show-locus.c
26495 (layout::should_print_annotation_line_p): Make private.
26496 (layout::print_annotation_line): Make private.
26497 (layout::annotation_line_showed_range_p): Make private.
26498 (layout::show_ruler): Make private.
26499 (layout::print_source_line): Make private. Pass in line and
26500 line_width, rather than calling location_get_source_line. Drop
26502 (layout::print_leading_fixits): New method.
26503 (layout::print_any_fixits): Rename to...
26504 (layout::print_trailing_fixits): ...this, and make private.
26505 Don't print newline fixits.
26506 (diagnostic_show_locus): Move logic for printing one row into...
26507 (layout::print_line): ...this new function. Move the
26508 location_get_source_line call and error-handling from
26509 print_source_line to here. Call print_leading_fixits, and rename
26510 print_any_fixits to print_trailing_fixits.
26511 (selftest::test_fixit_insert_containing_newline): Update now that
26512 newlines are partially supported.
26513 (selftest::test_fixit_insert_containing_newline_2): New test.
26514 (selftest::test_fixit_replace_containing_newline): Update comments.
26515 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26516 * edit-context.c (class added_line): New class.
26517 (class edited_line): Describe newline handling in comment.
26518 (edited_line::actually_edited_p): New method.
26519 (edited_line::print_content): Delete redundant decl.
26520 (edited_line::m_predecessors): New field.
26521 (edited_file::print_content): Call edited_line::print_content.
26522 (edited_file::print_diff): Update to support newlines.
26523 (edited_file::print_diff_hunk): Likewise.
26524 (edited_file::print_run_of_changed_lines): New function.
26525 (edited_file::print_diff_line): Convert to...
26526 (print_diff_line): ...this.
26527 (edited_file::get_effective_line_count): New function.
26528 (edited_line::edited_line): Initialize new field m_predecessors.
26529 (edited_line::~edited_line): Clean up m_predecessors.
26530 (edited_line::apply_fixit): Handle newlines.
26531 (edited_line::get_effective_line_count): New function.
26532 (edited_line::print_content): New function.
26533 (edited_line::print_diff_lines): New function.
26534 (selftest::test_applying_fixits_insert_containing_newline): New
26536 (selftest::test_applying_fixits_replace_containing_newline): New
26538 (selftest::insert_line): New function.
26539 (selftest::test_applying_fixits_multiple_lines): Add example of
26541 (selftest::edit_context_c_tests): Call the new tests.
26543 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26545 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26546 parameter cand. Update dump information.
26547 (get_computation_cost): Update uses.
26549 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26551 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26552 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26553 (get_computation_at, rewrite_use_address): Update use of
26554 get_computation_aff.
26556 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26558 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26559 (get_computation): Delete.
26560 (get_computation_cost): Implement like get_computation_cost_at.
26561 Use get_computation_at.
26562 (get_computation_cost_at): Delete.
26563 (rewrite_use_nonlinear_expr): Use get_computation_at.
26564 (rewrite_use_compare, remove_unused_ivs): Ditto.
26566 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26568 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26570 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26572 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26573 (ivopts_global_cost_for_size): Rename parameter and update uses.
26574 (iv_ca_recount_cost): Update uses.
26575 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26576 candidates seperately in n_invs and n_cands.
26577 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26579 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26581 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26582 (find_inv_vars_cb): New.
26583 (find_depends): Renamed to ...
26584 (find_inv_vars): ... this.
26585 (add_candidate_1, force_var_cost): Call find_inv_vars.
26586 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26588 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26590 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26591 inv_vars. Add inv_exprs.
26592 (struct iv_cand): Rename depends_on to inv_vars.
26593 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26594 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26595 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26596 (dump_cand): Dump inv_vars.
26597 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26598 (record_invariant, find_depends, add_candidate_1): Ditto.
26599 (set_group_iv_cost, force_var_cost): Ditto.
26600 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26601 (get_computation_cost_at, get_computation_cost): Ditto.
26602 (determine_group_iv_cost_generic): Ditto.
26603 (determine_group_iv_cost_address): Ditto.
26604 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26605 (determine_group_iv_costs): Ditto.
26606 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26607 (iv_ca_set_remove_invariants): Renamed to ...
26608 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26609 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26610 (iv_ca_set_add_invariants): Renamed to ...
26611 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26612 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26613 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26614 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26615 (create_new_ivs): Remove useless dump.
26617 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26619 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26621 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26622 (iv_ca_set_no_cp, create_new_iv): Ditto.
26624 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26626 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26628 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26630 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26631 function tree_check2.
26633 2017-05-02 Martin Liska <mliska@suse.cz>
26635 * doc/gcov.texi: Add missing preposition.
26636 * gcov.c (function_info::function_info): Properly fill up
26637 all member variables.
26639 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26641 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26643 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26645 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26647 2017-05-02 Martin Liska <mliska@suse.cz>
26650 * lto-streamer-in.c (lto_read_tree_1): Remove
26651 LTO_STREAMER_DEBUG.
26652 * lto-streamer.c (struct tree_hash_entry): Likewise.
26653 (struct tree_entry_hasher): Likewise.
26654 (tree_entry_hasher::hash): Likewise.
26655 (tree_entry_hasher::equal): Likewise.
26656 (lto_streamer_init): Likewise.
26657 (lto_orig_address_map): Likewise.
26658 (lto_orig_address_get): Likewise.
26659 (lto_orig_address_remove): Likewise.
26660 * lto-streamer.h: Likewise.
26661 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26662 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26664 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26666 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26667 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26668 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26669 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26670 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26671 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26672 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26673 (mm_maskz_sub_ss): New intrinsics.
26674 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26675 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26676 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26677 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26678 (__builtin_ia32_subss_mask_round): New builtins.
26679 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26680 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26681 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26683 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26684 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26686 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26689 2017-05-02 Martin Jambor <mjambor@suse.cz>
26691 PR tree-optimization/78687
26692 * tree-sra.c (access): New field parent.
26693 (process_subtree_disqualification): New function.
26694 (disqualify_candidate): Call it.
26695 (build_accesses_from_assign): Reset write flag if creating an
26697 (build_access_subtree): Fill in parent field and also prpagate
26698 down grp_write flag.
26699 (create_artificial_child_access): New parameter set_grp_write, set
26700 grp_write to its value.
26701 (propagate_subaccesses_across_link): Also propagate grp_write flag
26703 (propagate_all_subaccesses): Push the closest parent back to work
26704 queue if add_access_to_work_queue returned true.
26706 2017-05-02 Richard Biener <rguenther@suse.de>
26708 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26709 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26710 -fstrict-overflow documentation.
26711 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26712 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26713 flag_strict_overflow.
26714 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26715 * lto-opts.c (lto_write_options): Do not stream it.
26716 * lto-wrapper.c (merge_and_complain): Do not handle it.
26717 * opts.c (default_options_table): Do not set -fstrict-overflow.
26718 (finish_options): Likewise do not clear it when sanitizing.
26719 * simplify-rtx.c (simplify_const_relational_operation): Do not
26720 test flag_strict_overflow.
26722 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26724 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26725 using enabled attribute.
26726 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26727 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26728 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26729 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26730 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26731 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26732 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26733 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26734 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26735 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26737 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26739 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26741 2017-05-02 Richard Biener <rguenther@suse.de>
26743 PR tree-optimization/80591
26745 2017-04-10 Richard Biener <rguenther@suse.de>
26747 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26750 2017-05-02 Richard Biener <rguenther@suse.de>
26752 PR tree-optimization/80549
26753 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26754 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26755 headers if they do not have a preheader.
26757 2017-05-02 Martin Liska <mliska@suse.cz>
26760 * common.opt: Fix typo.
26761 * doc/invoke.texi: Likewise.
26763 2017-05-01 Jan Beulich <jbeulich@suse.com>
26765 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26766 swapping, add (x,x,m,x,n) alternative.
26768 2017-05-01 Nathan Sidwell <nathan@acm.org>
26770 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26771 unnecessary unadjusted_alignment check.
26773 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26776 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26778 * gimplify.c (gimplify_cilk_detach): New function.
26779 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26780 * tree-core.h: Document EXPR_CILK_SPAWN.
26781 * tree.h (EXPR_CILK_SPAWN): Define.
26783 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26785 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26786 to use new fixit_hint representation, using the "replace" logic.
26787 (get_line_span_for_fixit_hint): Likewise.
26788 (layout::print_any_fixits): Likewise.
26789 (selftest::test_one_liner_many_fixits): Rename to...
26790 (selftest::test_one_liner_many_fixits_1): ...this, and update
26791 comment and expected output to reflect that the multiple fix-it
26792 hints are now consolidated into one insertion.
26793 (selftest::test_one_liner_many_fixits_2): New test.
26794 (selftest::test_diagnostic_show_locus_one_liner): Update for
26796 (selftest::test_fixit_consolidation): Update for fix-it API
26798 * diagnostic.c (print_parseable_fixits): Likewise.
26799 * edit-context.c (edited_line::m_line_events): Convert from
26800 auto_vec <line_event *> to auto_vec <line_event>.
26801 (class line_event): Convert from abstract base class to a concrete
26802 class, taking over the role of replace_event.
26803 (class insert_event): Delete.
26804 (class replace_event): Rename to class line_event. Convert to
26806 (edit_context::add_fixits): Reimplement.
26807 (edit_context::apply_insert): Delete.
26808 (edit_context::apply_replace): Rename to...
26809 (edit_context::apply_fixit): ...this. Convert to half-open range.
26810 (edited_file::apply_insert): Delete.
26811 (edited_file::apply_replace): Rename to...
26812 (edited_file::apply_fixit): ...this.
26813 (edited_line::~edited_line): Drop deletion of events.
26814 (edited_line::apply_insert): Delete.
26815 (edited_line::apply_replace): Rename to...
26816 (edited_line::apply_fixit): ...this. Convert to half-open range.
26817 Update for change to type of m_line_events.
26818 * edit-context.h (edit_context::apply_insert): Delete.
26819 (edit_context::apply_replace): Rename to...
26820 (edit_context::apply_fixit): ...this.
26822 2017-05-01 Martin Sebor <msebor@redhat.com>
26824 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26827 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26830 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26831 __get_cpuid_max returns 0.
26832 (__get_cpuid_count): Ditto.
26834 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26836 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26837 replacement expression is another instance of one of its arguments.
26839 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26842 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26843 check for stack push/pop autoinc.
26844 * config/i386/i386.c (ix86_agi_dependent): Return false
26845 if the only reason why modified_in_p returned true is that
26846 addr is SP based and set_insn is a push or pop.
26848 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26850 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26853 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26856 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26857 (account_size_time): Use two predicates - exec_pred and
26859 (evaluate_conditions_for_known_args): Compute both clause and
26861 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26862 (inline_summary_t::duplicate): Update.
26863 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26865 (compute_inline_parameters): Likewise.
26866 (estimate_edge_size_and_time): Update caluclation of time.
26867 (estimate_node_size_and_time): Compute both time and nonspecialized
26869 (estimate_ipcp_clone_size_and_time): Update.
26870 (inline_merge_summary): Update.
26871 (do_estimate_edge_time): Update.
26872 (do_estimate_edge_size): Update.
26873 (do_estimate_edge_hints): Update.
26874 (inline_read_section, inline_write_summary): Stream both new predicates.
26875 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26877 (compute_inlined_call_time): Cleanup.
26878 (big_speedup_p): Update.
26879 (edge_badness): Update.
26880 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26881 (size_time_entry): Replace predicate by exec_predicate and
26882 nonconst_predicate.
26883 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26884 (estimate_edge_time): Return also nonspec_time.
26885 (reset_edge_growth_cache): Update.
26887 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26889 PR rtl-optimization/80491
26890 * ifcvt.c (noce_process_if_block): When looking for x setter
26891 with missing else_bb, don't check only the insn right before
26892 cond_earliest, but look for the last insn that x is modified in
26893 within the same bb.
26895 PR rtl-optimization/80491
26896 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26898 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26900 PR tree-optimization/80487
26901 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26903 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26905 PR tree-optimization/79697
26906 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26907 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26908 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26910 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26911 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26913 2017-04-28 Martin Sebor <msebor@redhat.com>
26915 PR tree-optimization/80523
26916 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26917 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26919 (maybe_warn, format_directive, parse_directive): Use new functions.
26920 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26922 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26924 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26926 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26928 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26929 target_header_dir): Set correctly.
26930 * configure: Regenerated.
26931 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26932 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26933 instead of SYSTEM_HEADER_DIR.
26935 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26937 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26938 (estimate_local_effects): Likewise.
26939 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26940 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26941 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26942 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26943 do_estimate_edge_time, estimate_edge_time): Likewise.
26944 * ipa-inline-analysis.c (estimate_node_size_and_time,
26945 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26946 (estimate_time_after_inlining): Remove.
26948 2017-04-28 Martin Liska <mliska@suse.cz>
26950 * doc/gcov.texi: Enhance documentation of gcov.
26952 2017-04-28 Martin Liska <mliska@suse.cz>
26954 * doc/gcov.texi: Sort options in alphabetic order.
26955 * doc/gcov-dump.texi: Likewise.
26956 * doc/gcov-tool.texi: Likewise.
26957 * gcov.c (print_usage): Likewise.
26958 * gcov-dump.c (print_usage): Likewise.
26959 * gcov-tool.c (print_merge_usage_message): Likewise.
26960 (print_rewrite_usage_message): Likewise.
26961 (print_overlap_usage_message): Likewise.
26963 2017-04-28 Martin Liska <mliska@suse.cz>
26965 PR gcov-profile/53915
26966 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26968 2017-04-28 Martin Liska <mliska@suse.cz>
26970 PR gcov-profile/79891
26971 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26972 is marked by compiler as living on a line.
26973 (get_cycles_count): Remove usage of the union.
26974 (output_intermediate_file): Likewise.
26975 (find_source): Fix GNU coding style.
26976 (accumulate_line_counts): Remove old non-all block mode.
26977 (output_lines): Remove usage of the union.
26978 * profile.c (output_location): Include all BBs, even if
26979 belonging to a same line (and file) as a previous BB.
26981 2017-04-28 Martin Liska <mliska@suse.cz>
26983 * gcov.c (process_args): Handle new argument 'w'.
26984 (read_graph_file): Assign ID to BBs.
26985 (output_branch_count): Display BB # if verbose flag is set.
26986 (output_lines): Likewise for arcs.
26987 (print_usage): Add '--verbose' option help.
26988 * doc/gcov.texi: Document --verbose (-w) option.
26990 2017-04-28 Martin Liska <mliska@suse.cz>
26992 * gcov.c (struct block_location_info): New struct.
26993 (process_file): Fill up the new structure.
26994 (read_graph_file): Replace usage of encoding by the newly added
26996 (add_line_counts): Likewise.
26997 (accumulate_line_counts): Remove usage of the union.
26998 (function_info::function_info): New function.
26999 (function_info::~function_info): Likewise.
27000 (process_file): Call delete instead of release_function.
27001 (release_function): Release the function.
27002 (release_structures): Call delete instead of release_function.
27003 (solve_flow_graph): Replace usage of num_blocks.
27004 (find_exception_blocks): Likewise.
27005 (output_lines): Fix GNU coding style.
27007 2017-04-28 Martin Liska <mliska@suse.cz>
27010 * coverage.c (coverage_remove_note_file): New function.
27011 * coverage.h: Declare the function.
27012 * toplev.c (finalize): Clean if an error has been seen.
27014 2017-04-28 Martin Liska <mliska@suse.cz>
27016 PR gcov-profile/80031
27017 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27018 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27019 * gcov.c (read_graph_file): Read just number of blocks.
27020 * profile.c (branch_prob): Do not stream 0 flags per a basic
27023 2017-04-28 Martin Liska <mliska@suse.cz>
27025 * gcov-dump.c (tag_*): Add new argument to declarations.
27026 (dump_gcov_file): Likewise.
27027 (tag_blocks): Add and use new argument depth.
27028 (tag_arcs): Likewise.
27029 (tag_lines): Likewise.
27030 (tag_counters): Likewise.
27031 (tag_summary): Likewise.
27032 (dump_working_sets): Use depth to do a proper indentation.
27034 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27037 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27038 * symtab.c (symtab_node::debug_symtab): Move definition here.
27040 2017-04-28 Richard Biener <rguenther@suse.de>
27042 * lto-streamer.h (LTO_major_version): Bump to 7.
27044 2017-04-28 Richard Biener <rguenther@suse.de>
27046 * tree-vrp.c (assert_info): New struct.
27047 (add_assert_info): New helper.
27048 (register_edge_assert_for_2): Refactor to add asserts to a vector
27050 (register_edge_assert_for_1): Likewise.
27051 (register_edge_assert_for): Likewise.
27052 (finish_register_edge_assert_for): New helper actually registering
27053 asserts where live on edge.
27054 (find_conditional_asserts): Adjust.
27055 (find_switch_asserts): Likewise.
27056 (evrp_dom_walker::try_find_new_range): Generalize.
27057 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27059 2017-04-27 Marek Polacek <polacek@redhat.com>
27062 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27063 arg10 and arg11 to itype.
27065 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27067 * doc/extend.texi (Object Size Checking): Improve grammar.
27069 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27072 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27073 that the logic for permitting reciprocal estimates matches that
27076 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27079 * tree.c (type_cache_hasher::equal): Only compare
27080 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27081 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27082 non-aggregate element types.
27083 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27084 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27086 2017-04-27 Richard Biener <rguenther@suse.de>
27088 PR middle-end/80533
27089 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27090 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27091 keeping a reference to a trailing array.
27093 2017-04-27 Richard Biener <rguenther@suse.de>
27095 PR middle-end/80539
27096 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27097 being in loop-closed SSA form conservatively.
27098 (chrec_fold_multiply_poly_poly): Likewise.
27100 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27102 PR middle-end/79665
27103 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27104 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27106 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27109 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27110 (aarch64_function_arg_alignment): Return unsigned int again, but still
27111 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27112 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27113 Don't emit -Wpsabi note.
27114 (aarch64_function_arg_boundary): Likewise.
27115 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27118 2017-04-26 Nathan Sidwell <nathan@acm.org>
27120 * tree.h (crc32_unsigned_n): Declare.
27121 (crc32_unsigned, crc32_unsigned): Make inline.
27122 * tree.c (crc32_unsigned_bits): Replace with ...
27123 (crc32_unsigned_n): ... this.
27124 (crc32_unsigned, crc32_byte): Remove.
27125 (crc32_string): Remove unnecessary braces.
27127 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27129 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27130 * ipa-inline-analysis.c (MAX_TIME): Remove.
27131 (account_size_time): Use sreal for time.
27132 (dump_inline_summary): Update.
27133 (estimate_function_body_sizes): Update.
27134 (estimate_edge_size_and_time): Update.
27135 (estimate_calls_size_and_time): Update.
27136 (estimate_node_size_and_time): Update.
27137 (inline_merge_summary): Update.
27138 (inline_update_overall_summary): Update.
27139 (estimate_time_after_inlining): Update.
27140 (inline_read_section): Update.
27141 (inline_write_summary): Update.
27142 * ipa-inline.c (compute_uninlined_call_time): Update.
27143 (compute_inlined_call_time): Update.
27144 (recursive_inlining): Update.
27145 (inline_small_functions): Update.
27146 (dump_overall_stats): Update.
27147 * ipa-inline.h: Include sreal.h.
27148 (size_time_entry): Turn time to sreal.
27149 (inline_summary): Turn self_time nad time to sreal.
27151 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27153 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27155 (sreal::stream_out, sreal::stream_in): New.
27156 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27158 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27160 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27163 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27166 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27168 (dimode_scalar_chain::compute_convert_gain): Ditto.
27169 (dimode_scalar_chain::make_vector_copies): Ditto.
27170 (dimode_scalar_chain::convert_reg): Ditto.
27171 (dimode_scalar_chain::convert_insn): Ditto.
27172 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27173 (VI248_AVX512BW_1): New mode iterator.
27174 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27175 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27178 2017-04-25 Martin Sebor <msebor@redhat.com>
27180 PR tree-optimization/80497
27181 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27182 constants are representable in HOST_WIDE_INT.
27183 (parse_directive): Ditto.
27185 2017-04-25 Martin Sebor <msebor@redhat.com>
27188 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27189 (new_zero_array): Adjust signature.
27190 (dom_info::dom_init): Used unsigned rather that size_t.
27191 (dom_info::dom_info): Same.
27193 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27194 Jakub Jelinek <jakub@redhat.com>
27197 * config/arm/arm.c: Include gimple.h.
27198 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27199 returns negative, increment ncrn only if it returned positive.
27200 (arm_needs_doubleword_align): Return int instead of bool,
27201 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27202 members, but if there is any such non-FIELD_DECL
27203 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27204 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27205 returns negative, increment nregs only if it returned positive.
27206 (arm_setup_incoming_varargs): Likewise.
27207 (arm_function_arg_boundary): Emit -Wpsabi note if
27208 arm_needs_doubleword_align returns negative, return
27209 DOUBLEWORD_ALIGNMENT only if it returned positive.
27211 2017-04-25 Marek Polacek <polacek@redhat.com>
27214 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27215 first argument to type.
27217 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27220 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27221 type checks to test for compatibility instead of equality.
27223 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27224 Jakub Jelinek <jakub@redhat.com>
27227 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27229 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27230 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27231 the alignment computation, but return their maximum in warn_alignment.
27232 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27233 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27235 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27236 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27239 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27241 * config/arc/simdext.md (dmpyh): Fix typo.
27243 2017-04-25 Richard Biener <rguenther@suse.de>
27245 PR tree-optimization/80492
27246 * alias.c (compare_base_decls): Handle registers with asm
27247 specification conservatively.
27248 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27249 compare_base_decls returning dont-know properly.
27251 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27253 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27254 (legitimate_offset_address_p): New function.
27255 (arc_legitimate_address_p): Use above function.
27257 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27259 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27261 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27263 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27264 ACCH registers whenever they are available.
27266 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27268 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27269 double regs fix when not used.
27271 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27273 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27275 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27276 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27278 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27280 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27281 when emitting short ADD instructions.
27283 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27285 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27287 (cmpsi_cc_c_insn): Likewise.
27288 (cbranchsi4_scratch): Compute proper instruction length using
27289 compact_hreg_operand.
27290 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27292 2017-04-25 Richard Biener <rguenther@suse.de>
27294 PR middle-end/80509
27295 * passes.c (pass_manager::pass_manager): Initialize
27296 m_name_to_pass_map.
27298 2017-04-25 Richard Biener <rguenther@suse.de>
27300 PR tree-optimization/79201
27301 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27303 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27306 * config/s390/vector.md: Split MEM->GPR vector moves for
27307 non-s_operand addresses.
27309 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27312 * config/s390/predicates.md (reload_const_wide_int_operand): New
27314 * config/s390/s390.md ("movti"): Remove d/P alternative.
27315 ("movti_bigconst"): New pattern definition.
27317 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27320 * s390-protos.h (s390_expand_cs_hqi): Removed.
27321 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27322 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27323 modes as well as CCZ1mode and CCZmode.
27324 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27325 signature of s390_emit_compare_and_swap.
27326 (s390_expand_cs_hqi): Likewise, make static.
27327 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27328 compare-and-swap, in some cases.
27329 (s390_expand_cs): Wrapper function.
27330 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27332 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27333 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27334 patterns for small and large integers. Forbid symref memory operands.
27335 Move expander to s390.c. Require cc register.
27336 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27337 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27338 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27339 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27340 symref memory operands. Remove CC mode and call s390_match_ccmode
27342 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27344 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27346 * config/s390/s390.md (define_peephole2): New peephole to help
27347 combining the load-and-test pattern with volatile memory.
27349 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27351 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27352 with CCZmode for TARGET_Z196.
27354 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27356 PR rtl-optimization/80501
27357 * combine.c (make_compound_operation_int): Set subreg_code to SET
27358 even for AND with mask of the sign bit of mode.
27360 PR rtl-optimization/80500
27361 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27362 sum's initial value.
27364 2017-04-25 Julian Brown <julian@codesourcery.com>
27365 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27367 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27369 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27371 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27373 2017-04-25 Julian Brown <julian@codesourcery.com>
27374 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27376 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27377 (thunderx2t99_sha): New Reservation.
27379 2017-04-25 Julian Brown <julian@codesourcery.com>
27380 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27382 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27383 type for 1-element load.
27385 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27387 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27389 2017-04-24 Martin Jambor <mjambor@suse.cz>
27391 PR tree-optimization/80293
27392 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27393 char arrays not totally scalarizable if it is false.
27394 (analyze_all_variable_accesses): Pass correct value in the new
27395 parameter. Add a statistics counter.
27397 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27399 PR middle-end/79931
27400 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27402 2017-04-24 Richard Biener <rguenther@suse.de>
27404 PR tree-optimization/80494
27405 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27406 out for complex types.
27408 2017-04-24 Richard Biener <rguenther@suse.de>
27410 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27411 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27412 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27413 (DFS): Adjust and never fail.
27414 (sccvn_dom_walker::fail): Remove.
27415 (sccvn_dom_walker::before_dom_children): Adjust.
27416 (run_scc_vn): Likewise and never fail.
27417 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27418 (pass_fre::execute): Likewise.
27420 2017-04-24 Richard Biener <rguenther@suse.de>
27422 PR tree-optimization/79725
27423 * tree-ssa-sink.c (statement_sink_location): Return whether
27424 failure reason was zero uses. Move that check later.
27425 (sink_code_in_bb): Deal with zero uses by removing the stmt
27428 2017-04-24 Richard Biener <rguenther@suse.de>
27431 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27432 pointer-based references.
27434 2017-04-24 Richard Biener <rguenther@suse.de>
27437 * pass_manager.h (pass_manager::operator new): Remove.
27438 (pass_manager::operator delete): Likewise.
27439 * passes.c (pass_manager::operator new): Remove.
27440 (pass_manager::operator delete): Likewise.
27441 (pass_manager::pass_manager): Zero individual pass members.
27443 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27446 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27447 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27448 Check "XEXP (src, 1)" operand here.
27449 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27450 Check "XEXP (src, 1)" operand here.
27451 (dimode_scalar_chain::make_vector_copies): Detect count register
27452 of a shift instruction. Zero extend count register from QImode
27453 to DImode to satisfy vector shift pattern count operand predicate.
27454 Substitute vector shift count operand with a DImode copy.
27455 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27458 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27460 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27461 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27462 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27463 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27464 (UNSPEC_NOREX_MEM): Remove definition.
27466 2017-04-21 Richard Biener <rguenther@suse.de>
27468 PR tree-optimization/79547
27469 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27470 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27471 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27472 without any constraints.
27474 2017-04-21 Richard Biener <rguenther@suse.de>
27476 PR tree-optimization/78847
27477 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27479 2017-04-21 Richard Biener <rguenther@suse.de>
27481 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27482 (build_distinct_type_copy): Likewise.
27483 (build_variant_type_copy): Likewise.
27484 * tree.c (build_qualified_type): Pass down mem-stat info.
27485 (build_distinct_type_copy): Likewise.
27486 (build_variant_type_copy): Likewise.
27488 2017-04-21 Richard Biener <rguenther@suse.de>
27490 PR tree-optimization/80237
27491 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27493 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27494 for a simplified result.
27496 2016-04-21 Richard Biener <rguenther@suse.de>
27498 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27499 sth as strict as a simple_iv but a chrec without symbols and an
27500 operand defined in the loop we are peeling (and not some subloop).
27501 (propagate_constants_for_unrolling): Propagate all constants.
27503 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27506 * config/i386/i386.c (print_reg): Remove assert for disalowed
27507 regno values, call output_operand_lossage instead.
27509 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27512 * config/i386/constraints.md (Yc): New register constraint.
27513 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27514 Use Yc constraint for alternative 2 of operand 0. Remove
27515 preferred_for_speed attribute.
27517 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27519 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27520 lastprivate clauses in SIMT case.
27522 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27524 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27526 2017-04-20 Richard Biener <rguenther@suse.de>
27528 PR tree-optimization/57796
27529 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27530 as N scalar stores.
27531 (vect_model_load_cost): Cost gathers as N scalar loads.
27533 2017-04-20 Richard Biener <rguenther@suse.de>
27535 * ggc-page.c (ggc_allocated_p): Rename to ...
27536 (safe_lookup_page_table_entry): ... this and return the lookup
27538 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27540 2017-04-20 Richard Biener <rguenther@suse.de>
27542 PR tree-optimization/80453
27543 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27544 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27545 from the conditions.
27546 (vn_phi_eq): Pass them down.
27547 (vn_phi_lookup): Record them.
27548 (vn_phi_insert): Likewise.
27550 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27552 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27553 uninitialized variable warning to avoid buffer overrun.
27555 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27558 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27559 is suppressed for '{ 0 }' in C.
27561 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27563 * BASE-VER: Set to 8.0.0.
27565 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27567 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27568 priority .init_array and .fini_array section with SECTION_NOTYPE
27571 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27573 PR middle-end/80423
27574 * tree.h (build_array_type): Add typeless_storage default argument.
27575 * tree.c (type_cache_hasher::equal): Also compare
27576 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27577 (build_array_type): Add typeless_storage argument, set
27578 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27580 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27581 (build_array_type): Likewise. Add typeless_storage argument.
27583 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27584 Jakub Jelinek <jakub@redhat.com>
27586 PR tree-optimization/80426
27587 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27588 operation on symbolic operands, also compute the overflow for the
27589 invariant part when the operation degenerates into a negation.
27591 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27594 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27595 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27598 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27600 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27603 * config/avr/avr.c (tree.h): Include it.
27604 (cgraph.h): Include it.
27605 (avr_encode_section_info): Don't warn for uninitialized progmem
27606 variable if it's just an alias.
27608 2017-04-19 Richard Biener <rguenther@suse.de>
27611 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27612 when needed by AutoPGO.
27614 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27617 * doc/lto.texi: Remove an extra 'that'.
27619 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27621 PR rtl-optimization/80429
27622 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27623 are only used in debug insns.
27625 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27626 Vladimir Makarov <vmakarov@redhat.com>
27628 * config/sparc/predicates.md (input_operand): Add comment. Return
27629 true for any memory operand when LRA is in progress.
27630 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27632 2017-04-18 Jeff Law <law@redhat.com>
27635 * mips.md ({return,simple_return}_internal): Do not overwrite
27638 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27640 PR tree-optimization/80443
27641 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27642 instead of adding 1, subtract -1 and similarly instead of subtracting
27645 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27647 PR rtl-optimization/80357
27648 * haifa-sched.c (tmp_bitmap): New variable.
27649 (model_recompute): Handle duplicate use records.
27650 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27651 (free_global_sched_pressure_data): Free it.
27653 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27656 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27657 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27658 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27659 instead of SYSTEM_HEADER_DIR.
27661 2017-04-18 Jeff Law <law@redhat.com>
27663 PR middle-end/80422
27664 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27665 predecessors after walking up the insn chain.
27667 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27670 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27671 sizetype type into debug info.
27673 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27676 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27677 unneeded test for TARGET_UPPER_REGS_SF.
27678 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27680 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27683 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27684 instead of gsi_after_labels.
27686 2017-04-18 Jeff Law <law@redhat.com>
27688 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27692 2017-04-13 Jeff Law <law@redhat.com>
27693 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27694 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27696 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27699 * config/avr/avr.c (intl.h): Include it.
27700 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27702 2017-04-18 Martin Liska <mliska@suse.cz>
27704 PR gcov-profile/78783
27705 * gcov-tool.c (gcov_output_files): Validate that destination
27706 file is either removed by the tool or by a user.
27708 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27709 Guy Benyei <guybe@mellanox.com>
27711 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27712 block, and do not negate it, the stored id is already negative.
27714 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27716 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27718 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27721 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27722 masks of options that should be turned off if the VSX vector
27723 options are turned off.
27724 (OTHER_P8_VECTOR_MASKS): Likewise.
27725 (OTHER_VSX_VECTOR_MASKS): Likewise.
27726 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27727 rs6000_disable_incompatible_switches to validate no type switches
27729 (rs6000_incompatible_switch): New function to disallow turning on
27730 other vector options if -mno-vsx, -mno-power8-vector, or
27731 -mno-power9-vector are specified.
27733 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27735 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27737 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27739 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27740 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27741 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27742 (ARG_POINTER_CFA_OFFSET): Likewise.
27744 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27746 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27747 conditions to take advantage of various optimizations.
27749 2017-04-13 Jeff Law <law@redhat.com>
27751 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27752 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27753 (zero_extendsidi2_dext): Likewise.
27755 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27758 * fold-const.c (fold_ternary_loc): Revert
27759 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27762 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27764 PR rtl-optimization/80343
27765 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27766 new scratch pseudo.
27768 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27771 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27772 to ubsan_encode_value.
27774 2017-04-13 Jeff Law <law@redhat.com>
27776 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27777 appearing in DEBUG_INSNs.
27779 2017-04-13 Martin Liska <mliska@suse.cz>
27781 PR gcov-profile/80413
27782 * gcov-io.c (gcov_write_string): Copy to buffer just when
27783 allocated size is greater than zero.
27785 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27788 * dwarf2out.c (decls_for_scope): Ignore declarations of
27789 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27791 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27794 * ipa-visibility.c (non_local_p): Fix typos.
27795 (localize_node): When localizing symbol in same comdat group,
27796 dissolve the group only when we know external symbols are going
27798 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27800 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27802 PR tree-optimization/79390
27803 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27804 order does not result in usable sequence, retry with reversed operand
27810 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27811 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27812 op0 instead of fold_convert_loc (loc, type, arg0).
27814 2017-04-12 Jeff Law <law@redhat.com>
27816 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27817 has a delay slot in the generated code.
27819 * config/cris/cris.md (cris_preferred_reload_class): Return
27820 GENNONACR_REGS rather than GENERAL_REGS.
27822 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27825 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27826 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27827 signedness of the result type.
27829 2017-04-12 Richard Biener <rguenther@suse.de>
27830 Jeff Law <law@redhat.com>
27832 PR tree-optimization/80359
27833 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27834 trim stores to TARGET_MEM_REFs.
27836 2017-04-12 Richard Biener <rguenther@suse.de>
27838 PR tree-optimization/79390
27839 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27840 threading case even more.
27842 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27845 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27846 for quad_address_p for TImode, instead of just not indexed_address.
27848 2017-04-12 Richard Biener <rguenther@suse.de>
27849 Bernd Edlinger <bernd.edlinger@hotmail.de>
27851 PR middle-end/79671
27852 * alias.c (component_uses_parent_alias_set_from): Handle
27853 TYPE_TYPELESS_STORAGE.
27854 (get_alias_set): Likewise.
27855 * tree-core.h (tree_type_common): Add typeless_storage flag.
27856 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27857 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27858 for types containing members with TYPE_TYPELESS_STORAGE.
27859 (place_field): Likewise.
27860 (layout_type): Likewise for ARRAY_TYPE.
27861 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27862 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27863 TYPE_TYPELESS_STORAGE.
27864 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27866 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27869 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27870 first argument to type.
27872 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27876 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27877 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27878 (rs6000_expand_binop_builtin): Likewise.
27879 (rs6000_expand_ternop_builtin): Likewise; also add missing
27880 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27881 vshasigma built-ins.
27882 * doc/extend.texi: Document that vec_xxpermdi's third argument
27883 must be a constant.
27885 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27887 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27888 Use shift_const cost parameter when calculating gain of STV shifts.
27890 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27892 PR rtl-optimization/70478
27893 * lra-constraints.c (process_alt_operands): Check memory for
27894 disfavoring memory insn operand.
27896 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27898 PR middle-end/80100
27899 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27900 left shift in unsigned HOST_WIDE_INT type.
27902 PR rtl-optimization/80385
27903 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27904 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27907 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27908 if they have any depend clauses.
27910 2017-04-11 Martin Liska <mliska@suse.cz>
27913 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27914 * ipa-split.c (split_function): Create a local comdat symbol
27915 if caller is in a comdat group.
27917 2017-04-11 Martin Liska <mliska@suse.cz>
27920 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27923 2017-04-11 Martin Sebor <msebor@redhat.com>
27925 PR middle-end/80364
27926 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27927 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27929 (directive::set_width, directive::set_precision, format_character):
27931 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27934 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27937 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27938 conflict, set target->arch_name instead of target->cpu_name.
27940 2017-04-11 Richard Biener <rguenther@suse.de>
27942 PR tree-optimization/80374
27943 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27944 build_zero_cst, remove fold_convertible_p check again.
27946 2017-04-11 Martin Liska <mliska@suse.cz>
27949 * ubsan.c (instrument_object_size): Do not instrument register
27952 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27955 * config/i386/i386-builtin-types.def
27956 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27957 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27958 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27959 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27960 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27961 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27962 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27963 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27964 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27965 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27966 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27967 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27968 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27969 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27970 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27971 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27972 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27973 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27974 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27975 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27976 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27977 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27978 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27979 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27980 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27981 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27982 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27983 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27984 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27985 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27986 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27987 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27988 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27989 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27990 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27991 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27992 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27993 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27994 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27995 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27996 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27997 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27998 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27999 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28000 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28001 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28003 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28004 flag to second_arg_count, handle 4 argument function type _COUNT
28005 aliases, handle second_arg_count on second argument rather than last.
28007 2017-04-10 Jeff Law <law@redhat.com>
28009 PR tree-optimization/80374
28010 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28011 record anything if we can not convert integer_zero_node to the
28014 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28017 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28018 Enhance special handling given to the TARGET_P9_MINMAX option in
28019 relation to certain other options.
28021 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28023 PR tree-optimization/80153
28024 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28025 remove POINTER_PLUS_EXPR's base part directly, rather than through
28028 2017-04-10 Richard Biener <rguenther@suse.de>
28029 Bin Cheng <bin.cheng@arm.com>
28031 PR tree-optimization/80153
28032 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28033 the first element of pointer type aff_tree. Build result expr in
28035 (add_elt_to_tree): Convert to type unconditionally. Remove other
28036 fold_convert calls.
28037 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28038 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28040 2017-04-10 Richard Biener <rguenther@suse.de>
28042 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28045 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28047 PR rtl-optimization/70478
28048 * lra-constraints.c (curr_small_class_check): New.
28049 (update_and_check_small_class_inputs): New.
28050 (process_alt_operands): Update curr_small_class_check. Disfavor
28051 alternative insn memory operands. Check available regs for small
28054 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28057 * config/mips/mips.opt (-mvirt): Update description.
28058 * doc/invoke.texi (-mvirt): Likewise.
28060 2017-04-10 Richard Biener <rguenther@suse.de>
28062 PR middle-end/80362
28063 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28064 looking for NEGATE_EXPR in -A / -B to A / B folding.
28066 2017-04-10 Martin Liska <mliska@suse.cz>
28068 PR gcov-profile/80224
28069 * gcov.c (print_usage): Fix usage string.
28070 (get_gcov_intermediate_filename): Remove.
28071 (output_gcov_file): Use both for normal and intermediate format.
28072 (generate_results): Do not initialize special file for
28073 intermediate format.
28075 2017-04-10 Richard Biener <rguenther@suse.de>
28077 PR tree-optimization/80304
28078 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28081 2017-04-10 Nathan Sidwell <nathan@acm.org>
28084 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28085 (rs6000_init_builtins): Use it.
28087 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28089 * config/arm/arm.md (<mrc>): Add mode to SET source.
28090 (<mrrc>): Likewise.
28092 2017-04-10 Richard Biener <rguenther@suse.de>
28094 PR middle-end/80344
28095 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28097 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28100 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28101 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28102 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28103 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28104 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28105 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28106 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28107 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28108 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28109 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28110 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28111 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28112 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28113 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28114 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28115 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28116 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28117 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28118 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28119 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28120 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28121 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28122 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28124 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28126 PR rtl-optimization/70478
28127 * lra-constraints.c: Reverse the last patch.
28129 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28131 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28132 Add comment for WCHAR_T.
28134 2017-04-08 Martin Liska <mliska@suse.cz>
28137 2017-04-07 Martin Liska <mliska@suse.cz>
28140 * ipa-split.c (split_function): Add function part to a same comdat
28143 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28146 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28148 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28150 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28151 (vec_pairing): ...to this.
28152 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28153 (rs6000_sched_init): Adjust for name change.
28154 (struct rs6000_sched_context): Likewise.
28155 (rs6000_init_sched_context): Likewise.
28156 (rs6000_set_sched_context): Likewise.
28158 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28164 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28166 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28167 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28168 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28170 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28172 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28174 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28176 PR rtl-optimization/70703
28177 * ira-color.c (update_conflict_hard_regno_costs): Use
28178 int64_t instead of HOST_WIDE_INT.
28180 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28182 PR rtl-optimization/70478
28183 * lra-constraints.c (process_alt_operands): Disfavor alternative
28184 insn memory operands.
28186 2017-04-07 Jeff Law <law@redhat.com>
28188 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28189 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28191 2017-04-07 Martin Liska <mliska@suse.cz>
28194 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28195 Show error message instead of an ICE.
28197 2017-04-07 Martin Liska <mliska@suse.cz>
28200 * ipa-split.c (split_function): Add function part to a same comdat
28203 2017-04-07 Richard Biener <rguenther@suse.de>
28205 PR middle-end/80341
28206 * tree.c (get_unwidened): Also handle ! for_type case for
28208 * convert.c (do_narrow): Split out from ...
28209 (convert_to_integer_1): ... here. Do not pass final truncation
28210 type to get_unwidened for TRUNC_DIV_EXPR.
28212 2017-04-07 Richard Biener <rguenther@suse.de>
28214 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28216 (aff_combination_const): Adjust.
28217 (aff_combination_scale): Likewise.
28218 (aff_combination_add_elt): Likewise.
28219 (aff_combination_add_cst): Likewise.
28220 (aff_combination_convert): Likewise.
28221 (add_elt_to_tree): Likewise. Remove unused argument.
28222 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28224 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28226 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28228 * config/arm/arm.c (arm_default_short_enums): Use
28229 ARM_DEFAULT_SHORT_ENUMS.
28230 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28232 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28235 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28236 members with redundant out-of-class redeclaration.
28238 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28241 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28242 * config/i386/i386.md (*zero_extendsidi2):
28243 Add (?*x,*x) and (?*v,*v) alternatives.
28245 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28248 * config/i386/i386.c (ix86_expand_builtin)
28249 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28250 mode from insn data. Convert operands to insn operand mode.
28251 Copy operands that don't satisfy insn predicate to a register.
28253 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28255 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28258 2017-04-06 Richard Biener <rguenther@suse.de>
28260 PR tree-optimization/80334
28261 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28262 preserve alignment of accesses.
28264 2017-04-06 Richard Biener <rguenther@suse.de>
28266 PR tree-optimization/80262
28267 * tree-sra.c (build_ref_for_offset): Preserve address-space
28269 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28270 Drop useless address-space information on MEM_REF offsets.
28272 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28274 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28276 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28278 PR rtl-optimization/70703
28279 * ira-color.c (update_conflict_hard_regno_costs): Use
28280 HOST_WIDE_INT instead of long.
28282 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28285 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28286 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28288 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28289 for x86_64 target. Handle -m3dnowa option.
28291 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28293 PR rtl-optimization/70703
28294 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28295 (update_conflict_hard_regno_costs): Use long instead of unsigned
28296 arithmetic for cost calculation.
28298 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28299 Bernd Edlinger <bernd.edlinger@hotmail.de>
28302 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28305 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28308 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28309 ptr_mode with Pmode throughout.
28310 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28311 into probe_stack_range and use DImode.
28313 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28316 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28317 call_eh_return is true.
28319 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28321 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28322 Initialize last_match_fntype_index.
28324 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28327 * tree-nvr.c: Include internal-fn.h.
28328 (pass_return_slot::execute): Ignore internal calls without
28331 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28332 Richard Biener <rguenther@suse.de>
28335 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28336 captures used multiple times, except for the last use.
28337 * generic-match-head.c: Include gimplify.h.
28339 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28341 PR tree-optimization/79390
28342 * target.h (struct noce_if_info): Declare.
28343 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28344 * target.def (noce_conversion_profitable_p): New target hook.
28345 * ifcvt.h (struct noce_if_info): New type, moved from ...
28346 * ifcvt.c (struct noce_if_info): ... here.
28347 (noce_conversion_profitable_p): Renamed to ...
28348 (default_noce_conversion_profitable_p): ... this. No longer
28350 (noce_try_store_flag_constants, noce_try_addcc,
28351 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28352 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28353 instead of noce_conversion_profitable_p.
28354 * config/i386/i386.c: Include ifcvt.h.
28355 (ix86_option_override_internal): Don't override
28356 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28357 (ix86_noce_conversion_profitable_p): New function.
28358 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28359 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28360 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28361 * doc/tm.texi: Regenerated.
28363 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28365 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28368 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28371 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28372 instructions for small multiply cores.
28374 2017-04-04 Jeff Law <law@redhat.com>
28376 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28378 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28379 that are not set by the loop over the elements.
28381 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28384 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28385 int mode, convert_modes it to mode as unsigned, otherwise use
28386 lowpart_subreg to mode rather than SImode.
28387 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28388 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28389 Use DImode instead of SImode for the shift count operand.
28390 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28393 2017-04-04 Richard Biener <rguenther@suse.de>
28395 PR middle-end/80281
28396 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28397 arithmetic done for the negate or the plus. Simplify.
28398 (A - (-B) -> A + B): Likewise.
28399 * fold-const.c (split_tree): Make sure to not negate pointers.
28401 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28403 PR rtl-optimization/60818
28404 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28405 a compare of comparisons with the thing compared if this results
28406 in a different machine mode.
28408 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28410 * alias.c (base_alias_check): Fix typo in comment.
28411 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28412 * cgraphunit.c (symbol_table::compile): Likewise.
28413 * collect2.c (maybe_run_lto_and_relink): Likewise.
28414 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28415 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28416 * config/avr/avr.c (avr_map_op_t): Likewise.
28417 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28418 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28419 * config/epiphany/epiphany.md (movcc): Likewise.
28420 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28421 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28423 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28424 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28425 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28426 * config/sh/sh.c (sh_rtx_costs): Likewise.
28427 * fold-const.c (fold_truth_andor): Likewise.
28428 * genautomata.c (collapse_flag): Likewise.
28429 * gengtype.h (struct type::u::s): Likewise.
28430 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28431 * input.c (FORMAT_AMOUNT): Likewise.
28432 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28433 (known_aggs_to_agg_replacement_list): Likewise.
28434 * ipa-inline-analysis.c: Likewise.
28435 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28436 * ipa-polymorphic-call.c
28437 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28438 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28439 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28441 * modulo-sched.c (apply_reg_moves): Likewise.
28442 * omp-expand.c (build_omp_regions_1): Likewise.
28443 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28444 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28445 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28446 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28447 * value-prof.c: Likewise.
28448 * var-tracking.c (val_reset): Likewise.
28450 2017-04-03 Richard Biener <rguenther@suse.de>
28452 PR tree-optimization/80275
28453 * fold-const.c (split_address_to_core_and_offset): Handle
28456 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28458 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28459 descriptors is at least equal to that of functions.
28461 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28463 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28465 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28468 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28469 (mov<IMOD4:mode>): New expander.
28470 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28472 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28474 PR rtl-optimization/79405
28475 * fwprop.c (propagations_left): New variable.
28476 (forward_propagate_into): Decrement it.
28477 (fwprop_init): Initialize it.
28478 (fw_prop): If the variable has reached zero, stop propagating.
28479 (fwprop_addr): Ditto.
28481 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28484 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28485 a FUNCTION_DECL, pass it as decl instead of origin to
28488 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28490 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28493 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28496 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28497 TARGET_VSX_SMALL_INTEGER.
28499 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28501 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28502 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28504 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28506 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28507 extraction from odd-numbered MSA register.
28509 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28511 PR middle-end/80173
28512 * expmed.c (store_bit_field_1): Don't attempt to create
28513 a word subreg out of hard registers wider than word if they
28514 have HARD_REGNO_NREGS of 1 for their mode.
28516 PR middle-end/80163
28517 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28518 conversions to integer types wider than word and pointer.
28521 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28522 (rtx_equal_for_cselib_p): Pass 0 to it.
28523 * cselib.c (cselib_hasher::equal): Likewise.
28524 (rtx_equal_for_cselib_1): Add depth argument. If depth
28525 is 128, don't look up VALUE locs and punt. Increment
28526 depth in recursive calls when walking VALUE locs.
28528 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28530 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28531 (make_gcov_file_name): Use the canonical path name for generating
28533 (read_line): Fix handling of files with ascii null bytes.
28535 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28537 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28538 to initialise a vector register instead
28539 of using a const_int.
28541 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28543 PR translation/80189
28544 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28545 diagnostic messages.
28547 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28550 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28551 (dfp_diex_<mode>): Update mode of operand 1.
28552 * doc/extend.texi (dxex, dxexq): Document change to return type.
28553 (diex, diexq): Document change to argument type.
28555 2017-03-30 Martin Jambor <mjambor@suse.cz>
28558 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28559 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28560 it reflects the signature changes performed at the callee side.
28561 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28562 to cgraph_build_function_type_skip_args.
28563 (build_function_decl_skip_args): Adjust call to the above function.
28565 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28568 * config/i386/sse.md
28569 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28570 register as dest whenever it is a MEM not rtx_equal_p to the
28571 corresponding dup operand, and when forcing into reg move the
28572 reg into the memory afterwards.
28573 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28574 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28575 for the force_reg mode.
28576 (avx512vl_vextractf128<mode>): Use register as dest either
28577 always when a MEM, or when it is a MEM not rtx_equal_p to the
28578 corresponding dup operand, or even not when it is a CONST_VECTOR
28579 depending on the mode and lo vs. hi.
28580 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28582 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28583 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28584 Likewise. Require that operands[2] is even.
28585 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28586 Remove extraneous parens. Require that operands[2] is a multiple
28588 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28589 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28590 disallow memory then.
28592 2017-03-30 Richard Biener <rguenther@suse.de>
28594 PR tree-optimization/77498
28595 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28596 to non-constants over backedges.
28598 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28600 PR rtl-optimization/80233
28601 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28602 as last_combined_insn. Do not test for BARRIER_P separately.
28604 2017-03-29 Andreas Schwab <schwab@suse.de>
28607 * calls.c (prepare_call_address): Convert funexp to Pmode before
28608 copying to temp reg.
28610 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28612 PR tree-optimization/80158
28613 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28614 Handle possible future case of more than one alternate
28616 (replace_rhs_if_not_dup): Likewise.
28617 (replace_one_candidate): Likewise.
28619 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28621 PR rtl-optimization/80193
28622 * ira.c (ira): Do not check allocation for LRA.
28624 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28626 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28627 (nvptx_output_simt_exit): Declare.
28628 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28629 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28630 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28631 (nvptx_declare_function_name): ...here. Emit declaration of local
28632 memory space buffer for omp_simt_enter insn.
28633 (nvptx_output_unisimt_switch): New.
28634 (nvptx_output_softstack_switch): New.
28635 (nvptx_output_simt_enter): New.
28636 (nvptx_output_simt_exit): New.
28637 * config/nvptx/nvptx.h (struct machine_function): New fields
28638 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28639 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28640 (UNSPECV_SIMT_EXIT): Ditto.
28641 (omp_simt_enter_insn): New insn.
28642 (omp_simt_enter): New expansion.
28643 (omp_simt_exit): New insn.
28644 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28646 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28647 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28648 (expand_GOMP_SIMT_EXIT): New.
28649 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28650 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28651 (GOMP_SIMT_EXIT): Ditto.
28652 * target-insns.def (omp_simt_enter): New insn.
28653 (omp_simt_exit): Ditto.
28654 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28656 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28657 (lower_rec_input_clauses): Likewise.
28658 (lower_lastprivate_clauses): Handle SIMT privatization.
28660 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28661 (ompdevlow_adjust_simt_enter): New.
28662 (find_simtpriv_var_op): New.
28663 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28664 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28666 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28667 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28668 (copy_decl_for_dup_finish): Ditto.
28670 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28672 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28675 * config/i386/i386.c (ix86_option_override_internal): Always
28676 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28678 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28680 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28682 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28684 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28685 mark new edge's irreducible flag accordign to it.
28686 (vect_do_peeling): Check loop preheader edge's irreducible flag
28687 and pass it to function slpeel_add_loop_guard.
28689 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28691 PR tree-optimization/80218
28692 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28693 Update block frequencies and counts.
28695 2017-03-28 Richard Biener <rguenther@suse.de>
28697 PR tree-optimization/78644
28698 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28699 of a simplification result we may not use it at all.
28701 2017-03-28 Richard Biener <rguenther@suse.de>
28704 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28705 without arguments, generate default definition of a SSA name.
28707 2017-03-28 Richard Biener <rguenther@suse.de>
28709 PR middle-end/80222
28710 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28711 TYPE_REF_CAN_ALIAS_ALL references.
28712 * fold-const.c (fold_indirect_ref_1): Likewise.
28714 2017-03-28 Martin Liska <mliska@suse.cz>
28717 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28718 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28720 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28721 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28723 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28724 (EXTRA_SPECS): Define.
28725 (SUBTARGET_EXTRA_SPECS): Likewise.
28726 (SUBTARGET_CPP_SPEC): Likewise.
28727 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28728 SUBTARGET_EXTRA_SPECS.
28729 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28731 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28733 * config/arc/simdext.md (vst64_insn): Update pattern.
28734 (vld32wh_insn): Likewise.
28735 (vld32wl_insn): Likewise.
28736 (vld64_insn): Likewise.
28737 (vld32_insn): Likewise.
28739 2017-03-28 Marek Polacek <polacek@redhat.com>
28742 * fold-const.c (fold_comparison): Use protected_set_expr_location
28743 instead of SET_EXPR_LOCATION.
28745 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28747 * tree.c (add_expr): Avoid name lookup warning.
28749 2017-03-27 Jeff Law <law@redhat.com>
28751 PR tree-optimization/80216
28752 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28753 function name. Limit recursion depth.
28754 (record_temporary_equivalences): Corresponding changes.
28756 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28758 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28761 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28764 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28766 * cfgcleanup.c (reg_note_cfa_p): New array.
28767 (insns_have_identical_cfa_notes): New function.
28768 (old_insns_match_p): Don't cross-jump in between /f
28769 and non-/f instructions. If both i1 and i2 are frame related,
28770 verify all CFA notes, their order and content.
28772 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28775 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28776 HImode and SImode with zero extend to DImode to one insn.
28777 (bswap<mode>2_extenddi): Likewise.
28778 (bswapsi2_extenddi): Likewise.
28779 (bswaphi2_extendsi): Likewise.
28780 (bswaphi2): Combine bswap HImode and SImode into one insn.
28781 Separate memory insns from swapping register.
28782 (bswapsi2): Likewise.
28783 (bswap<mode>2): Likewise.
28784 (bswaphi2_internal): Delete, no longer used.
28785 (bswapsi2_internal): Likewise.
28786 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28787 store, and gpr<-gpr swap insns.
28788 (bswap<mode>2_store): Likewise.
28789 (bswaphi2_reg): Register only splitter, combine with the splitter.
28790 (bswaphi2 splitter): Likewise.
28791 (bswapsi2_reg): Likewise.
28792 (bswapsi2 splitter): Likewise.
28793 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28794 the insns into load, store, and register/register insns.
28795 (bswapdi2_ldbrx): Likewise.
28796 (bswapdi2_load): Likewise.
28797 (bswapdi2_store): Likewise.
28798 (bswapdi2_reg): Likewise.
28800 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28802 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28803 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28805 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28808 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28810 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28811 special handling for target option conflicts between dform
28812 options (-mpower9-dform, -mpower9-dform-vector,
28813 -mpower9-dform-scalar) and -mno-direct-move.
28815 2017-03-27 Richard Biener <rguenther@suse.de>
28817 PR tree-optimization/80181
28818 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28820 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28822 * config/arc/predicates.md (move_double_src_operand): Replace the
28823 call to move_double_src_operand with a call to address_operand.
28825 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28827 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28828 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28829 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28831 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28833 * config/arc/predicates.md (long_immediate_loadstore_operand):
28834 Consider scaled addresses cases.
28836 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28838 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28839 restored when in interrupt.
28840 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28841 doesn't have delay slot.
28843 2017-03-27 Richard Biener <rguenther@suse.de>
28846 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28847 inlined thunk clones.
28849 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28852 * asan.c (instrument_derefs): Copy over last operand from
28853 original COMPONENT_REF to the new COMPONENT_REF with
28854 DECL_BIT_FIELD_REPRESENTATIVE.
28855 * ubsan.c (instrument_object_size): Likewise.
28857 2017-03-27 Richard Biener <rguenther@suse.de>
28859 PR tree-optimization/80170
28860 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28861 sure DR/SCEV didnt fold in constants we do not see when looking
28862 at the reference base alignment.
28864 2017-03-27 Richard Biener <rguenther@suse.de>
28866 PR middle-end/80171
28867 * gimple-fold.c (fold_ctor_reference): Properly guard against
28868 NULL return value from canonicalize_constructor_val.
28870 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28873 * config/i386/i386.c (ix86_expand_builtin)
28874 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28875 flags reg setting and flags reg using instructions.
28876 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28877 clobbering instructions to zero extend op2.
28879 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28881 * doc/install.texi (Configuration) <--with-aix-soname>:
28882 Update link to AIX ld.
28884 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28886 PR rtl-optimization/80160
28887 PR rtl-optimization/80159
28888 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28889 reg_alternate_class into account.
28891 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28894 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28895 to consider in curr_insn_transform.
28897 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28899 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28900 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28901 and emit_mode_inner.
28903 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28905 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28906 argument to the overloaded builtin variants. Use the new flag to
28907 deprecate certain builtin variants.
28908 * config/s390/s390-builtin-types.def: Add new builtin types.
28909 * config/s390/s390-builtins.h: Support new flags field for
28910 overloaded builtins.
28911 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28912 (s390_macro_to_expand): Enable vector float data type.
28913 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28914 builtins by incrementing the __VEC__ version number.
28915 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28917 (s390_resolve_overloaded_builtin): Emit error messages depending
28918 on the builtin flags.
28919 * config/s390/s390.c (s390_expand_builtin): Support additional
28920 flags argument. Change error message to match the messages
28921 emitted in s390-c.c.
28922 * config/s390/s390.md: New UNSPEC_* constants.
28923 (op_type): Add new instruction types.
28924 * config/s390/vecintrin.h: Add new builtins and test data class
28926 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28927 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28928 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28929 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28930 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28931 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28933 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28934 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28935 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28936 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28938 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28939 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28940 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28941 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28943 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28944 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28945 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28946 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28947 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28948 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28949 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28951 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28952 ("vec_scatter_element<V_HW_4:mode>_DI")
28953 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28954 ("vec_fpint<mode>", "vflls")
28955 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28956 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28957 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28958 ("*vec_cmphe<mode>_cc"): ... these.
28960 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28961 mode constant instead of magic value.
28963 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28965 * config/s390/s390.c (s390_expand_vec_compare): Support other
28966 vector floating point modes than just V2DF.
28967 (s390_expand_vcond): Likewise.
28968 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28969 (s390_cannot_change_mode_class): Prevent mode changes between TF
28970 and V1TF in vector registers.
28971 * config/s390/s390.md (DF, SF): New mode attributes.
28972 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28973 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28974 SFmode support for VRs.
28975 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28977 (VFT, VF_HW): New mode iterators.
28978 (vw, sdx): New mode attributes.
28979 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28980 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28981 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28982 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28983 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28984 also the new vector floating point modes. Renaming to ...
28986 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28987 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28988 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28989 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28990 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28991 ("vec_unordered<mode>"): ... these.
28993 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28994 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28995 ("*vec_extendv2df"): New insn definitions.
28997 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28999 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29000 ("mulditi3_2", "*muldi3_sign"): New patterns.
29001 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29002 rename the pattern definition.
29004 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29006 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29008 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29010 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29012 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29013 instruction if possible.
29014 * config/s390/vector.md (vec_halfnumelts): New mode
29016 ("*vec_vllezlf<mode>"): New pattern.
29018 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29020 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29021 ("popcountv4si2", "popcountv2di2"): Rename to ...
29022 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29023 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29025 ("popcount<mode>2_vxe"): New pattern.
29027 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29029 * common/config/s390/s390-common.c (processor_flags_table): Add
29031 * config.gcc: Add arch12.
29032 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29033 Default to arch12 for unknown CPU model numbers.
29034 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29035 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29036 PROCESSOR_max sanity check.
29037 * config/s390/s390-opts.h (enum processor_type): Add
29039 * config/s390/s390.c (processor_table): Add arch12.
29040 (s390_expand_builtin): Add check for B_VXE flag.
29041 (s390_issue_rate): Add PROCESSOR_ARCH12.
29042 (s390_get_sched_attrmask): Likewise.
29043 (s390_get_unit_mask): Likewise.
29044 (s390_sched_score): Enable z13 scheduling for arch12.
29045 (s390_sched_reorder): Likewise.
29046 (s390_sched_variable_issue): Likewise.
29047 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29049 (s390_tune_attr): Use z13 scheduling also for arch12.
29050 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29051 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29052 (TARGET_VXE_P): New macros.
29053 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29054 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29055 * config/s390/s390.opt: Add arch12 as processor_type.
29057 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29059 * config/s390/s390.md
29060 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29061 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29062 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29064 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29065 Rename expanders to ...
29067 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29068 ("fixuns_truncdddi2_emu"): ... these.
29070 ("fixuns_trunc<mode>si2_emu"): New expander.
29072 ("*fixuns_truncdfdi2_z13"): Rename to ...
29073 ("*fixuns_truncdfdi2_vx"): ... this.
29075 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29077 * config/s390/2964.md: Remove the single element vector compare
29078 instructions which are no longer used.
29079 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29081 (s390_canonicalize_comparison): Remove handling of DFmode
29083 (s390_expand_vec_compare_scalar): Remove function.
29084 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29085 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29087 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29089 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29091 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29093 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29094 will anyway by matched by mov<mode>_64dfp.
29096 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29098 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29099 vlef/vstef. Add missing operand to vleif.
29101 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29103 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29104 pair for all vector types with 64 bit elements.
29105 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29106 * config/s390/vector.md (V_HW_64): ... here.
29107 (V_128_NOSINGLE): New mode iterator.
29108 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29109 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29110 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29111 ("*vec_load_pairv2di"): Change to ...
29112 ("*vec_load_pair<mode>"): ... this one.
29114 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29116 * config/s390/constraints.md: Add comments.
29117 (jKK): Reject element sizes > 8 bytes.
29118 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29120 * config/s390/s390.md: Add the s_operand checks formerly in
29121 s390_split_ok_p to various splitters where they are still
29123 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29124 for 128 bit vectors. Plus two splitters.
29126 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29128 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29131 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29134 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29135 error if the boundary argument is not constant.
29137 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29139 PR rtl-optimization/80112
29140 * loop-doloop.c (doloop_condition_get): Don't check condition
29141 if cmp isn't SET with IF_THEN_ELSE src.
29143 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29145 PR tree-optimization/80158
29146 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29147 replacing a candidate statement, also replace it for the
29148 candidate's alternate interpretation.
29149 (replace_rhs_if_not_dup): Likewise.
29150 (replace_one_candidate): Likewise.
29152 2017-03-24 Richard Biener <rguenther@suse.de>
29154 PR tree-optimization/80167
29155 * graphite-isl-ast-to-gimple.c
29156 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29158 (translate_isl_ast_to_gimple::get_rename): Likewise.
29160 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29162 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29163 handling of certain combinations of target options, including the
29164 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29165 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29167 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29170 * config/arm/arm.md (*load_multiple): Add reload_completed to
29171 matching condition.
29173 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29174 Richard Biener <rguenth@suse.de>
29176 PR tree-optimization/79908
29177 PR tree-optimization/80136
29178 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29179 been cast away, gimplify_and_add suffices.
29181 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29183 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29185 2017-03-23 Richard Biener <rguenther@suse.de>
29187 PR tree-optimization/80032
29188 * gimplify.c (gimple_push_cleanup): Forced unconditional
29189 cleanups still have to go to the conditional_cleanups
29192 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29194 PR tree-optimization/80072
29195 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29197 (next_operand_entry_id): Change type to unsigned int.
29198 (sort_by_operand_rank): Make sure to return the right return value
29199 even if unsigned fields are bigger than INT_MAX.
29200 (struct oecount): Change cnt and id type to unsigned int.
29201 (oecount_hasher::equal): Formatting fix.
29202 (oecount_cmp): Make sure to return the right return value
29203 even if unsigned fields are bigger than INT_MAX.
29204 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29207 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29208 TREE_READONLY on result if writing it more than once.
29211 * doc/invoke.texi (-fsanitize=thread): Document that with
29212 -fnon-call-exceptions atomics are not able to throw
29216 * tsan.c: Include tree-eh.h.
29217 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29218 maybe_clean_or_replace_eh_stmt where needed.
29219 (instrument_memory_accesses): Add cfg_changed argument.
29220 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29221 if it returned true.
29222 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29224 PR rtl-optimization/63191
29225 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29226 wrapper function, moved the whole old content into ...
29227 (ix86_delegitimize_address_1): ... this. New inline function.
29228 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29229 true as last argument instead of ix86_delegitimize_address.
29231 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29233 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29234 cortexa57_branch_cost.
29236 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29238 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29240 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29243 * doc/md.texi (Constraints): Document wA constraint.
29244 * config/rs6000/constraints.md (wA): New.
29245 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29246 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29247 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29248 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29250 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29253 * gimplify.c (is_oacc_declared): New function.
29254 (oacc_default_clause): Use it to set default flags for acc declared
29255 variables inside parallel regions.
29256 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29257 declared variables.
29258 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29259 declare attribute to any decl as necessary.
29261 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29264 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29265 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29266 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29267 (arm_arch_lpae): This.
29268 * config/arm/arm.c (arm_arch7ve): Rename into ...
29269 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29270 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29273 2017-03-22 Martin Liska <mliska@suse.cz>
29276 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29277 error message instead of an ICE.
29279 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29281 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29283 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29285 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29287 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29290 2017-03-21 Martin Sebor <msebor@redhat.com>
29292 * doc/extend.texi: Use "cannot" instead of "can't."
29293 * doc/hostconfig.texi: Same.
29294 * doc/install.texi: Same.
29295 * doc/invoke.texi: Same.
29296 * doc/loop.texi: Same.
29297 * doc/md.texi: Same.
29298 * doc/objc.texi: Same.
29299 * doc/rtl.texi: Same.
29300 * doc/tm.texi: Same.
29301 * doc/tm.texi.in: Same.
29302 * doc/trouble.texi: Same.
29304 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29307 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29308 (collect_checksum_attributes): Set it.
29309 (die_checksum_ordered): Use it.
29311 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29313 PR tree-optimization/79908
29314 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29315 change: For a VA_ARG whose LHS has been cast away, use
29316 force_gimple_operand to construct the side effects.
29318 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29320 PR translation/80001
29321 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29322 more amenable to translation.
29323 (oacc_loop_auto_partitions): Likewise.
29325 2017-03-21 Marek Polacek <polacek@redhat.com>
29326 Martin Sebor <msebor@redhat.com>
29328 PR tree-optimization/80109
29329 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29330 on INTEGRAL_TYPE_P.
29332 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29333 Segher Boessenkool <segher@kernel.crashing.org>
29336 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29337 check reg_used_between_p between insn and one of succ or succ2
29338 depending on if succ is artificial insn not inserted into insn
29341 2017-03-21 Martin Liska <mliska@suse.cz>
29343 PR gcov-profile/80081
29344 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29345 * doc/gcc.texi: Include gcov-dump stuff.
29346 * doc/gcov-dump.texi: New file.
29348 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29350 PR rtl-optimization/79150
29351 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29352 conditional jump, if the jump is the last insn of the loop.
29354 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29355 Richard Biener <rguenth@suse.de>
29357 PR tree-optimization/79908
29358 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29359 been cast away, use force_gimple_operand to construct the side
29362 2017-03-21 Martin Liska <mliska@suse.cz>
29364 PR libfortran/79956
29365 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29368 2017-03-21 Brad Spengler <spender@grsecurity.net>
29371 * plugin.c (htab_hash_plugin): New function.
29372 (add_new_plugin): Use it and adjust.
29373 (parse_plugin_arg_opt): Adjust.
29374 (init_one_plugin): Likewise.
29376 2017-03-21 Richard Biener <rguenther@suse.de>
29378 PR tree-optimization/80032
29379 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29380 if set force the cleanup to happen unconditionally.
29381 (gimplify_target_expr): Push inserted clobbers with force_uncond
29382 to avoid them being removed by control-dependent DCE.
29384 2017-03-21 Richard Biener <rguenther@suse.de>
29386 PR tree-optimization/80122
29387 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29388 va_arg_pack_len when the inlined call stmt requires pack
29390 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29392 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29395 * tsan.c (instrument_builtin_call): If the memory model argument
29396 is not a constant, assume it is valid.
29399 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29402 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29404 PR rtl-optimization/79910
29405 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29406 if its dest is used by an insn before I2 (other than the combined
29407 insns themselves, which are properly handled already).
29409 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29412 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29414 * combine.c (record_used_regs): New static function.
29415 (try_combine): Handle situations where there is an additional
29416 instruction between I2 and I3 which needs to have a LOG_LINK
29420 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29422 * combine.c (try_combine): Delete redundant i1 test. Call
29423 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29425 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29428 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29429 alternatives 13/14.
29431 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29433 PR tree-optimization/80054
29434 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29435 the optimization if a PHI or any of its arguments is not dominated
29436 by the candidate's basis. Use gphi* rather than gimple* as
29438 (replace_profitable_candidates): Clean up a gimple* variable that
29439 should be a gphi* variable.
29441 2017-03-20 Martin Sebor <msebor@redhat.com>
29444 * doc/extend.texi (attribute constructor): Document present limitation.
29446 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29449 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29450 __POWER9_VECTOR__ #ifdef control, change template definition to
29451 use Power9-specific built-in function.
29452 (vec_any_eq): Likewise.
29453 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29454 to control outcomes from this test.
29455 (vector_ae_<mode>p): For VEC_F modes, likewise.
29457 2017-03-20 Ian Lance Taylor <iant@google.com>
29459 * config/i386/i386.c (ix86_function_regparm): Save an extra
29460 register for -fsplit-stack with DECL_STATIC_CHAIN.
29462 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29465 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29466 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29468 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29470 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29472 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29475 2017-03-20 Marek Polacek <polacek@redhat.com>
29478 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29480 2017-03-20 Richard Biener <rguenther@suse.de>
29482 PR tree-optimization/80113
29483 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29484 allocate extra SSA name for PHI def.
29485 (add_close_phis_to_outer_loops): Likewise.
29486 (add_close_phis_to_merge_points): Likewise.
29487 (copy_loop_close_phi_args): Likewise.
29488 (copy_cond_phi_nodes): Likewise.
29490 2017-03-20 Martin Liska <mliska@suse.cz>
29492 PR middle-end/79753
29493 * tree-chkp.c (chkp_build_returned_bound): Do not build
29494 returned bounds for a LHS that's not a BOUNDED_P type.
29496 2017-03-20 Martin Liska <mliska@suse.cz>
29500 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29501 COMPLEX_CST and VECTOR_CST.
29503 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29506 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29507 target operand. A new splitter adds the clobber statement in case
29508 the target operand is dead anyway.
29510 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29512 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29513 to age-old versions of binutils and glibc.
29515 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29517 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29519 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29521 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29523 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29525 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29526 requirement for binutils 2.13.
29528 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29530 * combine.c (try_combine): Delete redundant i1 test. Call
29531 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29533 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29535 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29536 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29538 <riscv64-*-elf>: Re-arrange section
29539 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29540 <riscv32-*-linux>: Likewise.
29541 <riscv64-*-elf>: Likewise
29542 <riscv64-*-linux>: Likewise.
29544 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29547 * aarch64.opt(verbose-cost-dump): Fix typo.
29549 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29552 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29553 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29555 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29557 * reload.c (find_reloads): When reloading a nonoffsettable address,
29558 use RELOAD_OTHER for it and its address reloads.
29560 PR rtl-optimization/79910
29561 * combine.c (record_used_regs): New static function.
29562 (try_combine): Handle situations where there is an additional
29563 instruction between I2 and I3 which needs to have a LOG_LINK
29566 2017-03-17 Jeff Law <law@redhat.com>
29568 PR tree-optimization/71437
29569 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29570 conditional in the hash table first.
29571 (vrp_dom_walker::before_dom_children): Extract condition from
29572 ASSERT_EXPR. Record condition, its inverion and any implied
29573 conditions as well.
29575 2017-03-17 Marek Polacek <polacek@redhat.com>
29576 Markus Trippelsdorf <markus@trippelsdorf.de>
29578 PR tree-optimization/80079
29579 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29582 2017-03-17 Richard Biener <rguenther@suse.de>
29584 PR middle-end/80075
29585 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29586 Properly verify the LHS before the RHS possibly claims to be
29588 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29591 2017-03-17 Martin Jambor <mjambor@suse.cz>
29593 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29594 (List of -O2 options): Likewise.
29595 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29598 2017-03-17 Tom de Vries <tom@codesourcery.com>
29600 * gcov-dump.c (print_usage): Print bug_report_url.
29602 2017-03-17 Richard Biener <rguenther@suse.de>
29604 PR middle-end/80050
29605 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29606 (parser::peek): Likewise.
29608 2017-03-17 Richard Biener <rguenther@suse.de>
29610 PR tree-optimization/80048
29611 * sese.c (free_sese_info): Properly release rename_map and
29612 copied_bb_map elements.
29614 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29616 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29617 Add linked-list forward and backlinks. Insert on
29618 construction, remove on destruction.
29619 (class pass_store_merging): Add m_stores_head field.
29620 (pass_store_merging::terminate_and_process_all_chains):
29621 Iterate over m_stores_head list.
29622 (pass_store_merging::terminate_all_aliasing_chains):
29624 (pass_store_merging::execute): Check for debug stmts first.
29625 Push new chains onto the m_stores_head stack.
29627 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29630 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29631 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29632 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29634 2017-03-16 Jeff Law <law@redhat.com>
29636 PR tree-optimization/71437
29637 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29638 member function. Implementation moved into after_dom_children
29639 member function and into the threader's thread_outgoing_edges
29641 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29642 some code into new thread_outgoing_edges.
29643 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29644 definition. Simplify marker handling (do it here). Assume we always
29645 have the available expression and the const/copies tables.
29646 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29648 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29649 * tree-vrp.c (equiv_stack): No longer file scoped.
29650 (vrp_dom_walker): New class.
29651 (vrp_dom_walker::before_dom_children): New member function.
29652 (vrp_dom_walker::after_dom_children): Likewise.
29653 (identify_jump_threads): Setup domwalker. Use it rather than
29654 walking edges in a random order by hand. Simplify setup/finalization.
29655 (finalize_jump_threads): Remove.
29656 (vrp_finalize): Do not call identify_jump_threads here.
29657 (execute_vrp): Do it here instead and call thread_through_all_blocks
29660 PR tree-optimization/71437
29661 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29663 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29665 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29666 (dom_opt_dom_walker::thread_across_edge): Remove
29667 handle_dominating_asserts argument. All callers changed.
29668 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29669 changes. Remove calls to lhs_of_dominating_assert. Other
29670 uses of handle_dominating_asserts turn into unconditional code
29671 (simplify_control_stmt_condition_1): Likewise.
29672 (simplify_control_stmt_condition): Likewise.
29673 (thread_through_normal_block, thread_across_edge): Likewise.
29674 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29675 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29676 object if it is not an SSA_NAME.
29677 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29678 before calling into the VRP specific simplifiers.
29679 (identify_jump_threads): Remove handle_dominating_asserts
29682 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29685 * tree-diagnostic.c (default_tree_printer): No longer static.
29686 * tree-diagnostic.h (default_tree_printer): New prototype.
29688 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29690 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29691 Change ins into fmov.
29693 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29695 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29696 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29697 Use h_con constraint for operand 1.
29698 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29699 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29701 2017-03-15 Jeff Law <law@redhat.com>
29703 PR tree-optimization/71437
29704 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29705 (record_temporary_equivalences): Use it.
29707 PR tree-optimization/71437
29708 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29709 tree-ssa-scopedtables.
29710 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29711 (record_conditions, record_cond, vuse_eq): Likewise.
29712 (record_edge_info): Adjust to API tweak of record_conditions.
29713 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29714 (record_temporary_equivalences, optimize_stmt): Likewise.
29715 (eliminate_redundant_computations): Likewise.
29716 (record_equivalences_from_stmt): Likewise.
29717 * tree-ssa-scopedtables.c: Include options.h and params.h.
29718 (vuse_eq): New function, moved from tree-ssa-dom.c
29719 (build_and_record_new_cond): Likewise.
29720 (record_conditions): Likewise. Accept vector of conditions rather
29721 than edge_equivalence structure for first argument.
29722 for the first argument.
29723 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29724 from tree-ssa-dom.c.
29725 (avail_exprs_stack::record_cond): Likewise.
29726 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29727 from tree-ssa-dom.c.
29728 (avail_exprs_stack): Add new member functions lookup_avail_expr
29730 (record_conditions): Declare.
29732 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29735 * lra-constraints.c (process_alt_operands): Increase reject for
29736 reloading an input/output operand.
29738 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29741 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29742 insns to convert from signed/unsigned char/short to IEEE 128-bit
29744 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29746 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29749 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29750 subreg of inner mode for values already in registers.
29752 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29754 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29755 iteration reg is used after the loop.
29757 2017-03-14 Martin Sebor <msebor@redhat.com>
29759 PR tree-optimization/79800
29760 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29761 precision in negative-positive range.
29762 (format_floating): Call non-const overload with adjusted precision.
29764 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29767 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29770 2017-03-14 Martin Sebor <msebor@redhat.com>
29772 PR middle-end/80020
29773 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29774 * builtins.def (aligned_alloc): Use it.
29777 * Makefile.in (GTFILES): Add calls.c.
29778 * calls.c: Include "gt-calls.h".
29780 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29782 PR rtl-optimization/79728
29783 * regs.h (struct target_regs): New field
29784 x_contains_allocatable_regs_of_mode.
29785 (contains_allocatable_regs_of_mode): New macro.
29786 * reginfo.c (init_reg_sets_1): Initialize it, and change
29787 contains_reg_of_mode so it includes global regs as well.
29788 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29789 rather than contains_regs_of_mode.
29791 2017-03-14 Martin Liska <mliska@suse.cz>
29793 * doc/invoke.texi: Document options that can't be combined with
29794 -fcheck-pointer-bounds.
29796 2017-03-14 Martin Liska <mliska@suse.cz>
29798 PR middle-end/79831
29799 * doc/invoke.texi (-Wchkp): Document the option.
29801 2017-03-14 Martin Liska <mliska@suse.cz>
29803 * Makefile.in: Install gcov-dump.
29805 2017-03-14 Martin Liska <mliska@suse.cz>
29807 * multiple_target.c (expand_target_clones): Bail out for
29808 an invalid attribute.
29810 2017-03-14 Richard Biener <rguenther@suse.de>
29812 * alias.c (struct alias_set_entry): Pack properly.
29813 * cfgloop.h (struct loop): Likewise.
29814 * cse.c (struct set): Likewise.
29815 * ipa-utils.c (struct searchc_env): Likewise.
29816 * loop-invariant.c (struct invariant): Likewise.
29817 * lra-remat.c (struct cand): Likewise.
29818 * recog.c (struct change_t): Likewise.
29819 * rtl.h (struct address_info): Likewise.
29820 * symbol-summary.h (function_summary): Likewise.
29821 * tree-loop-distribution.c (struct partition): Likewise.
29822 * tree-object-size.c (struct object_size_info): Likewise.
29823 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29824 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29825 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29826 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29827 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29828 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29829 (struct _stmt_vec_info): Likewise.
29831 2017-03-14 Martin Liska <mliska@suse.cz>
29834 * multiple_target.c (create_dispatcher_calls): Check that
29835 a target can create a function dispatcher.
29837 2017-03-14 Martin Liska <mliska@suse.cz>
29840 * multiple_target.c (expand_target_clones): Drop local.local
29841 flag for default implementation.
29843 2017-03-14 Richard Biener <rguenther@suse.de>
29845 PR tree-optimization/80030
29846 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29848 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29850 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29851 gcc_fallthrough() instead of __attribute__((fallthrough));
29853 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29855 * doc/gcc.texi: Remove "up" link to (DIR).
29856 * doc/gccint.texi: Ditto.
29858 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29860 * doc/install.texi (Specific) <avr>: Remove reference to
29863 2017-03-13 Jeff Law <law@redhat.com>
29865 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29866 attribute rather than comments.
29868 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29869 match_scratch operand is highest.
29871 2017-03-13 Martin Liska <mliska@suse.cz>
29873 PR middle-end/78339
29874 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29875 is a CHKP clone, use original declaration.
29877 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29879 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29880 (arc_conditional_register_usage): Use a different allocation order
29881 when optimizing for size.
29882 * common/config/arc/arc-common.c (arc_option_optimization_table):
29883 Section anchors default on when optimizing for size.
29885 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29887 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29889 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29891 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29892 * config/arc/arc.md (cpu_facility): Add cd variant.
29893 (*movqi_insn): Add code density variant.
29894 (*movhi_insn): Likewise.
29895 (*movqi_insn): Likewise.
29896 (*addsi3_mixed): Likewise.
29897 (subsi3_insn): Likewise.
29899 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29901 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29903 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29905 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29906 expressions with MINUS and UNARY ops.
29908 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29911 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29913 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29914 between vec_select and vector argument.
29915 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29916 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29917 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29918 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29919 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29920 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29922 2017-03-13 Richard Biener <rguenther@suse.de>
29925 * params.def (vect-max-peeling-for-alignment): Fix typo.
29927 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29929 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29930 issue that only occurred with binutils below 2.18.
29932 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29934 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29935 refer to binutils 2.11/2.12 minimum.
29937 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29939 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29940 ftp.kernel.org and simplify binutils requirement.
29942 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29944 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29946 (Optimize Options): Ditto. Also remove redundancy.
29948 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29950 PR translation/79848
29951 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29953 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29954 to G_ to avoid double translation.
29956 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29958 PR translation/79923
29959 * auto-profile.c (get_combined_location): Convert leading
29960 character of diagnostics to lower case and remove trailing period.
29961 (read_profile): Likewise for various diagnostics.
29962 * config/arm/arm.c (arm_option_override): Remove trailing period
29963 from various diagnostics.
29964 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29965 (msp430_expand_delay_cycles): Likewise.
29967 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29970 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29971 full command-line argument, rather than just "str".
29972 (aarch64_validate_march): Likewise.
29973 (aarch64_validate_mtune): Likewise.
29975 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29977 PR rtl-optimization/78911
29978 * lra-assigns.c (must_not_spill_p): New function.
29979 (spill_for): Use it.
29981 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29983 PR tree-optimization/79981
29984 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29985 ATOMIC_COMPARE_EXCHANGE ifn result.
29986 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29987 IFN_ATOMIC_COMPARE_EXCHANGE.
29989 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29992 * opts.c (parse_sanitizer_options): Add missing question mark to
29993 "did you mean" message.
29995 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29997 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29999 (VMULEUH_UNS): Likewise.
30000 (VMULOUB_UNS): Likewise.
30001 (VMULOUH_UNS): Likewise.
30002 * config/rs6000/rs6000.c (builtin_function_type): Remove
30003 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30005 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30008 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30009 x with result of extra_parsing_for_operand_code_0.
30010 (function_reader::extra_parsing_for_operand_code_0): Convert
30011 return type from void to rtx, returning x. When reading
30012 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30013 larger size containing struct block_symbol.
30015 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30017 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30018 -mfloat128-hardware without -m64.
30020 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30023 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30024 entries to the case statement that marks unsigned arguments to
30025 overloaded functions.
30027 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30029 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30030 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30032 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30035 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30036 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30038 2017-03-10 Martin Liska <mliska@suse.cz>
30042 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30043 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30046 2017-03-10 Olivier Hainque <hainque@adacore.com>
30048 * tree-switch-conversion (array_value_type): Start by resetting
30049 candidate type to it's main variant.
30051 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30053 PR rtl-optimization/79909
30054 * combine.c (try_combine): Use simplify_replace_rtx on individual
30055 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30056 of the whole CALL_INSN_FUNCTION_USAGE.
30058 PR tree-optimization/79972
30059 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30060 get_range_info on SSA_NAMEs. Formatting fixes.
30062 2017-03-10 Richard Biener <rguenther@suse.de>
30063 Jakub Jelinek <jakub@redhat.com>
30065 PR tree-optimization/77975
30066 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30067 edge to be constant.
30068 (get_val_for): For constant x return it. Formatting fix.
30069 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30070 would use the same bases as the current one.
30072 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30074 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30075 instead of vec_select for V1TImode.
30076 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30078 (VSX_LE_128): Add V1TI to this mode iterator.
30079 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30080 (*vsx_le_perm_store_<mode>): Likewise.
30081 (pre-reload splitter for VSX stores): Likewise.
30082 (post-reload splitter for VSX stores): Likewise.
30083 (*vsx_xxpermdi2_le_<mode>): Likewise.
30084 (*vsx_lxvd2x2_le_<mode>): Likewise.
30085 (*vsx_stxvd2x2_le_<mode>): Likewise.
30087 2017-03-09 Michael Eager <eager@eagercon.com>
30089 Correct failures with --enable-checking=yes,rtl.
30091 * config/microblaze/microblaze.c (microblaze_expand_shift):
30092 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30093 test for const0_rtx.
30094 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30095 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30097 2017-03-09 Richard Biener <rguenther@suse.de>
30099 PR tree-optimization/79977
30100 * graphite-scop-detection.c (scop_detection::merge_sese):
30101 Handle the case of extra exits to blocks dominating the entry.
30103 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30105 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30108 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30110 PR rtl-optimization/79949
30111 * lra-constraints.c (process_alt_operands): Check memory when
30112 trying to predict a cycle. Print about the overall increase.
30114 2017-03-09 Richard Biener <rguenther@suse.de>
30116 PR middle-end/79971
30117 * gimple-expr.c (useless_type_conversion_p): Preserve
30118 TYPE_SATURATING for fixed-point types.
30120 2017-03-09 Richard Biener <rguenther@suse.de>
30123 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30124 alignment of BLKmode params.
30126 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30129 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30130 (VALL_NO_V2Q): Likewise.
30132 * config/aarch64/aarch64-simd.md
30133 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30135 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30136 VALL_NO_V2Q mode iterator.
30137 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30139 2017-03-09 Martin Liska <mliska@suse.cz>
30141 PR tree-optimization/79631
30142 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30143 tree_int_cst_sign_bit just for INTEGER constants.
30145 2017-03-09 Martin Liska <mliska@suse.cz>
30149 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30152 2017-03-09 Marek Polacek <polacek@redhat.com>
30155 * tree.c (inchash::add_expr): Handle TREE_VEC.
30157 2017-03-09 Martin Liska <mliska@suse.cz>
30160 (chkp_narrow_size_and_offset): New function.
30161 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30162 (void chkp_parse_bit_field_ref): New function.
30163 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30164 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30166 2017-03-09 Martin Liska <mliska@suse.cz>
30169 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30170 (chkp_find_bounds_1): Remove gcc_unreachable.
30172 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30175 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30176 BUILT_IN_SYNC*, determine the access type from the size suffix and
30177 always build a MEM_REF with that type. Handle forgotten
30178 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30181 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30182 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30183 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30184 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30185 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30186 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30187 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30188 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30189 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30190 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30191 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30192 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30193 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30194 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30195 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30196 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30197 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30198 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30199 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30200 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30201 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30202 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30203 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30204 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30205 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30206 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30207 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30208 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30209 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30210 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30211 definitions outside of __OPTIMIZE__ guarded section.
30214 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30215 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30216 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30217 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30220 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30222 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30223 ("vfenez<mode>"): Add missing constraints.
30225 2017-03-08 Martin Sebor <msebor@redhat.com>
30228 * config/nds32/nds32.c (nds32_option_override):
30229 Fix misspelled diagnostic.
30231 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30234 * gimplify.c (gimplify_omp_for): Replace index var in outer
30235 taskloop statement with an artificial variable and add
30236 OMP_CLAUSE_PRIVATE clause for it.
30238 2017-03-08 Richard Biener <rguenther@suse.de>
30240 PR tree-optimization/79955
30241 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30242 for accesses that are completely outside of the variable.
30244 2017-03-08 Andrew Haley <aph@redhat.com>
30246 PR tree-optimization/79943
30247 * tree-ssa-loop-split.c (compute_new_first_bound): When
30248 calculating the new upper bound, (END-BEG) should be added, not
30251 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30253 * config/avr/avr.md (setmemhi): Make sure match_dup
30254 operand number comes before match_scratch.
30256 2017-03-08 Richard Biener <rguenther@suse.de>
30258 PR tree-optimization/79920
30259 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30260 with ncopies == 1 to ...
30261 (vect_transform_slp_perm_load): ... here. Properly compute
30262 all element loads by iterating VF times over the group. Do
30263 not handle ncopies (computed in a broken way) in
30264 vect_create_mask_and_perm.
30266 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30269 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30270 is a uniform vector, use uniform_vector_p return value instead of
30271 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30273 2017-03-07 Marek Polacek <polacek@redhat.com>
30275 PR middle-end/79809
30276 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30277 (alloca_call_type): Likewise.
30279 2017-03-07 Martin Liska <mliska@suse.cz>
30281 * gcov.c (process_args): Put comment to correct location.
30283 2017-03-07 Martin Liska <mliska@suse.cz>
30285 PR middle-end/68270
30286 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30287 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30288 (chkp_narrow_bounds_for_field): Likewise.
30289 (chkp_parse_array_and_component_ref): Pass one more argument to
30292 2017-03-07 Richard Biener <rguenther@suse.de>
30294 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30297 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30299 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30300 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30302 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30305 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30306 to end of description.
30307 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30309 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30311 PR rtl-optimization/79901
30312 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30314 (*avx512f_<code><mode>3<mask_name>): ... this.
30315 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30316 iterator instead of VI8_AVX2_AVX512BW.
30318 PR rtl-optimization/79901
30319 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30320 min/max expander, expand it using expand_vec_cond_expr.
30323 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30326 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30329 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30330 to void * for PCH reasons.
30331 * dwarf2out.c (output_loc_operands, output_die): Cast
30332 v.val_vec.array to unsigned char *.
30334 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30337 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30340 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30342 PR rtl-optimization/79571
30343 * lra-constraints.c (process_alt_operands): Calculate static
30344 reject and subtract it from overall when only addresses will be
30347 2017-03-06 Julia Koval <julia.koval@intel.com>
30350 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30351 incoming stack boundary to 128 for 64-bit targets.
30353 2017-03-06 Richard Biener <rguenther@suse.de>
30355 PR tree-optimization/79894
30356 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30357 to NULL after folding it.
30359 2017-03-06 Richard Biener <rguenther@suse.de>
30361 PR tree-optimization/79824
30362 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30363 check disabling peeling for gaps.
30365 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30367 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30368 attributes): Document gettimeofday.
30370 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30372 * config/s390/s390.c (s390_option_override_internal): Set
30373 PARAM_MIN_VECT_LOOP_BOUND
30375 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30377 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30378 * config/s390/s390.md: Likewise.
30380 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30383 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30384 (<avx2_avx512>_perm<mode>): Rename to ...
30385 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30387 (<avx512>_perm<mode>_mask): Rename to ...
30388 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30390 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30391 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30392 instead of VI8F_256_512.
30393 (avx512f_perm<mode>): New define_expand.
30394 (avx512f_perm<mode>_mask): Likewise.
30395 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30396 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30398 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30400 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30401 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30403 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30405 2017-03-06 Martin Liska <mliska@suse.cz>
30408 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30409 when having a SSA NAME w/o VAR_DECL assigned to it.
30411 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30413 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30414 msa_dpsub_<su>_d): Fix MODE for vec_select.
30416 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30418 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30420 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30422 2017-03-06 Richard Biener <rguenther@suse.de>
30424 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30425 * plugin.c (register_plugin_info): Likewise.
30426 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30428 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30430 * config/i386/sse.md (sse_storehps, sse_storelps,
30431 avx_<castmode><avxsizesuffix>_<castmode>,
30432 avx512f_<castmode><avxsizesuffix>_<castmode>,
30433 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30434 in condition that at least one operand is not a MEM.
30436 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30438 PR middle-end/79805
30439 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30440 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30442 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30443 gimple_call_nothrow_p flag based on whether original builtin can throw.
30444 If it can, emit following stmts on the fallthrough edge.
30445 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30446 don't create new bb if inserting just debug stmts on the edge, try to
30447 insert them on the fallthru bb or just reset debug stmts.
30449 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30452 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30453 restore recog_data (including the operand rtxes inside it) around
30454 the call to get_insn_template.
30456 2017-03-03 Martin Sebor <msebor@redhat.com>
30458 PR tree-optimization/79699
30459 * context.c (context::~context): Free MPFR caches to avoid
30460 a memory leak on program exit.
30462 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30464 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30465 Use wide_int::ulow () instead of .elt (0).
30467 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30469 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30470 (*pushxf): Limit oF constraint to 32bit targets and add oC
30471 constraint for 64bit targets.
30472 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30473 (*pushdf): Change rmF constraint to rmC.
30475 2017-03-03 Martin Liska <mliska@suse.cz>
30477 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30478 Remove unused variable.
30480 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30483 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30484 is a memory operand, increase num_memory.
30485 (ix86_expand_args_builtin): Likewise.
30487 2017-03-03 Jan Hubicka <jh@suse.cz>
30490 * ipa-devirt.c (maybe_record_node): Properly handle
30491 __cxa_pure_virtual visibility.
30493 2017-03-03 Martin Liska <mliska@suse.cz>
30495 PR tree-optimization/79803
30496 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30498 (pass_loop_prefetch::execute): Disabled optimization if an
30499 assumption about L1 cache size is not met.
30501 2017-03-03 Martin Liska <mliska@suse.cz>
30503 PR rtl-optimization/79574
30504 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30505 (hash_scan_set): Likewise.
30506 (dump_hash_table): Likewise.
30507 (hoist_code): Likewise.
30509 2017-03-03 Richard Biener <rguenther@suse.de>
30511 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30512 in place of uhigh ().
30513 (fixed_convert_from_real): Likewise.
30515 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30518 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30520 2017-03-03 Richard Biener <rguenther@suse.de>
30522 PR middle-end/79818
30523 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30524 TYPE_OVERFLOW_UNDEFINED check.
30526 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30528 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30530 (vector_ae_<mode>_p): Likewise.
30531 (vector_nez_<mode>_p): Likewise.
30532 (vector_ne_v2di_p): Likewise.
30533 (vector_ae_v2di_p): Likewise.
30534 (vector_ne_<mode>_p): Likewise.
30535 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30537 (vsx_tsqrt<mode>2_fe): Likewise.
30539 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30542 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30544 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30546 PR rtl-optimization/79780
30547 * cprop.c (one_cprop_pass): When second and further conditional trap
30548 in a single basic block is turned into an unconditional trap, turn it
30549 into a deleted note to avoid RTL verification failures.
30551 2017-03-02 Richard Biener <rguenther@suse.de>
30553 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30555 2017-03-02 Richard Biener <rguenther@suse.de>
30557 PR tree-optimization/79345
30559 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30560 param and abort the walk, returning -1 if it is hit.
30561 (walk_aliased_vdefs): Take a limit param and pass it on.
30562 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30563 defaulting to 0 and return a signed int.
30564 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30565 (check_defs): New helper.
30566 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30567 about uninitialized memory.
30568 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30569 bogus uninitialized warning.
30570 (fixed_convert_from_real): Likewise.
30572 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30574 PR tree-optimization/66768
30575 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30576 iv_use if base object can't be determined.
30578 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30580 PR tree-optimization/79345
30581 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30582 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30583 (get_pattern_stats): Initialize it.
30584 * genemit.c (gen_expand): Verify match_scratch numbers come after
30585 match_operand/match_dup numbers.
30586 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30587 match_scratch numbers.
30588 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30590 * config/s390/s390.md (trunctdsd2): Likewise.
30592 2017-03-02 Richard Biener <rguenther@suse.de>
30594 * wide-int.h (wide_int_storage::operator=): Implement in terms
30597 2017-03-02 Richard Biener <rguenther@suse.de>
30599 PR tree-optimization/79777
30600 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30601 the to insert expression to sth existing.
30603 2017-03-01 Martin Sebor <msebor@redhat.com>
30605 PR middle-end/79692
30606 * gimple-ssa-sprintf.c
30607 (directive::known_width_and_precision): New function.
30608 (format_integer): Use it.
30609 (get_mpfr_format_length): Consider the full range of precision
30610 when computing %g output with the # flag. Set the likely byte
30611 count to 3 rather than 1 when precision is indeterminate.
30612 (format_floating): Correct the lower bound of precision.
30614 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30616 * doc/invoke.texi: Document default code model for 64-bit Linux.
30618 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30621 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30622 udiv rather than div since input pattern is unsigned.
30624 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30626 * config/i386/i386.c (print_reg): Warn for values of
30627 unsupported size in integer register.
30629 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30632 * config/rs6000/predicates.md (current_file_function_operand): Do
30633 not allow self calls to be local if the function is replaceable.
30635 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30638 * config/rs6000/altivec.h (vec_ctz and others): Change the
30639 preprocessor macro that controls conditional compilation from
30640 _ARCH_PWR9 to __POWER9_VECTOR__.
30641 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30642 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30643 control (instead of _ARCH_PWR9 control) so that template
30644 definition uses power9-specific function.
30645 (vec_any_eq): Likewise.
30646 (vec_all_ne): Change macro definition to use a power9-specific
30647 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30648 _ARCH_PWR9 control).
30649 (vec_any_eq) Likewise.
30650 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30651 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30652 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30653 support for xvcmpnedp instruction.
30654 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30655 macro expansion so that Power9 implementation of vec_all_ne does
30656 not use the AltiVec predicate framework.
30657 (VCMPNEH_P): Likewise.
30658 (VCMPNEW_P): Likewise.
30659 (VCMPNED_P): Likewise.
30660 (VCMPNEFP_P): Likewise.
30661 (VCMPNEDP_P): Likewise.
30662 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30663 implementation of vec_any_eq to not use AltiVec predicate
30665 (VCMPAEH_P): Likewise.
30666 (VCMPAEW_P): Likewise.
30667 (VCMPAED_P): Likewise.
30668 (VCMPAEFP_P): Likewise.
30669 (VCMPAEDP_P): Likewise.
30670 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30671 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30672 not use the AltiVec predicate framework.
30673 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30674 of vec_any_eq to not use AltiVec predicate framework.
30675 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30676 support for predefined __POWER9_VECTOR__ macro to indicate that
30677 Power9 instruction selection is enabled.
30678 (altivec_overloaded_builtins): Remove extraneous
30679 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30680 function argument types RS6000_BTI_bool_V16QI and
30681 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30682 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30683 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30684 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30685 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30686 Power9 for implementations of vec_cmpne. Change the signature for
30687 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30688 (representing vec_all_ne) to remove the previously described first
30689 argument of type RS6000_BTI_INTSI, as this was an artifact of
30690 reliance on the AltiVec predicate framework, which is no longer
30691 used in the implementation of these functions. Add
30692 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30693 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30694 since, unlike the AltiVec predicate framework implementation, we
30695 do not share function descriptors between vec_alle and vec_anyeq.
30696 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30697 set of modes that receive special treatment even when
30698 TARGET_P9_VECTOR is true. The special treatment emits code that
30699 does not depend on Power9 instructions.
30700 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30701 define_expand to not rely on AltiVec predicate framework.
30702 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30704 (vector_ne_v2di_p): Change this define_expand to not rely on
30705 AltiVec predicate framework.
30706 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30708 (vector_ne_<mode>_p): Change this define_expand to not rely on
30709 AltiVec predicate framework.
30710 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30712 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30713 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30714 define_insn pattern.
30715 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30716 define_insn pattern because the xvcmpne<VSs>. instruction is not
30718 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30719 instruction is not supported.
30721 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30723 * config/nvptx/nvptx.c: Include intl.h.
30725 2017-03-01 Martin Jambor <mjambor@suse.cz>
30728 * ipa-prop.h (ipa_bits): Removed field known.
30729 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30730 to pointers. Adjusted their comments to warn about their sharing.
30731 (ipcp_transformation_summary): Change bits to a vector of pointers.
30732 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30733 (ipa_get_ipa_bits_for_value): Declare.
30734 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30735 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30736 (ipa_bits_hash_table): Likewise.
30737 (ipa_vr_ggc_hash_traits): Likewise.
30738 (ipa_vr_hash_table): Likewise.
30739 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30740 being pointers and vr_known being removed.
30741 (ipa_set_jf_unknown): Likewise.
30742 (ipa_get_ipa_bits_for_value): New function.
30743 (ipa_set_jfunc_bits): Likewise.
30744 (ipa_get_value_range): New overloaded functions.
30745 (ipa_set_jfunc_vr): Likewise.
30746 (ipa_compute_jump_functions_for_edge): Use the above functions to
30747 construct bits and vr parts of jump functions.
30748 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30749 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30751 (ipcp_grow_transformations_if_necessary): Also allocate
30752 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30754 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30755 them. Fix too long lines.
30756 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30757 vr_known being removed.
30758 (ipa_read_jump_function): Use new setter functions to construct bits
30759 and vr parts of jump functions or set them to NULL.
30760 (write_ipcp_transformation_info): Adjust for bits being pointers.
30761 (read_ipcp_transformation_info): Likewise.
30762 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30764 Include gt-ipa-prop.h.
30765 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30767 (ipcp_store_bits_results): Likewise.
30768 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30769 Do not write to existing jump functions but use a temporary instead.
30771 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30774 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30775 attempt to use its first operand as BIT_FIELD_REF base.
30777 2017-03-01 Richard Biener <rguenther@suse.de>
30779 PR middle-end/79721
30780 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30781 interpolating formula in wrapping arithmetic.
30782 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30784 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30786 PR tree-optimization/79734
30787 * tree-vect-generic.c (expand_vector_condition): Optimize
30788 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30789 Handle VEC_COND_EXPR where comparison has different inner width from
30790 type's inner width.
30792 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30794 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30795 markup, and similar issues. Remove @opindex entries for things
30796 that aren't options. Add missing -mmpy-option entries.
30798 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30800 PR tree-optimization/79737
30801 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30802 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30803 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30804 instead of byte_size. Formatting fix.
30805 (shift_bytes_in_array_right): Formatting fix.
30807 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30810 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30811 condition on optimize for the leaf function test.
30813 2017-02-28 Martin Liska <mliska@suse.cz>
30816 * read-rtl-function.c (function_reader::handle_unknown_directive):
30817 Bail out when one uses -flto.
30819 2017-02-28 Martin Liska <mliska@suse.cz>
30821 * common.opt: Replace space with tabular for options of <number>
30823 * config/i386/i386.opt: Show <number> value for
30824 -mlarge-data-threshold.
30825 * opts.c (print_filtered_help): Do not display number in hexadecimal
30828 2017-02-28 Martin Liska <mliska@suse.cz>
30830 * common.opt: Fix --help=option -Q for options which are of
30833 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30835 * config/i386/i386.c (print_reg): Error out for values
30836 of 8-bit size in invalid integer register.
30838 2017-02-28 Martin Sebor <msebor@redhat.com>
30840 PR tree-optimization/79691
30841 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30843 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30846 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30847 gcc_unreachable with output_operand_lossage.
30849 2017-02-28 Richard Biener <rguenther@suse.de>
30851 PR tree-optimization/79740
30852 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30854 (visit_nary_op): Insert the nary into the hashtable if we
30855 pattern-matched sth.
30856 * tree-ssa-pre.c (eliminate_insert): Robustify.
30858 2017-02-28 Richard Biener <rguenther@suse.de>
30860 PR middle-end/79731
30861 * fold-const.c (decode_field_reference): Reject out-of-bound
30864 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30866 * config/i386/i386.c: Include intl.h.
30867 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30868 instead of just cond ? "..." : "...".
30869 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30870 * coverage.c (read_counts_file): Likewise.
30871 * omp-offload.c: Include intl.h.
30872 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30873 of just cond ? "..." : "...".
30874 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30875 of just cond ? "..." : "...".
30877 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30880 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30882 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30883 'tune for' CPU name.
30884 * config/arm/arm-cpu-data.h: Regenerated.
30886 2017-02-28 Richard Biener <rguenther@suse.de>
30888 PR tree-optimization/79732
30889 * tree-inline.c (expand_call_inline): Do not shadow var.
30891 2017-02-28 Richard Biener <rguenther@suse.de>
30893 PR tree-optimization/79723
30894 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30895 address-space properly.
30897 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30899 * doc/optinfo.texi (Optimization groups): Fix option used for
30901 * doc/invoke.texi (-fopt-info): Document "omp".
30902 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30903 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30904 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30905 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30906 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30908 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30910 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30911 OPTGROUP_OMP with "omp".
30913 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30916 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30917 for arithmetic shift of unsigned V2DI.
30919 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30921 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30922 arc/linux.h headers.
30923 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30924 (LINK_SPEC): Likewise.
30925 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30926 (EXTRA_SPECS): Likewise.
30927 (STARTFILE_SPEC): Likewise.
30928 (ENDFILE_SPEC): Likewise.
30929 (LIB_SPEC): Likewise.
30930 (TARGET_SDATA_DEFAULT): Likewise.
30931 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30932 (MULTILIB_DEFAULTS): Likewise.
30933 (DWARF2_UNWIND_INFO): Likewise.
30934 * config/arc/big.h: New file.
30935 * config/arc/elf.h: Likewise.
30936 * config/arc/linux.h: Likewise.
30937 * config/arc/t-uClibc: Remove.
30939 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30941 PR tree-optimization/77536
30942 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30943 (tree_transform_and_unroll_loop): Use above function to compute the
30944 estimated niter of unrolled loop and use it when scaling profile.
30945 Also use count info rather than frequency if it's non-zero.
30946 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30947 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30948 (vect_transform_loop): Call above function.
30950 2017-02-27 Richard Biener <rguenther@suse.de>
30952 PR tree-optimization/45397
30953 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30954 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30955 (visit_nary_op): Add pattern matching for CSEing sign-changed
30956 or truncated operations with wider ones.
30958 2017-02-27 Richard Biener <rguenther@suse.de>
30960 PR tree-optimization/79690
30961 * tree-vect-stmts.c (vectorizable_store): Use vector type
30962 built from the DR with address-space.
30964 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30966 * doc/invoke.texi (Optimize Options): Refine the description
30967 of asan-use-after-return.
30969 2017-02-25 Alan Modra <amodra@gmail.com>
30971 PR rtl-optimization/79584
30972 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30973 base, not ad->base_term, the reg within base. Remove assertion
30974 that ad->base == ad->base_term. Replace gen_int_mode using
30975 bogus mode with const0_rtx.
30977 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30979 PR middle-end/79396
30980 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30981 FMA_EXPR like tcc_binary or tcc_unary.
30983 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30986 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30988 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30989 (output_loc_operands): Handle DW_OP_call_ref and
30990 DW_OP_GNU_variable_value.
30991 (struct variable_value_struct): New type.
30992 (struct variable_value_hasher): Likewise.
30993 (variable_value_hash): New variable.
30994 (string_types): Remove.
30995 (copy_loc_descr): New function.
30996 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30997 (prepend_loc_descr_to_each): New function.
30998 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30999 instead of add_loc_descr_to_each if the first argument is single
31000 location list and the second has multiple.
31001 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31002 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31003 when looking for variable value which doesn't have other location info.
31004 (loc_list_from_tree): Formatting fix.
31005 (gen_array_type_die): Simplify DW_AT_string_length handling.
31006 (adjust_string_types): Remove.
31007 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31008 string_types. Call resolve_variable_values.
31009 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31010 (resolve_addr_in_expr): Likewise. Add A argument.
31011 (copy_deref_exprloc): Remove deref argument. Adjust for the
31012 original expression being DW_OP_GNU_variable_value with optionally
31013 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31014 optionally after it.
31015 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31016 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31017 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31018 (variable_value_hasher::hash, variable_value_hasher::equal): New
31020 (resolve_variable_value_in_expr, resolve_variable_value,
31021 resolve_variable_values, note_variable_value_in_expr,
31022 note_variable_value): New functions.
31023 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31026 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31029 * opts.h (handle_generated_option): Add GENERATED_P argument.
31030 * opts-common.c (handle_option): Adjust function comment.
31031 (handle_generated_option): Add GENERATED_P argument, pass it to
31033 (control_warning_option): Pass false to handle_generated_option
31035 * opts.c (maybe_default_option): Pass true to handle_generated_option
31037 * optc-gen.awk: Likewise.
31039 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31041 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31042 a REG, look at the REG it is a SUBREG of.
31043 (splitter for cmpeqsi_t): Ditto.
31045 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31047 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31048 the special USEs with the pattern of the insn, not the insn itself.
31050 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31053 * doc/invoke.texi: Document -mload-store-pairs.
31055 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31056 Sandra Loosemore <sandra@codesourcery.com>
31058 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31059 argument isn't a CONST_INT.
31060 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31061 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31062 (nios2_validate_compare): Bypass alternate compare logic if *op2
31063 is not a CONST_INT.
31064 (ldstwm_operation_p): Return false if first_base is not a REG or
31065 if first_offset is not a CONST_INT.
31067 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31069 * config/cris/cris.md: Use correct operand in a define_peephole2.
31071 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31073 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31075 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31077 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31078 this_insn if it is an INSN or JUMP_INSN.
31079 (force_offsettable): Look at base, not at addr.
31080 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31081 on things that aren't necessarily CONST_INTs.
31083 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31085 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31086 -mfpmath=sse is the default also for x86-32 targets with SSE2
31087 instruction set when @option{-ffast-math} is enabled
31089 2017-02-24 Jeff Law <law@redhat.com>
31091 PR rtl-optimizatoin/79286
31092 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31095 2017-02-24 Richard Biener <rguenther@suse.de>
31097 PR tree-optimization/79389
31098 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31101 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31103 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31104 function comment to reflect reality.
31105 (loop_exits_before_overflow): Fix typo in function description.
31107 2017-02-24 Richard Biener <rguenther@suse.de>
31109 PR tree-optimization/79389
31110 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31111 properly that a threading opportunity exists. Detect conditional
31112 copy/constant propagation opportunities.
31114 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31116 * config/visium/visium.md (type): Add trap.
31117 (b): New mode attribute.
31118 (*btst): Rename into...
31119 (*btst<mode>): ...this and adjust.
31120 (*cbranchsi4_btst_insn): Rename into...
31121 (*cbranch<mode>4_btst_insn): ...this and adjust.
31122 (trap): New define_insn.
31124 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31126 PR tree-optimization/79389
31127 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31128 (noce_reversed_cond_code): New function.
31129 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31130 reversed_comparison_code. Formatting fix.
31131 (noce_try_store_flag): Test rev_cond != NULL in addition to
31132 reversed_comparison_code.
31133 (noce_try_store_flag_constants): Likewise.
31134 (noce_try_store_flag_mask): Likewise.
31135 (noce_try_addcc): Use rev_cond if non-NULL instead of
31136 reversed_comparison_code.
31137 (noce_try_cmove_arith): Likewise. Formatting fixes.
31138 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31139 (noce_find_if_block): Initialize rev_cond.
31140 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31141 instead of false as last argument never attempt to reverse it
31144 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31146 PR tree-optimization/79663
31147 * tree-predcom.c (combine_chains): Process refs in reverse order
31148 only for ZERO length chains, and add explaining comment.
31150 2017-02-23 Jeff Law <law@redhat.com>
31152 PR tree-optimization/79578
31153 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31154 in call to operand_equal_p.
31156 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31159 * config/i386/cpuid.h: Fix another undefined behavior.
31161 2017-02-23 Richard Biener <rguenther@suse.de>
31163 PR tree-optimization/79683
31164 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31165 vector types for data-refs.
31167 2017-02-23 Martin Liska <mliska@suse.cz>
31169 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31171 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31173 PR middle-end/79665
31174 * internal-fn.c (get_range_pos_neg): Moved to ...
31175 * tree.c (get_range_pos_neg): ... here. No longer static.
31176 * tree.h (get_range_pos_neg): New prototype.
31177 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31178 are known to be in between 0 and signed maximum inclusive, try to
31179 expand both unsigned and signed divmod and use the cheaper one from
31182 2017-02-22 Jeff Law <law@redhat.com>
31184 PR tree-optimization/79578
31185 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31186 to compare base operands.
31188 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31191 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31192 gpc_reg_operand instead of fpr_reg_operand.
31194 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31196 * config/mips/mips.c (mips_return_in_memory): Force FP
31197 vector types to be returned in memory for o32 ABI.
31199 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31201 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31202 instead of DW_TAG_member for static data member declarations and don't
31203 set no_linkage_name for static inline data members.
31204 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31207 2017-02-22 Martin Liska <mliska@suse.cz>
31209 * doc/invoke.texi: Replace inequality signs with square brackets
31212 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31215 * lra-constraints.c (simplify_operand_subreg): Handle
31216 WORD_REGISTER_OPERATIONS targets.
31218 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31221 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31222 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31223 elimination by swapping fld*.
31225 2017-02-22 Richard Biener <rguenther@suse.de>
31227 PR tree-optimization/79673
31228 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31229 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31230 irrelevant address-space qualifiers and avoiding a
31231 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31233 2017-02-22 Richard Biener <rguenther@suse.de>
31235 PR tree-optimization/79666
31236 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31237 to not symbolically negate if that may introduce undefined
31240 2017-02-22 Martin Liska <mliska@suse.cz>
31243 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31244 * data-streamer-out.c (streamer_write_gcov_count_stream):
31246 * value-prof.c (stream_out_histogram_value): Make assert more
31247 precise based on type of counter.
31249 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31252 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31253 Use nonimmediate_operand instead of memory_operand for operand 1.
31254 (float-extend standard_x87sse_constant_load splitter): Ditto.
31256 2017-02-21 Jeff Law <law@redhat.com>
31258 PR tree-optimization/79621
31259 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31260 blocks with edges to themselves.
31262 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31265 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31266 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31267 Use gimple_call_builtin_p.
31270 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31271 on temporarily removed DEBUG_INSNs.
31273 PR tree-optimization/79649
31274 * tree-loop-distribution.c (classify_partition): Give up on
31275 non-generic address space loads/stores.
31277 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31279 * doc/loop.texi (Loop manipulation): Remove nonexistent
31280 tree_ssa_loop_version from the documentation.
31281 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31283 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31286 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31287 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31288 * config/rs6000/rs6000.c: Include except.h.
31289 (rs6000_expand_split_stack_prologue): Call
31290 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31292 2017-02-21 Martin Jambor <mjambor@suse.cz>
31295 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31296 have been analyzed.
31298 2017-02-21 Martin Jambor <mjambor@suse.cz>
31300 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31301 for backward compatibility only.
31302 * doc/invoke.texi (Option Summary): Remove all references to
31303 -fipa-cp-alignment.
31305 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31309 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31311 * lra-constraints.c (curr_insn_transform): Handle
31312 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31314 2017-02-21 Martin Liska <mliska@suse.cz>
31316 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31318 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31320 PR translation/79638
31321 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31323 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31326 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31327 (arm_function_ok_for_sibcall): Return false for an indirect call by
31328 descriptor if all the argument registers are used.
31329 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31330 alignment of the function.
31332 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31334 PR tree-optimization/61441
31335 * simplify-rtx.c (simplify_const_unary_operation): For
31336 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31337 the sNaN unmodified.
31339 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31341 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31342 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31343 instead of SYSTEM_HEADER_DIR.
31345 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31346 Martin Liška <mliska@suse.cz>
31348 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31349 Fix typos and grammar, use active voice, and clarify.
31351 2017-02-20 Marek Polacek <polacek@redhat.com>
31353 PR middle-end/79537
31354 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31357 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31359 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31362 * config/i386/i386.c (ix86_expand_builtin): Handle
31363 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31364 ix86_builtins_isa[fcode].isa as a requirement of those
31365 flags and any other flag in the bitmask.
31366 (ix86_init_mmx_sse_builtins): Use 0 instead of
31367 ~OPTION_MASK_ISA_64BIT as mask.
31368 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31369 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31370 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31371 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31373 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31376 * lra-constraints.c (split_reg): Check requested split mode
31377 is supported by the register.
31379 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31381 * lra-constraints.c (simplify_operand_subreg): Remove early
31384 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31387 * lra-constraints.c (curr_insn_transform): Tighten condition
31388 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31390 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31393 * lra-constraints.c (curr_insn_transform): Handle
31394 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31396 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31399 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31401 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31403 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31406 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31409 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31411 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31414 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31416 * final.c (last_columnnum, override_columnnum): New variables.
31417 (final_start_function): Set last_columnnum, pass it to begin_prologue
31418 hook and pass 0 to dwarf2out_begin_prologue.
31419 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31420 to source_line debug hook.
31421 (notice_source_line): Compute last_columnnum and for debug_column_info
31422 return true on column changes.
31423 * debug.h (struct gcc_debug_hooks): Add column argument to
31424 source_line and begin_prologue hooks.
31425 (debug_nothing_int_charstar_int_bool): Remove prototype.
31426 (debug_nothing_int_int_charstar,
31427 debug_nothing_int_int_charstar_int_bool): New prototypes.
31428 (dwarf2out_begin_prologue): Add column argument.
31429 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31430 begin_prologue hooks.
31431 (debug_nothing_int_charstar_int_bool): Remove.
31432 (debug_nothing_int_int_charstar,
31433 debug_nothing_int_int_charstar_int_bool): New functions.
31434 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31435 through to dwarf2out_source_line.
31436 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31437 (dwarf2out_source_line): Add column argument, emit it if requested.
31438 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31440 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31441 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31442 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31443 through to dwarf2out_begin_prologue.
31444 (vmsdbgout_source_line): Add column argument, pass it through to
31445 dwarf2out_source_line.
31446 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31447 dbxout_source_line caller.
31448 (dbxout_source_line): Add column argument.
31450 * common.opt (gno-column-info, gcolumn-info): New options.
31451 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31452 (check_die): Also test for multiple DW_AT_decl_column attributes.
31453 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31454 DW_AT_decl_column if requested.
31455 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31457 (gen_variable_die): Likewise.
31458 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31459 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31462 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31463 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31464 (ix86_handle_option): Handle OPT_m3dnowa.
31465 * doc/invoke.texi (-m3dnowa): Document.
31466 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31467 -m3dnowa instead of -m3dnow -march=athlon.
31470 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31471 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31473 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31476 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31477 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31478 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31479 generator for vsx_xxpermdi_<mode>_be.
31480 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31481 force big-endian semantics.
31482 (vsx_xxpermdi_<mode>_be): New define_expand with same
31483 implementation as previous version of vsx_xxpermdi_<mode>.
31485 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31487 PR tree-optimization/79327
31488 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31489 variable, its initialization and use.
31491 2017-02-17 Julia Koval <julia.koval@intel.com>
31493 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31494 (OPTION_MASK_ISA_PKU_UNSET): New.
31495 (ix86_handle_option): Handle -mrdpid.
31496 * config/i386/cpuid.h (bit_RDPID): New.
31497 * config/i386/driver-i386.c (host_detect_local_cpu):
31498 Detect RDPID feature.
31499 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31500 * config/i386/i386-c.c (ix86_target_macros_internal):
31502 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31503 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31504 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31505 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31506 * config/i386/i386.md (define_insn "rdpid"): New.
31507 * config/i386/i386.opt Add -mrdpid.
31508 * config/i386/immintrin.h (_rdpid_u32): New.
31510 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31512 PR rtl-optimization/79541
31513 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31514 instead of transforming it into USE.
31516 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31518 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31519 If HONOR_SNANS (SFmode) force the input to a register.
31520 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31521 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31522 an frsp or similar insn.
31524 2017-02-17 Martin Liska <mliska@suse.cz>
31526 PR rtl-optimization/79577
31527 * params.def (selsched-max-sched-times): Increase minimum to 1.
31529 2017-02-17 Martin Liska <mliska@suse.cz>
31531 PR rtl-optimization/79574
31532 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31534 2017-02-17 Martin Liska <mliska@suse.cz>
31536 PR tree-optimization/79529
31537 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31538 ssa_defined_default_def_p to handle cases which are implicitly
31540 * tree-ssa.c (ssa_defined_default_def_p): New function.
31541 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31542 which are implicitly defined.
31543 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31545 2017-02-17 Richard Biener <rguenther@suse.de>
31547 PR middle-end/79576
31548 * params.def (max-ssa-name-query-depth): Limit to 10.
31550 2017-02-17 Richard Biener <rguenther@suse.de>
31552 PR tree-optimization/79552
31553 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31556 2017-02-17 Richard Biener <rguenther@suse.de>
31559 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31561 2017-02-17 Marek Polacek <polacek@redhat.com>
31563 PR middle-end/79536
31564 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31565 (fold_negate_expr): New wrapper.
31567 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31569 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31570 Correct terminology and de-emphasize pre-standard behavior.
31572 2017-02-16 Alan Modra <amodra@gmail.com>
31574 PR rtl-optimization/79286
31575 * ira.c (def_dominates_uses): New function.
31576 (update_equiv_regs): Don't create an equivalence for insns that
31577 may trap where the register def does not dominate the use.
31579 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31581 PR rtl-optimization/78127
31582 * lra.c (lra): Call lra_eliminate before finish the loop after
31585 2017-02-16 Richard Biener <rguenther@suse.de>
31587 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31589 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31590 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31591 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31592 (isl_val_int_from_wi): New function.
31593 (extract_affine_gmp): Rename to ...
31594 (extract_affine_wi): ... this, take a widest_int.
31595 (extract_affine_int): Just wrap extract_affine_wi.
31596 (add_param_constraints): Use isl_val_int_from_wi.
31597 (add_loop_constraints): Likewise, and extract_affine_wi.
31599 2017-02-15 Jeff Law <law@redhat.com>
31601 PR middle-end/79521
31602 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31603 ira_init_register_move_cost_if_necessary.
31605 2017-02-15 Martin Sebor <msebor@redhat.com>
31607 PR middle-end/32003
31608 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31609 removed in a prior commit.
31611 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31613 PR tree-optimization/79347
31614 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31615 counters during peeling.
31617 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31619 * Makefile.in (site.exp): Remove "set ISLVER".
31621 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31624 * real.c (real_from_integer): Call real_convert even for decimal.
31626 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31629 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31631 2017-02-14 Andrew Pinski <apinski@cavium.com>
31633 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31634 cores and change the partno/implementer to be correct.
31635 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31636 the 'B" as the implementer.
31637 * config/aarch64/aarch64-tune.md: Regenerate.
31639 2017-02-14 Carl Love <cel@us.ibm.com>
31641 * config/rs6000/rs6000.c: Add case statement entry to make the
31642 xvcvuxdsp built-in argument unsigned.
31643 * config/rs6000/vsx.md: Fix the source and return operand types so they
31644 match the instruction definitions from the ISA document. Fix typo
31645 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31648 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31651 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31652 member early_clobber_alts.
31653 * lra-lives.c (reg_early_clobber_p): New.
31654 (process_bb_lives): Use it.
31655 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31656 (debug_operand_data): Initialize early_clobber_alts.
31657 (setup_operand_alternative): Set up early_clobber_alts.
31658 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31659 alternatives to new_insn_reg.
31660 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31662 (lra_update_insn_regno_info): Pass the new arg.
31664 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31666 PR middle-end/79505
31667 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31668 (new_oacc_loop_raw): Don't clear already cleared fields.
31671 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31672 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31673 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31675 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31678 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31680 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31683 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31684 the extra instruction to the right place to store 128-bit constant
31687 2017-02-14 Martin Sebor <msebor@redhat.com>
31689 PR middle-end/79448
31690 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31691 warning for strings of unknown length.
31693 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31695 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31697 2017-02-14 Jeff Law <law@redhat.com>
31700 * ira-costs.c (scan_one_insn): Initialize register move costs
31701 for pseudos seen in USE/CLOBBER insns.
31703 PR tree-optimization/79095
31704 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31705 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31706 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31707 if the operands are known to be not equal, then the resulting range
31709 (intersect_ranges): If the new range is ~[0,0] and the old range is
31710 wide, then prefer ~[0,0].
31711 * tree-vrp.c (overflow_comparison_p_1): New function.
31712 (overflow_comparison_p): New function.
31713 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31714 if NAME is used in an overflow test.
31715 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31716 overflow check that can be expressed as an equality test, then adjust
31717 ops to be that equality test.
31719 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31721 * config/s390/s390-builtin-types.def: Remove flags argument.
31722 * config/s390/s390.c (s390_init_builtins): Likewise.
31724 2017-02-14 Martin Liska <mliska@suse.cz>
31726 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31727 vector. Fix trailing white spaces.
31729 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31731 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31734 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31736 PR rtl-optimization/68664
31737 * config/arm/arm.c (arm_sched_can_speculate_insn):
31738 New function. Declare prototype.
31739 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31741 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31743 PR rtl-optimization/68664
31744 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31746 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31748 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31750 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31751 max skip bytes for function, loop and jump.
31753 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31755 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31756 ABS_EXPR for gimple dump.
31758 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31761 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31763 PR tree-optimization/79408
31764 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31765 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31766 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31767 also if rhs1 is INTEGER_CST.
31769 2017-02-14 Richard Biener <rguenther@suse.de>
31771 PR middle-end/79432
31772 * tree-into-ssa.c (insert_phi_nodes): When the function can
31773 have abnormal edges rewrite SSA names with broken use-def
31774 dominance out of SSA and register them for PHI insertion.
31776 2017-02-13 Martin Sebor <msebor@redhat.com>
31778 PR middle-end/79496
31779 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31780 clearing info.nowrite flag when snprintf size argument is a range.
31782 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31784 * cprop.c (cprop_jump): Add missing space in string literal.
31785 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31786 (get_constraint_for_component_ref): Likewise.
31787 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31788 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31789 * lra-constraints.c (process_alt_operands): Likewise.
31790 * ipa-inline.c (inline_small_functions): Likewise.
31791 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31792 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31793 * trans-mem.c (diagnose_tm_1_op): Likewise.
31794 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31795 (grid_parallel_clauses_gridifiable): Likewise.
31797 * config/nvptx/mkoffload.c (process): Add space in between
31800 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31801 "MOD4_SSE_REGS" and "ALL_REGS".
31803 * spellcheck.c (test_data): Add , in between "foo" and "food".
31805 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31808 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31809 boundary crossing check and subsequent code generation agree.
31811 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31813 * config/aarch64/aarch64.c (has_memory_op): Delete.
31814 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31817 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31819 PR rtl-optimization/79388
31820 PR rtl-optimization/79450
31821 * combine.c (distribute_notes): When removing TEM_INSN for which
31822 corresponding dest has last value recorded, invalidate that last
31825 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31827 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31828 of explicit '@'. Add missing assembly comment marker on branch costs
31831 2017-02-13 Nathan Sidwell <nathan@acm.org>
31833 * gengtype-lex.l (<in_struct>): Add '/'.
31835 2017-02-13 Martin Liska <mliska@suse.cz>
31838 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31840 2017-02-13 Richard Biener <rguenther@suse.de>
31842 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31844 * configure: Re-generate.
31845 * config.in: Likewise.
31846 * graphite-dependences.c: Simplify as if
31847 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31848 * graphite-isl-ast-to-gimple.c: Likewise.
31849 * graphite-optimize-isl.c: Likewise.
31850 * graphite-poly.c: Likewise.
31851 * graphite-sese-to-poly.c: Likewise.
31852 * graphite.h: Likewise.
31853 * toplev.c: Include isl/version.h and use isl_version () for
31854 printing the ISL version.
31855 * doc/install.texi: Update ISL requirement.
31857 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31859 * doc/standards.texi (Standards): Update reference to
31862 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31864 * doc/extend.texi (Named Address Spaces): sourceware.org now
31866 * doc/install.texi (Binaries): Ditto.
31869 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31871 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31872 standard terminology "stringize"/"stringizing" throughout.
31873 * doc/cppinternals.texi: Likewise.
31875 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31877 * doc/extend.texi: Fix some spelling mistakes and typos.
31878 * doc/invoke.texi: Likewise.
31880 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31883 * params.def (inline-min-speedup) Change from 10 to 8.
31885 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31887 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31890 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31893 * ipa-inline-analysis.c (get_minimal_bb): New function.
31894 (record_modified): Use it.
31895 (remap_edge_change_prob): Handle also ancestor functions.
31897 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31899 * doc/contrib.texi (Contributors): Remove broken link into
31900 the Mauve CVS repository.
31902 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31904 PR middle-end/79454
31905 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31906 result computation whenever lhs doesn't have vector mode, not
31907 just when it has BLKmode.
31909 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31911 * doc/makefile.texi (profiledbootstrap): Refer to the
31912 installation instructions only in textual form.
31914 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31917 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31919 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31921 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31922 (Specific): Update mingw-w64 reference.
31924 (Specific): Remove broken link to Renesas RX processor.
31926 2017-02-10 Richard Biener <rguenther@suse.de>
31928 * toplev.c (process_options): Do not mention obsolete graphite
31929 options when printing sorry message about missing graphite support.
31930 Mention -floop-nest-optimize.
31932 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31934 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31935 (vtst_p16): Likewise.
31936 (vtstq_p8): Likewise.
31937 (vtstq_p16): Likewise.
31939 (vtstq_p64): Likewise.
31940 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31941 (vset_lane_p64): New.
31942 (vsetq_lane_p64): New.
31944 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31946 PR tree-optimization/79411
31947 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31948 stmt operands are SSA_NAMEs used in abnormal phis.
31949 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31952 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31955 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31958 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31960 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31962 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31964 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31968 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31969 not arbitrary TREE_CONSTANT.
31972 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31973 "omp declare target link" attribute unless is_global_var.
31974 * omp-offload.c (find_link_var_op): Likewise.
31976 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31977 Chung-Lin Tang <cltang@codesourcery.com>
31979 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31981 (gimplify_adjust_omp_clauses): Don't delete TILE.
31982 (gimplify_omp_for): Deal with TILE.
31983 * internal-fn.c (expand_GOACC_TILE): New function.
31984 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31986 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31987 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31989 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31990 avoid DIV for outermost collapse var.
31991 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31992 Remove out of date comments, fix whitespace.
31993 * omp-general.c (omp_extract_for_data): Deal with tiling.
31994 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31995 adjust OLF_DIM_BASE value.
31996 (struct omp_for_data): Add tiling field.
31997 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31998 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31999 for auto loops. Remove default auto determining, moved to
32000 oacc_loop_fixed_partitions.
32001 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32002 stmts, add e_mask field.
32003 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32004 (oacc_thread_numbers): Use oacc_dim_call.
32005 (oacc_xform_tile): New.
32006 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32007 (finish_oacc_loop): Adjust for ifns vector.
32008 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32009 add case for GOACC_TILE fns.
32010 (oacc_loop_xform_loop): Delete.
32011 (oacc_loop_process): Iterate over call list directly, and add
32012 handling for GOACC_TILE fns.
32013 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32015 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32016 vector partitioning to outer loops. Assign 2 partitions to loops
32017 when available. Add TILE handling.
32018 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32019 (execite_oacc_device_lower): Process GOACC_TILE fns,
32020 ignore unknown specs.
32021 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32022 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32023 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32025 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32027 * configure.ac (ACX_BUGURL): Update.
32028 * configure: Regenerate.
32030 2017-02-09 Richard Biener <rguenther@suse.de>
32032 PR tree-optimization/69823
32033 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32034 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32036 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32038 * config/arc/arc-c.def: Add __NPS400__ definition.
32039 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32040 (TARGET_NPS400): Define.
32042 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32044 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32046 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32047 pointer, arch_info.
32048 (arc_cpu_types): Fill the arch_info field with a pointer into the
32049 arc_arch_types table.
32050 (arc_selected_cpu): Declare.
32051 * config/arc/arc.c (arc_selected_cpu): Make global.
32052 (arc_selected_arch): Delete.
32053 (arc_base_cpu): Delete.
32054 (arc_override_options): Remove references to deleted variables,
32055 update access to arch information.
32056 (ARC_OPT): Update access to arch information.
32057 (ARC_OPTX): Likewise.
32058 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32059 (TARGET_ARC600): Update access to arch information.
32060 (TARGET_ARC601): Likewise.
32061 (TARGET_ARC700): Likewise.
32062 (TARGET_EM): Likewise.
32063 (TARGET_HS): Likewise.
32064 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32067 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32070 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32071 condition/operands for integer GE/LE/GEU/LEU operations.
32073 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32075 PR translation/79397
32076 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32079 2017-02-08 Martin Jambor <mjambor@suse.cz>
32082 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32083 whether allocation happened.
32084 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32085 nothing was allocated.
32087 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32089 PR tree-optimization/79408
32090 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32091 constant, but SSA_NAME with a known integer range, use the minimum
32092 of that range instead of op1 to determine if modulo can be replaced
32093 with its first operand.
32095 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32097 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32099 2017-02-08 Richard Biener <rguenther@suse.de>
32101 PR tree-optimization/71824
32102 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32103 Check all loops contained in the merged region.
32105 2017-02-07 Andrew Pinski <apinski@cavium.com>
32107 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32109 2017-02-07 Andrew Pinski <apinski@cavium.com>
32111 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32112 (thunderxt88): Likewise.
32113 (thunderxt81): Disable LSE and change v8.1 to v8.
32114 (thunderxt83): Likewise.
32116 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32117 Richard Biener <rguenther@suse.de>
32119 PR middle-end/79399
32120 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32121 type from int to size_t.
32122 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32124 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32126 PR rtl-optimization/79386
32127 * cprop.c (bypass_conditional_jumps): Initialize
32128 bypass_last_basic_block already before splitting bbs after
32129 unconditional traps...
32130 (bypass_conditional_jumps): ... rather than here.
32133 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32134 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32135 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32136 fix -masm=intel patterns.
32138 2017-02-07 Richard Biener <rguenther@suse.de>
32140 PR tree-optimization/79256
32141 PR middle-end/79278
32142 * builtins.c (get_object_alignment_2): Use min_align_of_type
32143 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32144 and ADJUST_FIELD_ALIGN.
32146 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32148 * doc/tm.texi: Regenerate.
32149 * stor-layout.c (layout_decl): Adjust.
32150 (update_alignment_for_field): Likewise.
32151 (place_field): Likewise.
32152 (min_align_of_type): Likewise.
32153 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32154 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32155 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32156 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32157 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32158 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32159 * config/i386/i386.c (x86_field_alignment): Likewise.
32160 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32161 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32162 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32163 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32164 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32165 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32169 2017-01-30 Richard Biener <rguenther@suse.de>
32171 PR tree-optimization/79256
32172 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32173 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32176 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32178 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32179 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32180 builtins to SImode and emit a zero-extend, if necessary.
32182 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32184 * docs/invoke.texi (RISC-V Options): Alphabetize.
32186 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32188 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32191 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32193 * config/riscv/riscv.c: New file.
32194 * common/config/riscv/riscv-common.c: Likewise.
32195 * config.gcc: Likewise.
32196 * config/riscv/constraints.md: Likewise.
32197 * config/riscv/elf.h: Likewise.
32198 * config/riscv/generic.md: Likewise.
32199 * config/riscv/linux.h: Likewise.
32200 * config/riscv/multilib-generator: Likewise.
32201 * config/riscv/peephole.md: Likewise.
32202 * config/riscv/pic.md: Likewise.
32203 * config/riscv/predicates.md: Likewise.
32204 * config/riscv/riscv-builtins.c: Likewise.
32205 * config/riscv/riscv-c.c: Likewise.
32206 * config/riscv/riscv-ftypes.def: Likewise.
32207 * config/riscv/riscv-modes.def: Likewise.
32208 * config/riscv/riscv-opts.h: Likewise.
32209 * config/riscv/riscv-protos.h: Likewise.
32210 * config/riscv/riscv.h: Likewise.
32211 * config/riscv/riscv.md: Likewise.
32212 * config/riscv/riscv.opt: Likewise.
32213 * config/riscv/sync.md: Likewise.
32214 * config/riscv/t-elf-multilib: Likewise.
32215 * config/riscv/t-linux: Likewise.
32216 * config/riscv/t-linux-multilib: Likewise.
32217 * config/riscv/t-riscv: Likewise.
32218 * configure.ac: Likewise.
32219 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32220 Waterman as RISC-V maintainers.
32221 * doc/install.texi: Add RISC-V entries.
32222 * doc/invoke.texi: Add RISC-V options section.
32223 * doc/md.texi: Add RISC-V constraints section.
32224 * configure: Regenerated.
32226 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32229 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32230 false values to be constant vectors with all 0 or all 1 bits set.
32231 (vcondu<mode><mode>): Likewise.
32232 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32234 (fpmask_comparison_operator): Update comment.
32235 (vecint_comparison_operator): New predicate.
32236 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32237 vector conditionals when the true and false values are constant
32238 vectors with all 0 bits or all 1 bits set.
32240 2017-02-06 Martin Sebor <msebor@redhat.com>
32242 PR tree-optimization/79376
32243 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32245 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32247 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32248 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32249 to simplify split condition.
32251 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32253 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32254 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32257 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32259 PR rtl-optimization/68664
32260 * target.def (can_speculate_insn): New hook.
32261 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32262 * doc/tm.texi: Regenerate.
32263 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32264 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32265 (rs6000_sched_can_speculate_insn): New function.
32267 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32269 PR tree-optimization/79284
32270 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32271 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32272 vectorizable_mask_load_store, vectorizable_operation,
32273 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32274 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32275 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32276 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32277 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32278 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32279 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32280 is_gimple_assign (stmt). Replace another such test with
32281 is_gimple_assign (stmt).
32283 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32286 * config/avr/avr.c (rtl-iter.h): Include it.
32287 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32288 (avr_legitimate_combined_insn): ...and implementation.
32290 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32292 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32293 * config/s390/s390.c (s390_const_operand_ok)
32294 (s390_canonicalize_comparison, s390_extract_part)
32295 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32296 (s390_contiguous_bitmask_p, s390_rtx_costs)
32297 (legitimize_pic_address): Likewise.
32298 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32299 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32300 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32301 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32302 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32304 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32306 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32307 REGNO($0) == REGNO($1).
32309 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32311 * config/s390/linux.h(SIZE_TYPE): Add comment.
32313 2017-02-06 Julian Brown <julian@codesourcery.com>
32314 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32315 Virendra Pathak <virendra.pathak@broadcom.com>
32317 * config/aarch64/aarch64-cores.def: Change the scheduler
32319 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32320 * config/aarch64/thunderx2t99.md: New file.
32322 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32324 * doc/standards.texi (Go Language): Update link to language
32327 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32329 * tree-eh.c (lower_resx): Sanitize profile.
32330 (cleanup_empty_eh_move_lp): Likewise.
32332 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32335 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32337 * cfgloopmanip.h (loop_version): Update prototype.
32338 * modulo-sched.c (sms_schedule): Update call of loop_version.
32339 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32340 * tree-parloops.c (gen_parallel_loop): Likewise.
32341 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32342 * tree-ssa-loop-split.c (split_loop): Likewise.
32343 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32344 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32346 2017-02-05 Martin Liska <mliska@suse.cz>
32349 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32351 (print_operand_address): Initialize a struct to zero.
32353 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32355 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32356 garbage collector only in textual form.
32358 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32360 * doc/extend.texi (x86 specific memory model extensions for
32361 transactional memory): Simplify a phrase.
32363 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32366 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32367 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32368 (atomic_storedi_1): Likewise.
32370 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32372 PR tree-optimization/79338
32373 * tree-parloops.c (gather_scalar_reductions): Don't call
32374 vect_analyze_loop_form for loop->inner before destroying loop's
32377 2017-02-03 Martin Sebor <msebor@redhat.com>
32379 PR tree-optimization/79327
32380 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32381 when precision has resulted in leading zeros.
32382 (format_integer): Adjust the likely counter to assume an unknown
32383 argument that may be zero is non-zero.
32385 2017-02-03 Jason Merrill <jason@redhat.com>
32388 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32389 avoid copying non-taken branch.
32391 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32393 PR tree-optimization/79340
32394 * tree-vect-loop.c (vectorizable_reduction): Release
32395 vec_defs elements after safe_splicing them into other vectors.
32398 PR tree-optimization/79327
32399 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32400 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32402 (format_integer): Use wide_int_to_tree instead of build_int_cst
32403 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32404 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32405 of shortest and longest sequence.
32407 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32409 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32410 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32412 2017-02-03 Walter Lee <walt@tilera.com>
32415 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32416 after initial stackframe link reg save.
32417 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32419 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32422 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32423 wu for stxssp alternative.
32425 2017-02-03 Martin Sebor <msebor@redhat.com>
32427 PR tree-optimization/79352
32428 * gimple-fold.c (get_range_strlen): Add argument.
32429 (get_range_strlen): Change return type to bool.
32430 (get_maxval_strlen): Pass in a dummy argument.
32431 * gimple-fold.h (get_range_strlen): Change return type to bool.
32432 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32433 * tree.h (array_at_struct_end_p): Add argument.
32434 * tree.c (array_at_struct_end_p): Handle it.
32436 2017-02-03 Martin Liska <mliska@suse.cz>
32439 * multiple_target.c (create_dispatcher_calls): Redirect edge
32440 from a caller of a dispatcher.
32441 (expand_target_clones): Make the clones local.
32442 (ipa_target_clone): Do both target clones and resolvers.
32443 (ipa_dispatcher_calls): Remove the pass.
32444 (pass_dispatcher_calls::gate): Likewise.
32445 (make_pass_dispatcher_calls): Likewise.
32446 * passes.def (pass_target_clone): Put as very first IPA early
32449 2017-02-03 Martin Liska <mliska@suse.cz>
32451 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32452 in case of a function with ifunc attribute.
32454 2017-02-03 Martin Liska <mliska@suse.cz>
32456 * cgraph.c (cgraph_node::dump): Dump function version info.
32457 * symtab.c (symtab_node::dump_base): Add missing new line.
32459 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32461 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32462 (ifcombine_ifandif): Use it.
32464 2017-02-03 Martin Liska <mliska@suse.cz>
32466 * doc/invoke.texi: Document default value for
32467 use-after-scope-direct-emission-threshold.
32469 2017-02-03 Martin Liska <mliska@suse.cz>
32471 PR tree-optimization/79339
32472 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32473 (format_floating): Likewise.
32475 2017-02-03 Martin Liska <mliska@suse.cz>
32478 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32480 (ipa_node_params_t::remove): Likewise.
32481 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32482 initialization from removed ipa_node_params_t::insert.
32483 (ipa_node_params::~ipa_node_params): Move from removed
32484 ipa_node_params_t::release.
32485 * symbol-summary.h (symbol_summary::m_released): New member.
32486 Do not release a summary twice. Do not allow to call finalizer
32487 for types of a summary that live in GGC memory.
32489 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32491 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32494 2017-02-02 Martin Sebor <msebor@redhat.com>
32496 PR middle-end/79275
32497 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32498 (format_string): Tighten up the range of output for non-constant
32499 strings and correct the expected range for wide non-constant strings.
32501 2017-02-02 Martin Sebor <msebor@redhat.com>
32503 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32505 PR middle-end/32003
32506 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32508 (-fdump-tree-@var): Add to index and document how to come up
32509 with pass-specific option and dump file names.
32510 (-fdump-passes): Clarify where to look for output.
32512 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32514 PR middle-end/77445
32515 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32516 statistics of the analyzed path; allow threading for speed when
32517 any of BBs along the path are optimized for speed.
32519 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32521 PR middle-end/78468
32522 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32523 settings of the virtual registers.
32526 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32528 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32529 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32532 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32534 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32535 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32537 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32539 * config/s390/s390.md: Add missing comments with the expanded
32541 * config/s390/vector.md: Likewise.
32542 * config/s390/vx-builtins.md: Likewise.
32544 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32547 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32548 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32549 conditions on a single line.
32551 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32553 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32554 __S390_VX__ to __VX__.
32556 2017-02-01 Andrew Pinski <apinski@cavium.com>
32558 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32559 stmt_info to record_stmt_cost.
32560 (vect_get_known_peeling_cost): Pass stmt_info if known to
32562 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32563 cpu_vector_cost field into
32564 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32565 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32566 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32567 splitting of scalar_stmt_cost and vec_stmt_cost.
32568 (thunderx_vector_cost): Likewise.
32569 (cortexa57_vector_cost): LIkewise.
32570 (exynosm1_vector_cost): Likewise.
32571 (xgene1_vector_cost): Likewise.
32572 (thunderx2t99_vector_cost): Improve after the splitting of the two
32574 (aarch64_builtin_vectorization_cost): Update for the splitting of
32575 scalar_stmt_cost and vec_stmt_cost.
32577 2017-02-01 Torvald Riegel <triegel@redhat.com>
32578 Richard Henderson <rth@redhat.com>
32580 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32581 conditional on existance of a fast atomic load.
32582 * optabs-query.c (can_atomic_load_p): New function.
32583 * optabs-query.h (can_atomic_load_p): Declare it.
32584 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32585 no fast atomic load is available for the particular size of access.
32586 (expand_atomic_compare_and_swap): Likewise.
32587 (expand_atomic_load): Likewise.
32588 (expand_atomic_store): Likewise.
32589 (expand_atomic_fetch_op): Likewise.
32590 * testsuite/lib/target-supports.exp
32591 (check_effective_target_sync_int_128): Remove x86 because it provides
32592 no fast atomic load.
32593 (check_effective_target_sync_int_128_runtime): Likewise.
32595 2017-02-01 Richard Biener <rguenther@suse.de>
32597 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32598 (graphite_transform_loops): Provide opt-info for optimized nests.
32599 * tree-parloop.c (parallelize_loops): Provide opt-info for
32600 parallelized loops.
32602 2017-02-01 Richard Biener <rguenther@suse.de>
32604 PR middle-end/79315
32605 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32606 was not set before.
32608 2017-02-01 Richard Biener <rguenther@suse.de>
32610 PR tree-optimization/71824
32611 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32612 Verify the loops are valid in the merged SESE region.
32613 (scop_detection::can_represent_loop_1): Check analyzing the
32614 evolution of the number of iterations in the region succeeds.
32616 2017-01-31 Ian Lance Taylor <iant@golang.org>
32618 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32619 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32621 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32623 PR preprocessor/79210
32624 * input.c (get_substring_ranges_for_loc): Replace line_width
32625 assertion with error-handling.
32627 2017-01-31 Richard Biener <rguenther@suse.de>
32629 PR tree-optimization/77318
32630 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32631 (create_pw_aff_from_tree): Take loop parameter.
32632 (add_condition_to_pbb): Pass loop of the condition to
32633 create_pw_aff_from_tree.
32635 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32637 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32638 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32640 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32644 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32646 (convert_int_to_float128): Likewise.
32647 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32648 (convert_int_to_float128): Likewise.
32649 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32650 (UNSPEC_IEEE128_CONVERT): Likewise.
32651 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32652 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32653 Use local variables for IBM extended format.
32654 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32655 (fix_trunc<mode>si2_fprs): Likewise.
32656 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32657 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32658 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32659 to know that we can now have integers of all sizes in vector
32661 (fix<uns>_<mode>di2_hw): Likewise.
32662 (float<uns>_<mode>si2_hw): Likewise.
32663 (fix_<mode>si2_hw): Likewise.
32664 (fixuns_<mode>si2_hw): Likewise.
32665 (float<uns>_<mode>di2_hw): Likewise.
32666 (float_<mode>di2_hw): Likewise.
32667 (float_<mode>si2_hw): Likewise.
32668 (floatuns_<mode>di2_hw): Likewise.
32669 (floatuns_<mode>si2_hw): Likewise.
32670 (xscvqp<su>wz_<mode>): Delete, no longer used.
32671 (xscvqp<su>dz_<mode>): Likewise.
32672 (xscv<su>dqp_<mode>): Likewise.
32673 (ieee128_mfvsrd_64bit): Likewise.
32674 (ieee128_mfvsrd_32bit): Likewise.
32675 (ieee128_mfvsrwz): Likewise.
32676 (ieee128_mtvsrw): Likewise.
32677 (ieee128_mtvsrd_64bit): Likewise.
32678 (ieee128_mtvsrd_32bit): Likewise.
32680 2017-01-31 Martin Liska <mliska@suse.cz>
32683 * ipa-prop.c (ipa_free_all_node_params): Call release method
32684 instead of ~sumbol_summary to not to trigger double times
32687 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32689 PR tree-optimization/71691
32690 * bitmap.h (class auto_bitmap): New.
32691 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32692 is_maybe_undefined instead of ssa_undefined_value_p.
32694 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32696 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32697 __S390_ARCH_LEVEL__ to __ARCH__.
32699 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32701 PR tree-optimization/79267
32702 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32703 if should_remove_lhs_p is true.
32705 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32708 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32709 (add_alignment_attribute): New.
32710 (base_type_die): Add alignment attribute.
32711 (subrange_type_die): Likewise.
32712 (modified_type_die): Likewise.
32713 (gen_array_type_die): Likewise.
32714 (gen_descr_array_type_die: Likewise.
32715 (gen_enumeration_type_die): Likewise.
32716 (gen_subprogram_die): Likewise.
32717 (gen_variable_die): Likewise.
32718 (gen_field_die): Likewise.
32719 (gen_ptr_to_mbr_type_die): Likewise.
32720 (gen_struct_or_union_type_die): Likewise.
32721 (gen_subroutine_type_die): Likewise.
32722 (gen_typedef_die): Likewise.
32723 (base_type_cmp): Compare alignment attribute.
32725 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32728 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32729 (setb_unsigned) New pattern for setb with CCUNS.
32730 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32731 subfc./subfe sequence to avoid overflow problems. Generate a
32732 shorter sequence with cmpld/setb for power9.
32733 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32734 for generating subfc. instruction.
32735 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32736 now uses this instruction.
32738 2017-01-30 Ian Lance Taylor <iant@google.com>
32741 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32742 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32744 2017-01-30 Martin Sebor <msebor@redhat.com>
32746 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32747 Move constant to the right of a relational operator.
32748 (get_mpfr_format_length, format_character, format_string): Ditto.
32749 (should_warn_p, maybe_warn): Same.
32751 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32753 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32756 * asan.c (get_translation_unit_decl): Remove function.
32757 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32759 2017-01-30 Martin Liska <mliska@suse.cz>
32761 PR gcov-profile/79259
32762 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32763 -fprofile-generate.
32765 2017-01-30 Martin Liska <mliska@suse.cz>
32768 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32769 Initialize variables with NULL value.
32771 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32774 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32776 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32778 2017-01-30 Richard Biener <rguenther@suse.de>
32780 PR tree-optimization/79276
32781 * tree-vrp.c (process_assert_insertions): Properly adjust common
32782 when removing a duplicate.
32784 2017-01-30 Richard Biener <rguenther@suse.de>
32786 PR tree-optimization/79256
32787 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32788 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32790 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32792 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32795 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32796 ("*r<noxa>sbg_<mode>_sll_bitmask")
32797 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32798 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32799 Use contiguous_bitmask_nowrap_operand.
32801 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32804 * config/rs6000/altivec.h (vec_xl): Revise #define.
32805 (vec_xst): Likewise.
32807 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32809 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32811 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32813 PR rtl-optimization/79194
32814 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32815 traps before call to bypass_conditional_jumps.
32817 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32819 PR tree-optimization/71374
32820 * lra-constraints.c (check_conflict_input_operands): New.
32821 (match_reload): Use it.
32823 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32826 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32827 account to calculate conflict_set.
32829 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32831 PR rtl-optimization/78559
32832 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32833 other_insn in combine.
32835 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32837 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32838 uint16_type_node for BT_UINT16.
32840 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32842 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32843 "RTL Tests" to menu.
32844 (GIMPLE Tests): New node.
32845 (RTL Tests): New node.
32847 2017-01-27 Richard Biener <rguenther@suse.de>
32849 PR tree-optimization/79245
32850 * tree-loop-distribution.c (distribute_loop): Apply cost
32851 modeling also to detected patterns.
32853 2017-01-27 Richard Biener <rguenther@suse.de>
32855 PR tree-optimization/71433
32856 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32857 (compare_assert_loc): New function.
32858 (process_assert_insertions): Sort and optimize assert locations
32859 to remove duplicates and push down identical assertions on
32860 edges to their destination block.
32862 2017-01-27 Richard Biener <rguenther@suse.de>
32864 PR tree-optimization/79244
32865 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32866 out SSA names even if abnormal.
32868 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32870 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32871 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32872 instead of MPFR_RNDN.
32874 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32877 * arm.c (arm_option_override): Don't call build_target_option_node
32878 until after doing all option overrides.
32879 (arm_valid_target_attribute_tree): Likewise.
32881 2017-01-27 Martin Liska <mliska@suse.cz>
32883 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32884 for {cd}tors and C++ {cd}tors.
32886 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32888 * config/s390/s390.md ("*setmem_long_and")
32889 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32891 2017-01-26 Martin Sebor <msebor@redhat.com>
32893 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32896 2017-01-26 Martin Sebor <msebor@redhat.com>
32898 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32899 HAVE_DFmode before using XFmode or DFmode.
32900 (parse_directive): Avoid using the z length modifier to avoid
32901 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32903 PR middle-end/78703
32904 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32905 to accept adjustment as an array.
32906 (get_int_range): New function.
32907 (struct directive): Make width and prec arrays.
32908 (directive::set_width, directive::set_precision): Call get_int_range.
32909 (format_integer, format_floating): Handle width and precision ranges.
32910 (format_string, parse_directive): Same.
32912 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32915 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32916 comdat_type_p set, just clone them, but keep the children in the
32920 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32921 which have direct callers with -fvar-tracking-assignments enabled
32923 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32924 inside of type units.
32926 2017-01-26 Martin Sebor <msebor@redhat.com>
32928 PR middle-end/78703
32929 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32931 (struct format_result): Replace number_chars, number_chars_min,
32932 and number_chars_max with a single member of struct result_range.
32934 (format_result::operator+=): Adjust.
32935 (struct fmtresult): Remove bounded. Handle likely and unlikely
32937 (fmtresult::adjust_for_width_or_precision): New function.
32938 (fmtresult:type_max_digits): New function.
32939 (bytes_remaining): Handle likely and unlikely counters.
32940 (min_bytes_remaining): Remove.
32941 (format_percent): Simplify.
32942 (format_integer, format_floating): Set likely and unlikely counters.
32943 (get_string_length, format_character, format_string): Same.
32944 (format_plain, should_warn_p): New function.
32945 (maybe_warn): Call should_warn_p. Update diagnostic messages
32946 and handle those for all directives, including plain strings.
32947 (format_directive): Handle likely and unlikely counters.
32948 Remove unnecessary quoting from diagnostics. Add an informational
32950 (add_bytes): Remove.
32951 (pass_sprintf_length::compute_format_length): Simplify.
32952 (try_substitute_return_value): Handle likely and unlikely counters.
32954 2017-01-26 Carl Love <cel@us.ibm.com>
32956 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32957 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32959 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32962 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32963 endianess for subregs into account.
32964 * lra-constraints.c (lra_constraints): Do risky transformations
32965 always on the first iteration.
32966 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32967 last_call_used_reg_set.
32968 (process_bb_lives): Define and use last_call_used_reg_set.
32969 * lra.c (lra): Always continue after lra_constraints on the first
32972 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32974 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32976 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32978 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32980 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32981 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32982 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32983 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32984 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32985 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32986 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32987 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32988 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32990 2017-01-26 Marek Polacek <polacek@redhat.com>
32993 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32994 for the third operand.
32996 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32998 PR middle-end/79236
32999 * omp-low.c (struct omp_context): Add simt_stmt field.
33000 (scan_omp_for): Return omp_context *.
33001 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33002 context to the _simt_ SIMD stmt.
33003 (lower_omp_for): For combined SIMD with sibling _simt_
33004 SIMD, make sure to use the same decls in _looptemp_
33005 clauses as in the sibling.
33007 2017-01-26 David Sherwood <david.sherwood@arm.com>
33009 PR middle-end/79212
33010 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33013 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33016 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33017 emit fld b; fld a; if possible.
33019 * brig-builtins.def: Update copyright years.
33020 * config/arm/arm_acle_builtins.def: Update copyright years.
33022 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33025 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33026 constraint instead of o for the stxsd instruction.
33028 2017-01-25 Carl Love <cel@us.ibm.com>
33030 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33031 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33033 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33035 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33037 2017-01-25 Richard Biener <rguenther@suse.de>
33039 PR tree-optimization/69264
33040 * target.def (vector_alignment_reachable): Improve documentation.
33041 * doc/tm.texi: Regenerate.
33042 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33044 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33045 earlier changes with respect to TYPE_USER_ALIGN.
33046 (vector_alignment_reachable_p): Likewise. Improve dumping.
33048 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33051 * config/arm/arm.md (xordi3): Force constant operand into a register
33054 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33056 * doc/invoke.texi (-fstore-merging): Correct default optimization
33057 levels at which it is enabled.
33058 (-O): Move -fstore-merging from list to...
33061 2017-01-25 Richard Biener <rguenther@suse.de>
33064 * omp-expand.c: Include debug.h.
33065 (expand_omp_taskreg): Make sure to generate early debug before
33066 outlining anything from a function.
33067 (expand_omp_target): Likewise.
33068 (grid_expand_target_grid_body): Likewise.
33070 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33073 * asan.c (get_translation_unit_decl): New function.
33074 (asan_add_global): Extract modules file name from globals
33075 TRANSLATION_UNIT_DECL name.
33077 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33080 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33081 for long calls with APCS frame and VFP.
33083 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33085 * cfg.c (original_copy_tables_initialized_p): New function.
33086 * cfg.h (original_copy_tables_initialized_p): New decl.
33087 * cfgrtl.c (relink_block_chain): Guard the call to
33088 free_original_copy_tables with a call to
33089 original_copy_tables_initialized_p.
33090 * cgraph.h (symtab_node::native_rtl_p): New decl.
33091 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33092 (symtab_node::needed_p): Don't assert for early assembly output
33093 for __RTL functions.
33094 (cgraph_node::finalize_function): Set "force_output" for __RTL
33096 (cgraph_node::analyze): Bail out early for __RTL functions.
33097 (analyze_functions): Update assertion to support __RTL functions.
33098 (cgraph_node::expand): Bail out early for __RTL functions.
33099 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33101 * function.h (struct function): Update comment for field
33103 * gimple-expr.c: Include "tree-pass.h".
33104 (gimple_has_body_p): Return false for __RTL functions.
33105 * Makefile.in (OBJS): Add run-rtl-passes.o.
33106 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33108 (gcc::pass_manager::get_clean_slate): New accessor.
33109 * passes.c: Include "insn-addr.h".
33110 (should_skip_pass_p): Add logging. Update logic for running
33111 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33112 property-provider override so it is only done for gimple passes.
33114 (skip_pass): New function.
33115 (execute_one_pass): Call skip_pass when skipping passes.
33116 * read-md.c (md_reader::read_char): Support filtering
33117 the input to a subset of line numbers.
33118 (md_reader::md_reader): Initialize fields
33119 m_first_line and m_last_line.
33120 (md_reader::read_file_fragment): New function.
33121 * read-md.h (md_reader::read_file_fragment): New decl.
33122 (md_reader::m_first_line): New field.
33123 (md_reader::m_last_line): New field.
33124 * read-rtl-function.c (function_reader::create_function): Only
33125 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33126 curr_properties. Set DECL_INITIAL to a dummy block.
33127 (read_rtl_function_body_from_file_range): New function.
33128 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33130 * run-rtl-passes.c: New file.
33131 * run-rtl-passes.h: New file.
33133 2017-01-24 Jeff Law <law@redhat.com>
33135 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33138 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33140 PR tree-optimization/79159
33141 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33142 (record_nonwrapping_iv): Improve boundary using above function if no
33143 value range information.
33145 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33146 Martin Jambor <mjambor@suse.cz>
33148 * brig-builtins.def: New file.
33149 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33150 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33151 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33152 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33153 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33154 * builtin-types.def (BT_INT8): New.
33155 (BT_INT16): Likewise.
33156 (BT_UINT8): Likewise.
33157 (BT_UINT16): Likewise.
33158 (BT_FN_ULONG): Likewise.
33159 (BT_FN_UINT_INT): Likewise.
33160 (BT_FN_UINT_ULONG): Likewise.
33161 (BT_FN_UINT_LONG): Likewise.
33162 (BT_FN_UINT_PTR): Likewise.
33163 (BT_FN_ULONG_PTR): Likewise.
33164 (BT_FN_INT8_FLOAT): Likewise.
33165 (BT_FN_INT16_FLOAT): Likewise.
33166 (BT_FN_UINT32_FLOAT): Likewise.
33167 (BT_FN_UINT16_FLOAT): Likewise.
33168 (BT_FN_UINT8_FLOAT): Likewise.
33169 (BT_FN_UINT64_FLOAT): Likewise.
33170 (BT_FN_UINT16_UINT32): Likewise.
33171 (BT_FN_UINT32_UINT16): Likewise.
33172 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33173 (BT_FN_INT_PTR_INT): Likewise.
33174 (BT_FN_UINT_PTR_UINT): Likewise.
33175 (BT_FN_LONG_PTR_LONG): Likewise.
33176 (BT_FN_ULONG_PTR_ULONG): Likewise.
33177 (BT_FN_VOID_UINT64_UINT64): Likewise.
33178 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33179 (BT_FN_INT8_INT8_INT8): Likewise.
33180 (BT_FN_INT16_INT16_INT16): Likewise.
33181 (BT_FN_INT_INT_INT): Likewise.
33182 (BT_FN_UINT_FLOAT_UINT): Likewise.
33183 (BT_FN_FLOAT_UINT_UINT): Likewise.
33184 (BT_FN_ULONG_UINT_UINT): Likewise.
33185 (BT_FN_ULONG_UINT_PTR): Likewise.
33186 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33187 (BT_FN_UINT_UINT_UINT): Likewise.
33188 (BT_FN_VOID_UINT_PTR): Likewise.
33189 (BT_FN_UINT_UINT_PTR: Likewise.
33190 (BT_FN_UINT32_UINT64_PTR): Likewise.
33191 (BT_FN_INT_INT_UINT_UINT): Likewise.
33192 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33193 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33194 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33195 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33196 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33197 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33198 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33199 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33200 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33201 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33202 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33203 * doc/frontends.texi: List BRIG FE.
33204 * doc/install.texi (Testing): Add BRIG tesring requirements.
33205 * doc/invoke.texi (Overall Options): Mention BRIG.
33206 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33208 2017-01-24 Richard Biener <rguenther@suse.de>
33210 PR translation/79208
33211 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33213 2017-01-24 Martin Jambor <mjambor@suse.cz>
33216 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33217 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33218 and known_contexts.
33220 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33222 PR middle-end/79123
33223 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33224 casts from signed to unsigned really don't have a range.
33226 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33228 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33229 GMP_RNDx for compatiblity.
33231 2017-01-24 Martin Liska <mliska@suse.cz>
33234 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33235 that would prevent us to call alloca with -1 as argument.
33237 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33239 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33240 Avoid -Wformat-security warning.
33242 2017-01-23 Andrew Pinski <apinski@cavium.com>
33244 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33247 2017-01-23 Martin Sebor <msebor@redhat.com>
33249 PR middle-end/78703
33250 * gimple-ssa-sprintf.c (warn_level): New global.
33251 (format_integer): Use it here and throughout the rest of the file.
33252 Use the same switch to compute sign as base.
33253 (maybe_warn): New function.
33254 (format_directive): Factor out warnings into maybe_warn.
33255 Add debugging output. Use warn_level.
33256 (add_bytes): Use warn_level.
33257 (pass_sprintf_length::compute_format_length): Add debugging output.
33258 (try_substitute_return_value): Same.
33259 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33261 PR middle-end/78703
33262 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33263 (struct fmtresult, format_integer, format_floating): Adjust.
33264 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33265 (get_string_length, format_string,format_directive): Same.
33266 (pass_sprintf_length::compute_format_length): Same.
33267 (try_substitute_return_value): Simplify slightly.
33269 PR middle-end/78703
33270 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33271 (fmtresult::operator+=): Outlined.
33272 (struct fmtresult): Add ctors.
33273 (struct conversion_spec): Rename...
33274 (struct directive): ...to this. Add and remove data members.
33275 (directive::set_width, directive::set_precision): New functions.
33276 (format_percent): Use fmtresult ctor.
33277 (get_width_and_precision): Remove.
33278 (format_integer): Make naming changes. Avoid computing width and
33280 (format_floating): Same. Adjust indentation.
33281 (format_character, format_none): New functions.
33282 (format_string): Moved character handling to format_character.
33283 (format_directive): Remove arguments, change return type.
33284 (parse_directive): New function.
33285 (pass_sprintf_length::compute_format_length): Move directive
33286 parsing to parse_directive.
33288 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33290 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33291 (assign_assembler_name_if_needed): ... this.
33292 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33293 (assign_assembler_name_if_needed): ... this.
33294 (free_lang_data_in_cgraph): Adjust callers.
33295 * cgraphunit.c (cgraph_node::analyze): Likewise.
33296 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33299 2017-01-23 Richard Biener <rguenther@suse.de>
33301 PR tree-optimization/79088
33302 PR tree-optimization/79188
33303 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33304 resetting loop bounds after last path deletion. Reset loop
33305 bounds of the target loop, make code match the comments.
33306 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33307 Make sure loops need no fixups.
33309 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33311 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33312 exponent support with double type for first argument.
33313 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33314 type returned by __builtin_vec_extract_sig,
33315 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33316 functions from "vector int" to "vector unsigned int" or from
33317 "vector long long int" to "vector unsigned long long int".
33318 Changed type returned by __builtin_vec_extract_exp,
33319 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33320 functions from "vector int" to "vector unsigned int" or from
33321 "vector long long int" to "vector unsigned long long int".
33322 Changed return type of __builtin_vec_test_data_class,
33323 __builtin_vec_test_data_class_sp, and
33324 __builtin_vec_test_data_class_dp from "vector int" to
33325 "vector bool int" or from "vector long long int" to "vector bool
33326 long long int" and changed second argument type from "unsigned
33327 int" to "int". Added new overloaded function forms "vector float
33328 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33329 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33330 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33331 double, vector unsigned long long int)" and "vector double
33332 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33333 long int)". Changed return type of
33334 __builtin_scalar_test_data_class and
33335 __builtin_scalar_test_data_class_sp and
33336 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33337 int" and changed second argument from "unsigned int" to "int".
33338 Changed type returned by __builtin_scalar_test_neg,
33339 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33340 from "int" to "bool int". Added new overloaded function form
33341 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33342 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33343 exponent double-precision with floating point first argument.
33344 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33345 documentation of scalar_test_data_class, scalar_test_neg,
33346 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33347 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33348 vec_test_data_class built-in functions to reflect refinements in
33349 their type signatures.
33351 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33353 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33355 (aarch64_elf_asm_destructor): Likewise.
33357 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33359 PR rtl-optimization/78634
33360 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33361 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33362 * ifcvt.c (noce_try_cmove): Add missing cost check.
33364 PR rtl-optimization/71724
33365 * combine.c (if_then_else_cond): Look for situations where it is
33366 beneficial to undo the work of one of the recursive calls.
33368 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33370 PR tree-optimization/70754
33371 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33372 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33373 combined stmt before it if not NULL.
33374 (combine_chains): Process refs reversely and compute dominance point
33377 2017-01-23 Martin Liska <mliska@suse.cz>
33379 PR tree-optimization/79196
33380 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33381 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33383 (strlen_optimize_stmt): Call the renamed function.
33385 2017-01-23 Michael Matz <matz@suse.de>
33387 PR tree-optimization/78384
33388 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33390 2017-01-23 Richard Biener <rguenther@suse.de>
33392 PR tree-optimization/79186
33393 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33394 both incoming edges before moving an assert.
33396 2017-01-23 Martin Jambor <mjambor@suse.cz>
33398 * ipa-prop.c (load_from_param_1): Removed.
33399 (load_from_unmodified_param): Bits from load_from_param_1 put back
33401 (load_from_param): Removed.
33402 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33403 with stmt. Reverted back to use of load_from_unmodified_param.
33405 2017-01-23 Martin Jambor <mjambor@suse.cz>
33408 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33409 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33410 field a pointer to garbage collected vector, mark lattices and
33411 ipcp_orig_node with GTY((skip)).
33412 (ipa_get_param_count): Adjust to descriptors being a pointer.
33413 (ipa_get_param): Likewise.
33414 (ipa_get_type): Likewise.
33415 (ipa_get_param_move_cost): Likewise.
33416 (ipa_set_param_used): Likewise.
33417 (ipa_get_controlled_uses): Likewise.
33418 (ipa_set_controlled_uses): Likewise.
33419 (ipa_is_param_used): Likewise.
33420 (ipa_node_params_t): Move into garbage collector. New methods insert
33422 (ipa_node_params_sum): Annotate wth GTY(()).
33423 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33425 (ipa_load_from_parm_agg): Adjust declaration.
33426 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33427 * ipa-profile.c (ipa_profile): Likewise.
33428 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33429 (ipa_populate_param_decls): Make descriptors parameter garbage
33431 (ipa_dump_param): Adjust to descriptors being a pointer.
33432 (ipa_alloc_node_params): Likewise.
33433 (ipa_initialize_node_params): Likewise.
33434 (load_from_param_1): Make descriptors parameter garbage collected.
33435 (load_from_unmodified_param): Likewise.
33436 (load_from_param): Likewise.
33437 (ipa_load_from_parm_agg): Likewise.
33438 (ipa_node_params::~ipa_node_params): Removed.
33439 (ipa_free_all_node_params): Remove call to delete operator.
33440 (ipa_node_params_t::insert): New.
33441 (ipa_node_params_t::remove): Likewise.
33442 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33443 copy known_csts and known_contexts vectors.
33444 (ipa_read_node_info): Adjust to descriptors being a pointer.
33445 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33447 (ipcp_transform_function): Make descriptors variable garbage
33450 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33452 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33453 * config/i386/avx512dqintrin.h: Ditto.
33454 * config/i386/avx512fintrin.h: Ditto.
33455 * config/i386/i386.c: Handle new builtins.
33456 * config/i386/i386-builtin.def: Add new builtins.
33457 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33458 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33460 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33461 Martin Liska <mliska@suse.cz>
33463 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33464 * asan.c (asan_expand_poison_ifn): Support stores and use
33465 appropriate ASAN report function.
33466 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33467 * internal-fn.def (ASAN_POISON_USE): Declare.
33468 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33469 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33470 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33471 ASAN_POISON calls w/o LHS.
33472 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33473 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33474 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33475 * gimplify.c (asan_poison_variables): Add attribute
33476 use_after_scope_memory to variables that really needs to live
33478 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33479 having the attribute.
33481 2017-01-23 Martin Liska <mliska@suse.cz>
33483 * asan.c (create_asan_shadow_var): New function.
33484 (asan_expand_poison_ifn): Likewise.
33485 * asan.h (asan_expand_poison_ifn): New declaration.
33486 * internal-fn.c (expand_ASAN_POISON): Likewise.
33487 * internal-fn.def (ASAN_POISON): New builtin.
33488 * sanopt.c (pass_sanopt::execute): Expand
33489 asan_expand_poison_ifn.
33490 * tree-inline.c (copy_decl_for_dup_finish): Make function
33492 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33493 * tree-ssa.c (is_asan_mark_p): New function.
33494 (execute_update_addresses_taken): Rewrite local variables
33495 (identified just by use-after-scope as addressable) into SSA.
33497 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33499 * doc/install.texi (Specific): opensource.apple.com uses https
33500 now. Remove trailing slash.
33502 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33504 * README.Portability: Remove note on an Irix compatibility issue.
33506 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33508 * gcov.c (INCLUDE_ALGORITHM): Define.
33509 (INCLUDE_VECTOR): Define.
33510 No longer include <vector> and <algorithm> directly.
33512 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33514 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33516 * doc/invoke.texi (Code Gen Options): Ditto.
33518 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33521 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33523 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33525 rtl-optimization/79125
33526 * cprop.c (local_cprop_pass): Handle cases where we make an
33527 unconditional trap.
33529 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33533 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33534 read from, for big endian.
33536 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33538 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33539 register pauth builtins for LP64 only.
33541 2017-01-20 Marek Polacek <polacek@redhat.com>
33544 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33547 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33549 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33551 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33552 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33553 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33555 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33558 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33559 in big-endian mode when they are not a single duplicated value.
33561 2017-01-20 Richard Biener <rguenther@suse.de>
33563 * BASE-VER: Bump to 7.0.1.
33565 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33567 * omp-low.c (omplow_simd_context): New struct. Use it...
33568 (lower_rec_simd_input_clauses): ...here and...
33569 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33570 references to idx, lane, max_vf, is_simt.
33572 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33574 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33577 2017-01-20 Martin Jambor <mjambor@suse.cz>
33579 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33580 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33582 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33583 (GTFILES): Rename hsa.c to hsa-common.c.
33584 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33585 * hsa-dump.c: Likewise.
33586 * hsa-gen.c: Likewise.
33587 * hsa-regalloc.c: Likewise.
33588 * ipa-hsa.c: Likewise.
33589 * omp-expand.c: Likewise.
33590 * omp-low.c: Likewise.
33591 * toplev.c: Likewise.
33593 2017-01-20 Marek Polacek <polacek@redhat.com>
33596 * doc/invoke.texi: Document -Wduplicated-branches.
33597 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33598 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33599 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33600 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33601 return 0 only when not OEP_LEXICOGRAPHIC.
33602 (fold_build_cleanup_point_expr): Use the expression
33603 location when building CLEANUP_POINT_EXPR.
33604 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33605 * tree.c (add_expr): Handle error_mark_node.
33607 2017-01-20 Martin Liska <mliska@suse.cz>
33610 * tree-profile.c (init_ic_make_global_vars): Do not call
33612 (gimple_init_gcov_profiler): Likewise.
33614 2017-01-20 Martin Liska <mliska@suse.cz>
33617 * cgraph.h (maybe_create_reference): Remove argument and
33619 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33621 * ipa-cp.c (create_specialized_node): Likewise.
33622 * symtab.c (symtab_node::maybe_create_reference): Handle
33623 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33625 2017-01-20 Martin Liska <mliska@suse.cz>
33627 * read-rtl-function.c (function_reader::create_function): Use
33628 build_decl instread of build_decl_stat.
33630 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33632 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33633 * config/i386/avx512dqintrin.h: Ditto.
33634 * config/i386/avx512fintrin.h: Ditto.
33635 * config/i386/i386-builtin-types.def: Add new types.
33636 * config/i386/i386.c: Handle new types.
33637 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33638 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33639 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33640 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33641 (__builtin_ia32_kshiftridi): New.
33642 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33644 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33648 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33649 define to rs6000_init_stack_protect_guard.
33650 (rs6000_init_stack_protect_guard): New function.
33652 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33653 Yunqiang Su <yunqiang.su@imgtec.com>
33655 * config.gcc (supported_defaults): Add madd4.
33656 (with_madd4): Add validation.
33657 (all_defaults): Add madd4.
33658 * config/mips/mips.opt (mmadd4): New option.
33659 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33661 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33663 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33664 (ISA_HAS_FUSED_MADD4): Likewise.
33665 * doc/invoke.texi (-mmadd4): Document the new option.
33666 * doc/install.texi (--with-madd4): Document the new option.
33668 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33670 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33671 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33672 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33673 (aarch64_init_pauth_hint_builtins): New.
33674 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33675 (aarch64_expand_builtin): Expand new builtins.
33677 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33679 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33680 * combine-stack-adj.c (no_unhandled_cfa): Handle
33681 REG_CFA_TOGGLE_RA_MANGLE.
33682 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33683 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33684 info for return address signing.
33685 (aarch64_expand_epilogue): Likewise.
33687 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33689 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33690 * config/aarch64/aarch64-protos.h
33691 (aarch64_return_address_signing_enabled): New declaration.
33692 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33694 (aarch64_expand_prologue): Sign return address before it's pushed onto
33696 (aarch64_expand_epilogue): Authenticate return address fetched from
33698 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33699 (aarch64_attributes): New function attributes for "sign-return-address".
33700 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33701 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33702 ("*do_return"): Generate combined instructions according to key index.
33703 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33704 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33706 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33707 * config/aarch64/aarch64.opt (msign-return-address=): New.
33708 * doc/extend.texi (AArch64 Function Attributes): Documents
33709 "sign-return-address=".
33710 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33712 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33714 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33715 overall option summary.
33717 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33719 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33720 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33721 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33722 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33724 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33726 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33727 -mpower9-minmax by default for -mcpu=power9.
33728 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33729 128-bit floating point.
33731 2017-01-20 Alan Modra <amodra@gmail.com>
33733 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33734 optimizing for size.
33736 2017-01-20 Alan Modra <amodra@gmail.com>
33739 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33740 for strcmp and strncmp from corresponding builtin decl.
33742 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33744 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33745 instead of i386/rtems-64.h.
33746 * config/i386/rtems-64.h: Remove.
33748 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33752 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33754 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33756 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33758 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33759 Change int to HOST_WIDE_INT.
33760 * config/aarch64/aarch64-protos.h
33761 (aarch64_simd_gen_const_vector_dup): Likewise.
33762 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33764 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33766 * langhooks-def.h (lhd_type_for_size): New decl.
33767 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33768 * langhooks.c (lhd_type_for_size): New function, taken from
33771 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33773 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33774 define_bypass for CR latency.
33775 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33776 (power9-alu2): Add define_bypass for CR latency.
33778 (power9-mul): Update insn latency.
33779 (power9-mul-compare): Update insn latency, bypass latency and remove
33782 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33784 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33786 * config/aarch64/aarch64.md
33787 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33788 aarch64_nopcrelative_literal_loads.
33789 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33791 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33793 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33794 TARGET_LOONGSON_3A.
33795 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33797 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33800 * config.gcc (supported_defaults): Add lxc1-sxc1.
33801 (with_lxc1_sxc1): Add validation.
33802 (all_defaults): Add lxc1-sxc1.
33803 * config/mips/mips.opt (mlxc1-sxc1): New option.
33804 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33806 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33807 __mips_no_lxc1_sxc1.
33808 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33809 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33810 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33812 2017-01-19 Richard Biener <rguenther@suse.de>
33814 PR tree-optimization/72488
33815 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33816 sure to restore SSA info.
33817 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33819 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33821 PR rtl-optimization/79121
33822 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33823 of the inner type when shifting an extended value.
33825 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33828 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33829 interposable aliases.
33831 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33834 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33835 Use the evmergelohi instruction.
33836 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33837 (mov_sitf_e500_subreg8_2_be): Likewise.
33838 (mov_sitf_e500_subreg12_2_le): Likewise.
33839 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33840 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33841 (mov_sitf_e500_subreg8_2_le): Likewise.
33842 (mov_sitf_e500_subreg12_2_be): Likewise.
33844 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33846 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33847 attribute from vecsimple to vecperm.
33848 (altivec_vbpermq2): Likewise.
33850 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33853 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33855 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33856 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33857 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33858 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33859 case where N arg is SIZE_MAX.
33860 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33861 (cmpstrsi): Add pattern.
33863 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33865 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33866 __builtin_vec_revb builtins.
33867 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33868 built-in functions to support generation of the ISA 3.0 XXBR<x>
33869 vector byte reverse instructions.
33870 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33871 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33872 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33873 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33874 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33875 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33876 (P9V_BUILTIN_VEC_REVB): Likewise.
33877 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33878 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33879 (p9_xxbrq_v16qi): Likewise.
33880 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33881 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33882 (p9_xxbrh_v8hi): Likewise.
33883 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33884 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33885 vec_revb built-in functions.
33887 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33889 PR rtl-optimization/78952
33890 * config/i386/i386.md (any_extract): New code iterator.
33891 (*insvqi_2): Use any_extract for source operand.
33892 (*insvqi_3): Use any_shiftrt for source operand.
33894 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33896 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33898 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33900 2017-01-18 Matthias Klose <doko@ubuntu.com>
33902 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33904 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33906 * config/rs6000/altivec.h (vec_bperm): Change #define.
33907 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33908 (altivec_vbpermq2): New define_insn.
33909 (altivec_vbpermd): Likewise.
33910 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33911 function interface.
33912 (VBPERMD): Likewise.
33913 (VBPERM): New polymorphic function interface.
33914 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33915 Add entries for P9V_BUILTIN_VEC_VBPERM.
33916 * doc/extend.texi: Add interfaces for vec_bperm.
33918 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33920 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33921 first letter of error messages.
33922 (s390_resolve_overloaded_builtin): Likewise.
33923 * config/s390/s390.c (s390_expand_builtin): Likewise.
33924 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33925 (s390_valid_target_attribute_inner_p): Likewise.
33926 * config/s390/s390.md ("tabort"): Likewise.
33928 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33930 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33931 (ISA_AVOID_DIV_HILO): New macro.
33932 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33933 (ISA_HAS_DDIV): Likewise.
33935 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33937 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33939 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33941 * doc/invoke.texi (fabi-version): Spelling fix.
33943 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33946 * doc/invoke.texi (fabi-version): Mention mangling fix for
33949 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33952 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33954 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33957 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33958 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33960 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33962 (stack_protect_test): Ditto.
33963 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33964 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33966 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33967 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33968 -mstack-protector-guard-offset=.
33969 (RS/6000 and PowerPC Options): Ditto.
33971 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33973 * config/i386/i386.h (MASK_CLASS_P): New define.
33974 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33975 there are no registers from different register sets also when
33976 mask registers are used. Update function comment.
33977 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33978 to (*k/*r) and (*k/*km) alternatives.
33980 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33982 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33983 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33984 (EH_RETURN_HANDLER_RTX): New define.
33985 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33986 Force frame pointer in EH return functions.
33987 (aarch64_expand_epilogue): Add barrier for eh_return.
33988 (aarch64_final_eh_return_addr): Remove.
33989 (aarch64_eh_return_handler_rtx): New function.
33990 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33992 (aarch64_eh_return_handler_rtx): New prototype.
33994 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33996 * config/rs6000/altivec.h (vec_rlmi): New #define.
33997 (vec_vrlnm): Likewise.
33998 (vec_rlnm): Likewise.
33999 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34000 (UNSPEC_VRLNM): Likewise.
34001 (VIlong): New mode iterator.
34002 (altivec_vrl<VI_char>mi): New define_insn.
34003 (altivec_vrl<VI_char>nm): Likewise.
34004 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34006 (VRLDNM): Likewise.
34007 (RLNM): New polymorphic function entry.
34008 (VRLWMI): New monomorphic function entry.
34009 (VRLDMI): Likewise.
34010 (RLMI): New polymorphic function entry.
34011 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34012 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34013 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34016 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34019 * dwarf2out.c (field_byte_offset): Restore the
34020 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34021 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34022 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34025 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34028 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34030 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34033 * dwarf2out.c (add_data_member_location_attribute): For constant
34034 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34035 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34036 DW_AT_byte_size attributes.
34038 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34040 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34041 after forcing to constant memory when the code model is medium.
34043 2017-01-17 Julia Koval <julia.koval@intel.com>
34046 * config/i386/avx512fintrin.h
34047 (_mm512_i32gather_ps): Change __addr type to void const*.
34048 (_mm512_mask_i32gather_ps): Ditto.
34049 (_mm512_i32gather_pd): Ditto.
34050 (_mm512_mask_i32gather_pd): Ditto.
34051 (_mm512_i64gather_ps): Ditto.
34052 (_mm512_mask_i64gather_ps): Ditto.
34053 (_mm512_i64gather_pd): Ditto.
34054 (_mm512_mask_i64gather_pd): Ditto.
34055 (_mm512_i32gather_epi32): Ditto.
34056 (_mm512_mask_i32gather_epi32): Ditto.
34057 (_mm512_i32gather_epi64): Ditto.
34058 (_mm512_mask_i32gather_epi64): Ditto.
34059 (_mm512_i64gather_epi32): Ditto.
34060 (_mm512_mask_i64gather_epi32): Ditto.
34061 (_mm512_i64gather_epi64): Ditto.
34062 (_mm512_mask_i64gather_epi64): Ditto.
34063 (_mm512_i32scatter_ps): Change __addr type to void*.
34064 (_mm512_mask_i32scatter_ps): Ditto.
34065 (_mm512_i32scatter_pd): Ditto.
34066 (_mm512_mask_i32scatter_pd): Ditto.
34067 (_mm512_i64scatter_ps): Ditto.
34068 (_mm512_mask_i64scatter_ps): Ditto.
34069 (_mm512_i64scatter_pd): Ditto.
34070 (_mm512_mask_i64scatter_pd): Ditto.
34071 (_mm512_i32scatter_epi32): Ditto.
34072 (_mm512_mask_i32scatter_epi32): Ditto.
34073 (_mm512_i32scatter_epi64): Ditto.
34074 (_mm512_mask_i32scatter_epi64): Ditto.
34075 (_mm512_i64scatter_epi32): Ditto.
34076 (_mm512_mask_i64scatter_epi32): Ditto.
34077 (_mm512_i64scatter_epi64): Ditto.
34078 (_mm512_mask_i64scatter_epi64): Ditto.
34079 * config/i386/avx512pfintrin.h
34080 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34081 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34082 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34083 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34084 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34085 (_mm512_prefetch_i32scatter_ps): Ditto.
34086 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34087 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34088 (_mm512_prefetch_i64scatter_pd): Ditto.
34089 (_mm512_prefetch_i64scatter_ps): Ditto.
34090 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34091 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34092 * config/i386/avx512vlintrin.h
34093 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34094 (_mm_mmask_i32gather_ps): Ditto.
34095 (_mm256_mmask_i32gather_pd): Ditto.
34096 (_mm_mmask_i32gather_pd): Ditto.
34097 (_mm256_mmask_i64gather_ps): Ditto.
34098 (_mm_mmask_i64gather_ps): Ditto.
34099 (_mm256_mmask_i64gather_pd): Ditto.
34100 (_mm_mmask_i64gather_pd): Ditto.
34101 (_mm256_mmask_i32gather_epi32): Ditto.
34102 (_mm_mmask_i32gather_epi32): Ditto.
34103 (_mm256_mmask_i32gather_epi64): Ditto.
34104 (_mm_mmask_i32gather_epi64): Ditto.
34105 (_mm256_mmask_i64gather_epi32): Ditto.
34106 (_mm_mmask_i64gather_epi32): Ditto.
34107 (_mm256_mmask_i64gather_epi64): Ditto.
34108 (_mm_mmask_i64gather_epi64): Ditto.
34109 (_mm256_i32scatter_ps): Change __addr type to void*.
34110 (_mm256_mask_i32scatter_ps): Ditto.
34111 (_mm_i32scatter_ps): Ditto.
34112 (_mm_mask_i32scatter_ps): Ditto.
34113 (_mm256_i32scatter_pd): Ditto.
34114 (_mm256_mask_i32scatter_pd): Ditto.
34115 (_mm_i32scatter_pd): Ditto.
34116 (_mm_mask_i32scatter_pd): Ditto.
34117 (_mm256_i64scatter_ps): Ditto.
34118 (_mm256_mask_i64scatter_ps): Ditto.
34119 (_mm_i64scatter_ps): Ditto.
34120 (_mm_mask_i64scatter_ps): Ditto.
34121 (_mm256_i64scatter_pd): Ditto.
34122 (_mm256_mask_i64scatter_pd): Ditto.
34123 (_mm_i64scatter_pd): Ditto.
34124 (_mm_mask_i64scatter_pd): Ditto.
34125 (_mm256_i32scatter_epi32): Ditto.
34126 (_mm256_mask_i32scatter_epi32): Ditto.
34127 (_mm_i32scatter_epi32): Ditto.
34128 (_mm_mask_i32scatter_epi32): Ditto.
34129 (_mm256_i32scatter_epi64): Ditto.
34130 (_mm256_mask_i32scatter_epi64): Ditto.
34131 (_mm_i32scatter_epi64): Ditto.
34132 (_mm_mask_i32scatter_epi64): Ditto.
34133 (_mm256_i64scatter_epi32): Ditto.
34134 (_mm256_mask_i64scatter_epi32): Ditto.
34135 (_mm_i64scatter_epi32): Ditto.
34136 (_mm_mask_i64scatter_epi32): Ditto.
34137 (_mm256_i64scatter_epi64): Ditto.
34138 (_mm256_mask_i64scatter_epi64): Ditto.
34139 (_mm_i64scatter_epi64): Ditto.
34140 (_mm_mask_i64scatter_epi64): Ditto.
34141 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34142 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34143 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34144 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34145 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34146 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34147 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34148 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34149 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34150 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34151 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34152 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34153 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34154 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34155 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34156 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34157 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34158 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34159 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34160 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34161 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34162 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34163 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34164 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34165 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34166 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34167 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34168 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34169 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34170 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34171 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34172 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34173 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34174 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34175 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34176 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34177 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34178 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34179 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34180 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34181 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34182 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34183 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34184 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34185 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34186 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34187 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34188 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34189 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34190 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34191 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34192 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34193 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34194 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34195 definitions accordingly.
34197 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34198 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34201 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34204 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34207 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34208 hard regs for inner regno.
34210 2017-01-17 Martin Liska <mliska@suse.cz>
34213 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34214 assumption and add comment.
34216 2017-01-17 Nathan Sidwell <nathan@acm.org>
34218 * ipa-visibility.c (localize_node): New function, broken out of ...
34219 (function_and_variable_visibility): ... here. Call it.
34221 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34223 PR middle-end/77445
34224 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34225 correctly set frequency of oudgoing edge.
34226 (duplicate_thread_path): Fix profile updating.
34228 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34231 * configure.ac: Add GCC_BASE_VER.
34232 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34233 version from BASE-VER file.
34234 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34235 (gcc.o): Depend on $(BASEVER).
34236 * common.opt (dumpfullversion): New option.
34237 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34238 * doc/invoke.texi: Document -dumpfullversion.
34239 * doc/install.texi: Document --with-gcc-major-version-only.
34240 * configure: Regenerated.
34242 2017-01-17 Richard Biener <rguenther@suse.de>
34244 PR tree-optimization/71433
34245 * tree-vrp.c (register_new_assert_for): Merge same asserts
34246 on all incoming edges.
34247 (process_assert_insertions_for): Handle insertions at the
34250 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34252 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34253 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34255 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34258 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34261 2017-01-17 Alan Modra <amodra@gmail.com>
34264 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34265 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34266 symbolic stack limit when pic.
34268 2017-01-16 Martin Sebor <msebor@redhat.com>
34270 PR tree-optimization/78608
34271 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34273 2017-01-16 Jeff Law <law@redhat.com>
34276 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34277 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34278 for several include directories that may be relative to sysroot.
34279 * config/i386/x-mingw32 (gplus_includedir): Define.
34280 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34281 (native_system_includedir): Likewise.
34282 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34283 override if TARGET_SYSTEM_ROOT is defined.
34284 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34286 PR tree-optimization/79090
34287 PR tree-optimization/33562
34288 PR tree-optimization/61912
34289 PR tree-optimization/77485
34290 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34291 and computed trims into the dump file.
34293 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34295 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34297 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34300 * gimplify.c (gimplify_init_constructor): If want_value and
34301 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34305 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34306 sequence. Formatting fixes.
34307 (doloop_optimize): Formatting fixes.
34310 * gcc.c (debug_level_greater_than_spec_func): New function.
34311 (static_spec_functions): Add debug-level-gt spec function.
34312 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34314 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34315 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34316 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34317 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34318 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34319 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34321 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34323 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34324 QImode fixups to general and mask registers only.
34326 2017-01-16 Carl Love <cel@us.ibm.com>
34328 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34329 for built-in functions
34330 vector signed char vec_nabs (vector signed char)
34331 vector signed short vec_nabs (vector signed short)
34332 vector signed int vec_nabs (vector signed int)
34333 vector signed long long vec_nabs (vector signed long long)
34334 vector float vec_nabs (vector float)
34335 vector double vec_nabs (vector double)
34336 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34338 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34339 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34340 * doc/extend.texi: Update the documentation file for the new built-in
34343 2017-01-16 Martin Sebor <msebor@redhat.com>
34345 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34348 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34350 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34351 UNSPEC_VSX__XXSPLTD to require special splat handling.
34353 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34356 * system.h: Poison strndup.
34358 2017-01-16 Alan Modra <amodra@gmail.com>
34361 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34364 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34366 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34368 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34370 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34371 call recog here. Assert that INSN_CODE (insn) is non-negative.
34373 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34376 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34378 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34379 in the currently scheduled RTL fragment.
34381 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34383 PR rtl-optimization/78751
34384 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34387 2017-01-14 Jeff Law <law@redhat.com>
34389 PR tree-optimization/79090
34390 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34391 variable length stores.
34392 (compute_trims): Delete dead assignment to *trim_tail.
34393 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34396 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34398 PR rtl-optimization/78626
34399 PR rtl-optimization/78727
34400 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34401 of a block, and split such blocks after everything else is finished.
34403 2017-01-14 Alan Modra <amodra@gmail.com>
34406 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34407 target legitimate_combined_insn.
34408 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34409 (rs6000_legitimate_combined_insn): New function.
34410 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34412 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34413 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34414 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34416 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34418 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34420 2017-01-13 Jeff Law <law@redhat.com>
34422 PR tree-optimization/33562
34423 PR tree-optimization/61912
34424 PR tree-optimization/77485
34425 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34427 (delete_dead_assignment): Likewise.
34428 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34429 statement to delete_dead_call and delete_dead_assignment.
34431 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34434 * substring-locations.c (format_warning_va): Strengthen case 1 so
34435 that both endpoints of the substring must be within the format
34436 range for just the substring to be printed.
34438 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34440 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34441 * config/i386/i386.c (ix86_target_string): Add missing options
34442 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34443 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34444 flags_other and ix86_target_other to flags2_other. Display unknown
34446 (ix86_valid_target_attribute_inner_p): Add missing options and
34447 reorder options by implied ISAs, as in ix86_target_string.
34449 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34451 * hash-table.h (hash_table::too_empty_p): New function.
34452 (hash_table::expand): Use it.
34453 (hash_table::traverse): Likewise.
34454 (hash_table::empty_slot): Use sizeof (value_type) instead of
34455 sizeof (PTR) to convert bytes to elements. Shrink the table
34456 if the current size is excessive for the current number of
34459 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34461 * ira-costs.c (record_reg_classes): Break from the inner loop
34462 early once alt_fail is known to be true. Update outer loop
34463 handling accordingly.
34465 2017-01-13 Jeff Law <law@redhat.com>
34467 * tree-ssa-dse.c (decrement_count): New function.
34468 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34469 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34470 when we know the partially dead statement is a mem* function.
34472 PR tree-optimization/61912
34473 PR tree-optimization/77485
34474 * tree-ssa-dse.c: Include expr.h.
34475 (maybe_trim_constructor_store): New function.
34476 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34478 PR tree-optimization/33562
34479 PR tree-optimization/61912
34480 PR tree-optimization/77485
34481 * doc/invoke.texi: Document new dse-max-object-size param.
34482 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34483 * tree-ssa-dse.c: Include params.h.
34484 (dse_store_status): New enum.
34485 (initialize_ao_ref_for_dse): New, partially extracted from
34487 (valid_ao_ref_for_dse, normalize_ref): New.
34488 (setup_live_bytes_from_ref, compute_trims): Likewise.
34489 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34490 (maybe_trim_partially_dead_store): Likewise.
34491 (maybe_trim_complex_store): Likewise.
34492 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34493 Track what bytes live from the original store. Return tri-state
34494 for dead, partially dead or live.
34495 (dse_dom_walker): Add constructor, destructor and new private members.
34496 (delete_dead_call, delete_dead_assignment): New extracted from
34498 (dse_optimize_stmt): Make a member of dse_dom_walker.
34499 Use initialize_ao_ref_for_dse.
34501 PR tree-optimization/33562
34502 PR tree-optimization/61912
34503 PR tree-optimization/77485
34504 * sbitmap.h (bitmap_count_bits): Prototype.
34505 (bitmap_clear_range, bitmap_set_range): Likewise.
34506 * sbitmap.c (bitmap_clear_range): New function.
34507 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34509 2017-01-13 Martin Liska <mliska@suse.cz>
34512 * function.c (set_cfun): Add new argument force.
34513 * function.h (set_cfun): Likewise.
34514 * ipa-inline-transform.c (inline_call): Use the function when
34515 strict alising from is dropped for function we inline to.
34517 2017-01-13 Richard Biener <rguenther@suse.de>
34519 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34520 for dumping GIMPLE INTEGER_CSTs.
34522 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34524 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34525 to 201112L since C++17.
34527 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34530 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34531 if -fsanitize=kernel-address is present.
34533 2017-01-13 Richard Biener <rguenther@suse.de>
34535 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34536 as _Literal ( type ) number in case usual suffixes do not
34537 preserve all information.
34539 2017-01-13 Richard Biener <rguenther@suse.de>
34541 PR tree-optimization/77283
34542 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34543 and ssa-iterators.h.
34544 (is_feasible_trace): Implement a cost model based on joiner
34547 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34550 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34551 char or short to __float128/_Float128 directly.
34553 2017-01-12 Martin Sebor <msebor@redhat.com>
34555 to -Wformat-overflow.
34556 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34557 (min_bytes_remaining): Same.
34558 (get_string_length): Same.
34559 (format_string): Same.
34560 (format_directive): Same.
34562 (pass_sprintf_length::handle_gimple_call): Same.
34564 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34566 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34567 info.nowrite calls with no lhs that can't throw. Return bool
34568 whether gsi_remove has been called or not.
34569 (pass_sprintf_length::handle_gimple_call): Return bool whether
34570 try_substitute_return_value called gsi_remove. Formatting fix.
34571 (pass_sprintf_length::execute): Don't use gsi_remove if
34572 handle_gimple_call returned true.
34575 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34576 be removed due to side-effects, don't remove following barrier nor
34577 turn the successor edge into fallthru edge.
34579 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34582 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34583 element-reversing loads and stores as not swappable.
34585 2017-01-12 Nathan Sidwell <nathan@acm.org>
34586 Nicolai Stange <nicstange@gmail.com>
34588 * combine.c (try_combine): Don't ignore result of overlap checking
34589 loop. Combine overlap & asm check into single loop.
34591 2017-01-12 Richard Biener <rguenther@suse.de>
34593 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34594 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34596 2017-01-12 Richard Biener <rguenther@suse.de>
34598 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34599 and TS_TARGET_OPTION directly derive from TS_BASE.
34600 * tree-core.h (tree_optimization_option): Derive from tree_base.
34601 (tree_target_option): Likewise.
34603 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34605 * config/i386/i386.c (memory_address_length): Increase len
34606 only when rip_relative_addr_p returns false.
34608 2017-01-11 Julia Koval <julia.koval@intel.com>
34610 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34611 (OPTION_MASK_ISA_SGX_SET): New.
34612 (ix86_handle_option): Handle OPT_msgx.
34613 * config.gcc: Added sgxintrin.h.
34614 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34615 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34616 * config/i386/i386.c (ix86_target_string): Add -msgx.
34618 (ix86_option_override_internal): Handle new options.
34619 (ix86_valid_target_attribute_inner_p): Add sgx.
34620 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34621 * config/i386/i386.opt: Add msgx.
34622 * config/i386/sgxintrin.h: New file.
34623 * config/i386/x86intrin.h: Add sgxintrin.h.
34625 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34628 * fold-const.c (maybe_nonzero_address): Return 1 for function
34630 (tree_single_nonzero_warnv_p): Don't handle function local objects
34634 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34637 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34640 * opts.c: Include "spellcheck.h"
34641 (struct string_fragment): New struct.
34642 (struct edit_distance_traits<const string_fragment &>): New
34644 (get_closest_sanitizer_option): New function.
34645 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34647 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34649 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34651 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34652 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34653 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34654 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34655 for initial die_offset if dwarf_split_debug_info.
34656 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34657 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34659 (output_skeleton_debug_sections): Formatting fix. Use
34660 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34661 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34663 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34665 * config/arm/cortex-a53.md: Add bypasses for
34666 cortex_a53_r2f_cvt.
34667 (cortex_a53_r2f): Only use for transfers.
34668 (cortex_a53_f2r): Likewise.
34669 (cortex_a53_r2f_cvt): Add reservation for conversions.
34670 (cortex_a53_f2r_cvt): Likewise.
34672 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34674 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34675 to all inlined functions, change static to extern.
34677 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34680 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34682 (arm_assemble_integer): Likewise.
34684 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34686 * config.gcc: Use new awk script to check CPU, FPU and architecture
34687 parameters for --with-... options.
34688 * config/arm/parsecpu.awk: New file
34689 * config/arm/arm-cpus.in: New file.
34690 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34692 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34694 * config/arm/t-arm: Update dependency rules.
34695 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34696 of processing .def files.
34697 * config/arm/genopt.sh: Deleted.
34698 * config/arm/gentune.sh: Deleted.
34699 * config/arm/arm-cores.def: Deleted.
34700 * config/arm/arm-arches.def: Deleted.
34701 * config/arm/arm-fpus.def: Deleted.
34702 * config/arm/arm-tune.md: Regenerated.
34703 * config/arm/arm-tables.opt: Regenerated.
34704 * config/arm/arm-cpu.h: New generated file.
34705 * config/arm/arm-cpu-data.h: New generated file.
34706 * config/arm/arm-cpu-cdata.h: New generated file.
34708 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34711 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34713 (input_varpool_node): Unpack dynamically_initialized bit.
34715 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34717 PR rtl-optimization/79032
34718 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34719 the alignment of the adjusted memory reference against that of MODE,
34720 instead of the alignment of the original memory reference.
34722 2017-01-11 Martin Jambor <mjambor@suse.cz>
34724 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34726 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34727 decorated functions.
34729 2017-01-11 Richard Biener <rguenther@suse.de>
34731 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34732 set range/nonnull info for PHI results. Do not set it on
34733 stmts marked for removal.
34735 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34737 * expr.c (store_field): In the bitfield case, fetch the return value
34738 from the registers before applying a single big-endian adjustment.
34739 Always do a final load for a BLKmode value not larger than a word.
34741 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34744 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34745 that we correctly handle column numbers greater than
34746 LINE_MAP_MAX_COLUMN_NUMBER.
34748 2017-01-10 Martin Sebor <msebor@redhat.com>
34750 PR middle-end/78245
34751 * gimple-ssa-sprintf.c (get_destination_size): Call
34752 {init,fini}object_sizes.
34753 * tree-object-size.c (addr_object_size): Adjust.
34754 (pass_through_call): Adjust.
34755 (pass_object_sizes::execute): Adjust.
34756 * tree-object-size.h (fini_object_sizes): Declare.
34758 2017-01-10 Martin Sebor <msebor@redhat.com>
34760 PR tree-optimization/78775
34761 * builtins.c (get_size_range): Move...
34762 * calls.c: ...to here.
34763 (alloc_max_size): Accept zero argument.
34764 (operand_signed_p): Remove.
34765 (maybe_warn_alloc_args_overflow): Call get_size_range.
34766 * calls.h (get_size_range): Declare.
34768 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34770 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34771 from TI's devices.csv file as of September 2016.
34772 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34774 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34776 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34777 * doc/invoke.texi: Likewise.
34778 * doc/md.texi: Likewise.
34779 * doc/objc.texi: Likewise.
34781 2017-01-10 Joshua Conner <joshconner@google.com>
34783 * config/arm/fuchsia-elf.h: New file.
34784 * config/fuchsia.h: New file.
34785 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34786 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34788 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34790 2016-01-10 Richard Biener <rguenther@suse.de>
34792 PR tree-optimization/79034
34793 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34794 Propagate out degenerate PHIs in the joiner.
34796 2017-01-10 Martin Liska <mliska@suse.cz>
34798 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34799 (sort_congruence_classes_by_decl_uid): Likewise.
34800 (sort_congruence_class_groups_by_decl_uid): Likewise.
34801 (sem_item_optimizer::merge_classes): Sort class, groups in these
34802 classes and members in the groups by DECL_UID of declarations.
34803 This would make merge operations stable.
34805 2017-01-10 Martin Liska <mliska@suse.cz>
34807 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34808 usage of m_classes_vec.
34809 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34810 (sem_item_optimizer::get_group_by_hash): Likewise.
34811 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34812 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34813 (sem_item_optimizer::verify_classes): Likewise.
34814 (sem_item_optimizer::process_cong_reduction): Likewise.
34815 (sem_item_optimizer::dump_cong_classes): Likewise.
34816 (sem_item_optimizer::merge_classes): Likewise.
34817 * ipa-icf.h (congruence_class_hash): Rename from
34818 congruence_class_group_hash. Remove declaration of m_classes_vec.
34820 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34822 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34823 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34824 * config.gcc: Add avx512vpopcntdqintrin.h.
34825 * config/i386/avx512vpopcntdqintrin.h: New.
34826 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34827 * config/i386/i386-builtin-types.def: Add new types.
34828 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34829 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34830 __builtin_ia32_vpopcountq_v8di_mask): New.
34831 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34832 __AVX512VPOPCNTDQ__.
34833 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34834 (PTA_AVX512VPOPCNTDQ): Define.
34835 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34836 TARGET_AVX512VPOPCNTDQ_P): Define.
34837 * config/i386/i386.opt: Add mavx512vpopcntdq.
34838 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34839 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34841 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34843 PR middle-end/77484
34844 * predict.def (PRED_CALL): Set to 67.
34846 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34848 * expr.c (store_field): In the bitfield case, if the value comes from
34849 a function call and is of an aggregate type returned in registers, do
34850 not modify the field mode; extract the value in all cases if the mode
34851 is BLKmode and the size is not larger than a word.
34853 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34856 * config/i386/cpuid.h: Fix undefined behavior.
34858 2017-01-04 Jeff Law <law@redhat.com>
34860 PR tree-optimization/79007
34861 PR tree-optimization/67955
34862 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34863 conservative for pt.null when flag_non_call_exceptions is on.
34865 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34867 PR translation/79019
34868 PR translation/79020
34869 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34870 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34871 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34873 * config/avr/avr.opt (maccumulate-args): Likewise.
34874 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34875 * common.opt (freport-bug): Likewise.
34876 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34877 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34878 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34879 translatable string.
34880 * config/i386/i386.c (function_value_32): Likewise.
34881 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34882 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34884 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34885 * common/config/msp430/msp430-common.c (msp430_handle_option):
34887 * symtab.c (symtab_node::verify_base): Likewise.
34888 * opts.c (set_debug_level): Likewise.
34889 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34890 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34891 missing whitespace to translatable strings.
34892 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34893 * config/sh/superh.h: Likewise.
34894 * config/i386/xopintrin.h: Likewise.
34895 * config/i386/znver1.md: Likewise.
34896 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34897 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34898 * double-int.h (struct double_int): Likewise.
34899 * double-int.c (div_and_round_double): Likewise.
34900 * wide-int.cc: Likewise.
34901 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34902 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34903 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34904 (crossjumps_occurred): ... this.
34905 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34908 PR tree-optimization/78899
34909 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34910 returning bool return struct loop *, NULL for failure and the new
34912 (versionable_outer_loop_p): Don't version outer loop if it has
34913 dont_vectorized bit set.
34914 (tree_if_conversion): When versioning outer loop, ensure
34915 tree_if_conversion is performed also on the inner loop of the
34916 non-vectorizable outer loop copy.
34917 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34918 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34919 prevent vectorization of it.
34920 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34921 the outer loop vectorization of the non-scalar version is attempted
34922 before vectorization of the inner loop in scalar version. If
34923 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34924 vectorization of its inner loop.
34925 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34926 has 2 inner loops, rename also on edges from bb whose single pred
34927 is outer_loop->header. Fix typo in function comment.
34929 2017-01-09 Martin Sebor <msebor@redhat.com>
34932 * asan.c (asan_emit_stack_protection): Increase local buffer size
34933 to avoid snprintf truncation warning.
34935 2017-01-09 Andrew Pinski <apinski@cavium.com>
34937 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34938 to reference thunderx2t99 for the tuning structure
34939 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34941 (thunderx2t99_extra_costs): This.
34942 * config/aarch64/aarch64-tune.md: Regenerate.
34943 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34944 (vulcan_addrcost_table): This.
34945 (vulcan_regmove_cost): Rename to ...
34946 (thunderx2t99_regmove_cost): This.
34947 (vulcan_vector_cost): Rename to ...
34948 (thunderx2t99_vector_cost): this.
34949 (vulcan_branch_cost): Rename to ...
34950 (thunderx2t99_branch_cost): This.
34951 (vulcan_tunings): Rename to ...
34952 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34953 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34955 2017-01-09 Martin Jambor <mjambor@suse.cz>
34959 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34960 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34961 (propagate_vr_accross_jump_function): Use the above function for all
34962 value range computations for pass-through jump functions and type
34963 converasion from explicit value range values.
34964 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34965 parameters from TYPE_ARG_TYPES.
34966 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34967 (ipa_write_node_info): Stream type of the actual argument.
34968 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34970 2017-01-09 Martin Liska <mliska@suse.cz>
34973 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34974 (lookup_compiler): Do not show error message with have_E.
34976 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34978 PR tree-optimization/78938
34979 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34980 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34981 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34982 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34985 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34987 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34990 2017-01-09 Richard Biener <rguenther@suse.de>
34992 PR tree-optimization/78997
34993 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34994 name condition properly.
34996 2017-01-09 Richard Biener <rguenther@suse.de>
34999 * dwarf2out.c (is_cxx): New overload with context.
35000 (is_naming_typedef_decl): Use it.
35002 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35004 * invoke.texi (Option Summary): Correct spacing in option lists
35005 and add line breaks to fix over-long lines.
35007 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35009 PR middle-end/17660
35011 * extend.texi (Common Variable Attributes): Add xref to GCC
35012 Internals manual to explain mode attribute keywords.
35014 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35017 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35018 and Preprocessor Options.
35019 (Options for Linking): Document -pthread here....
35020 (RS/6000 and PowerPC Options): ...not here.
35021 (Solaris 2 Options): ...or here.
35022 * doc/cppopts.texi: Document -pthread.
35024 2017-01-08 Martin Sebor <msebor@redhat.com>
35026 PR middle-end/77708
35027 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35028 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35029 New member functions.
35030 (format_directive): Used them.
35032 (pass_sprintf_length::handle_gimple_call): Same.
35033 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35034 to avoid truncation for any argument.
35035 (extract_affine_mul): Same.
35036 * tree.c (get_file_function_name): Same.
35038 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35040 PR middle-end/77484
35041 * predict.def (PRED_INDIR_CALL): Set to 86.
35043 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35045 PR preprocessor/54124
35046 * doc/cppopts.texi: Reformat -d subtable to list the full name
35047 of the options. Add cross-reference to the docs for the general
35048 compiler -d options.
35049 * doc/invoke.texi (Developer Options): Add cross-reference to the
35050 preprocessor-specific -d option documentation.
35052 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35054 PR preprocessor/13498
35055 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35056 redudant material, and reflect new command-line options.
35057 (System Headers): Likewise.
35059 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35061 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35062 -isystem, and -idirafter. Copy-edit.
35063 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35064 default for -ftrack-macro-expansion. Delete obsolete and
35065 badly-formatted implementation details about -fdebug-cpp output.
35066 * doc/cppwarnopts.texi: Copy-edit.
35068 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35071 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35072 that the transition from a max line width >= 1<<10 to narrower
35073 lines works correctly.
35075 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35077 * doc/options.texi (PerFunction): New.
35078 * opt-functions.awk (switch_flags): Map both Optimization and
35079 PerFunction to CL_OPTIMIZATION.
35080 * opth-gen.awk: Test for PerFunction flag along with
35082 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35083 it only when the latter is present. Skip those that don't in
35084 the hash function generator.
35085 * common.opt (fvar-tracking): Mark as PerFunction instead of
35087 (fvar-tracking-assignments): Likewise.
35088 (fvar-tracking-assignments-toggle): Likewise.
35089 (fvar-tracking-uninit): Likewise.
35091 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35093 PR translation/79018
35094 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35097 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35100 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35101 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35102 TARGET_LONG_JUMP_TABLE_OFFSETS.
35103 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35104 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35105 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35106 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35107 * config/m68k/m68k.md (tablejump expander): Likewise.
35108 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35109 TARGET_LONG_JUMP_TABLE_OFFSETS.
35110 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35111 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35113 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35114 David Holsgrove <david.holsgrove@xilinx.com>
35116 * common/config/microblaze/microblaze-common.c
35117 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35118 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35120 * config/microblaze/microblaze.c (microblaze_must_save_register)
35121 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35123 (microblaze_eh_return): New function.
35124 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35125 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35126 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35127 * config/microblaze/microblaze.md (eh_return): New pattern.
35129 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35131 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35132 GCC_DIAGNOSTIC_STRINGIFY): Define.
35134 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35136 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35138 * config/arm/arm.md (<mcrr>): New.
35140 * config/arm/arm.c (arm_arch5te): New.
35141 (arm_option_override): Set arm_arch5te.
35142 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35144 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35145 (arm_mcrr_qualifiers): ... this. New.
35146 (MRRC_QUALIFIERS): Define to...
35147 (arm_mrrc_qualifiers): ... this. New.
35148 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35150 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35151 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35152 (MRRCI, mrrc, MRRC): New.
35153 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35154 VUNSPEC_MRRC2): New.
35156 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35158 * config/arm/arm.md (<mcr>): New.
35160 * config/arm/arm.c (arm_coproc_builtin_available): Add
35161 support for mcr, mrc, mcr2 and mrc2.
35162 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35163 (arm_mcr_qualifiers): ... this. New.
35164 (MRC_QUALIFIERS): Define to ...
35165 (arm_mrc_qualifiers): ... this. New.
35166 (MCR_QUALIFIERS): Define to ...
35167 (arm_mcr_qualifiers): ... this. New.
35168 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35170 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35171 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35172 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35173 VUNSPEC_MRC2): New.
35175 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35177 * config/arm/arm.md (*ldc): New.
35181 * config/arm/arm.c (arm_coproc_builtin_available): Add
35182 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35183 (arm_coproc_ldc_stc_legitimate_address): New.
35184 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35185 'qualifier_const_pointer'.
35186 (LDC_QUALIFIERS): Define to...
35187 (arm_ldc_qualifiers): ... this. New.
35188 (STC_QUALIFIERS): Define to...
35189 (arm_stc_qualifiers): ... this. New.
35190 * config/arm/arm-protos.h
35191 (arm_coproc_ldc_stc_legitimate_address): New.
35192 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35193 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35194 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35195 stc2, stcl, stc2l): New.
35196 * config/arm/constraints.md (Uz): New.
35197 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35198 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35199 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35200 VUNSPEC_STC2L): New.
35202 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35204 * config/arm/arm.md (<cdp>): New.
35205 * config/arm/arm.c (neon_const_bounds): Rename this ...
35206 (arm_const_bounds): ... this.
35207 (arm_coproc_builtin_available): New.
35208 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35209 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35210 (CDP_QUALIFIERS): Define to...
35211 (arm_cdp_qualifiers): ... this. New.
35213 (arm_expand_builtin_args): Add case for 6 arguments.
35214 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35215 (arm_const_bounds): ... this.
35216 (arm_coproc_builtin_available): New.
35217 * config/arm/arm_acle.h (__arm_cdp): New.
35219 * config/arm/arm_acle_builtins.def (cdp): New.
35221 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35222 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35223 'arm_const_bounds'.
35224 * config/arm/types.md (coproc): New.
35225 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35226 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35227 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35228 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35230 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35232 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35233 (UBINOP_QUALIFIERS): New.
35235 (acle_builtin_data): New. Change comment.
35236 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35237 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35238 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35239 arm_acle_builtins.def.
35240 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35241 (arm_init_acle_builtins): New.
35242 (CRC32_BUILTIN): Remove.
35243 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35244 crc32cb, crc32ch and crc32cw.
35245 (arm_init_crc32_builtins): Remove.
35246 (arm_init_builtins): Use arm_init_acle_builtins rather
35247 than arm_init_crc32_builtins.
35248 (arm_expand_acle_builtin): New.
35249 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35250 * config/arm/arm_acle_builtins.def: New.
35252 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35254 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35255 (arm_builtin_datum): ... this.
35256 (arm_init_neon_builtin): Rename to ...
35257 (arm_init_builtin): ... this. Add a new parameters PREFIX
35258 and USE_SIG_IN_NAME.
35259 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35260 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35261 'arm_builtin_datum'.
35262 (arm_init_vfp_builtins): Likewise.
35263 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35264 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35265 (arm_expand_neon_args): Rename to ...
35266 (arm_expand_builtin_args): ... this. Rename builtin_arg
35267 enum values and differentiate between ARG_BUILTIN_MEMORY
35268 and ARG_BUILTIN_NEON_MEMORY.
35269 (arm_expand_neon_builtin_1): Rename to ...
35270 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35271 values, arm_expand_builtin_args and add bool parameter NEON.
35272 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35273 (arm_expand_vfp_builtin): Likewise.
35274 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35276 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35278 PR middle-end/77484
35279 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35280 * predict.c (tree_estimate_probability_bb): Reverse direction of
35281 polymorphic call predictor.
35283 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35285 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35286 (determine_pass_name_match): ...this new function and...
35287 (should_skip_pass_p): ...this new function.
35289 2017-01-06 Nathan Sidwell <nathan@acm.org>
35291 * ipa-visibility.c (function_and_variable_visibility): Reformat
35292 comments and long lines. Remove extrneous if.
35293 * symtab.c (symtab_node::make_decl_local): Fix code format.
35294 (symtab_node::set_section_for_node): Fix comment typo.
35296 2017-01-06 Martin Liska <mliska@suse.cz>
35299 * lra-constraints.c: Rename invariant to lra_invariant.
35300 * predict.c (set_even_probabilities): Initialize e to NULL.
35302 2017-01-05 Martin Sebor <msebor@redhat.com>
35304 PR tree-optimization/78910
35305 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35306 (format_integer): Correct off-by-one error in the handling
35307 of precision with negative numbers in signed conversions..
35309 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35311 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35313 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35315 PR tree-optimization/71016
35316 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35317 factor_out_conditional_conversion. Formatting fix.
35318 (factor_out_conditional_conversion): Add cond_stmt argument.
35319 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35320 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35323 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35325 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35326 read-rtl-function.o, and selftest-rtl.o.
35327 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35328 (selftest::aarch64_test_loading_full_dump): New function.
35329 (selftest::aarch64_run_selftests): New function.
35330 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35331 selftest::aarch64_run_selftests.
35332 * config/i386/i386.c
35333 (selftest::ix86_test_loading_dump_fragment_1): New function.
35334 (selftest::ix86_test_loading_call_insn): New function.
35335 (selftest::ix86_test_loading_full_dump): New function.
35336 (selftest::ix86_test_loading_unspec): New function.
35337 (selftest::ix86_run_selftests): Call the new functions.
35338 * emit-rtl.c (maybe_set_max_label_num): New function.
35339 * emit-rtl.h (maybe_set_max_label_num): New decl.
35340 * function.c (instantiate_decls): Guard call to
35341 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35342 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35344 * gensupport.c (gen_reader::gen_reader): Pass "false"
35345 for new "compact" param of rtx_reader.
35346 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35347 rather than an empty string for NULL strings.
35348 * read-md.c: Potentially include config.h rather than bconfig.h.
35349 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35350 (have_error): New global, copied from errors.c.
35351 (md_reader::read_name): Rename to...
35352 (md_reader::read_name_1): ...this, adding "out_loc" param,
35353 and converting "missing name or number" to returning false, rather
35355 (md_reader::read_name): Reimplement in terms of read_name_1.
35356 (md_reader::read_name_or_nil): New function.
35357 (md_reader::read_string): Handle "(nil)" by returning NULL.
35358 (md_reader::md_reader): Add new param "compact".
35359 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35360 (md_reader::read_file): New method.
35361 * read-md.h (md_reader::md_reader): Add new param "compact".
35362 (md_reader::read_file): New method.
35363 (md_reader::is_compact): New accessor.
35364 (md_reader::read_name): Convert return type from void to file_location.
35365 (md_reader::read_name_or_nil): New decl.
35366 (md_reader::read_name_1): New decl.
35367 (md_reader::m_compact): New field.
35368 (noop_reader::noop_reader): Pass "false" for new "compact" param
35370 (rtx_reader::rtx_reader): Add new "compact" param.
35371 (rtx_reader::read_rtx_operand): Make virtual and convert return
35372 type from void to rtx.
35373 (rtx_reader::read_until): New decl.
35374 (rtx_reader::handle_any_trailing_information): New virtual function.
35375 (rtx_reader::postprocess): New virtual function.
35376 (rtx_reader::finalize_string): New virtual function.
35377 (rtx_reader::m_in_call_function_usage): New field.
35378 (rtx_reader::m_reuse_rtx_by_id): New field.
35379 * read-rtl-function.c: New file.
35380 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35381 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35382 (selftest::verify_three_block_rtl_cfg): New decl.
35383 * read-rtl-function.h: New file.
35384 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35385 For host, include function.h, memmodel.h, and emit-rtl.h.
35386 (one_time_initialization): New function.
35387 (struct compact_insn_name): New struct.
35388 (compact_insn_names): New array.
35389 (find_code): Handle insn codes in compact dumps.
35390 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35391 (bind_subst_iter_and_attr): Likewise.
35392 (add_condition_to_string): Likewise.
35393 (add_condition_to_rtx): Likewise.
35394 (apply_attribute_uses): Likewise.
35395 (add_current_iterators): Likewise.
35396 (apply_iterators): Likewise.
35397 (initialize_iterators): Guard usage of apply_subst_iterator with
35398 #ifdef GENERATOR_FILE.
35399 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35400 (md_reader::read_mapping): Likewise.
35401 (add_define_attr_for_define_subst): Likewise.
35402 (add_define_subst_attr): Likewise.
35403 (read_subst_mapping): Likewise.
35404 (check_code_iterator): Likewise.
35405 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35407 (one_time_initialization): New function.
35408 (rtx_reader::read_until): New method.
35409 (read_flags): New function.
35410 (parse_reg_note_name): New function.
35411 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35412 Handle reuse_rtx ids.
35413 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35414 Add parsing support for RTL dumps, mirroring the special-cases in
35415 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35416 values, and calling handle_any_trailing_information.
35417 (rtx_reader::read_rtx_operand): Convert return type from void
35418 to rtx, returning return_rtx. Handle case 'e'. Call
35419 finalize_string on XSTR and XTMPL fields.
35420 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35421 "(nil)" values were omitted. Call the postprocess vfunc on the
35423 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35424 class ctor. Initialize m_in_call_function_usage. Call
35425 one_time_initialization.
35426 * rtl-tests.c (selftest::test_uncond_jump): Call
35427 set_new_first_and_last_insn.
35428 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35429 * selftest-rtl.c: New file.
35430 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35431 (selftest::get_insn_by_uid): New decl.
35432 * selftest-run-tests.c (selftest::run_tests): Call
35433 read_rtl_function_c_tests.
35434 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35435 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35438 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35440 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35441 operands in a special way. Assert that pos+len <= mode precision.
35443 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35445 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35446 3 argument Alias with unlimited for the negative form.
35447 (fno-vect-cost-model): Removed.
35449 2017-01-05 Martin Liska <mliska@suse.cz>
35451 * hsa-gen.c (gen_hsa_divmod): New function.
35452 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35454 2017-01-05 Martin Liska <mliska@suse.cz>
35457 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35460 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35462 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35463 small constant length operands.
35465 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35467 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35468 between loop iterations.
35470 2017-01-05 Martin Liska <mliska@suse.cz>
35473 * gimplify.c (gimplify_decl_expr): Compare to
35474 asan_poisoned_variables instread of checking flags.
35475 (gimplify_target_expr): Likewise.
35476 (gimplify_expr): Likewise.
35477 (gimplify_function_tree): Conditionally initialize
35478 asan_poisoned_variables.
35480 2017-01-04 Jeff Law <law@redhat.com>
35482 PR tree-optimizatin/78812
35483 * rtl.h (contains_mem_rtx_p): Prototype.
35484 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35485 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35486 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35487 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35488 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35490 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35492 * input.c (assert_char_at_range): Default-initialize actual_range.
35494 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35496 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35497 to match the caller.
35499 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35501 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35502 insns after final jump in test to emit dummy move.
35504 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35506 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35507 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35509 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35511 * multiple_target.c (create_dispatcher_calls): Init e_next.
35512 * tree-ssa-loop-split.c (split_loop): Init border.
35513 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35516 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35521 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35522 (altivec_register_operand): Do not return true if the operand
35523 contains a SUBREG mixing SImode and SFmode.
35524 (vsx_register_operand): Likewise.
35525 (vsx_reg_sfsubreg_ok): New predicate.
35526 (vfloat_operand): Do not return true if the operand contains a
35527 SUBREG mixing SImode and SFmode.
35528 (vint_operand): Likewise.
35529 (vlogical_operand): Likewise.
35530 (gpc_reg_operand): Likewise.
35531 (int_reg_operand): Likewise.
35532 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35533 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35534 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35536 (rs6000_emit_move_si_sf_subreg): New helper function.
35537 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35538 fixup SUBREGs involving SImode and SFmode.
35539 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35540 numbers for the new peephole2 optimization.
35541 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35542 the GLIBC math library that do AND/IOR/XOR operations on single
35543 precision floating point.
35544 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35545 target macros to say whether we need to avoid SUBREGs mixing
35547 (TARGET_ALLOW_SF_SUBREG): Likewise.
35548 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35549 (UNSPEC_SI_FROM_SF): Likewise.
35550 (iorxor): Change spacing.
35551 (and_ior_xor): New iterator for AND, IOR, and XOR.
35552 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35553 (movdi_from_sf_zero_ext): Likewise.
35554 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35555 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35556 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35557 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35558 (fms<mode>4): Likewise.
35559 (fnma<mode>4): Likewise.
35560 (fnms<mode>4): Likewise.
35561 (nfma<mode>4): Likewise.
35562 (nfms<mode>4): Likewise.
35564 2017-01-04 Marek Polacek <polacek@redhat.com>
35567 * doc/invoke.texi: Document -Wpointer-compare.
35569 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35571 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35574 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35575 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35578 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35581 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35582 documentation of the powerpc_popcntb_ok attribute.
35583 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35584 code to issue warning messages if a requested CPU configuration is
35585 not supported by the binary (assembler and loader) toolchain.
35586 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35587 made to define a built-in function that has been disabled.
35588 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35589 made to define a built-in function that has been disabled.
35590 (altivec_init_builtins): Add comment explaining why definition
35591 of the DST built-in functions is not preceded by an assertion
35592 check. Add assertions to prevent ICE if attempts are made to
35593 define an altivec predicate or an abs* built-in function that has
35595 (htm_init_builtins): Add comment explaining why definition of the
35596 htm built-in functions is not preceded by an assertion check.
35598 2017-01-04 Jeff Law <law@redhat.com>
35600 PR tree-optimizatin/67955
35601 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35602 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35603 the points-to solution does not include pt_null. Use DECL_PT_UID
35606 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35608 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35609 Use gen_int_mode instead of gen_lopwart for const_int operands.
35611 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35613 PR tree-optimization/71563
35614 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35615 out of range value - has low bits known to be zero.
35617 2017-01-04 Alan Modra <amodra@gmail.com>
35619 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35620 * configure: Regenerate.
35621 * config.in: Regenerate.
35623 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35626 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35627 a substring of the message, but strcmp with the whole message. Ifdef
35628 ENABLE_NLS, translate the message first using dgettext.
35630 2017-01-03 Jeff Law <law@redhat.com>
35632 PR tree-optimizatin/78856
35633 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35634 (mark_threaded_blocks): Remove code to truncate thread paths that
35635 cross multiple loop headers. Instead invalidate the cached loop
35636 iteration information and handle case of a thread path walking
35637 into an irreducible region.
35639 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35642 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35643 assertions. Add support for doing the signbit if the IEEE 128-bit
35644 floating point value is in a GPR.
35645 * config/rs6000/rs6000.md (Fsignbit): Delete.
35646 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35647 Update the length attribute if the value is in a GPR.
35648 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35649 the sign or zero extension instruction, since the value is always 0/1.
35650 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35653 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35654 extracting SImode to a GPR register so that we can generate a
35655 store, limit the vector to be in a traditional Altivec register
35656 for the vextuwrx instruction.
35658 2017-01-03 Ian Lance Taylor <iant@google.com>
35660 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35662 2017-01-03 Martin Sebor <msebor@redhat.com>
35664 PR tree-optimization/78696
35665 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35666 precision. Use MPFR for %f for greater fidelity. Correct handling
35668 (pass_sprintf_length::compute_format_length): Set width and precision
35669 specified by asrerisk to void_node for vararg functions.
35670 (try_substitute_return_value): Adjust dump output.
35672 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35674 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35676 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35678 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35679 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35680 -mlra/-mno-lra was passed to the compiler.
35682 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35684 PR rtl-optimization/65618
35685 * emit-rtl.c (try_split): Move initialization of "before" and
35686 "after" to just before the call to emit_insn_after_setloc.
35688 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35690 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35692 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35694 * dwarf2out.c (gen_enumeration_type_die): When
35695 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35697 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35699 PR tree-optimization/78965
35700 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35701 Change first argument from const call_info & to call_info &. For %n
35702 set info.nowrite to false.
35704 PR middle-end/78901
35705 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35706 possibly throwing calls.
35708 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35709 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35710 and fns handling, rather than in a separate case SSA_NAME.
35712 2017-01-02 Jeff Law <law@redhat.com>
35714 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35715 fixes for first_period and second_period variables.
35717 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35720 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35721 (*insvqi_1): New insn pattern.
35722 (*insvqi_1_mem_rex64): Ditto.
35723 (*insvqi_2): Ditto.
35724 (*insvqi_3): Rename from *insvqi.
35726 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35728 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35730 * doc/cfg.texi (Edges): Remove reference to Java.
35731 (Maintaining the CFG): Ditto.
35733 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35735 PR middle-end/77674
35736 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35737 transparent aliases.
35739 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35741 PR middle-end/77484
35742 * predict.def (PRED_CALL): Update hitrate.
35743 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35744 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35745 into direct/indirect/polymorphic variants.
35747 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35749 Update copyright years.
35751 * gcc.c (process_command): Update copyright notice dates.
35752 * gcov-dump.c (print_version): Ditto.
35753 * gcov.c (print_version): Ditto.
35754 * gcov-tool.c (print_version): Ditto.
35755 * gengtype.c (create_file): Ditto.
35756 * doc/cpp.texi: Bump @copying's copyright year.
35757 * doc/cppinternals.texi: Ditto.
35758 * doc/gcc.texi: Ditto.
35759 * doc/gccint.texi: Ditto.
35760 * doc/gcov.texi: Ditto.
35761 * doc/install.texi: Ditto.
35762 * doc/invoke.texi: Ditto.
35764 Copyright (C) 2017 Free Software Foundation, Inc.
35766 Copying and distribution of this file, with or without modification,
35767 are permitted in any medium without royalty provided the copyright
35768 notice and this notice are preserved.