1 2017-11-20 Tom de Vries <tom@codesourcery.com>
3 PR rtl-optimization/82020
4 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
5 IF_THEN_ELSE condition.
7 2017-11-19 Jeff Law <law@redhat.com>
9 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
10 of degenerates resulting from ignoring an edge.
12 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
15 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
17 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
20 * profile-count.c (profile_count::to_sreal_scale): Fix return value
21 for uninitialied counts.
23 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
26 * tree-inline.c (estimate_num_insns): Set to 1 at least.
28 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
31 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane
34 2017-11-19 Tom de Vries <tom@codesourcery.com>
36 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
39 2017-11-19 Tom de Vries <tom@codesourcery.com>
41 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
44 2017-11-19 Tom de Vries <tom@codesourcery.com>
46 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
49 2017-11-19 Tom de Vries <tom@codesourcery.com>
51 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
53 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
55 2017-11-19 Tom de Vries <tom@codesourcery.com>
57 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
59 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
60 semicolon after MCORE_EXPORT_NAME call.
62 2017-11-19 Tom de Vries <tom@codesourcery.com>
65 * vmsdbgout.c (vmsdbgout_early_finish): New function.
66 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
68 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
70 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
71 (cgraph_edge::clone): Cleanup updating of profile.
72 * ipa-cp.c (update_profiling_info): Likewise.
73 * ipa-inline-transform.c (inline_transform): Likewise.
74 * ipa-inline.c (inline_small_functions): Add missing space to dump.
75 * ipa-split.c (execute_split_functions): Do not split when function
77 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
78 * profile-count.c (profile_count::dump): Add global0.
79 (profile_count::to_cgraph_frequency): Do not ICE when entry is
81 (profile_count::to_sreal_scale): Likewise.
82 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
83 (profile_count::combine_with_ipa_count): New function.
84 * profile-count.h (profile_guessed_global0adjusted): New.
85 (profile_count::adjusted_zero): New.
86 (profile_count::global0adjusted): New.
87 (profile_count::combine_with_ipa_count): New.
88 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
89 correct profile of return block of split functions.
90 (copy_cfg_body): Remove unused profile_count.
91 (copy_body): Likewise.
92 (expand_call_inline): Update.
93 (tree_function_versioning): Update.
95 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
97 * hash-set.h (hash_set::empty): New.
98 * tree-ssa-threadbackward.h: Delete.
99 * tree-ssa-threadbackward.c (class thread_jumps): New.
100 Move max_threaded_paths into class.
101 (fsm_find_thread_path): Remove arguments that are now in class.
102 (profitable_jump_thread_path): Rename to...
103 (thread_jumps::profitable_jump_thread_path): ...this.
104 (convert_and_register_jump_thread_path): Rename to...
105 (thread_jumps::convert_and_register_current_path): ...this.
106 (check_subpath_and_update_thread_path): Rename to...
107 (thread_jumps::check_subpath_and_update_thread_path): ...this.
108 (register_jump_thread_path_if_profitable): Rename to...
109 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
110 (handle_phi): Rename to...
111 (thread_jumps::handle_phi): ...this.
112 (handle_assignment): Rename to...
113 (thread_jumps::handle_assignment): ...this.
114 (fsm_find_control_statement_thread_paths): Rename to...
115 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
116 (find_jump_threads_backwards): Rename to...
117 (thread_jumps::find_jump_threads_backwards): ...this.
118 Initialize path local data.
119 (pass_thread_jumps::execute): Call find_jump_threads_backwards
120 from within thread_jumps class.
121 (pass_early_thread_jumps::execute): Same.
123 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
125 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
127 2017-11-17 Steve Ellcey <sellcey@cavium.com>
129 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
130 to canonical location.
132 2017-11-17 Steve Ellcey <sellcey@cavium.com>
135 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
137 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
139 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
141 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
142 rcrt1.o%s/grcrt1.o%s for -static-pie.
144 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
146 * i386.c (ix86_multiplication_cost, ix86_division_cost,
147 ix86_shift_rotate_cost): Break out from ...
148 (ix86_rtx_costs): ... here.
149 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
152 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
154 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
155 when entry block was promoted unlikely.
156 (estimate_bb_frequencies): Increase frequency scale.
157 * profile-count.h (profile_count): Export precision info.
159 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
161 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
162 disturbin profile of entry block.
164 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
166 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
169 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
171 * ipa-cp.c (update_profiling_info): Handle conversion to local
173 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
175 2017-11-17 Jeff Law <law@redhat.com>
177 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
178 from evrp_dom_walker class. Various methods moved into new class.
179 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
180 (evrp_range_analyzer::enter): New method.
181 (evrp_range_analyzer::leave): New method.
182 (evrp_dom_walker): Remove delegators no longer needed by this class.
183 Replace vr_values data member with evrp_range_analyzer
185 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
186 method extracted from evrp_dom_walker::before_dom_children.
187 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
188 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
190 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
191 Add private copy constructor and move assignment operators.
192 Privatize methods and class data where trivially possible.
193 (evrp_dom_walker::cleanup): New function, extracted from
194 execute_early_vrp. Simplify access to class data.
196 * vr-values.h (get_output_for_vrp): Prototype.
197 * vr-values.c (get_output_for_vrp): New function extracted from
198 vrp_visit_assignment_or_call and extract_range_from_stmt.
199 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
201 2017-11-17 Luis Machado <luis.machado@linaro.org>
203 * config/aarch64/aarch64.c
204 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
205 (qdf24xx_tunings) <autoprefetcher_model>: Set to
206 tune_params::AUTOPREFETCHER_WEAK.
208 2017-11-17 Tamar Christina <tamar.christina@arm.com>
211 * config/arm/arm.c (arm_valid_target_attribute_rec):
212 Parse "arch=" and "+<ext>".
213 (arm_valid_target_attribute_tree): Re-init global options.
214 (arm_option_override): Make non-static.
215 (arm_options_perform_arch_sanity_checks): Make errors fatal.
216 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
217 (__ARM_FEATURE_CRC32): Support undef.
218 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
219 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
221 2017-11-17 David Malcolm <dmalcolm@redhat.com>
223 * gdbinit.in (break-on-diagnostic): New command.
225 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
227 * config/i386/i386.c (ix86_expand_epilogue): Change simple
228 return to indirect jump for EH return if control-flow
229 protection is enabled. Change explicit 'false' argument in
230 pro_epilogue_adjust_stack with a value of
232 * config/i386/i386.md (simple_return_indirect_internal):
233 Remove SImode restriction to support 64-bit.
235 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
237 * combine.c (added_notes_insn): New.
238 (try_combine): Handle added_notes_insn like added_links_insn.
239 Rewrite return value code.
240 (distribute_notes): Set added_notes_insn to the earliest insn we added
243 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
245 PR rtl-optimization/82621
246 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
247 dest of one of those SETs is unused.
249 2017-11-17 Richard Biener <rguenther@suse.de>
252 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
253 * tree-pretty-print.c (dump_generic_node): Handle
254 annot_expr_parallel_kind.
255 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
256 * gimplify.c (gimple_boolify): Likewise.
258 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
260 * config.gcc (extra_headers): Add cet.h for x86 targets.
261 * config/i386/cet.h: New file.
262 * doc/install.texi: Add --enable-cet/--disable-cet.
264 2017-11-17 Richard Biener <rguenther@suse.de>
266 PR tree-optimization/83017
267 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
268 (gen_parallel_loop): Properly count iterations.
269 (parallelize_loops): Handle loop->can_be_parallel independent
270 of flag_loop_parallelize_all. Make static profitability test match
272 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
273 * invoke.texi (parloops-min-per-thread): Document.
275 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
277 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
278 upstreaming review comments.
280 2017-11-17 Tamar Christina <tamar.christina@arm.com>
282 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
283 with fast unaligned access.
284 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
286 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
288 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
289 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
291 2017-11-17 Richard Biener <rguenther@suse.de>
293 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
294 folding of references.
296 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
299 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
300 of replacing call to strncmp with corresponding call to strcmp when
303 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
305 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
306 option prefer-avx256 for skylake-avx512 configuration.
307 * config/i386/i386.c (ix86_option_override_internal): Ditto.
308 (get_builtin_code_for_version): Ditto.
310 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
311 Monk Chiang <sh.chiang04@gmail.com>
313 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
314 (FIXED_REGISTERS): Reserve more register numbers.
315 (CALL_USED_REGISTERS): Likewise.
316 (REG_ALLOC_ORDER): Likewise.
317 (REG_CLASS_CONTENTS): Likewise.
318 (REGISTER_NAMES): Likewise.
320 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
321 Kito Cheng <kito.cheng@gmail.com>
323 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
325 * config/nds32/iterators.md: Add vector mode iterators and attributes.
327 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
329 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
330 parameter list for vec_splats.
332 2017-11-16 Joseph Myers <joseph@codesourcery.com>
334 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
336 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
339 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
342 * doc/install.texi: Document incompatibility of Perl >=5.6.26
343 with the required version of automake 1.11.6.
345 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
347 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
349 (DU_C2_3_power9): Correct reservation combinations.
350 (FP_DIV_power9, VEC_DIV_power9): New.
351 (power9-alu): Split out rotate/shift...
352 (power9-rot): ...to here, correct dispatch resource.
353 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
355 (power9-fp): Correct latency.
356 (power9-sdiv): Add div/sqrt resource.
357 (power9-ddiv): Correct latency, add div/sqrt resource.
358 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
359 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
361 (power9-qpdiv, power9-qpmul): Adjust resource usage.
363 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
365 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
366 switch statement mapping KF built-ins to TF built-ins if we don't
367 have the proper ISA 3.0 assembler support.
369 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
371 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
372 (lower_emutls_function_body): Do not compute it.
374 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
376 * ipa-split.c (split_bb_info): Turn time to sreal.
377 (split_point): Likewise.
378 (dump_split_point): Likewise.
379 (fine_split_points): Likewise.
380 (execute_split_functions): Only zero split_bbs; turn time to sreals.
382 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
384 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
387 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
389 * predict.c (combine_predictions_for_bb): Preserve zero predicted
391 (expensive_function_p): Remove useless assert.
392 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
394 2017-11-16 Martin Sebor <msebor@redhat.com>
396 PR tree-optimization/82588
397 PR tree-optimization/82583
398 * tree-vrp.c (check_array_ref): Handle flexible array members,
399 string literals, and inner indices.
400 (search_for_addr_array): Add detail to diagnostics.
402 2017-11-16 Nathan Sidwell <nathan@acm.org>
406 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
407 (SET_DECL_ASSEMBLER_NAME): Forward to
408 overwrite_decl_assembler_name.
409 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
410 (overwrite_decl_assembler_name): Declare.
411 * tree.c (overwrite_decl_assembler_name): New.
412 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
413 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
414 (LANG_HOOKS_INITIALIZER): Add it.
415 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
416 * langhooks.c (lhd_set_decl_assembler_name): Use
417 SET_DECL_ASSEMBLER_NAME.
418 (lhd_overwrite_decl_assembler_name): Default implementation.
420 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
421 Jackson Woodruff <jackson.woodruff@arm.com>
423 PR tree-optimization/71026
424 * match.pd: Canonicalize constant multiplies in division.
426 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
428 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
429 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
430 Remove OPT_fomit_frame_pointer entry.
431 * common/config/arc/arc-common.c: Likewise.
432 * common/config/arm/arm-common.c: Likewise.
433 * common/config/avr/avr-common.c: Likewise.
434 * common/config/c6x/c6x-common.c: Likewise.
435 * common/config/cr16/cr16-common.c: Likewise.
436 * common/config/cris/cris-common.c: Likewise.
437 * common/config/epiphany/epiphany-common.c: Likewise.
438 * common/config/fr30/fr30-common.c: Likewise.
439 * common/config/frv/frv-common.c: Likewise.
440 * common/config/ia64/ia64-common.c: Likewise.
441 * common/config/iq2000/iq2000-common.c: Likewise.
442 * common/config/lm32/lm32-common.c: Likewise.
443 * common/config/m32r/m32r-common.c: Likewise.
444 * common/config/mcore/mcore-common.c: Likewise.
445 * common/config/microblaze/microblaze-common.c: Likewise.
446 * common/config/mips/mips-common.c: Likewise.
447 * common/config/mmix/mmix-common.c: Likewise.
448 * common/config/mn10300/mn10300-common.c: Likewise.
449 * common/config/nios2/nios2-common.c: Likewise.
450 * common/config/pa/pa-common.c: Likewise.
451 * common/config/pdp11/pdp11-common.c: Likewise.
452 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
453 * common/config/riscv/riscv-common.c: Likewise.
454 * common/config/rs6000/rs6000-common.c: Likewise.
455 * common/config/rx/rx-common.c: Likewise.
456 * common/config/s390/s390-common.c: Likewise.
457 * common/config/sh/sh-common.c: Likewise.
458 * common/config/sparc/sparc-common.c: Likewise.
459 * common/config/tilegx/tilegx-common.c: Likewise.
460 * common/config/tilepro/tilepro-common.c: Likewise.
461 * common/config/v850/v850-common.c: Likewise.
462 * common/config/visium/visium-common.c: Likewise.
463 * common/config/xstormy16/xstormy16-common.c: Likewise.
464 * common/config/xtensa/xtensa-common.c: Likewise.
465 * invoke.texi (-fomit-frame-pointer): Update documentation.
467 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
469 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
471 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
473 * cfg.c (scale_bbs_frequencies_int,
474 cale_bbs_frequencies_gcov_type): Remove.
475 * cfg.h (scale_bbs_frequencies_int,
476 cale_bbs_frequencies_gcov_type): Remove.
478 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
480 * tree-ssa-loop-manip.c
481 (scale_dominated_blocks_in_loop): Update to profile counts.
482 (tree_transform_and_unroll_loop): Likewise.
484 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
486 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
487 scale_bbs_frequencies_int.
489 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
491 * final.c (compute_alignments): Use counts rather than frequencies.
493 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
495 * cfgloopanal.c: Include sreal.h
496 (average_num_loop_insns): Use counts and sreal for accounting.
498 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
500 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
503 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
504 Kito Cheng <kito.cheng@gmail.com>
506 * config/nds32/constraints.md: Provide more constraints.
507 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
508 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
509 support constraints usage.
511 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
513 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
515 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
516 Kito Cheng <kito.cheng@gmail.com>
518 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
519 * config/nds32/nds32.opt: Refine the layout.
520 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
521 TARGET_EXT_STRING): Support new options.
522 * config/nds32/nds32.h: Likewise.
523 * config/nds32/nds32.md: Likewise.
524 * config/nds32/nds32-predicates.c: Likewise.
525 * config/nds32/constraints.md: Likewise.
526 * common/config/nds32/nds32-common.c: Likewise.
528 2017-11-16 Julia Koval <julia.koval@intel.com>
531 * config/i386/gfniintrin.h: Add sse check.
532 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
534 2017-11-16 Julia Koval <julia.koval@intel.com>
536 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
537 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
538 (ix86_handle_option): Handle -mavx512vbmi2.
539 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
540 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
541 * config/i386/i386-c.c (__AVX512VBMI2__): New.
542 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
543 (ix86_valid_target_attribute_inner_p): Ditto.
544 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
545 * config/i386/i386.opt (mavx512vbmi2): New option.
546 * doc/invoke.texi: Add new option.
548 [2017-11-16 Julia Koval <julia.koval@intel.com>
550 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
551 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
552 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
553 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
554 _mm512_gf2p8mul_epi8): New intrinsics.
555 * config/i386/i386-builtin-types.def
556 (V64QI_FTYPE_V64QI_V64QI): New type.
557 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
558 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
559 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
560 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
561 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
562 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
564 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
566 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
567 explicitly as a stream of bytes.
569 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
571 * config/rs6000/altivec.h (vec_xst_be): New #define.
572 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
573 and externalize from *altivec_vperm_<mode>_internal.
574 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
576 (XL_BE_V8HI): Likewise.
577 (XL_BE_V4SI): Likewise.
578 (XL_BE_V4SI): Likewise.
579 (XL_BE_V2DI): Likewise.
580 (XL_BE_V4SF): Likewise.
581 (XL_BE_V2DF): Likewise.
582 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
583 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
584 all array entries with these keys: VSX_BUILTIN_VEC_XL,
585 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
586 VSX_BUILTIN_VEC_XST_BE.
587 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
588 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
590 (altivec_init_builtins): Replace conditional calls to def_builtin
591 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
592 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
593 with unconditional calls. Remove calls to def_builtin for
594 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
595 __builtin_vec_xst_be.
596 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
597 to define_expand, and add alternate RTL generation for P8.
598 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
600 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
601 add alternate RTL generation for P8.
602 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
603 vsx_ld_elemrev_v16qi.
604 (vsx_st_elemrev_v8hi): Convert define_insn
605 to define_expand, and add alternate RTL generation for P8.
606 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
608 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
609 add alternate RTL generation for P8.
610 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
611 vsx_st_elemrev_v16qi.
613 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
616 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
617 TARGET_AVX512ER check.
618 (ix86_option_override_internal): Set MASK_VZEROUPPER if
619 neither -mzeroupper nor -mno-zeroupper is used and
620 TARGET_EMIT_VZEROUPPER is set.
621 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
622 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
624 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
626 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
627 folding of vector compares.
628 (fold_build_vec_cmp): New helper function.
629 (fold_compare_helper): New helper function.
630 (builtin_function_type): Add compare builtins to the list of functions
631 having unsigned arguments. Cosmetic updates to comment indentation.
632 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
633 the not+eq combination.
635 2017-11-15 Bin Cheng <bin.cheng@arm.com>
637 PR tree-optimization/82726
638 PR tree-optimization/70754
639 * tree-predcom.c (order_drefs_by_pos): New function.
640 (combine_chains): Move code setting has_max_use_after to...
641 (try_combine_chains): ...here. New parameter. Sort combined chains
642 according to position information.
643 (tree_predictive_commoning_loop): Update call to above function.
644 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
646 2017-11-15 Bin Cheng <bin.cheng@arm.com>
648 PR tree-optimization/82726
650 2017-01-23 Bin Cheng <bin.cheng@arm.com>
652 PR tree-optimization/70754
653 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
654 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
655 combined stmt before it if not NULL.
656 (combine_chains): Process refs reversely and compute dominance point
660 2017-02-23 Bin Cheng <bin.cheng@arm.com>
662 PR tree-optimization/79663
663 * tree-predcom.c (combine_chains): Process refs in reverse order
664 only for ZERO length chains, and add explaining comment.
666 2017-11-15 Tamar Christina <tamar.christina@arm.com>
668 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
669 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
670 * doc/invoke.texi (ARM Options): Add armv8.3-a.
672 2017-11-15 Tamar Christina <tamar.christina@arm.com>
674 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
676 2017-11-15 Martin Liska <mliska@suse.cz>
678 * tree-cfg.c (pass_warn_function_return::execute):
679 Compare warn_return_type for greater than zero.
681 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
685 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
686 to return true on Xeon and not on Xeon Phi.
687 (ix86_check_avx256_register): Changed to ...
688 (ix86_check_avx_upper_register): ... this. Add extra check for
689 VALID_AVX512F_REG_OR_XI_MODE.
690 (ix86_avx_u128_mode_needed): Changed
691 ix86_check_avx256_register to ix86_check_avx_upper_register.
692 (ix86_check_avx256_stores): Changed to ...
693 (ix86_check_avx_upper_stores): ... this. Changed
694 ix86_check_avx256_register to ix86_check_avx_upper_register.
695 (ix86_avx_u128_mode_after): Changed
696 avx_reg256_found to avx_upper_reg_found. Changed
697 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
698 (ix86_avx_u128_mode_entry): Changed
699 ix86_check_avx256_register to ix86_check_avx_upper_register.
700 (ix86_avx_u128_mode_exit): Ditto.
701 * config/i386/i386.h: (host_detect_local_cpu): New define.
703 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
705 * config/arm/xgene1.md (xgene1): Split into automatons
706 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
707 (xgene1_f_load): Adjust reservations and/or types.
708 (xgene1_f_store): Likewise.
709 (xgene1_load_pair): Likewise.
710 (xgene1_store_pair): Likewise.
711 (xgene1_fp_load1): Likewise.
712 (xgene1_load1): Likewise.
713 (xgene1_store1): Likewise.
714 (xgene1_move): Likewise.
715 (xgene1_alu): Likewise.
716 (xgene1_simd): Likewise.
717 (xgene1_bfm): Likewise.
718 (xgene1_neon_load1): Likewise.
719 (xgene1_neon_store1): Likewise.
720 (xgene1_neon_logic): Likewise.
721 (xgene1_neon_st1): Likewise.
722 (xgene1_neon_ld1r): Likewise.
723 (xgene1_alu_cond): Added.
724 (xgene1_shift_reg): Likwise.
725 (xgene1_bfx): Likewise.
726 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
728 2017-11-15 Jakub Jelinek <jakub@redhat.com>
731 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
733 (can_widen_mult_without_libcall): New function.
734 (expand_mul_overflow): If only checking unsigned mul overflow,
735 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
736 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
737 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
738 (expand_DIVMOD): Formatting fix.
739 * expmed.h (expand_mult): Add NO_LIBCALL argument.
740 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
741 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
743 PR tree-optimization/82977
744 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
745 constructed temporary to strlen_to_stridx.put.
747 2017-11-15 Martin Liska <mliska@suse.cz>
749 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
750 * configure: Regenerate.
752 2017-11-15 Martin Liska <mliska@suse.cz>
755 * config/sh/sh-mem.cc: Use proper probability for
758 2017-11-14 Jeff Law <law@redhat.com>
760 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
761 the red zone for stack_clash_protection_final_dynamic_probe targets
762 when the total dynamic stack size is zero bytes.
764 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
765 blocks is post order.
767 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
769 * dumpfile.h (TDF_COMPARE_DEBUG): New.
770 * final.c (rest_of_clean_state): Set it for the
771 -fcompare-debug dump.
772 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
773 class when TDF_COMPARE_DEBUG is set.
775 * dwarf2out.c (gen_producer_string): Discard
778 2017-11-15 Joseph Myers <joseph@codesourcery.com>
781 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
782 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
783 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
784 (__TGMATH_REAL_2_3): Remove macros.
785 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
786 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
788 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
790 (remquo): Define using __TGMATH_REAL_3.
792 2017-11-14 Jeff Law <law@redhat.com>
794 * vr-values.c: New file with contents extracted from tree-vrp.c.
795 * Makefile.in (OBJS): Add vr-values.o
796 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
797 (set_value_range, set_and_canonicalize_value_range): Likewise.
798 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
799 (value_range_constant_singleton, symbolic_range_p): Likewise.
800 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
801 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
802 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
803 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
804 (find_case_label_range, find_case_label_index): Likewise.
805 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
806 (range_int_cst_singleton_p, value_inside_range): Likewise.
807 (get_single_symbol): Likewise.
808 (switch_update): Move structure definition here.
809 (to_remove_edges, to_update_switch_stmts): Provide externs.
810 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
811 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
812 (vrp_val_is_min, set_value_range): Likewise.
813 (set_and_canonicalize_value_range, copy_value_range): Likewise.
814 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
815 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
816 (range_is_nonnull, range_int_cst_p): Likewwise.
817 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
818 (get_single_symbol, operand_less_p): Likewise
819 (compare_values_warnv, compare_values): Likewise.
820 (value_inside_range, value_range_constant_singleton): Likewise.
821 (zero_nonzero_bitgs_from_vr): Likewise.
822 (extract_range_from_binary_expr_1): Likewise.
823 (overflow_comparison_p): Likewise.
824 (to_remove_edges, to_update_switch_stmts): Likewise.
825 (find_case_label-index, find_case_label_range): Likewise.
826 (switch_update, set_value_range_to_nonnegative): Remove.
827 (set_value_range_to_truthvalue): Likewise.
828 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
829 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
830 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
831 (find_case_label_ranges, test_for_singularity): Likewise.
832 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
833 (x_vr_values): Move to its remaining use site.
835 2017-11-10 Jeff Law <law@redhat.com>
837 * vr-values.h (VR_INITIALIZER): Move #define here.
838 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
839 * Makefile.in (OBJS): Add tree-evrp.o
840 * tree-vrp.h (assert_info): Move structure definition here.
841 (set_value_range_to_varying): Prototype.
842 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
843 (infer_value_range, register_edge_assert_for): Likewise.
844 (stmt_interesting_for_vrp): Likewise.
845 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
846 (set_value_range_to_varying): No longer static.
847 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
848 (infer_value_range, register_edge_assert_for): Likewise.
850 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
852 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
853 generate the XXBRD instruction.
855 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
856 (rs6000_builtin_type_compatible): Treat _Float128 and long double
857 as being compatible if -mabi=ieeelongdouble.
858 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
859 to setup float128 built-ins with hardware support.
860 (BU_FLOAT128_HW_2): Likewise.
861 (BU_FLOAT128_HW_3): Likewise.
862 (BU_FLOAT128_HW_VSX_1): Likewise.
863 (BU_FLOAT128_HW_VSX_2): Likewise.
864 (scalar_extract_expq): Change float128 built-in functions to
865 accommodate having both KFmode and TFmode functions. Use the
866 KFmode variant as the default.
867 (scalar_extract_sigq): Likewise.
868 (scalar_test_neg_qp): Likewise.
869 (scalar_insert_exp_q): Likewise.
870 (scalar_insert_exp_qp): Likewise.
871 (scalar_test_data_class_qp): Likewise.
872 (sqrtf128_round_to_odd): Delete processing the round to odd
873 built-in functions as special built-in functions, and define them
874 as float128 built-ins. Use the KFmode variant as the default.
875 (truncf128_round_to_odd): Likewise.
876 (addf128_round_to_odd): Likewise.
877 (subf128_round_to_odd): Likewise.
878 (mulf128_round_to_odd): Likewise.
879 (divf128_round_to_odd): Likewise.
880 (fmaf128_round_to_odd): Likewise.
881 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
882 support for KFmode and TFmode xststdcqp calls.
883 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
884 point, switch the built-in handlers for the get/set float128
885 exponent, get float128 mantissa, float128 test built-ins, and the
886 float128 round to odd built-in functions. Eliminate creating the
887 float128 round to odd built-in functions as special built-ins.
888 (rs6000_init_builtins): Eliminate special creation of the float128
889 round to odd built-in functions.
890 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
891 function insns to support both TFmode and KFmode variants.
892 (xsxsigqp_<mode>): Likewise.
893 (xsiexpqpf_<mode>): Likewise.
894 (xsiexpqp_<mode>): Likewise.
895 (xststdcqp_<mode>): Likewise.
896 (xststdcnegqp_<mode>): Likewise.
897 (xststdcqp_<mode>): Likewise.
899 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
901 * tree-ssa-threadupdate.c (compute_path_counts): Remove
902 unused path_in_freq_ptr parameter.
903 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
905 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
907 * ipa-inline.c (edge_badness): Dump sreal frequency.
908 (compute_inlined_call_time): Match natural implementaiton ...
909 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
910 forgotten division by CGRAPH_FREQ_BASE.
912 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
914 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
915 Solaris 11. Update comment.
916 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
918 * config/sol2.h (STARTFILE_SPEC): Likewise.
919 * configure: Regenerate.
921 2017-11-14 Carl Love <cel@us.ibm.com>
923 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
924 le_ and be_ prefixes to swap* variables. Remove
925 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
928 2017-11-14 Jason Merrill <jason@redhat.com>
930 Support GTY((cache)) on hash_map.
931 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
932 (ggc_cache_remove): Override it instead of ggc_mx.
933 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
934 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
935 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
936 (simple_cache_map_traits): Override maybe_mx.
937 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
938 (hash_map): Friend gt_cleare_cache.
939 (gt_cleare_cache): New.
940 * tree.h (tree_cache_traits): New hash_map traits class.
941 (tree_cache_map): New typedef.
943 2017-11-14 Richard Biener <rguenther@suse.de>
945 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
946 paramter and handling.
947 (cleanup_control_flow_bb): Likewise.
948 (cleanup_control_flow_pre): New helper performing a DFS walk
949 to call cleanup_control_flow_bb in PRE order.
950 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
951 via cleanup_control_flow_pre.
953 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
955 * config/aarch64/aarch64-simd.md
956 (aarch64_simd_bsl<mode>_internal): Remove DImode.
957 (*aarch64_simd_bsl<mode>_alt): Likewise.
958 (aarch64_simd_bsldi_internal): New.
959 (aarch64_simd_bsldi_alt): Likewise.
961 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
963 * tracer.c (better_p): Do not compare frequencies.
964 * reg-stack.c (better_edge): Likewise.
965 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
968 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
970 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
971 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
972 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
973 * lto-streamer-in.c (input_function): Use update_max_bb_count.
974 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
975 * predict.c (maybe_hot_frequency_p): Inline to ...
976 (maybe_hot_count_p): ... here; rewrite to counts.
977 (counts_to_freqs): Rename to ...
978 (update_max_bb_count): ... this one.
979 (expensive_function_p): Use counts.
980 (estimate_bb_frequencies): Update.
981 (rebuild_frequencies): Update.
982 * predict.h (counts_to_freqs): Rename to ...
983 (update_max_bb_count): ... this one.
984 * profile.c (compute_branch_probabilities): Add debug info
985 * tree-inline.c (expand_call_inline): Update debug info.
986 (optimize_inline_calls): Use update_max_bb_count..
987 (tree_function_versioning): Use update_max_bb_count..
988 * value-prof.c (gimple_value_profile_transformations):
989 Do not use update_max_bb_count.
991 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
993 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
994 always use frequencies.
996 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
998 * bb-reorder.c: Remove frequencies from comments.
999 (better_edge_p): Use profile counts.
1000 (find_traces): Dump profile counts.
1001 (rotate_loop): Use profile counts.
1002 (find_traces_1_round): Likewise.
1003 (connect_better_edge_p): Use counts instead of probabilities for
1005 (copy_bb_p): Drop early check for non-0 frequency.
1006 (sanitize_hot_paths): Update comments.
1008 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1010 * ipa-split.c (struct split_point): Add count.
1011 (consider_split): Do not compute incoming frequency; compute incoming
1012 count and store it to split_point.
1013 (split_function): Set count of the call to split part correctly.
1015 2017-11-13 Carl Love <cel@us.ibm.com>
1017 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1019 2017-11-13 Tom Tromey <tom@tromey.com>
1021 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1023 2017-11-13 Carl Love <cel@us.ibm.com>
1025 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1026 Add support for builtins:
1027 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1028 vector {un,}signed {char,int,short},
1029 vector {un,}signed {char,int,short}) arguments.
1030 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1031 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1032 Add BU_P9V_AV_2 expansions for the builtins.
1033 * config/rs6000/altivec.h (vec_first_match_index,
1034 vec_first_mismatch_index, vec_first_match_or_eos_index,
1035 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1036 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1037 new extern declaration.
1038 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1039 * config/rs6000/vsx.md (first_match_index_<mode>,
1040 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1041 first_mismatch_or_eos_index_<mode>): Add define expand.
1042 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1043 * doc/extend.texi: Update the built-in documenation file for the new
1046 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1048 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1049 into the min/max operations for _Float<N> and _Float<N>X types.
1051 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1054 * dwarf2out.c (do_eh_frame): New static variable.
1055 (dwarf2out_begin_prologue): Set it.
1056 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1058 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1060 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1062 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1064 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1067 * debug.h (dwarf2out_do_eh_frame): Declare.
1068 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1069 (dwarf2out_do_frame): Use it.
1070 (dwarf2out_do_cfi_asm): Likewise.
1071 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1072 (dwarf2out_assembly_start): Likewise.
1073 (dwarf2out_begin_prologue): Fix comment.
1074 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1075 if the target needs either debug or unwind DWARF2 info.
1076 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1077 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1078 -ftrapping-math, -ftrapv and -fwrapv.
1080 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1082 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1083 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1084 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1085 (estimate_edge_size_and_time): Likewise.
1086 (ipa_merge_fn_summary_after_inlining): Likewise.
1087 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1088 (compute_uninlined_call_time): Use sreal_frequency.
1089 (compute_inlined_call_time): Likewise.
1090 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1091 * profile-count.c: Include sreal.h.
1092 (profile_count::to_sreal_scale): New.
1093 * profile-count.h: Forward declare sreal.
1094 (profile_count::to_sreal_scale): Declare.
1096 2017-11-13 Nathan Sidwell <nathan@acm.org>
1098 * diagnostic.c (maybe_line_and_column): New.
1099 (diagnostic_get_location_text): Use it.
1100 (diagnostic_report_current_module): Likewise.
1101 (test_diagnostic_get_location_text): Add tests.
1103 2017-11-13 Luis Machado <luis.machado@linaro.org>
1105 * doc/md.texi (Specifying processor pipeline description): Fix
1106 incorrect latency for the div instruction example.
1108 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1110 PR tree-optimization/78821
1111 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1112 that bit_not_p is the same.
1113 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1114 (split_group): Count precisely bit_not_p bits in each statement.
1115 (invert_op): New function.
1116 (imm_store_chain_info::output_merged_store): Use invert_op to
1117 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1118 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1120 2017-11-13 Martin Liska <mliska@suse.cz>
1122 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1123 (struct source_info): Likewise.
1124 (add_branch_counts): Likewise.
1125 (add_line_counts): Likewise.
1126 (function_summary): Likewise.
1127 (output_intermediate_line): Likewise.
1128 (generate_results): Likewise.
1130 2017-11-13 Martin Liska <mliska@suse.cz>
1132 * gcov.c (struct block_info): Remove typedef for block_t.
1133 (struct line_info): Likewise.
1134 (line_info::has_block): Likewise.
1135 (EXIT_BLOCK): Likewise.
1136 (unblock): Likewise.
1137 (circuit): Likewise.
1138 (get_cycles_count): Likewise.
1139 (process_file): 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.
1147 2017-11-13 Martin Liska <mliska@suse.cz>
1149 * gcov.c (struct arc_info): Remove typedef for arc_t.
1150 (struct line_info): Likewise.
1151 (add_branch_counts): Likewise.
1152 (output_branch_count): Likewise.
1153 (function_info::~function_info): Likewise.
1154 (circuit): Likewise.
1155 (output_intermediate_line): Likewise.
1156 (read_graph_file): Likewise.
1157 (solve_flow_graph): Likewise.
1158 (find_exception_blocks): Likewise.
1159 (add_line_counts): Likewise.
1160 (accumulate_line_info): Likewise.
1161 (output_line_details): Likewise.
1162 (output_function_details): Likewise.
1164 2017-11-13 Martin Liska <mliska@suse.cz>
1166 * gcov.c (struct function_info): Remove typedef for function_t.
1167 (struct source_info): Likewise.
1168 (source_info::get_functions_at_location): Likewise.
1169 (solve_flow_graph): Likewise.
1170 (find_exception_blocks): Likewise.
1171 (add_line_counts): Likewise.
1172 (output_intermediate_file): Likewise.
1173 (process_file): Likewise.
1174 (generate_results): Likewise.
1175 (release_structures): Likewise.
1176 (read_graph_file): Likewise.
1177 (read_count_file): Likewise.
1178 (accumulate_line_counts): Likewise.
1179 (output_lines): Likewise.
1181 2017-11-13 Martin Liska <mliska@suse.cz>
1183 * gcov.c (function_info::function_info): Remove num_counts
1184 and add vector<gcov_type>.
1185 (function_info::~function_info): Use the vector.
1186 (process_file): Likewise.
1187 (read_graph_file): Likewise.
1188 (read_count_file): Likewise.
1189 (solve_flow_graph): Likewise.
1191 2017-11-13 Martin Liska <mliska@suse.cz>
1193 * gcov.c (function_info::is_artificial): New function.
1194 (process_file): Erase all artificial early.
1195 (generate_results): Skip as all artificial are already
1198 2017-11-13 Martin Liska <mliska@suse.cz>
1200 * gcov.c (read_graph_file): Store to global vector of functions.
1201 (read_count_file): Iterate the vector.
1202 (process_file): Likewise.
1203 (generate_results): Likewise.
1204 (release_structures): Likewise.
1206 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1208 PR tree-optimization/82954
1209 * gimple-ssa-store-merging.c
1210 (imm_store_chain_info::coalesce_immediate_stores): If
1211 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1213 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1215 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1216 Upddate call to ENDIAN_LANE_N.
1217 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1218 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1219 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1220 and use aarch64_endian_lane_rtx.
1221 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1223 2017-11-12 Tom de Vries <tom@codesourcery.com>
1225 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1227 2017-11-12 Tom de Vries <tom@codesourcery.com>
1229 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1231 2017-11-12 Tom de Vries <tom@codesourcery.com>
1233 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1235 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1236 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1237 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1239 2017-11-11 Martin Sebor <msebor@redhat.com>
1242 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1245 * prefic.c (translate_name): Replace strncpy with memcpy to
1246 avoid -Wstringop-truncation.
1248 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1250 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1252 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1254 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1256 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1258 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1261 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1263 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1264 enable generating XXBRH if the value is in a vector register.
1265 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1266 value is in a vector register.
1267 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1268 register to register bswap64's instead of doing the GPR sequence
1269 used on previous machines.
1270 (bswapdi2_xxbrd): New insn.
1271 (bswapdi2_reg): Disallow on ISA 3.0.
1272 (register to register bswap64 splitter): Do not split the insn on
1273 ISA 3.0 systems that use XXBRD.
1275 2017-11-10 Martin Sebor <msebor@redhat.com>
1278 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1280 (find_subframework_file): Same.
1282 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1284 * auto-profile.c (afdo_indirect_call): Drop frequency.
1285 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1286 (cgraph_node::create_edge): Drop frequency argument.
1287 (cgraph_node::create_indirect_edge): Drop frequency argument.
1288 (cgraph_edge::make_speculative): Drop frequency arguments.
1289 (cgraph_edge::resolve_speculation): Do not update frequencies
1290 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1291 (cgraph_node::dump): Check consistency in IPA mode.
1292 (cgraph_edge::maybe_hot_p): Use IPA counter.
1293 (cgraph_edge::verify_count_and_frequency): Rename to ...
1294 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1295 (cgraph_node::verify_node): Update.
1296 * cgraph.h (struct cgraph_edge): Drop frequency.
1297 (cgraph_edge::frequency): New function.
1298 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1300 (cgraph_edge::rebuild_edges): Likewise.
1301 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1302 (duplicate_thunk_for_node): Do not pass frequency.
1303 (cgraph_node::create_clone): Scale only counts.
1304 (cgraph_node::create_virtual_clone): Do not pass frequency.
1305 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1306 (cgraph_node::create_version_clone): Do not pass frequency.
1307 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1308 (cgraph_node::expand_thunk): Do not pass frequency.
1309 (cgraph_node::create_wrapper): Do not pass frequency.
1310 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1312 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1313 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1314 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1315 (gather_caller_stats): Use frequency function.
1316 (ipcp_cloning_candidate_p): Use frequency function.
1317 (ipcp_propagate_stage): Use frequency function.
1318 (get_info_about_necessary_edges): Use frequency function.
1319 (update_profiling_info): Update only IPA profile.
1320 (update_specialized_profile): Use frequency functoin.
1321 (perhaps_add_new_callers): Update only IPA profile.
1322 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1323 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1324 (dump_ipa_call_summary): Use frequency function.
1325 (estimate_edge_size_and_time): Use frequency function.
1326 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1327 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1328 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1329 (update_noncloned_counts): ... ths one; scale counts only.
1330 (clone_inlined_nodes): Do not scale frequency.
1331 (inline_call): Do not pass frequency.
1332 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1333 (compute_inlined_call_time): Use IPA profile.
1334 (want_inline_small_function_p): Use IPA profile.
1335 (want_inline_self_recursive_call_p): Use IPA profile.
1336 (edge_badness): Use IPA profile.
1337 (lookup_recursive_calls): Use IPA profile.
1338 (recursive_inlining): Do not pass frequency.
1339 (resolve_noninline_speculation): Do not update frequency.
1340 (inline_small_functions): Collect max of IPA profile.
1341 (dump_overall_stats): Dump IPA porfile.
1342 (dump_inline_stats): Dump IPA porfile.
1343 (ipa_inline): Collect IPA stats.
1344 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1345 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1346 (ipa_propagate_frequency): Use frequency function.
1347 (ipa_profile): Cleanup.
1348 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1349 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1350 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1351 (input_node): Do not stream frequency.
1352 (input_edge): Do not stream frequency.
1353 (merge_profile_summaries): Scale only IPA profiles.
1354 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1355 * predict.c (drop_profile): Do not recompute frequency.
1356 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1357 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1358 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1359 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1360 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1361 * tree-inline.c (copy_bb): Do not scale frequency.
1362 (expand_call_inline): Do not scale frequency.
1363 (tree_function_versioning): Do not scale frequency.
1364 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1366 2017-11-10 Julia Koval <julia.koval@intel.com>
1368 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1369 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1370 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1371 (_mm256_maskz_gf2p8affine_epi64_epi8)
1372 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1373 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1374 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1375 (__builtin_ia32_vgf2p8affineqb_v32qi)
1376 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1377 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1379 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1383 (arm_option_override): Refactor.
1384 (arm_option_reconfigure_globals): New.
1385 (arm_options_perform_arch_sanity_checks): New.
1386 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1388 (arm_options_perform_arch_sanity_checks): Likewise
1390 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1392 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1393 (power9-qpmul): New.
1394 * rs6000/rs6000.md ("type" attr): Add qmul.
1395 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1396 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1397 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1399 2017-11-10 Martin Sebor <msebor@redhat.com>
1402 * builtins.c (compute_objsize): Handle arrays that
1403 compute_builtin_object_size likes to fail for. Make extern.
1404 * builtins.h (compute_objsize): Declare.
1405 (check_strncpy_sizes): New function.
1406 (expand_builtin_strncpy): Call check_strncpy_sizes.
1407 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1408 -Wstringop-truncation.
1409 (gimple_fold_builtin_strncat): Same.
1410 * gimple.c (gimple_build_call_from_tree): Set call location.
1411 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1412 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1413 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1414 (handle_builtin_strlen): Use strlen_to_stridx.
1415 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1417 Use strlen_to_stridx.
1418 (pass_strlen::execute): Release strlen_to_stridx.
1419 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1420 (-Wstringop-truncation): Document new option.
1422 2017-11-10 Martin Liska <mliska@suse.cz>
1424 PR gcov-profile/82702
1425 * gcov.c (main): Handle intermediate files in a different
1427 (get_gcov_intermediate_filename): New function.
1428 (output_gcov_file): Remove support of intermediate files.
1429 (generate_results): Allocate intermediate file.
1430 (release_structures): Clean-up properly fn_end.
1431 (output_intermediate_file): Start iterating with line 1.
1433 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1435 PR tree-optimization/82929
1436 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1437 ops_swapped_p non-static data member.
1438 (store_immediate_info::store_immediate_info): Clear it.
1439 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1440 ops set ops_swapped_p.
1441 (count_multiple_uses): Handle ops_swapped_p.
1443 2017-11-10 Martin Liska <mliska@suse.cz>
1445 * coverage.c (coverage_init): Stream information about
1446 support of has_unexecuted_blocks.
1447 * doc/gcov.texi: Document that.
1448 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1449 * gcov.c (read_graph_file): Likewise.
1450 (output_line_beginning): Fix a small issue with
1453 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1455 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1456 reference of trivial component.
1458 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1461 * gimple-ssa-store-merging.c
1462 (pass_store_merging::terminate_all_aliasing_chains): For
1463 gimple_store_p stmts also call refs_output_dependent_p.
1465 PR rtl-optimization/82913
1466 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1469 2017-11-09 Jeff Law <law@redhat.com>
1471 * vr-values.h: New file with vr_values class.
1472 * tree-vrp.c: Include vr-values.h
1473 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1474 data objects into the vr_values class.
1475 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1476 (get_value_range): Make it a member function within vr_values class.
1477 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1478 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1479 (op_with_constant_singleton_value_range): Likewise.
1480 (extract_range_for_var_from_comparison_expr): Likewise.
1481 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1482 (extract_range_from_binary_expr): Likewise.
1483 (extract_range_from_unary_expr): Likewise.
1484 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1485 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1486 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1487 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1488 (compare_name_with_value, compare_names): Likewise.
1489 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1490 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1491 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1492 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1493 (extract_range_from_phi_node): Likewise.
1494 (simplify_truth_ops_using_ranges): Likewise.
1495 (simplify_div_or_mod_using_ranges): Likewise.
1496 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1497 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1498 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1499 (simplify_float_conversion_using_ranges): Likewise.
1500 (simplify_internal_call_using_ranges): Likewise.
1501 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1502 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1503 poitner via x_vr_values for calls into gimple folder.
1504 (vrp_initialize_lattice): Make this the vr_values ctor.
1505 (vrp_free_lattice): Make this the vr_values dtor.
1506 (set_vr_value): New function.
1507 (class vrp_prop): Add vr_values data member. Add various member
1508 functions as well as member functions that delegate to vr_values.
1509 (check_array_ref): Make a member function within vrp_prop class.
1510 (search_for_addr_array, vrp_initialize): Likewise.
1511 (vrp_finalize): Likewise. Revamp to avoid direct access to
1512 vr_value, values_propagated, etc.
1513 (check_array_bounds): Extract vrp_prop class instance pointer from
1514 walk info structure. Use it to call member functions.
1515 (check_all_array_refs): Make a member function within vrp_prop class.
1516 Smuggle class instance pointer via walk info structure.
1517 (x_vr_values): New local static.
1518 (vrp_valueize): Use x_vr_values to get class instance.
1519 (vr_valueize_1): Likewise.
1520 (class vrp_folder): Add vr_values data member. Add various member
1521 functions as well as member functions that delegate to vr_values.
1522 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1523 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1524 class instance from vr_values. Use it to call member functions.
1525 (vrp_dom_walker): Add vr_values data member.
1526 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1527 instance via x_vr_values.
1528 (identify_jump_threads): Accept vr_values as argument. Store
1529 it into the walker structure.
1530 (evrp_dom_walker): Add vr_values class data member. Add various
1532 (evrp_dom_walker::try_find_new_range): Use vr_values data
1533 member to access the memory allocator.
1534 (evrp_dom_walker::before_dom_children): Store vr_values class
1535 instance into the vrp_folder class.
1536 (evrp_dom_walker::push_value_range): Rework to avoid direct
1537 access to num_vr_values and vr_value.
1538 (evrp_dom_walker::pop_value_range): Likewise.
1539 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1540 Use vr_values to get to dump_all_value_ranges member function.
1541 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1542 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1543 Pass vr_values class instance down to identify_jump_threads.
1544 Remove call to vrp_free_lattice.
1545 (debug_all_value_ranges): Remove.
1547 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1548 (vrp_folder): Likewise.
1550 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1551 Get it from the existing bitmap instead.
1552 (vrp_intersect_ranges_1): Likewise.
1554 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1556 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1558 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1559 set bit_not_p to it.
1560 (imm_store_chain_info::coalesce_immediate_stores): Break group
1561 if bit_not_p is different.
1562 (count_multiple_uses, split_group,
1563 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1564 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1565 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1566 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1567 to store_immediate_info ctor.
1569 2017-11-09 Jim Wilson <jimw@sifive.com>
1571 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1572 (scan_prog_file): Likewise.
1574 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1576 * bb-reorder.c (max_entry_frequency): Remove.
1577 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1578 connect_traces, push_to_next_round_p): Remove prototypes.
1579 (find_traces_1_round): Use counts only.
1580 (push_to_next_round_p): Likewise.
1581 (find_traces): Likewise.
1582 (rotate_loop): Likewise.
1583 (find_traces_1_round): Likewise.
1584 (connect_traces): Likewise.
1585 (edge_order): Likewise.
1587 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1589 * config/arm/arm.c (output_return_instruction): Add comments to
1590 indicate requirement for cmse_nonsecure_entry return to account
1591 for the size of clearing instruction output here.
1592 (thumb_exit): Likewise.
1593 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1594 return in hardfloat mode.
1596 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1598 * config/rs6000/rs6000.c (machine_function): Add a bool,
1599 "toc_is_wrapped_separately".
1600 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1601 if it wasn't explicitly set or unset, we are optimizing for speed, and
1602 doing separate shrink-wrapping.
1603 (rs6000_get_separate_components): Enable the TOC component if
1604 saving the TOC register in the prologue.
1605 (rs6000_components_for_bb): Handle the TOC component.
1606 (rs6000_emit_prologue_components): Store the TOC register where needed.
1607 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1608 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1610 2017-11-09 Martin Jambor <mjambor@suse.cz>
1612 * ipa-param-manipulation.c: New file.
1613 * ipa-param-manipulation.h: Likewise.
1614 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1615 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1616 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1617 (ipa_parm_adjustment): Likewise.
1618 (ipa_parm_adjustment_vec): Likewise.
1619 (ipa_get_vector_of_formal_parms): Moved declaration to
1620 ipa-param-manipulation.h.
1621 (ipa_get_vector_of_formal_parm_types): Likewise.
1622 (ipa_modify_formal_parameters): Likewise.
1623 (ipa_modify_call_arguments): Likewise.
1624 (ipa_combine_adjustments): Likewise.
1625 (ipa_dump_param_adjustments): Likewise.
1626 (ipa_modify_expr): Likewise.
1627 (ipa_get_adjustment_candidate): Likewise.
1628 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1629 ipa-param-manipulation.c.
1630 (ipa_get_vector_of_formal_parm_types): Likewise.
1631 (ipa_modify_formal_parameters): Likewise.
1632 (ipa_modify_call_arguments): Likewise.
1633 (ipa_modify_expr): Likewise.
1634 (get_ssa_base_param): Likewise.
1635 (ipa_get_adjustment_candidate): Likewise.
1636 (index_in_adjustments_multiple_times_p): Likewise.
1637 (ipa_combine_adjustments): Likewise.
1638 (ipa_dump_param_adjustments): Likewise.
1639 * tree-sra.c: Also include ipa-param-manipulation.h
1640 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1643 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1644 Alan Hayward <alan.hayward@arm.com>
1645 David Sherwood <david.sherwood@arm.com>
1647 * doc/sourcebuild.texi (vect_masked_store): Document.
1649 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1650 Alan Hayward <alan.hayward@arm.com>
1651 David Sherwood <david.sherwood@arm.com>
1653 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1655 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1656 Alan Hayward <alan.hayward@arm.com>
1657 David Sherwood <david.sherwood@arm.com>
1659 * doc/sourcebuild.texi (vect_variable_length): Document.
1661 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1662 Alan Hayward <alan.hayward@arm.com>
1663 David Sherwood <david.sherwood@arm.com>
1665 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1667 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1668 Alan Hayward <alan.hayward@arm.com>
1669 David Sherwood <david.sherwood@arm.com>
1671 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1673 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1674 Alan Hayward <alan.hayward@arm.com>
1675 David Sherwood <david.sherwood@arm.com>
1677 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1678 previously undocumented selectors.
1679 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1681 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1683 * doc/rtl.texi (const_vector): Say that elements can be
1684 const_wide_ints too.
1685 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1686 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1687 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1688 * optabs.c (expand_vector_broadcast): Likewise.
1690 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1691 Alan Hayward <alan.hayward@arm.com>
1692 David Sherwood <david.sherwood@arm.com>
1694 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1695 scaled index even if the unscaled address was invalid.
1696 Don't increase the complexity of using a scale in that case.
1698 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1699 Alan Hayward <alan.hayward@arm.com>
1700 David Sherwood <david.sherwood@arm.com>
1702 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1703 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1704 UNITS_PER_WORD bytes.
1705 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1706 based on REGMODE_NATURAL_SIZE of the inner mode.
1707 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1708 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1709 * expmed.c (lowpart_bit_field_p): Divide the value up into
1710 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1711 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1712 whether something is likely to occupy more than one register.
1714 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1717 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1718 profile_count::adjust_for_ipa_scaling.
1719 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1720 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1722 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1724 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1726 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1727 (split_group): Add total_orig and total_new arguments, estimate the
1728 number of statements related to the store group without store merging
1729 and with store merging.
1730 (imm_store_chain_info::output_merged_store): Adjust split_group
1731 callers, punt if estimated number of statements with store merging
1732 is not smaller than estimated number of statements without it.
1734 (handled_load): Remove has_single_use checks.
1735 (pass_store_merging::process_store): Likewise.
1737 2017-11-09 Richard Biener <rguenther@suse.de>
1739 PR tree-optimization/82902
1740 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1742 2017-11-09 Martin Liska <mliska@suse.cz>
1745 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1748 2017-11-09 Martin Liska <mliska@suse.cz>
1750 PR tree-optimization/82669
1751 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1753 2017-11-09 Martin Liska <mliska@suse.cz>
1755 PR gcov-profile/48463
1756 * coverage.c (coverage_begin_function): Output also end locus
1757 of a function and information whether the function is
1759 * gcov-dump.c (tag_function): Parse and print the information.
1760 * gcov.c (INCLUDE_MAP): Add include.
1761 (INCLUDE_SET): Likewise.
1762 (struct line_info): Move earlier in the source file because
1763 of vector<line_info> in function_info structure.
1764 (line_info::line_info): Likewise.
1765 (line_info::has_block): Likewise.
1766 (struct source_info): Add new member index.
1767 (source_info::get_functions_at_location): New function.
1768 (function_info::group_line_p): New function.
1769 (output_intermediate_line): New function.
1770 (output_intermediate_file): Use the mentioned function.
1771 (struct function_start): New.
1772 (struct function_start_pair_hash): Likewise.
1773 (process_file): Add code that identifies group functions.
1774 Assign lines either to global or function scope.
1775 (generate_results): Skip artificial functions.
1776 (find_source): Assign index for each source file.
1777 (read_graph_file): Read new flag artificial and end_line.
1778 (add_line_counts): Assign it either to global of function scope.
1779 (accumulate_line_counts): Isolate core of the function to
1780 accumulate_line_info and call it for both function and global
1782 (accumulate_line_info): New function.
1783 (output_line_beginning): Fix GNU coding style.
1784 (print_source_line): New function.
1785 (output_line_details): Likewise.
1786 (output_function_details): Likewise.
1787 (output_lines): Iterate both source (global) scope and function
1789 (struct function_line_start_cmp): New class.
1790 * doc/gcov.texi: Reflect changes in documentation.
1792 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1795 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1796 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1797 and similarly for not instead of neg.
1799 2017-11-08 Andi Kleen <ak@linux.intel.com>
1801 * config/i386/i386.opt: Add -mforce-indirect-call.
1802 * config/i386/predicates.md: Check for flag_force_indirect_call.
1803 * doc/invoke.texi: Document -mforce-indirect-call
1805 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1807 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1809 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1810 config/riscv/riscv.c (predict.h): New include.
1811 (riscv_slow_unaligned_access_p): No longer static.
1812 (riscv_block_move_straight): Add require.
1813 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1815 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1818 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1819 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1820 nonimmediate_operand predicate for operand 1 instead of
1823 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1825 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1827 * config/aarch64/constraints.md (Uml): New constraint.
1828 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1831 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1833 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1834 of two vec_duplicates into a vec_concat.
1836 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1838 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1839 Simplify vec_merge of vec_duplicate and vec_concat.
1840 * config/aarch64/constraints.md (Utq): New constraint.
1841 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1844 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1846 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1847 Simplify vec_merge of vec_duplicate and const_vector.
1848 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1850 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1851 mode iterator. Update predicate on operand 1 to
1852 handle non-const_vec constants. Delete constraints.
1853 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1855 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1857 PR tree-optimization/78821
1858 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1860 (store_operand_info::store_operand_info): Initialize it to false.
1861 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1862 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1863 store in the group, and if chain_info is non-NULL, to ignore altogether
1865 (compatible_load_p): Fail if bit_not_p does not match.
1866 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1867 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1868 (pass_store_merging::process_store): Adjust
1869 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1870 call terminate_all_aliasing_chains newly when adding a store into
1871 a chain with non-NULL chain_info.
1873 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1875 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1877 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1879 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1881 (aarch64_layout_frame): Initialise emit_frame_chain.
1882 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1883 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1885 2017-11-08 Martin Liska <mliska@suse.cz>
1887 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1888 of gimple_call_internal_p.
1890 2017-11-07 Tom de Vries <tom@codesourcery.com>
1892 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1894 2017-11-07 Tom de Vries <tom@codesourcery.com>
1896 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1899 2017-11-08 Martin Liska <mliska@suse.cz>
1902 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1904 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1906 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1908 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1911 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1912 (*cmp<mode>_ccz_1): New insn with $k alternative.
1915 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1916 RTX_COMM_COMPARE as commutative as well.
1917 (ix86_binary_operator_ok): Formatting fix.
1918 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1919 *<code><mode>3<mask_name><round_saeonly_name>,
1920 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1921 *<s>mul<mode>3_highpart<mask_name>,
1922 *vec_widen_umult_even_v16si<mask_name>,
1923 *vec_widen_umult_even_v8si<mask_name>,
1924 *vec_widen_umult_even_v4si<mask_name>,
1925 *vec_widen_smult_even_v16si<mask_name>,
1926 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1927 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1928 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1929 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1930 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1931 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1932 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1933 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1934 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1935 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1936 ix86_binary_operator_ok. Formatting fixes.
1937 (*<plusminus_insn><mode>3<mask_name><round_name>,
1938 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1941 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1943 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1944 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1945 eq and ne if TARGET_ISEL.
1946 (cmp): New code_iterator.
1947 (UNS, UNSU_, UNSIK): New code_attrs.
1948 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1949 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1951 ("eq<mode>3"): ... this.
1952 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1954 ("ne<mode>3"): ... this.
1956 2017-11-07 Julia Koval <julia.koval@intel.com>
1959 * common/config/i386/i386-common.c
1960 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1961 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1962 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1963 * config/i386/i386.opt: Ditto.
1964 * config/i386/i386.c (ix86_target_string): Ditto.
1965 (ix86_option_override_internal): Ditto.
1966 (ix86_init_mpx_builtins): Move MPX to args2.
1967 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1968 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1969 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1970 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1971 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1972 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1973 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1975 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1978 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1979 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1980 (zero-extendsidi peephole2): Remove peephole.
1982 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1985 * stor-layout.c: Include attribs.h.
1986 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1987 explicit lookup of "aligned" attribute.
1989 2017-11-07 Andrew Waterman <andrew@sifive.com>
1991 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1992 (riscv_expand_block_move): Likewise.
1993 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1995 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1996 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1997 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1998 (riscv_adjust_block_mem): Likewise.
1999 (riscv_block_move_loop): Likewise.
2000 (riscv_expand_block_move): Likewise.
2001 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2003 2017-11-07 Michael Clark <michaeljclark@mac.com>
2005 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2006 (MUSL_DYNAMIC_LINKER): Likewise.
2008 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2010 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2011 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2014 2017-11-07 Richard Biener <rguenther@suse.de>
2016 * match.pd: Fix build.
2018 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2019 Jackson Woodruff <jackson.woodruff@arm.com>
2021 PR tree-optimization/71026
2022 * match.pd: Canonicalize negate in division.
2024 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2027 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2029 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2031 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2032 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2034 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2036 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2037 non-scalar integral types.
2038 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2039 (-(A-B), -(~A)): New transformations.
2041 2017-11-07 Tom de Vries <tom@codesourcery.com>
2043 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2044 semicolon after "do {} while (0)".
2045 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2046 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2047 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2048 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2049 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2051 2017-11-07 Tom de Vries <tom@codesourcery.com>
2053 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2054 after "do {} while (0)".
2055 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2056 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2057 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2058 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2059 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2061 2017-11-07 Tom de Vries <tom@codesourcery.com>
2064 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2066 (arm_rtx_costs_internal): Add missing semicolon after
2067 HANDLE_NARROW_SHIFT_ARITH call.
2069 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2071 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2072 disable isel if it was not set explicitly.
2074 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2076 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
2078 (add_type_attribute) likewise.
2080 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2082 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2083 of struct ix86_frame.
2084 (ix86_initial_elimination_offset): Likewise.
2085 (ix86_expand_split_stack_prologue): Likewise.
2087 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2089 * tree-vrp.h (enum value_range_type): Update stale comment.
2091 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2092 Alan Hayward <alan.hayward@arm.com>
2093 David Sherwood <david.sherwood@arm.com>
2095 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2096 (aarch64_expand_vec_perm_const): Take the number of units too.
2097 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2098 (aarch64_expand_vec_perm_const): Likewise.
2099 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2100 (vec_perm<mode>): Update accordingly.
2102 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2103 Alan Hayward <alan.hayward@arm.com>
2104 David Sherwood <david.sherwood@arm.com>
2106 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2107 Take the number of units too.
2108 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2109 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2110 but check for a vector mode before rather than after the call.
2111 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2112 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2113 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2114 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2115 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2116 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2117 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2118 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2119 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2120 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2121 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2122 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2123 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2124 (aarch64_sqdmull2_n<mode>): Update accordingly.
2126 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2127 Alan Hayward <alan.hayward@arm.com>
2128 David Sherwood <david.sherwood@arm.com>
2130 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2131 the number of units too.
2132 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2133 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2134 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2135 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2136 (vec_store_lanesxi<mode>): Update accordingly.
2138 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2139 Alan Hayward <alan.hayward@arm.com>
2140 David Sherwood <david.sherwood@arm.com>
2142 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2143 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2144 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2145 of units rather than the mode.
2146 * config/aarch64/iterators.md (nunits): New mode attribute.
2147 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2148 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2149 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2150 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2151 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2152 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2153 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2154 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2155 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2156 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2157 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2158 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2159 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2160 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2161 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2162 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2163 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2164 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2165 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2166 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2167 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2168 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2169 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2170 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2171 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2172 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2173 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2174 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2175 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2176 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2177 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2178 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2179 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2180 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2181 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2182 (aarch64_simd_vec_setv2di): Likewise.
2184 2017-11-06 Carl Love <cel@us.ibm.com>
2186 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2188 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2189 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2191 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2193 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2194 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2195 Add power 8 macro expansions.
2196 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2197 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2198 power 8 instructions. (VSX_XXBR): Add iterator.
2200 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2202 * config/arm/arm.md (predicable_short_it): Change default to "no",
2203 improve documentation, remove uses that are identical to the default.
2204 (enabled_for_depr_it): Rename to enabled_for_short_it.
2205 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2206 Remove default uses.
2207 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2208 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2209 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2210 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2212 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2215 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2216 float128 helper macros, which are no longer used after deleting
2217 the old 'q' built-in functions, and moving the round to odd
2218 built-in functions to being special built-in functions.
2219 (BU_FLOAT128_2): Likewise.
2220 (BU_FLOAT128_1_HW): Likewise.
2221 (BU_FLOAT128_2_HW): Likewise.
2222 (BU_FLOAT128_3_HW): Likewise.
2223 (FABSQ): Delete old 'q' built-in functions.
2224 (COPYSIGNQ): Likewise.
2225 (SQRTF128_ODD): Move round to odd built-in functions to be
2226 special built-in functions, so that we can handle
2227 -mabi=ieeelongdouble.
2228 (TRUNCF128_ODD): Likewise.
2229 (ADDF128_ODD): Likewise.
2230 (SUBF128_ODD): Likewise.
2231 (MULF128_ODD): Likewise.
2232 (DIVF128_ODD): Likewise.
2233 (FMAF128_ODD): Likewise.
2234 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2235 built-in names to 'f128'.
2236 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2237 old 'q' built-in functions, as the machine independent code for
2238 'f128' built-in functions handles this.
2239 (rs6000_expand_builtin): Add expansion for float128 round to odd
2240 functions, keying off on -mabi=ieeelongdouble of whether to use
2241 the KFmode or TFmode variant.
2242 (rs6000_init_builtins): Initialize the _Float128 round to odd
2244 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2245 _Float128 'q' built-in functions are now mapped into the new
2246 'f128' built-in functions.
2248 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2250 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2251 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2253 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2256 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2257 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2258 * toplev.c: Include "ipa-fnsummary.h".
2259 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2261 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2263 PR tree-optimization/82838
2264 * gimple-ssa-store-merging.c
2265 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2266 on a separate gimple_seq which is then appended to seq.
2268 2017-11-06 Jeff Law <law@redhat.com>
2271 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2272 (get_probe_interval): New functions.
2273 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2274 (ix86_adjust_stack_and_probe): Likewise.
2275 (output_adjust_stack_and_probe): Likewise.
2276 (ix86_emit_probe_stack_range): Likewise.
2277 (ix86_expand_prologue): Likewise.
2279 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2281 PR tree-optimization/82816
2282 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2283 if the modes of the two types are the same.
2284 (convert_plusminus_to_widen): Likewise.
2286 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2288 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2290 (usadv16qi): New define_expand.
2291 (usadv8hi): New define_expand.
2293 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2296 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2298 (inline_transform): Likewise.
2299 * predict.c (counts_to_freqs): Remove useless conditional.
2300 * profile-count.h (profile_count::apply_scale): Move sanity check.
2301 * tree-inline.c (copy_bb): Always scale.
2302 (copy_cfg_body): Likewise.
2304 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2307 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2309 (*cmp_ite0): Add enabled_for_depr_it attribute.
2310 (*cmp_ite1): Likewise.
2312 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2314 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2317 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2319 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2320 return the value by pointer.
2321 (extract_range_from_multiplicative_op_1): Update accordingly.
2322 Return as soon as an operation fails.
2324 2017-11-05 Tom de Vries <tom@codesourcery.com>
2327 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2328 (DEF_SANITIZER_BUILTIN): ... here.
2329 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2330 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2332 2017-11-05 Tom de Vries <tom@codesourcery.com>
2335 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2337 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2338 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2339 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2341 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2342 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2343 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2345 2017-11-05 Tom de Vries <tom@codesourcery.com>
2348 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2351 2017-11-04 Michael Clark <michaeljclark@mac.com>
2353 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2354 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2356 (*addsi3_extended): Likewise.
2357 (*addsi3_extended2): Likewise.
2358 (<optab>si3): Likewise.
2359 (<optab>di3): Likewise.
2360 (<optab><mode>3): Likewise.
2361 (<*optabe>si3_internal): Likewise.
2362 (zero_extendqi<SUPERQI:mode>2): Likewise.
2363 (*add<mode>hi3): Likewise.
2364 (*xor<mode>hi3): Likewise.
2365 (<optab>di3): Likewise.
2366 (*<optab>si3_extend): Likewise.
2367 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2368 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2369 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2371 2017-11-04 Andrew Waterman <andrew@sifive.com>
2373 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2374 TARGET_STRICT_ALIGN based upon -mtune argument.
2376 2017-11-04 Andrew Waterman <andrew@sifive.com>
2378 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2380 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2382 config/i386/i386.c (choose_basereg): Use optional scratch
2383 register and add assertion.
2384 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2385 needed, and don't allocate stack.
2386 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2387 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2388 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2390 2017-11-03 Jeff Law <law@redhat.com>
2392 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2393 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2394 to probe at the start of a noreturn function.
2396 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2398 PR tree-optimization/78821
2399 * gimple-ssa-store-merging.c: Update the file comment.
2400 (MAX_STORE_ALIAS_CHECKS): Define.
2401 (struct store_operand_info): New type.
2402 (store_operand_info::store_operand_info): New constructor.
2403 (struct store_immediate_info): Add rhs_code and ops data members.
2404 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2405 and op1r arguments to the ctor, initialize corresponding data members.
2406 (struct merged_store_group): Add load_align_base and load_align
2408 (merged_store_group::merged_store_group): Initialize them.
2409 (merged_store_group::do_merge): Update them.
2410 (merged_store_group::apply_stores): Pick the constant for
2411 encode_tree_to_bitpos from one of the two operands, or skip
2412 encode_tree_to_bitpos if neither operand is a constant.
2413 (class pass_store_merging): Add process_store method decl. Remove
2414 bool argument from terminate_all_aliasing_chains method decl.
2415 (pass_store_merging::terminate_all_aliasing_chains): Remove
2416 var_offset_p argument and corresponding handling.
2417 (stmts_may_clobber_ref_p): New function.
2418 (compatible_load_p): New function.
2419 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2420 if there is overlap and rhs_code is not INTEGER_CST. For
2421 non-overlapping stores terminate group if rhs is not mergeable.
2422 (get_alias_type_for_stmts): Change first argument from
2423 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2424 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2425 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2427 (get_location_for_stmts): Change first argument from
2428 auto_vec<gimple *> & to vec<gimple *> &.
2429 (struct split_store): Remove orig_stmts data member, add orig_stores.
2430 (split_store::split_store): Create orig_stores rather than orig_stmts.
2431 (find_constituent_stmts): Renamed to ...
2432 (find_constituent_stores): ... this. Change second argument from
2433 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2434 to info structures rather than the statements.
2435 (split_group): Rename ALLOW_UNALIGNED argument to
2436 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2437 it. Adjust find_constituent_stores caller.
2438 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2439 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2440 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2441 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2442 (mem_valid_for_store_merging): New function.
2443 (handled_load): New function.
2444 (pass_store_merging::process_store): New method.
2445 (pass_store_merging::execute): Use process_store method. Adjust
2446 terminate_all_aliasing_chains caller.
2448 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2450 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2451 Return true for more constants, symbols and label references.
2452 (aarch64_valid_floating_const): Remove unused function.
2454 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2457 * config/aarch64/aarch64.c (aarch64_layout_frame):
2458 Undo forcing of LR at bottom of frame.
2460 2017-11-03 Jeff Law <law@redhat.com>
2463 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2464 for int_registers_saved.
2466 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2467 extracted from tree-ssa-dom.c.
2468 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2469 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2470 (record_equivalences_from_incoming_edge): Add additional argument
2471 to single_pred_edge_ignoring_loop_edges call.
2472 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2473 (uncprop_dom_walker::before_dom_children): Add additional argument
2474 to single_pred_edge_ignoring_loop_edges call.
2475 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2476 single_pred_edge_ignoring_loop_edges rather than open coding.
2477 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2479 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2481 * match.pd (-(-A)): Rewrite.
2483 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2485 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2486 (rs6000_emit_int_cmove): New declaration.
2487 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2488 (rs6000_emit_sISEL): Delete.
2489 (rs6000_emit_int_cmove): Make non-static.
2490 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2491 instead of rs6000_emit_sISEL.
2493 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2495 * asan.c (create_cond_insert_point): Maintain profile.
2496 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2498 * basic-block.h (struct basic_block_def): Remove frequency.
2499 (EDGE_FREQUENCY): Use to_frequency
2500 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2502 (find_traces): Update to use to_frequency.
2503 (find_traces_1_round): Likewise; use only IPA counts.
2504 (bb_to_key): Likewise.
2505 (connect_traces): Use IPA counts only.
2506 (copy_bb_p): Update to use to_frequency.
2507 (fix_up_crossing_landing_pad): Likewise.
2508 (sanitize_hot_paths): Likewise.
2509 * bt-load.c (basic_block_freq): Likewise.
2510 * cfg.c (init_flow): Set count_max to uninitialized.
2511 (check_bb_profile): Remove frequencies; check counts.
2512 (dump_bb_info): Do not dump frequencies.
2513 (update_bb_profile_for_threading): Update counts only.
2514 (scale_bbs_frequencies_int): Likewise.
2515 (MAX_SAFE_MULTIPLIER): Remove.
2516 (scale_bbs_frequencies_gcov_type): Update counts only.
2517 (scale_bbs_frequencies_profile_count): Update counts only.
2518 (scale_bbs_frequencies): Update counts only.
2519 * cfg.h (struct control_flow_graph): Add count-max.
2520 (update_bb_profile_for_threading): Update prototype.
2521 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2522 (find_many_sub_basic_blocks): Likewise.
2523 * cfgcleanup.c (try_forward_edges): Likewise.
2524 (try_crossjump_to_edge): Likewise.
2525 * cfgexpand.c (expand_gimple_cond): Likewise.
2526 (expand_gimple_tailcall): Likewise.
2527 (construct_init_block): Likewise.
2528 (construct_exit_block): Likewise.
2529 * cfghooks.c (verify_flow_info): Check consistency of counts.
2530 (dump_bb_for_graph): Do not dump frequencies.
2531 (split_block_1): Do not update frequencies.
2532 (split_edge): Do not update frequencies.
2533 (make_forwarder_block): Do not update frequencies.
2534 (duplicate_block): Do not update frequencies.
2535 (account_profile_record): Do not update frequencies.
2536 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2537 for global heuristics.
2538 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2539 (expected_loop_iterations_unbounded): Use counts only.
2540 * cfgloopmanip.c (scale_loop_profile): Simplify.
2541 (create_empty_loop_on_edge): Simplify
2543 (duplicate_loop_to_header_edge): Simplify
2544 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2545 (update_br_prob_note): Take care of removing note when profile
2547 (relink_block_chain): Do not dump frequency.
2548 (rtl_account_profile_record): Use to_frequency.
2549 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2550 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2551 (cgraph_update_edges_for_call_stmt_node): Likewise.
2552 (cgraph_edge::verify_count_and_frequency): Update.
2553 (cgraph_node::verify_node): Temporarily disable frequency verification.
2554 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2555 to_cgraph_frequency.
2556 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2557 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2559 (cgraph_node::expand_thunk): Update profile.
2560 * except.c (dw2_build_landing_pads): Do not update frequency.
2561 * final.c (compute_alignments): Use to_frequency.
2562 (dump_basic_block_info): Do not dump frequency.
2563 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2564 (dump_gimple_bb_header): Do not dump frequency.
2565 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2567 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2568 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2569 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2570 (init_before_recovery): Do not update frequency.
2571 (sched_create_recovery_edges): Do not update frequency.
2572 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2573 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2574 (ipa_cp_c_finalize): Set max_count to uninitialized.
2575 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2576 (param_change_prob): Use counts.
2577 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2579 * ipa-split.c (consider_split): Use to_frequency.
2580 (split_function): Use to_frequency.
2581 * ira-build.c (loop_compare_func): Likewise.
2582 (mark_loops_for_removal): Likewise.
2583 (mark_all_loops_for_removal): Likewise.
2584 * loop-doloop.c (doloop_modify): Do not update frequency.
2585 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2587 * lto-streamer-in.c (input_function): Update count_max.
2588 * omp-expand.c (expand_omp_taskreg): Update count_max.
2589 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2590 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2591 (maybe_hot_count_p): Use ipa counts only.
2592 (maybe_hot_bb_p): Simplify.
2593 (maybe_hot_edge_p): Simplify.
2594 (probably_never_executed): Do not take frequency argument.
2595 (probably_never_executed_bb_p): Do not pass frequency.
2596 (probably_never_executed_edge_p): Likewise.
2597 (combine_predictions_for_bb): Check that profile is nonzero.
2598 (propagate_freq): Do not set frequency.
2599 (drop_profile): Simplify.
2600 (counts_to_freqs): Simplify.
2601 (expensive_function_p): Use to_frequency.
2602 (propagate_unlikely_bbs_forward): Simplify.
2603 (determine_unlikely_bbs): Simplify.
2604 (estimate_bb_frequencies): Add hack to silence graphite issues.
2605 (compute_function_frequency): Use ipa counts.
2606 (pass_profile::execute): Update.
2607 (rebuild_frequencies): Use counts only.
2608 (force_edge_cold): Use counts only.
2609 * profile-count.c (profile_count::dump): Dump new count types.
2610 (profile_count::differs_from_p): Check compatiblity.
2611 (profile_count::to_frequency): New function.
2612 (profile_count::to_cgraph_frequency): New function.
2613 * profile-count.h (struct function): Declare.
2614 (enum profile_quality): Add profile_guessed_local and
2615 profile_guessed_global0.
2616 (class profile_proability): Decrease number of bits to 29;
2617 update from_reg_br_prob_note and to_reg_br_prob_note.
2618 (class profile_count: Update comment; decrease number of bits
2619 to 61. Check compatibility.
2620 (profile_count::compatible_p): New private member function.
2621 (profile_count::ipa_p): New member function.
2622 (profile_count::operator<): Handle global zero correctly.
2623 (profile_count::operator>): Handle global zero correctly.
2624 (profile_count::operator<=): Handle global zero correctly.
2625 (profile_count::operator>=): Handle global zero correctly.
2626 (profile_count::nonzero_p): New member function.
2627 (profile_count::force_nonzero): New member function.
2628 (profile_count::max): New member function.
2629 (profile_count::apply_scale): Handle IPA scalling.
2630 (profile_count::guessed_local): New member function.
2631 (profile_count::global0): New member function.
2632 (profile_count::ipa): New member function.
2633 (profile_count::to_frequency): Declare.
2634 (profile_count::to_cgraph_frequency): Declare.
2635 * profile.c (OVERLAP_BASE): Delete.
2636 (compute_frequency_overlap): Delete.
2637 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2638 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2639 * sched-ebb.c (rank): Use counts only.
2640 * shrink-wrap.c (handle_simple_exit): Use counts only.
2641 (try_shrink_wrapping): Use counts only.
2642 (place_prologue_for_one_component): Use counts only.
2643 * tracer.c (find_best_predecessor): Use to_frequency.
2644 (find_trace): Use to_frequency.
2645 (tail_duplicate): Use to_frequency.
2646 * trans-mem.c (expand_transaction): Do not update frequency.
2647 * tree-call-cdce.c: Do not update frequency.
2648 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2649 (gimple_merge_blocks): Likewise.
2650 (gimple_split_edge): Likewise.
2651 (gimple_duplicate_sese_region): Likewise.
2652 (gimple_duplicate_sese_tail): Likewise.
2653 (move_sese_region_to_fn): Likewise.
2654 (gimple_account_profile_record): Likewise.
2655 (insert_cond_bb): Likewise.
2656 * tree-complex.c (expand_complex_div_wide): Likewise.
2657 * tree-eh.c (lower_resx): Update profile.
2658 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2660 (initialize_cfun): Do not initialize frequencies
2661 (freqs_to_counts): Delete.
2662 (copy_cfg_body): Ignore count parameter.
2663 (copy_body): Update.
2664 (expand_call_inline): Update count_max.
2665 (optimize_inline_calls): Update count_max.
2666 (tree_function_versioning): Update count_max.
2667 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2668 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2670 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2671 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2672 (try_peel_loop): Likewise.
2673 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2675 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2676 (tree_transform_and_unroll_loop): Do not use frequencies
2677 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2678 Use reliable prediction only.
2679 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2680 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2681 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2682 probability scaling.
2683 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2684 not update frequency
2685 (any_remaining_duplicated_blocks): Likewise.
2686 (update_profile): Likewise.
2687 (estimated_freqs_path): Delete.
2688 (freqs_to_counts_path): Delete.
2689 (clear_counts_path): Delete.
2690 (ssa_fix_duplicate_block_edges): Likewise.
2691 (duplicate_thread_path): Likewise.
2692 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2693 * tree-tailcall.c (decrease_profile): Do not update frequency.
2694 (eliminate_tail_call): Likewise.
2695 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2696 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2697 (optimize_mask_stores): Likewise.
2698 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2699 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2700 (ubsan_expand_ptr_ifn): Update profile.
2701 * value-prof.c (gimple_ic): Simplify.
2702 * value-prof.h (gimple_ic): Update prototype.
2703 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2704 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2706 (want_inline_self_recursive_call_p): Likewise.
2707 (resolve_noninline_speculation): Only cummulate defined counts.
2708 (inline_small_functions): Use nonzero_p.
2709 (ipa_inline): Do not access freed node.
2711 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2713 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2714 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2716 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2718 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2719 non-legitimate address.
2721 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2723 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2724 (*lt0_<mode>di, *lt0_<mode>si): New.
2726 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2728 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2729 TARGET_PAIRED_FLOAT.
2731 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2732 Jim Wilson <jim.wilson@linaro.org>
2734 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2735 * config/aarch64/aarch64-tune.md: Regenerated.
2736 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2737 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2739 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2741 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2742 (arc_expand_prologue): Restore blink for millicode.
2743 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2745 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2748 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2749 gen_vec_duplicate after forcing the scalar into a register.
2751 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2753 * combine (try_combine): Print the insns input to try_combine to the
2756 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2759 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2760 Remove second argument from aarch64_process_target_attr call.
2761 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2763 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2765 (aarch64_handle_attr_arch): Remove second argument.
2766 (aarch64_handle_attr_cpu): Ditto.
2767 (aarch64_handle_attr_tune): Ditto.
2768 (aarch64_handle_attr_isa_flags): Ditto.
2769 (aarch64_process_one_target_attr): Ditto.
2770 (aarch64_process_target_attr): Ditto.
2771 (aarch64_option_valid_attribute_p): Remove second argument.
2772 on aarch64_process_target_attr call.
2774 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2776 * diagnostic.c: Include "selftest-diagnostic.h".
2777 (selftest::assert_location_text): New function.
2778 (selftest::test_diagnostic_get_location_text): New function.
2779 (selftest::diagnostic_c_tests): Call it.
2781 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2783 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2784 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2785 (class selftest::test_diagnostic_context): Move to...
2786 * selftest-diagnostic.c: New file.
2787 * selftest-diagnostic.h: New file.
2789 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2791 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2792 offset range for FT32B.
2793 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2794 * config/ft32/ft32.md: Add TARGET_NOPM.
2795 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2796 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2798 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2800 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2802 2017-11-02 Jeff Law <law@redhat.com>
2804 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2805 virtual keyword on FINAL OVERRIDE members.
2807 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2808 virtuals together. Add virtual destructor.
2809 (substitute_and_fold_engine): Similarly.
2811 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2813 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2815 2017-11-02 Richard Biener <rguenther@suse.de>
2817 PR tree-optimization/82795
2818 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2820 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2822 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2824 (gcc_GAS_CHECK_FEATURE): Remove.
2825 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2826 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2827 * configure: Regenerate.
2829 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2831 * config/arc/arc.c (hwloop_optimize): Account for empty
2834 2017-11-02 Richard Biener <rguenther@suse.de>
2837 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2838 Truncate ARRAY_REF index and element size.
2840 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2842 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2844 2017-11-01 Jeff Law <law@redhat.com>
2846 * tree-ssa-ccp.c (ccp_folder): New class derived from
2847 substitute_and_fold_engine.
2848 (ccp_folder::get_value): New member function.
2849 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2850 (ccp_fold_stmt): Remove prototype.
2851 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2852 * tree-ssa-copy.c (copy_folder): New class derived from
2853 substitute_and_fold_engine.
2854 (copy_folder::get_value): Renamed from get_value.
2855 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2856 * tree-vrp.c (vrp_folder): New class derived from
2857 substitute_and_fold_engine.
2858 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2859 (vrp_folder::get_value): New member function.
2860 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2861 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2862 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2863 provide a class interface to folder/substitute routines.
2864 (ssa_prop_fold_stmt_fn): Remove typedef.
2865 (ssa_prop_get_value_fn): Likewise.
2866 (subsitute_and_fold): Remove prototype.
2867 (replace_uses_in): Likewise.
2868 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2869 Renamed from replace_uses_in. Call the virtual member function
2870 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2871 (substitute_and_fold_dom_walker): Remove initialization of
2872 data member entries for calbacks. Add substitute_and_fold_engine
2873 member and initialize it.
2874 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2875 member functions for get_value, replace_phi_args_in c
2876 replace_uses_in, and fold_stmt calls.
2877 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2878 substitute_and_fold. Remove assert. Update ctor call.
2880 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2881 (ssa_prop_visit_phi_fn): Likewise.
2882 (class ssa_propagation_engine): New class to provide an interface
2884 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2886 (ssa_prop_visit_phi): Likewise.
2887 (ssa_propagation_engine::simulate_stmt): Moved into class.
2888 Call visit_phi/visit_stmt from the class rather than via
2889 file scoped static variables.
2890 (ssa_propagation_engine::simulate_block): Moved into class.
2891 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2892 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2893 set file scoped statics for the visit_stmt/visit_phi callbacks.
2894 * tree-complex.c (complex_propagate): New class derived from
2895 ssa_propagation_engine.
2896 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2897 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2898 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2900 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2901 ssa_propagation_engine.
2902 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2903 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2904 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2905 * tree-ssa-copy.c (copy_prop): New class derived from
2906 ssa_propagation_engine.
2907 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2908 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2909 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2910 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2911 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2912 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2913 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2915 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2917 PR rtl-optimization/82778
2918 PR rtl-optimization/82597
2919 * compare-elim.c (struct comparison): Add in_a_setter field.
2920 (find_comparison_dom_walker::before_dom_children): Remove killed
2921 bitmap and df_simulate_find_defs call, instead walk the defs.
2922 Compute last_setter and initialize in_a_setter. Merge definitions
2923 with first initialization for a few variables.
2924 (try_validate_parallel): Use insn_invalid_p instead of
2925 recog_memoized. Return insn rather than just the pattern.
2926 (try_merge_compare): Fix up comment. Don't uselessly test if
2927 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2929 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2932 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2933 Alan Hayward <alan.hayward@arm.com>
2934 David Sherwood <david.sherwood@arm.com>
2936 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2937 aarch64_hard_regno_nregs to get the number of registers
2940 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2941 Alan Hayward <alan.hayward@arm.com>
2942 David Sherwood <david.sherwood@arm.com>
2944 * config/aarch64/constraints.md (Upl): Rename to...
2946 * config/aarch64/aarch64.md
2947 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2950 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2951 Alan Hayward <alan.hayward@arm.com>
2952 David Sherwood <david.sherwood@arm.com>
2954 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2955 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2958 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2959 Alan Hayward <alan.hayward@arm.com>
2960 David Sherwood <david.sherwood@arm.com>
2962 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2963 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2964 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2966 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2967 of a single element.
2968 * config/aarch64/iterators.md: Add a comment above the permute
2969 unspecs to say that they are generated directly by
2970 aarch64_expand_vec_perm_const.
2971 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2973 2017-11-01 Nathan Sidwell <nathan@acm.org>
2975 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2977 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2979 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2980 and medany code models, and describe what they do.
2982 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2984 Revert accidental duplicate:
2986 * combine.c (can_change_dest_mode): Reject changes in
2987 REGMODE_NATURAL_SIZE.
2989 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2991 PR rtl-optimization/64682
2992 PR rtl-optimization/69567
2993 PR rtl-optimization/69737
2994 PR rtl-optimization/82683
2995 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2996 register mentioned in the note, drop the note, unless it came from I3,
2997 in which case it should go to I3 again.
2999 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3001 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3002 and return false if not.
3003 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3005 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3007 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3008 range is known to be empty.
3010 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3011 Alan Hayward <alan.hayward@arm.com>
3012 David Sherwood <david.sherwood@arm.com>
3014 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3015 and CONST_VECTOR_NUNITS instead of computing the number of units from
3016 the byte sizes of the vector and element.
3017 (simplify_binary_operation_1): Likewise.
3018 (simplify_const_binary_operation): Likewise.
3019 (simplify_ternary_operation): Likewise.
3021 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3022 Alan Hayward <alan.hayward@arm.com>
3023 David Sherwood <david.sherwood@arm.com>
3025 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3026 (int_mem_offset): ...this new function.
3027 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3028 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3029 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3032 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3033 Alan Hayward <alan.hayward@arm.com>
3034 David Sherwood <david.sherwood@arm.com>
3036 * lower-subreg.c (interesting_mode_p): New function.
3037 (compute_costs, find_decomposable_subregs, decompose_register)
3038 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3039 (resolve_clobber, dump_choices): Use it.
3041 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3042 Alan Hayward <alan.hayward@arm.com>
3043 David Sherwood <david.sherwood@arm.com>
3045 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3047 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3048 Alan Hayward <alan.hayward@arm.com>
3049 David Sherwood <david.sherwood@arm.com>
3051 * alias.c (find_base_value, find_base_term): Only process integer
3052 truncations. Check the precision rather than the size.
3054 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3055 Alan Hayward <alan.hayward@arm.com>
3056 David Sherwood <david.sherwood@arm.com>
3058 * machmode.h (is_narrower_int_mode): New function
3059 * optabs.c (expand_float, expand_fix): Use it.
3060 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3062 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3063 Alan Hayward <alan.hayward@arm.com>
3064 David Sherwood <david.sherwood@arm.com>
3066 * rtl.h (narrower_subreg_mode): New function.
3067 * ira-color.c (update_costs_from_allocno): Use it.
3069 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3070 Alan Hayward <alan.hayward@arm.com>
3071 David Sherwood <david.sherwood@arm.com>
3073 * optabs-query.h (convert_optab_p): New function, split out from...
3074 (convert_optab_handler): ...here.
3075 (widening_optab_handler): Delete.
3076 (find_widening_optab_handler): Remove permit_non_widening parameter.
3077 (find_widening_optab_handler_and_mode): Likewise. Provide an
3078 override that operates on mode class wrappers.
3079 * optabs-query.c (widening_optab_handler): Delete.
3080 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3081 parameter. Assert that the two modes are the same class and that
3082 the "from" mode is narrower than the "to" mode. Use
3083 convert_optab_handler instead of widening_optab_handler.
3084 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3085 instead of widening_optab_handler.
3086 * expr.c (expand_expr_real_2): Update calls to
3087 find_widening_optab_handler.
3088 * optabs.c (expand_widen_pattern_expr): Likewise.
3089 (expand_binop_directly): Take the insn_code as a parameter.
3090 (expand_binop): Only call find_widening_optab_handler for
3091 conversion optabs; use optab_handler otherwise. Update calls
3092 to find_widening_optab_handler and expand_binop_directly.
3093 Use convert_optab_handler instead of widening_optab_handler.
3094 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3095 find_widening_optab_handler and use scalar_mode rather than
3097 (convert_plusminus_to_widen): Likewise.
3099 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3100 Alan Hayward <alan.hayward@arm.com>
3101 David Sherwood <david.sherwood@arm.com>
3103 * machmode.h (fixed_size_mode): New class.
3104 * rtl.h (get_pool_mode): Return fixed_size_mode.
3105 * gengtype.c (main): Add fixed_size_mode.
3106 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3107 (get_raw_arg_mode): Likewise.
3108 * doc/tm.texi: Regenerate.
3109 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3110 * targhooks.c (default_get_reg_raw_mode): Likewise.
3111 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3112 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3113 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3114 (msp430_get_raw_result_mode): Likewise.
3115 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3116 * dbxout.c (dbxout_parms): Require fixed-size modes.
3117 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3118 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3119 * omp-low.c (lower_oacc_reductions): Likewise.
3120 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3121 (simplify_subreg): Update accordingly.
3122 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3123 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3124 that aren't fixed-size.
3125 (get_pool_mode): Return a fixed_size_mode.
3126 (output_constant_pool_2): Take a fixed_size_mode.
3128 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3129 Alan Hayward <alan.hayward@arm.com>
3130 David Sherwood <david.sherwood@arm.com>
3132 * doc/rtl.texi (vec_series): Document.
3133 (const): Say that the operand can be a vec_series.
3134 * rtl.def (VEC_SERIES): New rtx code.
3135 * rtl.h (const_vec_series_p_1): Declare.
3136 (const_vec_series_p): New function.
3137 * emit-rtl.h (gen_const_vec_series): Declare.
3138 (gen_vec_series): Likewise.
3139 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3140 (gen_vec_series): Likewise.
3141 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3142 * simplify-rtx.c (simplify_unary_operation): Handle negations
3144 (simplify_binary_operation_series): New function.
3145 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3146 (test_vector_ops_series): New function.
3147 (test_vector_ops): Call it.
3148 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3149 gen_const_vec_series.
3150 (altivec_lvsr): Likewise.
3151 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3153 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3154 Alan Hayward <alan.hayward@arm.com>
3155 David Sherwood <david.sherwood@arm.com>
3157 * doc/rtl.texi (const): Update description of address constants.
3158 Say that vector constants are allowed too.
3159 * common.md (E, F): Use CONSTANT_P instead of checking for
3161 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3162 checking for CONST_VECTOR.
3163 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3165 * expr.c (expand_expr_real_2): Check for vector modes instead
3166 of checking for CONST_VECTOR.
3167 * rtl.h (const_vec_p): New function.
3168 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3169 (unwrap_const_vec_duplicate): Handle them here too.
3171 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3172 David Malcolm <dmalcolm@redhat.com>
3173 Alan Hayward <alan.hayward@arm.com>
3174 David Sherwood <david.sherwood@arm.com>
3176 * rtl.h (vec_duplicate_p): New function.
3177 * selftest-rtl.c (assert_rtx_eq_at): New function.
3178 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3179 (assert_rtx_eq_at): Declare.
3180 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3181 * selftest-run-tests.c (selftest::run_tests): Call it.
3182 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3183 (simplify_unary_operation_1): Recursively handle vector duplicates.
3184 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3186 (simplify_subreg): Handle subregs of vector duplicates.
3187 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3188 (selftest::simplify_rtx_c_tests): New functions.
3190 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3191 Alan Hayward <alan.hayward@arm.com>
3192 David Sherwood <david.sherwood@arm.com>
3194 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3195 (gen_vec_duplicate): Likewise.
3196 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3198 (gen_const_vector): ...here.
3199 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3200 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3201 whose elements are all equal.
3202 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3203 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3204 (simplify_relational_operation): Likewise.
3205 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3207 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3208 (aarch64_expand_vector_init): Likewise.
3209 * config/arm/arm.c (neon_vdup_constant): Likewise.
3210 (neon_expand_vector_init): Likewise.
3211 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3212 (arm_block_set_unaligned_vect): Likewise.
3213 (arm_block_set_aligned_vect): Likewise.
3214 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3215 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3216 (expand_vec_perm_even_odd_pack): Likewise.
3217 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3218 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3219 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3220 gen_const_vec_duplicate.
3221 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3222 * config/mips/mips.c (mips_gen_const_int_vector): Use
3223 gen_const_vec_duplicate.
3224 (mips_expand_vector_init): Use CONST0_RTX.
3225 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3226 (define_split): Use gen_const_vec_duplicate.
3227 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3228 (define_split): Use gen_const_vec_duplicate.
3229 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3230 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3231 * config/spu/spu.c (spu_const): Likewise.
3233 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3234 Alan Hayward <alan.hayward@arm.com>
3235 David Sherwood <david.sherwood@arm.com>
3237 * combine.c (can_change_dest_mode): Reject changes in
3238 REGMODE_NATURAL_SIZE.
3240 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3242 * configure.ac (--enable-libssp): New.
3243 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3244 trying to determine target-specific default. Adjust indentation.
3245 * configure: Regenerated.
3246 * doc/install.texi (Configuration): Expand --disable-libssp
3249 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3251 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3254 2017-10-31 Martin Jambor <mjambor@suse.cz>
3257 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3259 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3261 * auto-profile.c (autofdo_source_profile::read): Use
3262 UNKNOWN_LOCATION rather than 0.
3263 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3264 (warning_at): ...this overload.
3265 (warning_at_rich_loc_n): Rename to...
3266 (warning_n): ...this overload.
3267 (error_at_rich_loc): Rename to...
3268 (error_at): ...this overload.
3269 (pedwarn_at_rich_loc): Rename to...
3270 (pedwarn): ...this overload.
3271 (permerror_at_rich_loc): Rename to...
3272 (permerror): ...this overload.
3273 (inform_at_rich_loc): Rename to...
3274 (inform): ...this overload.
3275 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3276 (diagnostic_n_impl_richloc): Rename to...
3277 (diagnostic_n_impl): ...this rich_location *-based decl.
3278 (inform_at_rich_loc): Rename to...
3279 (inform): ...this, and add an assertion.
3280 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3281 (warning_at_rich_loc): Rename to...
3282 (warning_at): ...this, and add an assertion.
3283 (warning_at_rich_loc_n): Rename to...
3284 (warning_n): ...this, and add an assertion.
3285 (warning_n): Update location_t-based implementation for removal of
3286 location_t-based diagnostic_n_impl.
3287 (pedwarn_at_rich_loc): Rename to...
3288 (pedwarn): ...this, and add an assertion.
3289 (permerror_at_rich_loc): Rename to...
3290 (permerror): ...this, and add an assertion.
3291 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3292 (error_at_rich_loc): Rename to...
3293 (error_at): ...this, and add an assertion.
3294 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3295 (driver::do_spec_on_infiles): Likewise.
3296 * substring-locations.c (format_warning_va): Update for renaming
3297 of inform_at_rich_loc.
3299 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3301 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3302 _Float<N>X built-in functions so that the variant without the
3303 "__builtin_" prefix is only enabled for the GNU C and Objective C
3304 languages when they are in non-strict ANSI/ISO mode.
3305 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3306 * target.def (floatn_builtin_p): Add a target hook to control
3307 whether _Float<N> and _Float<N>X built-in functions without the
3308 "__builtin_" prefix are enabled, and return true for C and
3309 Objective C in the default hook. Include langhooks.h in
3311 * targhooks.h (default_floatn_builtin_p): Likewise.
3312 * targhooks.c (default_floatn_builtin_p): Likewise.
3313 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3314 floatn_builtin_p target hook.
3315 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3317 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3318 Eric Botcazou <ebotcazou@adacore.com>
3320 PR rtl-optimization/81803
3321 * lra-constraints.c (curr_insn_transform): Also reload the whole
3322 register for a strict subreg no wider than a word if this is for
3323 a WORD_REGISTER_OPERATIONS target.
3325 2017-10-31 Jason Merrill <jason@redhat.com>
3327 * gdbinit.in: Skip over inlines from timevar.h.
3329 2017-10-31 Martin Liska <mliska@suse.cz>
3331 * doc/gcov.texi: Document new option.
3332 * gcov.c (print_usage): Likewise print it.
3333 (process_args): Support the argument.
3334 (format_count): New function.
3335 (format_gcov): Use the function.
3337 2017-10-31 Martin Liska <mliska@suse.cz>
3339 * gcov.c (struct name_map): do not use typedef.
3340 Define operator== and operator<.
3341 (name_search): Remove.
3342 (name_sort): Remove.
3343 (main): Do not allocate names.
3344 (process_file): Add vertical space.
3345 (generate_results): Use std::find.
3346 (release_structures): Do not release memory.
3347 (find_source): Use std::find.
3349 2017-10-31 Martin Liska <mliska@suse.cz>
3351 * gcov.c (struct line_info): Remove it's typedef.
3352 (line_info::line_info): Add proper ctor.
3353 (line_info::has_block): Do not use a typedef.
3354 (struct source_info): Do not use typedef.
3355 (circuit): Likewise.
3356 (get_cycles_count): Likewise.
3357 (output_intermediate_file): Iterate via vector iterator.
3358 (add_line_counts): Use std::vector methods.
3359 (accumulate_line_counts): Likewise.
3360 (output_lines): Likewise.
3362 2017-10-31 Martin Liska <mliska@suse.cz>
3364 * gcov.c (struct source_info): Remove typedef.
3365 (source_info::source_info): Add proper ctor.
3366 (accumulate_line_counts): Use struct, not it's typedef.
3367 (output_gcov_file): Likewise.
3368 (output_lines): Likewise.
3369 (main): Do not allocate an array.
3370 (output_intermediate_file): Use size of vector container.
3371 (process_file): Resize the vector.
3372 (generate_results): Do not preallocate, use newly added vector
3374 (release_structures): Do not release sources.
3375 (find_source): Use vector methods.
3376 (add_line_counts): Do not use typedef.
3378 2017-10-31 Martin Liska <mliska@suse.cz>
3380 * doc/gcov.texi: Document that.
3381 * gcov.c (add_line_counts): Mark lines with a non-executed
3383 (output_line_beginning): Handle such lines.
3384 (output_lines): Pass new argument.
3385 (output_intermediate_file): Print it in intermediate format.
3387 2017-10-31 Martin Liska <mliska@suse.cz>
3389 * color-macros.h: New file.
3390 * diagnostic-color.c: Factor out color related to macros to
3392 * doc/gcov.texi: Document -k option.
3393 * gcov.c (INCLUDE_STRING): Include string.h.
3394 (print_usage): Add -k option.
3395 (process_args): Parse it.
3396 (pad_count_string): New function.
3397 (output_line_beginning): Likewise.
3398 (DEFAULT_LINE_START): New macro.
3399 (output_lines): Support color output.
3401 2017-10-31 Martin Liska <mliska@suse.cz>
3403 PR gcov-profile/82633
3404 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3405 their interaction with GCOV infrastructure.
3406 * configure.ac: Add -fkeep-{inline,static}-functions to
3408 * configure: Regenerate.
3410 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3413 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3415 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3418 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3419 into a register if it does not fit into an immediate offset field.
3421 2017-10-31 Olivier Hainque <hainque@adacore.com>
3423 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3425 2017-10-31 Julia Koval <julia.koval@intel.com>
3427 * config.gcc: Add gfniintrin.h.
3428 * config/i386/gfniintrin.h: New.
3429 * config/i386/i386-builtin-types.def
3430 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3431 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3432 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3433 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3434 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3435 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3436 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3437 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3438 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3439 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3440 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3441 * config/i386/immintrin.h: Include gfniintrin.h.
3442 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3444 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3446 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3448 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3450 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3451 (arm_ashldi3_1bit): Remove pattern.
3452 (ashrdi3): Remove shift by 1 expansion.
3453 (arm_ashrdi3_1bit): Remove pattern.
3454 (lshrdi3): Remove shift by 1 expansion.
3455 (arm_lshrdi3_1bit): Remove pattern.
3456 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3457 cost of ashldi3 by 1.
3458 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3459 (<shift>di3_neon): Likewise.
3461 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3463 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3464 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3465 and (*aarch64_simd_mov<VQ:mode>).
3466 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3467 pattern alternative.
3468 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3469 attributes to match pattern alternative.
3471 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3473 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3474 * config/rs6000/emmintrin.h: New file.
3475 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3477 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3479 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3480 * (movdi_vfp_cortexa8): Remove pattern.
3482 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3484 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3486 (Specific, alpha*-dec-osf5.1): Remove.
3487 (Specific, mips-sgi-irix5): Remove.
3488 (Specific, mips-sgi-irix6): Remove.
3490 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3493 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3494 arguments to clear_bit_region_be.
3496 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3498 * gimplify.c: Include memmodel.h.
3500 2017-10-30 Martin Jambor <mjambor@suse.cz>
3502 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3503 condition whether loop should be executed at all.
3505 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3507 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3508 gimple folding of vec_madd() intrinsics.
3509 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3510 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3511 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3513 2017-10-30 Richard Biener <rguenther@suse.de>
3515 PR tree-optimization/82762
3517 2017-10-23 Richard Biener <rguenther@suse.de>
3519 PR tree-optimization/82129
3521 2017-08-01 Richard Biener <rguenther@suse.de>
3523 PR tree-optimization/81181
3524 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3525 (compute_antic): ... end of iteration here.
3527 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3529 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3530 -std=iso9899:2017 and -std=gnu17.
3531 * doc/standards.texi (C Language): Document C17 support.
3532 * doc/cpp.texi (Overview): Mention -std=c17.
3533 (Standard Predefined Macros): Document C11 and C17 values of
3534 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3535 * doc/extend.texi (Inline): Do not list individual options for
3536 standards newer than C99.
3537 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3539 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3542 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3544 * asan.c (asan_finish_file): Align asan globals array by shadow
3547 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3550 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3551 (struct store_immediate_info): Add bitregion_start and bitregion_end
3553 (store_immediate_info::store_immediate_info): Add brs and bre
3554 arguments and initialize bitregion_{start,end} from those.
3555 (struct merged_store_group): Add bitregion_start, bitregion_end,
3556 align_base and mask fields. Drop unnecessary struct keyword from
3557 struct store_immediate_info. Add do_merge method.
3558 (clear_bit_region_be): Use memset instead of loop storing zeros.
3559 (merged_store_group::do_merge): New method.
3560 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3561 stores as long as the surrounding bitregions have no gaps.
3562 (merged_store_group::merge_overlapping): Use do_merge.
3563 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3564 is byte aligned, rather than requiring that start and width are
3565 byte aligned. Drop unnecessary struct keyword from
3566 struct store_immediate_info. Allocate and populate also mask array.
3567 Make start of the arrays relative to bitregion_start rather than
3568 start and size them according to bitregion_{end,start} difference.
3569 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3570 struct store_immediate_info.
3571 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3572 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3573 struct keyword from struct store_immediate_info.
3574 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3575 between stores as long as the surrounding bitregions have no gaps.
3577 (struct split_store): Add orig non-static data member.
3578 (split_store::split_store): Initialize orig to false.
3579 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3580 if there is exactly a single original stmt. Change stmts argument
3581 to pointer from reference, if NULL, don't push anything to it. Add
3582 first argument, use it to optimize skipping over orig stmts that
3583 are known to be before bitpos already. Simplify.
3584 (split_group): Return unsigned int count how many stores are or
3585 would be needed rather than a bool. Add allow_unaligned argument.
3586 Change split_stores argument from reference to pointer, if NULL,
3587 only do a dry run computing how many stores would be produced.
3588 Rewritten algorithm to use both alignment and misalign if
3589 !allow_unaligned and handle bitfield stores with gaps.
3590 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3591 from bitregion_start instead of start. Compute allow_unaligned
3592 here, if true, do 2 split_group dry runs to compute which one
3593 produces fewer stores and prefer aligned if equal. Punt if
3594 new count is bigger or equal than original before emitting any
3595 statements, rather than during that. Remove no longer needed
3596 new_ssa_names tracking. Replace num_stmts with
3597 split_stores.length (). Use 32-bit stack allocated entries
3598 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3599 if possible. Handle bitfields with gaps.
3600 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3601 Compute bitregion_{start,end} for the stores and construct
3602 store_immediate_info with that. Formatting fixes.
3604 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3607 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3608 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3610 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3612 * gimplify.c: Include tm_p.h.
3614 * common.opt (gcoff): Re-add as ignored option.
3615 (gcoff1, gcoff2, gcoff3): Likewise.
3617 * Makefile.in (OBJS): Delete sdbout.o.
3618 (GTFILES): Delete $(srcdir)/sdbout.c.
3619 * debug.h: Delete sdb_debug_hooks.
3620 * final.c: Delete sdbout.h include.
3621 (final_scan_insn): Delete SDB_DEBUG check.
3622 (rest_of_clean_state): Likewise.
3623 * output.h: Delete sdb_begin_function_line.
3626 * toplev.c: Delete sdbout.h include.
3627 (process_options): Delete SDB_DEBUG check.
3628 * tree-core.h (tree_type_common): Delete pointer field of
3630 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3631 TYPE_SYMTAB_POINTER.
3632 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3633 (TYPE_SYMTAB_IS_POINTER): Delete.
3634 (TYPE_SYMTAB_IS_DIE): Renumber.
3635 * xcoffout.c: Refer to former sdbout.c file.
3636 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3638 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3639 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3640 (Debugging Options): Delete -gcoff.
3641 (-gstabs): Delete SDB reference.
3643 (-gcoff@var{level}): Delete.
3644 * doc/passes.texi (Debugging information output): Delete SDB and
3645 sdbout.c references.
3646 * doc/tm.texi: Regenerate.
3647 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3648 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3650 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3651 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3652 (SDB_DEBUGGING_INFO): Delete.
3653 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3654 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3655 * target.def (output_source_filename): Delete COFF reference.
3657 * common.opt (gcoff): Delete.
3658 (gxcoff+): Update Negative chain.
3659 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3661 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3662 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3663 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3665 (expand_function_start): Change sdb reference to past tense.
3666 (expand_function_end): Change sdb reference to past tense.
3667 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3668 * opts.c (debug_type_names): Delete coff entry.
3669 (common_handle_option): Delete OPT_gcoff case.
3670 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3672 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3673 * config/cris/cris.h: Delete SDB reference in comment.
3674 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3675 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3676 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3677 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3679 (ix86_expand_prologue): Likewise.
3680 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3681 * config/ia64/ia64.h: Likewise.
3682 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3683 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3685 * config/mmix/mmix.h: Likewise.
3686 * config/nds32/nds32.c: Likewise.
3687 * config/stormy/storym16.h: Likewise.
3688 * config/visium/visium.h: Likewise.
3689 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3691 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3693 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3694 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3695 FRAME_POINTER_REGNUM point at high end of local var area.
3697 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3699 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3700 Move comment around. Do not reset best_edge for a copiable
3701 destination if the copy would cause a partition change.
3702 (better_edge_p): Remove redundant check.
3704 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3706 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3708 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3710 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3711 for math functions that have _Float<N> and _Float<N>X variants.
3712 (mathfn_built_in_2): Add support for math functions that have
3713 _Float<N> and _Float<N>X variants.
3714 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3715 (expand_builtin_mathfn_ternary): Add support for fma with
3716 _Float<N> and _Float<N>X variants.
3717 (expand_builtin): Likewise.
3718 (fold_builtin_3): Likewise.
3719 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3720 create math function _Float<N> and _Float<N>X variants as external
3722 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3723 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3724 the __builtin_ prefix and if not strict ansi, without the prefix.
3725 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3726 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3727 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3728 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3729 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3730 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3731 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3732 function signatures for fma _Float<N> and _Float<N>X variants.
3733 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3734 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3735 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3736 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3737 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3738 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3739 * gencfn-macros.c (print_case_cfn): Add support for math functions
3740 that have _Float<N> and _Float<N>X variants.
3741 (print_define_operator_list): Likewise.
3742 (fltfn_suffixes): Likewise.
3744 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3745 for math functions that have _Float<N> and _Float<N>X variants.
3746 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3747 and _Float<N>X variants.
3748 (COPYSIGN): Likewise.
3751 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3752 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3754 (integer_valued_read_call_p): Likewise.
3755 * fold-const-call.c (fold_const_call_ss): Likewise.
3756 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3757 _Float<N> and _Float<N>X variants.
3758 (fold_const_call_ssss): Add support for fma _Float<N> and
3759 _Float<N>X variants.
3760 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3761 support for copysign and fma _Float<N> and _Float<N>X variants.
3762 (backprop::process_builtin_call_use): Likewise.
3763 * tree-call-cdce.c (can_test_argument_range); Add support for
3764 sqrt _Float<N> and _Float<N>X variants.
3765 (edom_only_function): Likewise.
3766 (get_no_error_domain): Likewise.
3767 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3768 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3769 copysign _Float<N> and _Float<N>X variants.
3770 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3771 handled by machine independent code.
3772 (FMAF128): Likewise.
3773 * doc/cpp.texi (Common Predefined Macros): Document defining
3774 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3775 fma _Float<N> and _Float<N>X variants.
3777 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3780 * config/i386/i386-modes.def (CCFPU): Remove definition.
3781 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3782 (ix86_cc_modes_compatible): Ditto.
3783 (ix86_expand_carry_flag_compare): Ditto.
3784 (ix86_expand_int_movcc): Ditto.
3785 (ix86_expand_int_addcc): Ditto.
3786 (ix86_reverse_condition): Ditto.
3787 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3788 Return true/false for unordered/ordered fp comparisons.
3789 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3790 (ix86_prepare_fp_compare_args): Update for rename.
3791 (ix86_expand_fp_compare): Update for rename. Generate unordered
3792 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3793 (ix86_expand_sse_compare_and_jump): Ditto.
3794 * config/i386/predicates.md (fcmov_comparison_operator):
3795 Remove CCFPU mode handling.
3796 (ix86_comparison_operator): Ditto.
3797 (ix86_carry_flag_operator): Ditto.
3798 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3799 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3800 (*cmpu<mode>_cc_i387): Ditto.
3801 (FPCMP): Remove mode iterator.
3802 (unord): Remove mode attribute.
3803 (unord_subst): New define_subst transformation
3804 (unord): New define_subst attribute.
3806 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3807 (*cmpi<unord>xf_i387): Ditto.
3808 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3809 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3810 using unord_subst transformation.
3811 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3812 (round_saeonly): Also handle CCFP mode.
3813 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3814 Remove UNSPEC_SAHF unspec handling.
3816 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3818 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3820 2017-10-27 Jeff Law <law@redhat.com>
3822 * gimple-ssa-sprintf.c: Include domwalk.h.
3823 (class sprintf_dom_walker): New class, derived from dom_walker.
3824 (sprintf_dom_walker::before_dom_children): New function.
3825 (struct call_info): Moved into sprintf_dom_walker class
3826 (compute_formath_length, handle_gimple_call): Likewise.
3827 (sprintf_length::execute): Call the dominator walker rather
3828 than walking the statements.
3830 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3831 gimple statement locations.
3832 (check_array_bounds): Corresponding changes. Get the statement's
3833 location directly from wi->stmt.
3835 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3838 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3840 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3842 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3843 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3845 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3848 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3849 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3850 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3851 maybe_get_pool_constant.
3852 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3855 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3857 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3858 2.26 caveat. Update gas and gld versions.
3859 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3862 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3864 * cgraph.h (set_malloc_flag): Declare.
3865 * cgraph.c (set_malloc_flag_1): New function.
3866 (set_malloc_flag): Likewise.
3867 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3868 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3870 (read_ipa_call_summary): Add support for reading is_return_callee.
3871 (write_ipa_call_summary): Stream is_return_callee.
3872 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3873 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3874 ipa-prop.h, ipa-fnsummary.h.
3875 (pure_const_names): Change to static.
3876 (malloc_state_e): Define.
3877 (malloc_state_names): Define.
3878 (funct_state_d): Add field malloc_state.
3879 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3880 (check_retval_uses): New function.
3881 (malloc_candidate_p): Likewise.
3882 (analyze_function): Add support for malloc attribute.
3883 (pure_const_write_summary): Stream malloc_state.
3884 (pure_const_read_summary): Add support for reading malloc_state.
3885 (dump_malloc_lattice): New function.
3886 (propagate_malloc): New function.
3887 (warn_function_malloc): New function.
3888 (ipa_pure_const::execute): Call propagate_malloc and
3889 ipa_free_fn_summary.
3890 (pass_local_pure_const::execute): Add support for malloc attribute.
3891 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3892 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3894 2017-10-27 Martin Liska <mliska@suse.cz>
3896 PR gcov-profile/82457
3897 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3898 for fork-like functions to be properly instrumented.
3900 2017-10-27 Richard Biener <rguenther@suse.de>
3903 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3904 info when we redirected EH.
3906 2017-10-26 Michael Collison <michael.collison@arm.com>
3908 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3910 (<optab>_trunchf<GPI:mode>2: New pattern.
3911 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3912 * config/aarch64/iterators.md (wv): New mode attribute.
3913 (vf, VF): New mode attributes.
3914 (vgp, VGP): New mode attributes.
3915 (s): Update attribute with SImode and DImode prefixes.
3917 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3919 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3920 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3921 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3922 (nios2_option_overide): Initialize it. Don't allow R0-relative
3923 addressing with PIC.
3924 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3925 (nios2_symbolic_constant_p): Likewise.
3926 (nios2_legitimate_address_p): Likewise.
3927 (nios2_r0rel_section_name_p): New.
3928 (nios2_symbol_ref_in_r0rel_data_p): New.
3929 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3930 (r0rel_constant_p): New.
3931 (nios2_print_operand_address): Handle r0rel_constant_p.
3932 (nios2_cdx_narrow_form_p): Likewise.
3933 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3934 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3935 (Nios II Options): Document -mr0rel-sec.
3937 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3939 * config/nios2/nios2.c: Include xregex.h.
3940 (nios2_gprel_sec_regex): New.
3941 (nios2_option_overide): Initialize it. Don't allow GP-relative
3942 addressing with PIC.
3943 (nios2_small_section_name_p): Check for regex match.
3944 * config/nios2/nios2.opt (mgprel-sec=): New option.
3945 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3946 (Nios II Options): Document -mgprel-sec.
3948 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3950 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3952 2017-10-26 Tom de Vries <tom@codesourcery.com>
3954 PR tree-optimization/82707
3955 * gimple.c (gimple_copy): Fix unsharing of
3956 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3958 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3960 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3961 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3962 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3963 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3964 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3965 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3966 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3967 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3968 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3969 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3970 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3971 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3972 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3973 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3974 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3975 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3976 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3978 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3980 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3982 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3983 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3984 warning to rs6000.c. Remove the Undocumented flag, since it has
3986 (-mabi=ibmlongdouble): Likewise.
3987 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3988 already set, set the default format for long double.
3989 (rs6000_debug_reg_global): Print whether long double is IBM or
3991 (rs6000_option_override_internal): Rework setting long double
3992 format. Only warn if the user is changing the long double default
3993 and they did not use -Wno-psabi.
3994 * doc/invoke.texi (PowerPC options): Update the documentation for
3995 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3997 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3998 Alan Hayward <alan.hayward@arm.com>
3999 David Sherwood <david.sherwood@arm.com>
4001 * rtl.h (wider_subreg_mode): New function.
4002 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4003 rather than an unsigned int *.
4004 * ira-color.c (regno_max_ref_width): Replace with...
4005 (regno_max_ref_mode): ...this new variable.
4006 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4007 Use wider_subreg_mode.
4008 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4009 rather than an unsigned int *.
4010 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4011 (process_alt_operands): Likewise.
4012 (invariant_p): Likewise.
4013 * lra-spills.c (assign_mem_slot): Likewise.
4014 (add_pseudo_to_slot): Likewise.
4015 * lra.c (collect_non_operand_hard_regs): Likewise.
4016 (add_regs_to_insn_regno_info): Likewise.
4017 * reload1.c (regno_max_ref_width): Replace with...
4018 (regno_max_ref_mode): ...this new variable.
4019 (reload): Update accordingly. Update call to
4020 ira_sort_regnos_for_alter_reg.
4021 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4022 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4023 (scan_paradoxical_subregs): Likewise.
4025 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4027 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4028 (aarch64_frame): Add emit_frame_chain boolean.
4029 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4030 Move eh_return case to aarch64_layout_frame.
4031 (aarch64_layout_frame): Initialize emit_frame_chain.
4032 (aarch64_expand_prologue): Use emit_frame_chain.
4034 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4036 * config/aarch64/aarch64.c (aarch64_layout_frame):
4037 Ensure LR is always stored at the bottom of the callee-saves.
4038 Remove rarely used frame layout which saves callee-saves at top of
4039 frame, so the store of LR can be used as a valid probe in all cases.
4041 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4043 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4044 Improve unaligned TImode/TFmode base/offset split.
4046 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4047 Alan Hayward <alan.hayward@arm.com>
4048 David Sherwood <david.sherwood@arm.com>
4050 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4051 * combine.c (find_single_use_1): Likewise.
4052 (expand_field_assignment): Likewise.
4053 (move_deaths): Likewise.
4054 * lra-constraints.c (simplify_operand_subreg): Likewise.
4055 (curr_insn_transform): Likewise.
4056 * lra.c (collect_non_operand_hard_regs): Likewise.
4057 (add_regs_to_insn_regno_info): Likewise.
4058 * rtlanal.c (reg_referenced_p): Likewise.
4059 (covers_regno_no_parallel_p): Likewise.
4061 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4063 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4064 Don't print any bits outside the precision of the value.
4065 * wide-int.cc (test_printing): Add some new tests.
4067 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4069 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4070 supports -xbrace_comment option.
4071 * configure: Regenerate.
4072 * config.in: Regenerate.
4073 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4074 (ASM_CPU_SPEC): Use it.
4076 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4078 * target.def (static_rtx_alignment): New hook.
4079 * targhooks.h (default_static_rtx_alignment): Declare.
4080 * targhooks.c (default_static_rtx_alignment): New function.
4081 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4082 * doc/tm.texi: Regenerate.
4083 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4084 instead of targetm.constant_alignment. Remove call to
4086 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4087 (cris_preferred_mininum_alignment): New function, split out from...
4088 (cris_constant_alignment): ...here.
4089 (cris_static_rtx_alignment): New function.
4090 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4092 (ix86_constant_alignment): ...here.
4093 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4094 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4095 (mmix_static_rtx_alignment): New function.
4096 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4097 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4099 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4102 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4103 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4105 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4107 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4108 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4109 mark as a sign-extending load.
4110 (local_pic_load_u): Define.
4112 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4115 * fold-const.c (operand_equal_for_comparison_p): Also return true
4116 if ARG0 is a simple variant of ARG1 with narrower precision.
4117 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4119 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4121 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4123 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4124 scatter_static, scatter_per_elt.
4125 * x86-tune-costs.h: Add new cost entries.
4127 2017-10-25 Richard Biener <rguenther@suse.de>
4129 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4130 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4131 class pass_fre): Move to ...
4132 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4133 class pass_fre): ... here and adjust for statistics.
4135 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4138 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4139 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4140 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4141 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4144 2017-10-25 Richard Biener <rguenther@suse.de>
4146 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4147 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4148 eliminate_push_avail, eliminate_insert): Move inside...
4149 (class eliminate_dom_walker): ... this class in preparation
4151 (fini_eliminate): Remove by merging with ...
4152 (eliminate): ... this function. Adjust for class changes.
4153 (pass_pre::execute): Remove fini_eliminate call.
4154 (pass_fre::execute): Likewise.
4156 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4159 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4160 (VPERMI2, VPERMI2I): New mode iterators.
4161 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4162 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4164 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4165 mode iterator. Remove 3 old define_insn patterns.
4166 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4167 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4168 VPERMI2 mode iterator, remove the other two expanders.
4169 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4170 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4171 instructions, remove the other two patterns.
4172 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4173 mode iterator, remove the other two patterns.
4174 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4175 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4176 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4177 and adjust argument order accordingly.
4178 (ix86_expand_vec_perm): Adjust caller.
4179 (expand_vec_perm_1): Likewise.
4180 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4181 (expand_vec_perm_vpermt2_vpshub2): ... this.
4182 (ix86_expand_vec_perm_const_1): Adjust caller.
4183 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4186 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4187 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4188 (vec_shl_<mode>): Remove unused expander.
4189 (avx512bw_<shift_insn><mode>3): New define_insn.
4190 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4191 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4193 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4196 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4199 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4201 PR rtl-optimization/82396
4202 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4203 (autopref_multipass_init): Simplify initialization.
4204 (autopref_rank_data): Simplify sort order.
4205 * gcc/sched-int.h (autopref_multipass_data_): Remove
4206 multi_mem_insn_p, min_offset and max_offset.
4208 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4211 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4212 Check special value of flag_omit_frame_pointer.
4213 (aarch64_can_eliminate): Likewise.
4214 (aarch64_override_options_after_change_1): Simplify handling of
4215 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4217 2017-10-24 Richard Biener <rguenther@suse.de>
4219 PR tree-optimization/82697
4220 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4221 zero for conditional load and unconditional store.
4223 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4225 * doc/install.texi: Document bootstrap-cet.
4227 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4230 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4231 ENDBR instruction at function entrance if function is only
4234 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4237 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4238 patterns to better describe from which operation the CF is computed.
4239 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4240 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4241 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4242 is 0, use _0 suffixed expanders instead of emitting a comparison
4245 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4247 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4248 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4250 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4253 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4254 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4255 * loop-iv.c (iv_get_reaching_def): Likewise.
4256 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4257 variable is promoted and the partition contains undefined values.
4259 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4261 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4263 (nios2_address_cost): Define.
4264 (nios2_legitimize_address): Recognize (exp + constant) directly.
4265 (TARGET_ADDRESS_COST): Define.
4267 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4269 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4270 (nios2_symbolic_memory_operand_p): Declare.
4271 (nios2_split_large_constant): Declare.
4272 (nios2_split_symbolic_memory_operand): Declare.
4273 * config/nios2/nios2.c: Adjust includes.
4274 (nios2_symbolic_constant_allowed): New.
4275 (nios2_symbolic_constant_p): New.
4276 (nios2_plus_symbolic_constant_p): New.
4277 (nios2_valid_addr_expr_p): Recognize addresses involving
4279 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4280 (nios2_symbolic_memory_operand_p): New.
4281 (nios2_large_constant_p): New.
4282 (nios2_split_large_constant): New.
4283 (nios2_split_plus_large_constant): New.
4284 (nios2_split_symbolic_memory_operand): New.
4285 (nios2_legitimize_address): Code refactoring. Handle addresses
4286 involving symbolic constants.
4287 (nios2_emit_move_sequence): Likewise.
4288 (nios2_print_operand): Improve error output.
4289 (nios2_print_operand_address): Handle LO_SUM.
4290 (nios2_cdx_narrow_form_p): Likewise.
4291 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4292 operands involving symbolic constants.
4293 (movhi_internal, movsi_internal): Likewise.
4294 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4295 (extendhisi2, extendqi<mode>2): Likewise.
4297 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4299 * tree-pass.h (PROP_rtl_split_insns): Define.
4300 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4302 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4304 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4306 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4309 * target.def (const_not_ok_for_debug_p): Default to
4310 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4311 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4312 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4313 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4314 which targetm.const_not_ok_for_debug_p returned true.
4315 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4317 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4319 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4320 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4321 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4323 (ix86_const_not_ok_for_debug_p): New function.
4324 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4325 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4327 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4330 * system.h: Conditionally include "unique-ptr.h" if
4331 INCLUDE_UNIQUE_PTR is defined.
4332 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4333 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4335 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4337 * config/rl78/rl78.md: New define_expand "subdi3".
4339 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4342 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4343 DF_REF_INSN if DF_REF_INSN_INFO is false.
4345 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4347 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4348 xmm_move instead of sse_move.
4349 (sse_store_index): New function.
4350 (ix86_register_move_cost): Be more sensible about mismatch stall;
4351 model AVX moves correctly; make difference between sse->integer and
4353 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4354 moves; make difference between SSE and AVX.
4355 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4356 and zmm_move. Increase size of sse load and store tables;
4357 add unaligned load and store tables; add ssemmx_to_integer.
4358 * x86-tune-costs.h: Update all entries according to real
4359 move latencies from Agner Fog's manual and chip documentation.
4361 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4364 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4365 * config/i386/constraints.md (Wf): New constraint.
4366 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4367 (cmp<dwi>_doubleword): Removed.
4368 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4369 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4370 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4371 expand with cmp<dwi>_doubleword. For LTU and GEU use
4372 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4374 * common.opt (gcolumn-info): Enable by default.
4375 * doc/invoke.texi (gcolumn-info): Document new default.
4377 2017-10-23 Richard Biener <rguenther@suse.de>
4379 PR tree-optimization/82672
4380 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4381 Fold the stmt if we propagated into it.
4383 2017-10-23 Richard Biener <rguenther@suse.de>
4385 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4386 (bitmap_remove_expr_from_set): ... this. All callers call this
4387 for non-constant values.
4388 (bitmap_set_subtract): Rename to...
4389 (bitmap_set_subtract_expressions): ... this. Adjust and
4391 (bitmap_set_contains_value): Remove superfluous check.
4392 (bitmap_set_replace_value): Inline into single caller ...
4393 (bitmap_value_replace_in_set): ... here and simplify.
4394 (dependent_clean): Merge into ...
4395 (clean): ... this using an overload. Adjust.
4396 (prune_clobbered_mems): Adjust.
4397 (compute_antic_aux): Likewise.
4398 (compute_partial_antic_aux): Likewise.
4400 2017-10-23 Richard Biener <rguenther@suse.de>
4402 PR tree-optimization/82129
4404 2017-08-01 Richard Biener <rguenther@suse.de>
4406 PR tree-optimization/81181
4407 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4408 (compute_antic): ... end of iteration here.
4410 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4412 * target.def (starting_frame_offset): New hook.
4413 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4414 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4415 * doc/tm.texi.in: Regenerate.
4416 * hooks.h (hook_hwi_void_0): Declare.
4417 * hooks.c (hook_hwi_void_0): New function.
4418 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4419 STARTING_FRAME_OFFSET.
4420 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4421 * reload1.c (reload): Likewise.
4422 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4423 instead of STARTING_FRAME_OFFSET.
4424 * function.c (try_fit_stack_local): Likewise.
4425 (assign_stack_local_1): Likewise
4426 (instantiate_virtual_regs): Likewise.
4427 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4428 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4429 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4430 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4431 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4432 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4433 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4434 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4435 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4436 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4437 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4438 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4439 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4440 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4441 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4442 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4443 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4444 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4445 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4446 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4447 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4448 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4449 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4450 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4451 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4452 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4453 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4454 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4455 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4456 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4457 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4458 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4459 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4460 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4461 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4462 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4463 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4464 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4465 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4466 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4467 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4468 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4469 return a HOST_WIDE_INT.
4470 (avr_builtin_setjmp_frame_value): Use it instead of
4471 STARTING_FRAME_OFFSET.
4472 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4473 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4474 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4476 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4477 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4478 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4479 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4480 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4481 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4482 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4483 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4484 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4485 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4486 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4487 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4489 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4490 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4491 * config/mips/mips.c (mips_compute_frame_info): Refer to
4492 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4493 (mips_starting_frame_offset): New function.
4494 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4495 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4496 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4497 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4498 and return a HOST_WIDE_INT.
4499 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4500 (mmix_initial_elimination_offset): Refer to
4501 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4502 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4503 * config/pa/pa.c (pa_starting_frame_offset): New function.
4504 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4505 (pa_expand_prologue): Likewise.
4506 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4507 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4508 !FRAME_GROWS_DOWNWARD handling to...
4509 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4510 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4511 !FRAME_GROWS_DOWNWARD handling to...
4512 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4513 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4514 !FRAME_GROWS_DOWNWARD handling to...
4515 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4516 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4518 (rs6000_starting_frame_offset): New function.
4519 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4520 !FRAME_GROWS_DOWNWARD handling to...
4521 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4522 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4523 !FRAME_GROWS_DOWNWARD handling to...
4524 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4525 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4526 !FRAME_GROWS_DOWNWARD handling to...
4527 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4528 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4529 (rs6000_starting_frame_offset): New function.
4530 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4531 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4532 * config/vax/vax.c (vax_starting_frame_offset): New function.
4533 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4534 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4535 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4536 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4537 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4538 * system.h (STARTING_FRAME_OFFSET): Poison.
4540 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4542 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4543 SCALAR_TYPE_MODE instead of TYPE_MODE.
4545 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4546 Alan Hayward <alan.hayward@arm.com>
4547 David Sherwood <david.sherwood@arm.com>
4549 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4551 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4552 Alan Hayward <alan.hayward@arm.com>
4553 David Sherwood <david.sherwood@arm.com>
4555 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4557 2017-10-23 Richard Biener <rguenther@suse.de>
4559 PR tree-optimization/82129
4560 * tree-ssa-pre.c (bitmap_set_and): Remove.
4561 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4562 canonicalizing expressions in the set to those with lowest
4563 ID rather than taking that from the first edge.
4565 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4567 * combine.c (rtx_equal_for_field_assignment_p): Use
4568 byte_lowpart_offset.
4570 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4571 Alan Hayward <alan.hayward@arm.com>
4572 David Sherwood <david.sherwood@arm.com>
4574 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4575 to a SUBREG_PROMOTED_VAR.
4577 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4578 Alan Hayward <alan.hayward@arm.com>
4579 David Sherwood <david.sherwood@arm.com>
4581 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4582 (expand_debug_source_expr): Likewise.
4583 * combine.c (combine_simplify_rtx): Likewise.
4584 * cse.c (fold_rtx): Likewise.
4585 * optabs.c (expand_float): Likewise.
4586 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4587 (simplify_binary_operation_1): Likewise.
4589 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4590 Alan Hayward <alan.hayward@arm.com>
4591 David Sherwood <david.sherwood@arm.com>
4593 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4594 (record_promoted_value): Likewise.
4595 * expr.c (expand_expr_real_2): Likewise.
4596 * ree.c (update_reg_equal_equiv_notes): Likewise.
4597 (combine_set_extension): Likewise.
4598 * rtlanal.c (low_bitmask_len): Likewise.
4599 * simplify-rtx.c (neg_const_int): Likewise.
4600 (simplify_binary_operation_1): Likewise.
4602 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4603 Alan Hayward <alan.hayward@arm.com>
4604 David Sherwood <david.sherwood@arm.com>
4606 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4607 * regcprop.c (maybe_mode_change): Likewise.
4608 * reload1.c (alter_reg): Likewise.
4610 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4612 * inchash.h (inchash::hash::add_wide_int): New function.
4613 * lto-streamer-out.c (hash_tree): Use it.
4615 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4617 * inchash.h (inchash::hash::add_wide_int): Rename to...
4618 (inchash::hash::add_hwi): ...this.
4619 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4620 (polymorphic_call_target_hasher::hash): Likewise.
4621 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4622 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4623 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4624 * lto-streamer-out.c (hash_tree): Likewise.
4625 * optc-save-gen.awk: Likewise.
4626 * tree.c (add_expr): Likewise.
4628 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4631 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4632 for ordered inequality comparisons even with TARGET_IEEE_FP.
4634 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4637 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4638 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4639 Expand with cmp<dwi>_doubleword.
4641 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4643 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4644 List CET intrinsics.
4645 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4646 specific to -fcf-protection option.
4648 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4650 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4651 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4652 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4653 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4654 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4655 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4656 (extra_objs): Add cet.o for Linux/x86 targets.
4657 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4658 * config/i386/cet.c: New file.
4659 * config/i386/cetintrin.h: Likewise.
4660 * config/i386/t-cet: Likewise.
4661 * config/i386/cpuid.h (bit_SHSTK): New.
4662 (bit_IBT): Likewise.
4663 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4664 pass IBT and SHSTK bits.
4665 * config/i386/i386-builtin-types.def
4666 (VOID_FTYPE_UNSIGNED_PVOID): New.
4667 (VOID_FTYPE_UINT64_PVOID): Likewise.
4668 * config/i386/i386-builtin.def: Add CET intrinsics.
4669 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4670 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4671 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4672 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4674 * config/i386/i386.c (rest_of_insert_endbranch): New.
4675 (pass_data_insert_endbranch): Likewise.
4676 (pass_insert_endbranch): Likewise.
4677 (make_pass_insert_endbranch): Likewise.
4678 (ix86_notrack_prefixed_insn_p): Likewise.
4679 (ix86_target_string): Add -mibt, -mshstk flags.
4680 (ix86_option_override_internal): Add flag_cf_protection
4682 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4683 (ix86_print_operand): Add 'notrack' prefix output.
4684 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4685 (ix86_expand_builtin): Expand CET intrinsics.
4686 (x86_output_mi_thunk): Add 'endbranch' instruction.
4687 * config/i386/i386.h (TARGET_IBT): New.
4688 (TARGET_IBT_P): Likewise.
4689 (TARGET_SHSTK): Likewise.
4690 (TARGET_SHSTK_P): Likewise.
4691 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4692 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4693 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4694 (builtin_setjmp_setup): New pattern.
4695 (builtin_longjmp): Likewise.
4696 (rdssp<mode>): Likewise.
4697 (incssp<mode>): Likewise.
4698 (saveprevssp): Likewise.
4699 (rstorssp): Likewise.
4700 (wrss<mode>): Likewise.
4701 (wruss<mode>): Likewise.
4702 (setssbsy): Likewise.
4703 (clrssbsy): Likewise.
4704 (nop_endbr): Likewise.
4705 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4707 * config/i386/immintrin.h: Include <cetintrin.h>.
4708 * config/i386/linux-common.h
4709 (file_end_indicate_exec_stack_and_cet): New prototype.
4710 (TARGET_ASM_FILE_END): New.
4712 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4714 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4715 latencies instead of having separate table; make difference between
4716 integer and float costs.
4717 * i386.h (processor_costs): Remove scalar_stmt_cost,
4718 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4719 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4721 * x86-tune-costs.h: Remove entries which has been removed in
4722 procesor_costs from all tables; make cond_taken_branch_cost
4723 and cond_not_taken_branch_cost COST_N_INSNS based.
4725 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4727 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4729 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4731 * config/i386/i386.md (isa): Remove fma_avx512f.
4732 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4733 <avx512>_fmadd_<mode>_mask3<round_name>,
4734 <avx512>_fmsub_<mode>_mask<round_name>,
4735 <avx512>_fmsub_<mode>_mask3<round_name>,
4736 <avx512>_fnmadd_<mode>_mask<round_name>,
4737 <avx512>_fnmadd_<mode>_mask3<round_name>,
4738 <avx512>_fnmsub_<mode>_mask<round_name>,
4739 <avx512>_fnmsub_<mode>_mask3<round_name>,
4740 <avx512>_fmaddsub_<mode>_mask<round_name>,
4741 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4742 <avx512>_fmsubadd_<mode>_mask<round_name>,
4743 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4744 (*vec_widen_umult_even_v16si<mask_name>,
4745 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4746 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4748 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4750 * extend.texi: Add 'nocf_check' documentation.
4751 * gimple.texi: Add second parameter to
4752 gimple_build_call_from_tree.
4753 * invoke.texi: Add -fcf-protection documentation.
4754 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4756 2017-10-20 Richard Biener <rguenther@suse.de>
4758 PR tree-optimization/82473
4759 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4760 the largest input type.
4762 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4764 * c-attribs.c (handle_nocf_check_attribute): New function.
4765 (c_common_attribute_table): Add 'nocf_check' handling.
4766 * gimple-parser.c: Add second argument NULL to
4767 gimple_build_call_from_tree.
4768 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4769 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4771 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4772 * common.opt: Add fcf-protection flag.
4773 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4774 * flag-types.h: Add enum cf_protection_level.
4775 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4776 Add 'nocf_check' attribute propagation to gimple call.
4777 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4778 (gimple_build_call_from_tree): Update prototype.
4779 (gimple_call_nocf_check_p): New function.
4780 (gimple_call_set_nocf_check): Likewise.
4781 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4782 * ipa-icf.c: Add nocf_check attribute in statement hash.
4783 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4784 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4785 * toplev.c (process_options): Add flag_cf_protection handling.
4787 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4789 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4791 2017-10-20 Richard Biener <rguenther@suse.de>
4793 PR tree-optimization/82603
4794 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4795 remove false predicated stores.
4797 2017-10-20 Richard Biener <rguenther@suse.de>
4799 * graphite-isl-ast-to-gimple.c
4800 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4801 Remove return value and simplify, dump copied stmt after lhs
4803 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4804 Reduce dump verbosity.
4805 (gsi_insert_earliest): Likewise.
4806 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4807 * graphite.c (print_global_statistics): Adjust dumping.
4808 (print_graphite_scop_statistics): Likewise.
4809 (print_graphite_statistics): Do not dump loops here.
4810 (graphite_transform_loops): But here.
4812 2017-10-20 Nicolas Roche <roche@adacore.com>
4814 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4815 * configure: Regenerate.
4817 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4820 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4821 functions when optimizing replace GIMPLE_RETURN stmts with
4822 calls to __builtin_unreachable ().
4825 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4826 for -fsanitize=thread link of executables.
4827 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4828 link of executables.
4831 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4833 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4834 define_insns for logical vector shifts to use VI248_AVX512BW
4835 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4836 condition, useless isa and prefix attributes. Change the first
4837 2 of these define_insns to ...
4838 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4839 define_insn for avx512vl.
4840 (<shift_insn><mode>3): ... and this, new define_insn without
4841 masking for non-avx512vl.
4844 * config/i386/sse.md (*andnot<mode>3,
4845 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4846 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4847 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4848 not applied use empty suffix even for TARGET_AVX512VL.
4849 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4850 is applied, supply evex,evex or evex,evex,evex instead of just
4853 2017-10-20 Julia Koval <julia.koval@intel.com>
4855 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4856 (OPTION_MASK_ISA_GFNI_UNSET): New.
4857 (ix86_handle_option): Handle OPT_mgfni.
4858 * config/i386/cpuid.h (bit_GFNI): New.
4859 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4860 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4861 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4862 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4863 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4864 * config/i386/i386.opt: Add mgfni.
4866 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4868 * config/msp430/msp430.c (msp430_option_override): Disable
4869 -fdelete-null-pointer-checks.
4870 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4872 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4874 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4875 of x87 and SSE instructions.
4877 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4879 * asan.c (create_cond_insert_point): Do not update edge count.
4880 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4881 (afdo_propagate_circuit): Likewise.
4882 (afdo_calculate_branch_prob): Likewise.
4883 (afdo_annotate_cfg): Likewise.
4884 * basic-block.h (struct edge_def): Remove count.
4885 (edge_def::count): New accessor.
4886 * bb-reorder.c (rotate_loop): Update.
4887 (find_traces_1_round): Update.
4888 (connect_traces): Update.
4889 (sanitize_hot_paths): Update.
4890 * cfg.c (unchecked_make_edge): Update.
4891 (make_single_succ_edge): Update.
4892 (check_bb_profile): Update.
4893 (dump_edge_info): Update.
4894 (update_bb_profile_for_threading): Update.
4895 (scale_bbs_frequencies_int): Update.
4896 (scale_bbs_frequencies_gcov_type): Update.
4897 (scale_bbs_frequencies_profile_count): Update.
4898 (scale_bbs_frequencies): Update.
4899 * cfganal.c (connect_infinite_loops_to_exit): Update.
4900 * cfgbuild.c (compute_outgoing_frequencies): Update.
4901 (find_many_sub_basic_blocks): Update.
4902 * cfgcleanup.c (try_forward_edges): Update.
4903 (try_crossjump_to_edge): Update
4904 * cfgexpand.c (expand_gimple_cond): Update
4905 (expand_gimple_tailcall): Update
4906 (construct_exit_block): Update
4907 * cfghooks.c (verify_flow_info): Update
4908 (redirect_edge_succ_nodup): Update
4909 (split_edge): Update
4910 (make_forwarder_block): Update
4911 (duplicate_block): Update
4912 (account_profile_record): Update
4913 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4914 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4915 * cfgloopmanip.c (scale_loop_profile): Update.
4917 (lv_adjust_loop_entry_edge): Update.
4918 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4919 (force_nonfallthru_and_redirect): Update.
4920 (purge_dead_edges): Update.
4921 (rtl_flow_call_edges_add): Update.
4922 * cgraphunit.c (init_lowered_empty_function): Update.
4923 (cgraph_node::expand_thunk): Update.
4924 * gimple-pretty-print.c (dump_probability): Update.
4925 (dump_edge_probability): Update.
4926 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4927 * haifa-sched.c (sched_create_recovery_edges): Update.
4928 * hsa-gen.c (convert_switch_statements): Update.
4929 * ifcvt.c (dead_or_predicable): Update.
4930 * ipa-inline-transform.c (inline_transform): Update.
4931 * ipa-split.c (split_function): Update.
4932 * ipa-utils.c (ipa_merge_profiles): Update.
4933 * loop-doloop.c (add_test): Update.
4934 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4935 * lto-streamer-in.c (input_cfg): Update.
4936 (input_function): Update.
4937 * lto-streamer-out.c (output_cfg): Update.
4938 * modulo-sched.c (sms_schedule): Update.
4939 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4940 * predict.c (maybe_hot_edge_p): Update.
4941 (unlikely_executed_edge_p): Update.
4942 (probably_never_executed_edge_p): Update.
4943 (dump_prediction): Update.
4944 (drop_profile): Update.
4945 (propagate_unlikely_bbs_forward): Update.
4946 (determine_unlikely_bbs): Update.
4947 (force_edge_cold): Update.
4948 * profile.c (compute_branch_probabilities): Update.
4949 * reg-stack.c (better_edge): Update.
4950 * shrink-wrap.c (handle_simple_exit): Update.
4951 * tracer.c (better_p): Update.
4952 * trans-mem.c (expand_transaction): Update.
4953 (split_bb_make_tm_edge): Update.
4954 * tree-call-cdce.c: Update.
4955 * tree-cfg.c (gimple_find_sub_bbs): Update.
4956 (gimple_split_edge): Update.
4957 (gimple_duplicate_sese_region): Update.
4958 (gimple_duplicate_sese_tail): Update.
4959 (gimple_flow_call_edges_add): Update.
4960 (insert_cond_bb): Update.
4961 (execute_fixup_cfg): Update.
4962 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4963 * tree-complex.c (expand_complex_div_wide): Update.
4964 * tree-eh.c (lower_resx): Update.
4965 (unsplit_eh): Update.
4966 (cleanup_empty_eh_move_lp): Update.
4967 * tree-inline.c (copy_edges_for_bb): Update.
4968 (freqs_to_counts): Update.
4969 (copy_cfg_body): Update.
4970 * tree-ssa-dce.c (remove_dead_stmt): Update.
4971 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4972 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4973 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4974 (unloop_loops): Update.
4975 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4976 * tree-ssa-loop-split.c (connect_loops): Update.
4977 (split_loop): Update.
4978 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4979 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4980 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4981 * tree-ssa-reassoc.c (branch_fixup): Update.
4982 * tree-ssa-tail-merge.c (replace_block_by): Update.
4983 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4984 (compute_path_counts): Update.
4985 (update_profile): Update.
4986 (recompute_probabilities): Update.
4987 (update_joiner_offpath_counts): Update.
4988 (estimated_freqs_path): Update.
4989 (freqs_to_counts_path): Update.
4990 (clear_counts_path): Update.
4991 (ssa_fix_duplicate_block_edges): Update.
4992 (duplicate_thread_path): Update.
4993 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4994 (case_bit_test_cmp): Update.
4995 (collect_switch_conv_info): Update.
4996 (gen_inbound_check): Update.
4997 (do_jump_if_equal): Update.
4998 (emit_cmp_and_jump_insns): Update.
4999 * tree-tailcall.c (decrease_profile): Update.
5000 (eliminate_tail_call): Update.
5001 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5002 (vect_do_peeling): Update.
5003 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5004 * ubsan.c (ubsan_expand_null_ifn): Update.
5005 (ubsan_expand_ptr_ifn): Update.
5006 * value-prof.c (gimple_divmod_fixed_value): Update.
5007 (gimple_mod_pow2): Update.
5008 (gimple_mod_subtract): Update.
5009 (gimple_ic): Update.
5010 (gimple_stringop_fixed_value): Update.
5012 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5015 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5017 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5019 PR rtl-optimization/82395
5020 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5021 based on non_spilled_static_chain_regno_p.
5023 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5025 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5026 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5027 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5029 2017-10-19 Martin Sebor <msebor@redhat.com>
5031 PR tree-optimization/82596
5032 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5034 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5036 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5037 (get_mem_refs_of_builtin_call): Likewise.
5038 * builtins.c (expand_builtin_apply): Adjust call to
5039 allocate_dynamic_stack_space.
5040 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5041 the third argument to allocate_dynamic_stack_space, otherwise -1.
5042 (expand_builtin): Deal with all alloca variants.
5043 (is_inexpensive_builtin): Likewise.
5044 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5045 * calls.c (special_function_p): Deal with all alloca variants.
5046 (initialize_argument_information): Adjust call to
5047 allocate_dynamic_stack_space.
5048 (expand_call): Likewise.
5049 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5050 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5051 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5052 use it for the stack usage computation.
5053 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5054 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5055 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5056 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5057 (in_loop_p): Remove first argument and useless check.
5058 (pass_walloca::execute): Remove useless test and adjust call to above.
5059 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5060 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5061 (gimplify_call_expr): Deal with all alloca variants.
5062 * hsa-gen.c (gen_hsa_alloca): Likewise.
5063 (gen_hsa_insns_for_call): Likewise.
5064 * ipa-pure-const.c (special_builtin_state): Likewise.
5065 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5066 * tree-object-size.c (alloc_object_size): Likewise.
5067 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5068 (call_may_clobber_ref_p_1): Likewise.
5069 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5070 (ccp_fold_stmt): Likewise.
5071 (optimize_stack_restore): Likewise.
5072 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5073 (mark_all_reaching_defs_necessary_1): Likewise.
5074 (propagate_necessity): Likewise.
5075 (eliminate_unnecessary_stmts): Likewise.
5076 * tree.c (build_common_builtin_nodes): Build
5077 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5078 (build_alloca_call_expr): New function.
5079 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5080 (CASE_BUILT_IN_ALLOCA): Likewise.
5081 (build_alloca_call_expr): Declare.
5082 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5084 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5087 * dwarf2out.c (new_die_raw): New static inline function.
5088 (new_die): Use it to create the DIE.
5089 (add_AT_external_die_ref): Likewise.
5090 (clone_die): Likewise.
5091 (clone_as_declaration): Likewise.
5092 (dwarf2out_vms_debug_main_pointer): Likewise.
5093 (base_type_die): Likewise. Remove early return for corner cases.
5094 Do not call add_pubtype on the DIE here.
5095 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5096 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5097 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5098 native order exists for base types, attach the DIE manually and call
5099 add_pubtype on it. Do not equate a reverse order DIE to the type.
5101 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5103 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5104 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5106 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5108 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5110 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5111 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5112 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5113 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5114 * opts.c (sanitizer_opts): Add builtin.
5115 * ubsan.c (instrument_builtin): New function.
5116 (pass_ubsan::execute): Call it.
5117 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5118 * doc/invoke.texi: Document -fsanitize=builtin.
5120 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5121 builtins, store max (log2 (align), 0) into uchar field instead of
5122 align into uptr field.
5123 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5124 store uchar 0 field instead of uptr 0 field.
5125 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5126 instead of passing one address of struct with 2 locations pass
5127 two addresses of structs with 1 location each.
5128 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5129 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5130 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5131 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5132 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5133 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5134 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5135 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5137 2017-10-19 Martin Liska <mliska@suse.cz>
5140 * file-find.c (remove_prefix): Remove.
5141 * file-find.h (remove_prefix): Likewise.
5142 * gcc-ar.c: Remove smartness of lookup.
5144 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5146 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5147 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5148 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5150 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5153 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5154 (setcc + and to xor + setcc): New peephole2.
5156 2017-10-19 Tom de Vries <tom@codesourcery.com>
5158 * doc/sourcebuild.texi (Test Directives, Variants of
5159 dg-require-support): Add dg-require-stack-size.
5161 2017-10-19 Martin Liska <mliska@suse.cz>
5164 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5165 that have a large alignment.
5166 (gimplify_target_expr): Likewise.
5168 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5170 PR rtl-optimization/82602
5171 * ira.c (rtx_moveable_p): Return false for volatile asm.
5173 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5176 * config/i386/i386-modes.def (CCGZ): New CC mode.
5177 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5178 * config/i386/predicates.md (ix86_comparison_operator):
5180 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5181 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5182 with double-word subtraction.
5183 (put_condition_code): Handle CCGZmode.
5185 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5187 * wide-int.cc (debug (const wide_int &)): New.
5188 (debug (const wide_int *)): New.
5189 (debug (const widest_int &)): New.
5190 (debug (const widest_int *)): New.
5192 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5195 * lra-constraints.c (curr_insn_transform): Use non-input operand
5196 instead of output one for matched reload.
5198 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5200 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5201 (tree-ssa-loop-ivopts.h): New header file.
5202 (struct builtin_info): New fields.
5203 (classify_builtin_1): Compute and record base and offset parts for
5204 memset builtin partition by calling strip_offset.
5205 (offset_cmp, fuse_memset_builtins): New functions.
5206 (finalize_partitions): Fuse adjacent memset partitions by calling
5208 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5209 Expose the interface.
5210 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5212 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5214 PR tree-optimization/82574
5215 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5216 that data reference must be executed exactly once per iteration
5217 against the outermost loop in nest.
5218 (classify_partition): Update call to above function.
5220 2017-10-18 Richard Biener <rguenther@suse.de>
5222 PR tree-optimization/82591
5223 * graphite.c (graphite_transform_loops): Move code gen message
5225 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5226 Here. Handle scop_to_isl_ast failing.
5227 (scop_to_isl_ast): Limit the number of ISL operations.
5229 2017-10-18 Richard Biener <rguenther@suse.de>
5231 * graphite-isl-ast-to-gimple.c
5232 (translate_isl_ast_to_gimple::set_rename): Simplify.
5233 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5234 (graphite_copy_stmts_from_block): ... here.
5235 (copy_bb_and_scalar_dependences): Simplify.
5236 (add_parameters_to_ivs_params): Canonicalize.
5237 (generate_entry_out_of_ssa_copies): Simplify.
5238 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5239 by passing in ISL dimension.
5240 (parameter_index_in_region_1): Rename to ...
5241 (parameter_index_in_region): ... this.
5242 (extract_affine): Adjust assert, pass down parameter index.
5243 (add_param_constraints): Use range-info when available.
5244 (build_scop_context): Adjust.
5245 * sese.c (new_sese_info): Adjust.
5246 (free_sese_info): Likewise.
5247 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5248 Remove unused typedefs.
5249 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5251 2017-10-18 Martin Liska <mliska@suse.cz>
5253 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5255 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5257 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5258 (s390_sched_init): Do not reset s390_sched_state if we entered the
5259 current basic block via a fallthru edge and all others are unlikely.
5261 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5263 * config/s390/s390.c (NUM_SIDES): New variable.
5264 (LONGRUNNING_THRESHOLD): New variable.
5265 (LATENCY_FACTOR): New variable.
5266 (s390_sched_score): Decrease score for long-running instructions on
5268 (s390_sched_variable_issue): Perform bookkeeping for long-running
5271 2017-10-18 Richard Biener <rguenther@suse.de>
5273 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5274 Simplify with removal of the parameter rename map.
5275 (set_rename): Likewise.
5276 (should_copy_to_new_region): Likewise.
5277 (graphite_copy_stmts_from_block): Likewise.
5278 (copy_bb_and_scalar_dependences): Remove initialization of
5279 unused copied_bb_map.
5281 (copy_internal_parameters): Likewise.
5282 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5283 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5284 Use INTEGRAL_TYPE_P.
5285 (parameter_index_in_region_1): Rename to ...
5286 (assign_parameter_index_in_region): ... this. Assert we have
5287 a parameter we handle.
5288 (scan_tree_for_params): Adjust.
5289 * sese.h (parameter_rename_map_t): Remove.
5290 (struct sese_info_t): Remove unused parameter_rename_map and
5291 copied_bb_map members.
5292 * sese.c (new_sese_info): Adjust.
5293 (free_sese_info): Likewise.
5295 2017-10-18 Martin Liska <mliska@suse.cz>
5298 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5299 on an abnormal edge.
5301 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5303 * doc/invoke.texi (ffunction-sections and fdata-sections):
5306 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5308 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5309 the use statement can throw internally.
5311 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5313 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5314 any RTX present on the RHS of a SET.
5315 * compare-elim.c (try_eliminate_compare): Restore comment.
5317 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5319 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5320 may be also called on tcc_type nodes.
5321 * langhooks.c (lhd_tree_size): Likewise.
5323 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5325 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5326 format_warning_at_substring.
5327 (maybe_warn): Convert source_range * param to a location_t. Pass
5328 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5329 (format_directive): Remove code to extract source_ranges and
5330 source_range * in favor of just a location_t.
5331 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5333 * substring-locations.c (format_warning_va): Convert
5334 source_range * param to a location_t.
5335 (format_warning_at_substring): Likewise.
5336 * substring-locations.h (format_warning_va): Likewise.
5337 (format_warning_at_substring): Likewise.
5339 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5341 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5343 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5344 and scatter/gather ops.
5346 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5347 vec_gather_load and vec_scatter_store.
5348 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5349 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5350 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5351 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5352 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5353 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5355 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5357 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5358 Detect FTST instruction and handle its register pops. Only pop
5359 second operand if can_pop_second_op is true.
5360 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5361 set can_pop_second_op to false in the compare_for_stack_reg call.
5363 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5364 output_fp_compare for stack register operands.
5365 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5366 instructions here. Do not emit stack register pops here. Assert
5367 that FCOMPP pops next to top stack register. Rewrite function.
5369 2017-10-17 Nathan Sidwell <nathan@acm.org>
5372 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5373 use DECL_ASSEMBLER_NAME_RAW.
5376 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5379 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5380 Wilco Dijkstra <wilco.dijkstra@arm.com>
5382 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5383 converstion to Pmode from the buf_addr.
5385 2017-10-17 Richard Biener <rguenther@suse.de>
5387 * graphite-dependences.c (scop_get_reads_and_writes): Change
5388 output parameters to references.
5390 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5392 PR 71026/tree-optimization
5393 * fold-const.c (distribute_real_division): Removed.
5394 (fold_binary_loc): Remove calls to distribute_real_divison.
5396 2017-10-17 Richard Biener <rguenther@suse.de>
5398 * graphite-scop-detection.c
5399 (scop_detection::stmt_has_simple_data_refs_p): Always use
5400 the full nest as region.
5401 (try_generate_gimple_bb): Likewise.
5402 * sese.c (scalar_evolution_in_region): Simplify now that
5403 SCEV can handle instantiation in regions.
5404 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5405 in the non-loop part of a function if requested.
5407 2017-10-17 Richard Biener <rguenther@suse.de>
5409 PR tree-optimization/82563
5410 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5412 (graphite_regenerate_ast_isl): Call it.
5413 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5415 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5417 PR tree-optimization/82549
5418 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5419 Formatting fixes. Instead of calling make_bit_field_ref with negative
5422 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5424 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5425 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5426 _mm_maskz_reduce_ss): New.
5427 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5428 __builtin_ia32_reducess_mask): Ditto..
5429 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5430 * config/i386/sse.md (reduces<mode>): Renamed to ...
5431 (reduces<mode><mask_scalar_name>): ... this.
5432 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5434 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5435 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5437 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5439 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5440 * selftest-run-tests.c (selftest::run_tests): Call
5441 selftest::unique_ptr_tests_cc_tests.
5442 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5443 * unique-ptr-tests.cc: New file.
5445 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5448 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5450 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5451 (make_hard_regno_born, make_hard_regno_dead): Update
5452 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5454 2017-10-16 Jeff Law <law@redhat.com>
5456 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5458 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5460 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5462 2017-10-16 Olivier Hainque <hainque@adacore.com>
5464 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5465 with_cpu if we were configured for an e500v2 target cpu name.
5467 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5469 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5470 * doc/invoke.texi: Document +nodsp as a valid extension for
5473 2017-10-16 Martin Liska <mliska@suse.cz>
5475 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5476 (test_set_range): Likewise.
5477 (test_range_functions): Rename to ...
5478 (test_bit_in_range): ... this.
5479 (sbitmap_c_tests): Add new test.
5481 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5483 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5485 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5486 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5488 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5490 * config/aarch64/aarch64-builtins.c
5491 (aarch64_types_quadopu_lane_qualifiers): New.
5492 (TYPES_QUADOPU_LANE): New.
5493 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5494 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5495 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5496 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5497 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5498 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5499 (Vdottype, DOTPROD): New.
5500 (sur): Add SDOT and UDOT.
5502 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5504 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5505 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5506 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5508 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5509 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5510 Enable TARGET_DOTPROD.
5511 (cortex-a75.cortex-a55): Likewise.
5512 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5514 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5516 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5517 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5519 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5521 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5522 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5523 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5524 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5525 * config/arm/types.md (neon_dot, neon_dot_q): New.
5526 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5528 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5530 * config/arm/arm.h (TARGET_DOTPROD): New.
5531 * config/arm/arm.c (arm_arch_dotprod): New.
5532 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5533 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5534 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5535 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5536 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5537 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5538 * doc/invoke.texi (armv8.2-a): Document dotprod
5540 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5542 * i386.c (ix86_vec_cost): New function.
5543 (ix86_rtx_costs): Handle vector operations better.
5544 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5545 * x86-tune-costs.h: Add new costs to all tables.
5547 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5549 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5551 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5552 divsd, sqrtss and sqrtsd
5553 * x86-tune-costs.h: Add new entries to all costs.
5554 (znver1_cost): Fix to match real instruction latencies.
5556 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5557 Michael Collison <michael.collison@arm.com>
5559 * compare-elim.c: Include emit-rtl.h.
5560 (can_merge_compare_into_arith): New function.
5561 (try_validate_parallel): Likewise.
5562 (try_merge_compare): Likewise.
5563 (try_eliminate_compare): Call the above when no previous clobber
5565 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5568 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5572 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5573 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5577 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5578 to be any operand_equal_p operands. For & (B - 1) require
5579 B to be power of 2. Recognize
5580 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5582 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5585 * optabs.c (expand_memory_blockage): Fix call of
5586 targetm.have_memory_blockage.
5588 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5591 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5592 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5593 objects to extra_objs instead of overwriting it.
5595 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5597 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5598 Use any_fp_register_operand as operand[3] predicate. Simplify
5599 equality test for operands[2] and operands[4] memory location.
5600 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5601 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5602 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5603 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5604 any_fp_register_operand as operand[1] predicate. Simplify
5605 equality test for operands[0] and operands[3] memory location.
5606 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5607 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5608 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5610 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5612 * target-insns.def: Add memory_blockage.
5613 * optabs.c (expand_memory_blockage): New function.
5614 (expand_asm_memory_barrier): Rename ...
5615 (expand_asm_memory_blockage): ... to this.
5616 (expand_mem_thread_fence): Call expand_memory_blockage
5617 instead of expand_asm_memory_barrier.
5618 (expand_mem_singnal_fence): Ditto.
5619 (expand_atomic_load): Ditto.
5620 (expand_atomic_store): Ditto.
5621 * doc/md.texi (Standard Pattern Names For Generation):
5622 Document memory_blockage instruction pattern.
5624 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5626 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5627 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5628 * config/rl78/rl78.md: New define_expand "adddi3".
5630 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5632 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5635 2017-10-13 Jeff Law <law@redhat.com>
5637 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5639 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5642 * internal-fn.c (expand_mul_overflow): If both operands have
5643 the same highpart of -1 or 0 and the topmost bit of lowpart
5644 is different, overflow is if res <= 0 rather than res < 0.
5646 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5648 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5649 TARGET_P9_VECTOR code for unaligned_load case.
5651 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5653 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5655 2017-10-13 Nathan Sidwell <nathan@acm.org>
5657 * tree-core.h (tree_contains_struct): Make bool.
5658 * tree.c (tree_contains_struct): Likewise.
5659 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5660 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5661 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5662 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5663 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5665 2017-10-13 Richard Biener <rguenther@suse.de>
5667 * graphite-isl-ast-to-gimple.c
5668 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5669 parameters and dominance check.
5670 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5671 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5672 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5673 Do not update SSA form here or do intermediate IL verification.
5674 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5675 (graphite_initialize): Remove check on the number of loops in
5676 the function and inline into graphite_transform_loops.
5677 (graphite_finalize): Inline into graphite_transform_loops.
5678 (graphite_transform_loops): Perform SSA update and IL verification
5680 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5682 2017-10-13 Richard Biener <rguenther@suse.de>
5684 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5685 graphite_expression_type_precision): Avoid global constructor
5687 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5688 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5689 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5690 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5691 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5692 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5694 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5697 * config/i386/i386.h (ix86_red_zone_size): New.
5698 * config/i386/i386.md (push peephole2s): Replace
5699 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5701 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5702 Alan Hayward <alan.hayward@arm.com>
5703 David Sherwood <david.sherwood@arm.com>
5705 * combine.c (can_change_dest_mode): Reject changes in
5706 REGMODE_NATURAL_SIZE.
5708 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5709 Alan Hayward <alan.hayward@arm.com>
5710 David Sherwood <david.sherwood@arm.com>
5712 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5713 (expand_debug_source_expr): Likewise.
5714 * combine.c (combine_simplify_rtx): Likewise.
5715 * cse.c (fold_rtx): Likewise.
5716 * fwprop.c (canonicalize_address): Likewise.
5717 * targhooks.c (default_shift_truncation_mask): Likewise.
5719 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5720 Alan Hayward <alan.hayward@arm.com>
5721 David Sherwood <david.sherwood@arm.com>
5723 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5724 (widened_mode): Likewise.
5725 (expand_unop): Likewise.
5726 * ree.c (transform_ifelse): Likewise.
5727 (merge_def_and_ext): Likewise.
5728 (combine_reaching_defs): Likewise.
5729 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5731 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5732 Alan Hayward <alan.hayward@arm.com>
5733 David Sherwood <david.sherwood@arm.com>
5735 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5736 * combine.c (gen_lowpart_for_combine): Likewise.
5737 * dwarf2out.c (rtl_for_decl_location): Likewise.
5738 * final.c (alter_subreg): Likewise.
5739 * rtlhooks.c (gen_lowpart_general): Likewise.
5740 (gen_lowpart_if_possible): Likewise.
5742 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5743 Alan Hayward <alan.hayward@arm.com>
5744 David Sherwood <david.sherwood@arm.com>
5746 * calls.c (expand_call): Use subreg_lowpart_offset.
5747 * cse.c (cse_insn): Likewise.
5748 * regcprop.c (copy_value): Likewise.
5749 (copyprop_hardreg_forward_1): Likewise.
5751 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5754 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5755 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5756 =Q constraints to +Q and into insn condition add check
5757 that operands[0] and operands[1] are equal.
5758 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5759 =Q constraints to +Q and into insn condition add check
5760 that operands[0] is equal to either operands[1] or operands[2].
5763 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5764 instead of handling MINUS_EXPR twice (once for each argument),
5765 canonicalize operand order and handle just once, use rtype where
5766 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5769 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5770 any values of __C while still being pattern recognizable as a simple
5773 2017-10-13 Richard Biener <rguenther@suse.de>
5775 PR tree-optimization/82451
5777 2017-10-02 Richard Biener <rguenther@suse.de>
5779 PR tree-optimization/82355
5780 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5781 a mapping for the enclosing loop but avoid generating one for
5783 (copy_bb_and_scalar_dependences): Remove premature codegen
5784 error on PHIs in blocks duplicated into multiple places.
5785 * graphite-scop-detection.c
5786 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5787 in the region use it as loop and nest to analyze the DR in.
5788 (try_generate_gimple_bb): Likewise.
5789 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5790 (add_loop_constraints): For blocks in a loop not in the region
5791 create a dimension with a single iteration.
5792 * sese.h (gbb_loop_at_index): Remove assert.
5794 * cfgloop.c (loop_preheader_edge): For the loop tree root
5795 return the single successor of the entry block.
5796 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5797 Reset the SCEV hashtable and niters.
5798 * graphite-scop-detection.c
5799 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5800 assert that we only have POLYNOMIAL_CHREC that vary in loops
5801 contained in the region.
5802 (scop_detection::graphite_can_represent_expr): Adjust.
5803 (scop_detection::stmt_has_simple_data_refs_p): For loops
5804 not in the region set loop to NULL. The nest is now the
5805 entry edge to the region.
5806 (try_generate_gimple_bb): Likewise.
5807 * sese.c (scalar_evolution_in_region): Adjust for
5808 instantiate_scev change.
5809 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5810 Make nest parameter the edge into the region.
5811 (create_data_ref): Likewise.
5812 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5813 entry edge into a region and adjust instantiate_scev calls.
5814 (create_data_ref): Likewise.
5815 (graphite_find_data_references_in_stmt): Likewise.
5816 (find_data_references_in_stmt): Pass the loop preheader edge
5817 from the nest argument.
5818 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5819 parameter the edge into the region.
5820 (instantiate_parameters): Use the loop preheader edge as entry.
5821 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5823 (get_instantiated_value_entry): Make instantiate_below parameter
5824 the edge into the region.
5825 (instantiate_scev_name): Likewise. Adjust dominance checks,
5826 when we cannot use loop-based instantiation instantiate by
5827 walking use-def chains.
5828 (instantiate_scev_poly): Adjust.
5829 (instantiate_scev_binary): Likewise.
5830 (instantiate_scev_convert): Likewise.
5831 (instantiate_scev_not): Likewise.
5832 (instantiate_array_ref): Remove.
5833 (instantiate_scev_3): Likewise.
5834 (instantiate_scev_2): Likewise.
5835 (instantiate_scev_1): Likewise.
5836 (instantiate_scev_r): Do not blindly handle N-operand trees.
5837 Do not instantiate array-refs. Handle all constants and invariants.
5838 (instantiate_scev): Make instantiate_below parameter
5839 the edge into the region.
5840 (resolve_mixers): Use the loop preheader edge for the region
5841 parameter to instantiate_scev_r.
5842 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5844 2017-10-13 Richard Biener <rguenther@suse.de>
5846 PR tree-optimization/82525
5847 * graphite-isl-ast-to-gimple.c
5848 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5850 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5851 Fail code generation when we cannot represent the isl integer.
5852 (binary_op_to_tree): Elide modulo operations that are no-ops
5853 in the type we code generate. Remove now superfluous code
5856 2017-10-13 Richard Biener <rguenther@suse.de>
5858 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5859 (scop_detection::harmful_loop_in_region): Remove premature
5860 IV type restriction.
5861 (scop_detection::graphite_can_represent_scev): We can handle
5862 pointer IVs just fine.
5864 2017-10-13 Alan Modra <amodra@gmail.com>
5866 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5867 "Clobbers and Scratch Registers". Add paragraph on
5868 alternative to clobbers for scratch registers and OpenBLAS
5871 2017-10-13 Alan Modra <amodra@gmail.com>
5873 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5874 example of a memory input for a string of known length. Move
5875 commentary out of table. Add a number of new examples
5876 covering array memory inputs.
5878 2017-10-12 Martin Liska <mliska@suse.cz>
5880 PR tree-optimization/82493
5881 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5882 (test_range_functions): New function.
5883 (sbitmap_c_tests): Likewise.
5884 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5885 * selftest.h (sbitmap_c_tests): New function.
5887 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5889 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5891 * config/rs6000/amo.h: Fix spacing issue.
5893 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5896 * config/i386/i386.md (*ashl<mode>3_mask_1,
5897 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5898 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5901 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5903 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5904 (profile_probability): Set max_probability
5905 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5907 * profile-count.c (profile_probability::differs_from_p): Do not
5908 rely on max_probaiblity == 10000
5910 2017-10-12 Jeff Law <law@redhat.com>
5912 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5915 2017-10-12 Martin Sebor <msebor@redhat.com>
5919 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5920 (handle_alias_pairs): Call it.
5921 * common.opt (-Wattribute-alias): New option.
5922 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5923 * doc/invoke.texi (-Wattribute-alias): Document.
5925 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5928 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5930 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5932 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5933 (make_hard_regno_born, make_hard_regno_dead): Update
5934 bb_killed_pseudos and bb_gen_pseudos.
5936 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5938 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5940 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5942 * config/alpha/alpha.c (alpha_split_conditional_move):
5943 Use std::swap instead of manually swapping.
5944 (alpha_stdarg_optimize_hook): Ditto.
5945 (alpha_canonicalize_comparison): Ditto.
5947 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5949 * tree-loop-distribution.c (struct builtin_info): New struct.
5950 (struct partition): Refactor fields into struct builtin_info.
5951 (partition_free): Free struct builtin_info.
5952 (build_size_arg_loc, build_addr_arg_loc): Delete.
5953 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5954 information from struct builtin_info.
5955 (find_single_drs): New function refactored from classify_partition.
5956 Also moved builtin validity checks to this function.
5957 (compute_access_range, alloc_builtin): New functions.
5958 (classify_builtin_st, classify_builtin_ldst): New functions.
5959 (classify_partition): Refactor code into functions find_single_drs,
5960 classify_builtin_st and classify_builtin_ldst.
5961 (distribute_loop): Don't do runtime alias check when distributing
5963 (find_seed_stmts_for_distribution): New function.
5964 (pass_loop_distribution::execute): Refactor code finding seed
5965 stmts into above function. Support distribution for the innermost
5966 two-level loop nest. Adjust dump information.
5968 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5970 * tree-loop-distribution.c: Adjust the general comment.
5971 (NUM_PARTITION_THRESHOLD): New macro.
5972 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5973 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5974 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5975 in call to build_partition_graph.
5976 (finalize_partitions): New parameter. Make loop distribution more
5977 conservative by fusing more partitions.
5978 (distribute_loop): Don't do runtime alias check in case of loop nest
5980 (find_seed_stmts_for_distribution): New function.
5981 (prepare_perfect_loop_nest): New function.
5982 (pass_loop_distribution::execute): Refactor code finding seed stmts
5983 and loop nest into above functions. Support loop nest distribution.
5984 Adjust dump information accordingly.
5986 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5988 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5989 and set PTYPE_SEQUENTIAL for merged partition.
5991 2017-10-12 Richard Biener <rguenther@suse.de>
5993 PR tree-optimization/69728
5995 2017-09-19 Richard Biener <rguenther@suse.de>
5997 PR tree-optimization/69728
5998 * graphite-sese-to-poly.c (schedule_error): New global.
5999 (add_loop_schedule): Handle empty domain by failing the
6001 (build_original_schedule): Handle schedule_error.
6003 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6004 domain by returning an unchanged schedule.
6006 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6008 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6009 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6011 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6013 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6016 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6019 * expr.c (store_field): Don't optimize away bitsize == 0 store
6020 from CALL_EXPR with addressable return type.
6022 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6024 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6025 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6026 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6027 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6028 TARGET_ISEL instead of TARGET_ISEL<sel>.
6030 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6032 * config/rs6000/rs6000.c
6033 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6035 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6037 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6039 (reg_or_cint_operand): Fix comment.
6040 (reg_or_zero_operand): New predicate.
6041 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6042 * config/rs6000/rs6000.c (output_isel): Delete.
6043 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6044 instead of reg_or_cint_operand. Output instruction directly (not via
6046 (isel_unsigned_<mode>): Ditto.
6047 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6048 gpc_reg_operand. Add an instruction alternative for this. Output
6049 instruction directly.
6050 (*isel_reversed_unsigned_<mode>): Ditto.
6052 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6054 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6055 (TARGET_CANONICALIZE_COMPARISON): Define.
6057 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6060 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6061 Check whether the dest is REG before adding REG_EQUIV note.
6063 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6066 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6068 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6069 (make_hard_regno_born, make_hard_regno_dead): Update
6070 bb_killed_pseudos and bb_gen_pseudos.
6072 2017-10-11 Nathan Sidwell <nathan@acm.org>
6074 * incpath.h (enum incpath_kind): Name enum, prefix values.
6075 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6076 * incpath.c (heads, tails): Use INC_MAX.
6077 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6078 (merge_include_chains, split_quote_chain,
6079 register_include_chains): Update incpath_kind names.
6080 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6081 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6083 (add_framework_path, darwin_register_objc_includes): Likewise.
6084 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6086 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6088 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6089 Do not use float_operator operator predicate.
6090 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6091 * config/i386/predicates.md (float_operator): Remove predicate.
6093 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6095 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6096 (*jccxf_i387): Ditto.
6097 (*jcc<mode>_i387): Ditto.
6098 (*jccu<mode>_i387): Ditto.
6099 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6100 (*jcc_*_i387 splitters): Remove.
6101 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6102 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6103 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6106 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6108 * profile-count.h (slow_safe_scale_64bit): New function.
6109 (safe_scale_64bit): New inline.
6110 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6111 * profile-count.c: Include wide-int.h
6112 (slow_safe_scale_64bit): New.
6114 2017-10-11 Nathan Sidwell <nathan@acm.org>
6116 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6117 HAS_DECL_ASSEMBLER_NAME_P.
6118 * gimple-expr.c (gimple_decl_printable_name: Check
6119 HAS_DECL_ASSEMBLER_NAME_P too.
6120 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6121 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6122 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6123 * passes.c (rest_of_decl_compilation): Check
6124 HAS_DECL_ASSEMBLER_NAME_P too.
6125 * recog.c (verify_changes): Likewise.
6126 * tree-pretty-print.c (dump_decl_name): Likewise.
6127 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6129 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6130 (SET_DECL_ASSEMBLER_NAME): Use it.
6131 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6132 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6133 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6135 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6137 * config.gcc (i386, x86_64): Add extra objects.
6138 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6139 (ix86_min_insn_size): Declare.
6140 (ix86_issue_rate): Declare.
6141 (ix86_adjust_cost): Declare.
6142 (ia32_multipass_dfa_lookahead): Declare.
6143 (ix86_macro_fusion_p): Declare.
6144 (ix86_macro_fusion_pair_p): Declare.
6145 (ix86_bd_has_dispatch): Declare.
6146 (ix86_bd_do_dispatch): Declare.
6147 (ix86_core2i7_init_hooks): Declare.
6148 (ix86_atom_sched_reorder): Declare.
6149 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6150 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6151 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6152 (rip_relative_addr_p): Rename to ...
6153 (ix86_rip_relative_addr_p): ... this one; export.
6154 (memory_address_length): Update.
6155 (ix86_issue_rate): Move to x86-tune-sched.c.
6156 (ix86_flags_dependent): Move to x86-tune-sched.c.
6157 (ix86_agi_dependent): Move to x86-tune-sched.c.
6158 (exact_dependency_1): Move to x86-tune-sched.c.
6159 (exact_store_load_dependency): Move to x86-tune-sched.c.
6160 (ix86_adjust_cost): Move to x86-tune-sched.c.
6161 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6162 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6163 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6164 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6165 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6166 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6167 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6168 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6169 (min_insn_size): Rename to ...
6170 (ix86_min_insn_size): ... this one; export.
6171 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6172 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6173 (core2i7_first_cycle_multipass_backtrack): Move to
6174 x86-tune-sched-core.c.
6175 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6176 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6177 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6178 (ix86_avoid_jump_mispredicts): Update.
6179 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6180 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6181 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6182 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6183 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6184 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6185 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6186 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6187 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6188 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6189 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6190 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6191 (BIG): Move to ix86-tune-sched-bd.c.
6192 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6193 (enum insn_path): Move to ix86-tune-sched-bd.c.
6194 (get_mem_group): Move to ix86-tune-sched-bd.c.
6195 (is_cmp): Move to ix86-tune-sched-bd.c.
6196 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6197 (is_branch): Move to ix86-tune-sched-bd.c.
6198 (is_prefetch): Move to ix86-tune-sched-bd.c.
6199 (init_window): Move to ix86-tune-sched-bd.c.
6200 (allocate_window): Move to ix86-tune-sched-bd.c.
6201 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6202 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6203 (process_end_window): Move to ix86-tune-sched-bd.c.
6204 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6205 (find_constant): Move to ix86-tune-sched-bd.c.
6206 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6207 (has_immediate): Move to ix86-tune-sched-bd.c.
6208 (get_insn_path): Move to ix86-tune-sched-bd.c.
6209 (get_insn_group): Move to ix86-tune-sched-bd.c.
6210 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6211 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6212 (add_insn_window): Move to ix86-tune-sched-bd.c.
6213 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6214 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6215 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6216 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6217 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6218 (do_dispatch): Move to ix86-tune-sched-bd.c.
6219 (has_dispatch): Move to ix86-tune-sched-bd.c.
6220 * config/i386/t-i386: Add new object files.
6221 * config/i386/x86-tune-costs.h: New file.
6222 * config/i386/x86-tune-sched-atom.c: New file.
6223 * config/i386/x86-tune-sched-bd.c: New file.
6224 * config/i386/x86-tune-sched-core.c: New file.
6225 * config/i386/x86-tune-sched.c: New file.
6227 2017-10-11 Liu Hao <lh_mouse@126.com>
6229 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6230 the generic version below instead.
6231 (should_colorize): Recognize Windows consoles as terminals
6233 * pretty-print.c [__MINGW32__] (write_all): New function.
6234 [__MINGW32__] (find_esc_head): Likewise.
6235 [__MINGW32__] (find_esc_terminator): Likewise.
6236 [__MINGW32__] (eat_esc_sequence): Likewise.
6237 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6239 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6242 2017-10-11 Richard Biener <rguenther@suse.de>
6244 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6245 Properly call analyze_scalar_evolution with the loop of the stmt.
6247 2017-10-11 Richard Biener <rguenther@suse.de>
6249 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6250 * tree-core.h (tree_base): Add chrec_var union member.
6251 * tree.h (CHREC_VAR): Remove.
6252 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6253 * tree-chrec.h (build_polynomial_chrec): Adjust.
6254 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6255 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6257 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6259 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6260 * match.pd: ... here.
6261 ((T) X == (T) Y): Relax condition.
6263 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6265 PR tree-optimization/82472
6266 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6268 (break_alias_scc_partitions): Update postorder number.
6270 2017-10-11 Martin Liska <mliska@suse.cz>
6273 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6275 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6277 2017-10-11 Martin Liska <mliska@suse.cz>
6279 * print-rtl.c (print_insn): Move declaration of idbuf
6280 to same scope as name.
6282 2017-10-11 Martin Liska <mliska@suse.cz>
6287 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6290 2017-10-11 Martin Liska <mliska@suse.cz>
6293 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6296 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6298 * config.gcc (default_gnu_indirect_function): Default to yes for
6299 arm*-*-linux* with glibc.
6301 2017-10-11 Richard Biener <rguenther@suse.de>
6303 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6304 default-defs and types we do not want to analyze.
6305 (interpret_loop_phi): Replace unreachable code with an assert.
6306 (compute_scalar_evolution_in_loop): Remove and inline ...
6307 (analyze_scalar_evolution_1): ... here, replacing condition with
6308 what makes the intent clearer. Remove handling of cases
6309 get_scalar_evolution now handles.
6311 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6313 PR rtl-optimization/81434
6314 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6315 comment for main loop. In sched_group_found if, also add checks for
6316 pass and min_cost_group.
6318 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6320 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6321 (rs6000_insn_cost): New function.
6322 * config/rs6000/rs6000.md (cost): New attribute.
6324 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6325 H.J. Lu <hongjiu.lu@intel.com>
6329 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6330 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6331 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6332 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6333 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6334 OPTION_MASK_ISA_MMX and some other bit set require both
6335 mmx and the ISAs without the mmx bit.
6336 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6337 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6338 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6339 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6340 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6341 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6342 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6343 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6344 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6345 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6346 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6347 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6348 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6349 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6350 Add OPTION_MASK_ISA_MMX.
6352 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6354 * config.gcc (armv7*-*-freebsd*): New target.
6355 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6357 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6359 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6360 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6363 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6365 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6366 a template parameter.
6367 (WIDE_INT_REF_FOR): Update accordingly.
6368 * tree.h (wi::int_traits <const_tree>): Delete.
6369 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6370 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6371 (wi::tree_to_wide_ref): New typedef.
6372 (wi::to_wide): New function.
6373 * calls.c (get_size_range): Use wi::to_wide when operating on
6375 * cgraph.c (cgraph_node::create_thunk): Likewise.
6376 * config/i386/i386.c (ix86_data_alignment): Likewise.
6377 (ix86_local_alignment): Likewise.
6378 * dbxout.c (stabstr_O): Likewise.
6379 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6380 * expr.c (const_vector_from_tree): Likewise.
6381 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6382 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6383 (fold_negate_expr_1, int_const_binop_1, const_binop)
6384 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6385 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6386 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6387 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6388 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6389 (fold_not_const, round_up_loc): Likewise.
6390 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6391 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6392 (alloca_call_type): Likewise.
6393 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6394 * godump.c (go_output_typedef): Likewise.
6395 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6396 * internal-fn.c (get_min_precision): Likewise.
6397 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6398 * ipa-polymorphic-call.c
6399 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6400 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6401 (ipa_modify_call_arguments): Likewise.
6402 * match.pd: Likewise.
6403 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6404 * print-tree.c (print_node_brief, print_node): Likewise.
6405 * stmt.c (expand_case): Likewise.
6406 * stor-layout.c (layout_type): Likewise.
6407 * tree-affine.c (tree_to_aff_combination): Likewise.
6408 * tree-cfg.c (group_case_labels_stmt): Likewise.
6409 * tree-data-ref.c (dr_analyze_indices): Likewise.
6410 (prune_runtime_alias_test_list): Likewise.
6411 * tree-dump.c (dequeue_and_dump): Likewise.
6412 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6413 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6414 * tree-pretty-print.c (dump_generic_node): Likewise.
6415 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6416 (simple_iv_with_niters): Likewise.
6417 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6418 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6419 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6420 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6421 (refine_value_range_using_guard, number_of_iterations_ne_max)
6422 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6423 (get_cst_init_from_scev, record_nonwrapping_iv)
6424 (scev_var_range_cant_overflow): Likewise.
6425 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6426 * tree-ssa-pre.c (compute_avail): Likewise.
6427 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6428 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6429 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6430 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6431 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6432 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6433 (dump_case_nodes, try_switch_expansion): Likewise.
6434 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6435 (vect_do_peeling): Likewise.
6436 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6437 * tree-vect-stmts.c (vectorizable_load): Likewise.
6438 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6439 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6440 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6441 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6442 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6443 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6444 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6445 (evrp_dom_walker::before_dom_children): Likewise.
6446 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6447 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6448 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6449 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6450 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6451 (get_range_pos_neg): Likewise.
6452 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6453 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6454 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6455 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6456 * config/avr/avr.c (avr_fold_builtin): Likewise.
6457 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6458 * config/msp430/msp430.c (msp430_attr): Likewise.
6459 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6460 * config/powerpcspe/powerpcspe-c.c
6461 (altivec_resolve_overloaded_builtin): Likewise.
6462 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6463 (rs6000_expand_ternop_builtin): Likewise.
6464 * config/rs6000/rs6000-c.c
6465 (altivec_resolve_overloaded_builtin): Likewise.
6466 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6467 (rs6000_expand_ternop_builtin): Likewise.
6468 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6470 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6472 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6473 when copying loop nest with only one inner loop.
6475 2017-10-10 Richard Biener <rguenther@suse.de>
6477 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6478 blocks if SCEV is active.
6479 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6481 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6482 (scev_initialize): Assert we are not yet initialized.
6484 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6486 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6487 inner loop's exit stmt by making it always exit the loop, otherwise
6488 we would generate an infinite empty loop.
6490 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6492 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6493 renaming variables in new preheader if it's deleted.
6495 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6497 * tree-loop-distribution.c (struct partition): Remove unused field
6498 loops of the structure.
6499 (partition_alloc, partition_free): Ditto.
6500 (build_rdg_partition_for_vertex): Ditto.
6502 2017-10-09 Jeff Law <law@redhat.com>
6504 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6505 return type to match prototype and documentation.
6507 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6509 * config/rs6000/rs6000.c (processor_costs): Move to ...
6510 * config/rs6000/rs6000.h: ... here.
6511 (rs6000_cost): Declare.
6513 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6515 * except.c (setjmp_fn): New global variable.
6516 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6517 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6518 if DONT_USE_BUILTIN_SETJMP is defined.
6520 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6522 * target.def (insn_cost): New hook.
6523 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6524 * doc/tm.texi: Regenerate.
6525 * rtlanal.c (insn_cost): Use the new hook.
6527 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6529 * combine.c (combine_validate_cost): Compute the new insn_cost,
6530 not just pattern_cost.
6531 (try_combine): Adjust comment.
6533 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6535 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6537 * combine.c (uid_insn_cost): Adjust comment.
6538 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6540 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6541 * dse.c (find_shift_sequence): Ditto.
6542 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6543 (bb_valid_for_noce_process_p): Use pattern_cost.
6544 * rtl.h (insn_rtx_cost): Delete.
6545 (pattern_cost): New prototype.
6546 (insn_cost): New prototype.
6547 * rtlanal.c (insn_rtx_cost): Rename to...
6548 (pattern_cost): ... this.
6551 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6553 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6554 (*jcc<mode>_0_r_i387): Ditto.
6555 (*jccxf_r_i387): Ditto.
6556 (*jcc<mode>_r_i387): Ditto.
6557 (*jccu<mode>_r_i387): Ditto.
6558 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6559 (*jcc): Rename from *jcc_1.
6561 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6563 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6564 deferred rescans after the lvx/stvx recombination pre-pass.
6566 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6568 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6569 memory operation instruction support.
6570 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6571 (rs6000-ibm-aix[789]*): Likewise.
6572 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6573 Document new functions.
6575 2017-10-09 Richard Biener <rguenther@suse.de>
6577 PR tree-optimization/82397
6578 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6579 equality only for semantically equal trees.
6581 2017-10-09 Richard Biener <rguenther@suse.de>
6583 PR tree-optimization/82449
6584 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6585 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6586 allow constant addresses.
6587 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6590 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6592 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6595 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6598 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6601 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6604 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6606 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6609 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6610 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6612 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6614 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6615 (WI_BINARY_PREDICATE_RESULT): Likewise.
6616 (wi::binary_traits::operator_result): New type.
6617 (wi::binary_traits::predicate_result): Likewise.
6618 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6619 (generic_wide_int::operator==, generic_wide_int::operator!=)
6620 (generic_wide_int::operator&, generic_wide_int::and_not)
6621 (generic_wide_int::operator|, generic_wide_int::or_not)
6622 (generic_wide_int::operator^, generic_wide_int::operator+
6623 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6625 (operator~, unary operator-, operator==, operator!=, operator&)
6626 (operator|, operator^, operator+, binary operator-, operator*): New
6628 * expr.c (get_inner_reference): Use wi::bit_and_not.
6629 * fold-const.c (fold_binary_loc): Likewise.
6630 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6631 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6632 (bit_value_binop): Likewise.
6633 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6634 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6635 (extract_range_from_binary_expr_1): Likewise.
6636 (masked_increment): Likewise.
6637 (simplify_bit_ops_using_ranges): Likewise.
6639 2017-10-09 Martin Jambor <mjambor@suse.cz>
6642 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6643 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6644 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6646 (hsa_fixup_mov_insn_type): New function.
6647 (hsa_op_with_type::get_in_type): Use it.
6648 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6649 immediates in an assert.
6650 (hsa_op_with_type::extend_int_to_32bit): New method.
6651 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6652 types. Convert to dest type if necessary.
6653 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6654 (reg_for_gimple_ssa): Pass false as min32int to
6655 hsa_type_for_scalar_tree_type.
6656 (gen_hsa_addr): Fixup type when creating addresable temporary.
6657 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6658 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6659 necessary. Call hsa_fixup_mov_insn_type.
6660 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6661 extend operands and convert to dest type if necessary.
6662 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6663 to dest type if necessary.
6664 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6665 if conversion nt necessary and size matches.
6666 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6667 to dest type if necessary.
6668 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6669 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6671 (gen_hsa_clrsb): Likewise.
6672 (gen_hsa_ffs): Likewise.
6673 (gen_hsa_divmod): Extend operands and convert to dest type if
6675 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6677 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6679 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6680 Remove empty default arguments. Use a brace block as output
6682 (conditional return): Ditto.
6684 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6685 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6687 (group_ending_nop): Ditto.
6688 (doloop_end): Ditto.
6689 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6690 (splitters for those): Ditto.
6692 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6694 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6695 a conditional jump (and the compare for it) so that pc_rtx is the
6697 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6698 for the deleted and renamed ctr<mode>_internal[234] patterns.
6699 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6700 Delete second conditional return pattern.
6701 (ctr<mode>_internal2): Delete this second bdnz pattern.
6702 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6703 (ctr<mode>_internal4): Delete this second bdz pattern.
6705 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6707 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6708 (always_initialized_rtx_for_ssa_name_p): New predicate.
6709 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6710 (finish_out_of_ssa): Free new field of SA.
6711 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6712 * tree-ssa-coalesce.c: Include tree-ssa.h.
6713 (get_parm_default_def_partitions): Remove extern keyword.
6714 (get_undefined_value_partitions): New function.
6715 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6716 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6718 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6720 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6722 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6724 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6726 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6727 for targets that preffer 128bit.
6729 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6731 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6733 2017-10-08 Olivier Hainque <hainque@adacore.com>
6735 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6736 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6738 (thumb_set_return_address): Likewise.
6740 2017-10-08 Olivier Hainque <hainque@adacore.com>
6742 * common/config/arm/arm-common.c (arm_except_unwind_info):
6743 Handle DWARF2_UNWIND_INFO.
6745 2017-10-07 Michael Collison <michael.collison@arm.com>
6747 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6750 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6752 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6753 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6754 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6755 defined, force the creation of a new block for a dispatch label.
6757 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6759 * invoke.texi (Wsuggest-attribute=cold): Document.
6760 * common.opt (Wsuggest-attribute=cold): New
6761 * ipa-pure-const.c (warn_function_cold): New function.
6762 * predict.c (compute_function_frequency): Use it.
6763 * predict.h (warn_function_cold): Declare.
6765 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6767 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6770 2017-10-06 Martin Liska <mliska@suse.cz>
6772 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6773 keyword for member functions.
6774 (struct sanopt_tree_couple): New struct.
6775 (struct sanopt_tree_couple_hash): New function.
6776 (struct sanopt_ctx): Add new hash_map.
6777 (has_dominating_ubsan_ptr_check): New function.
6778 (record_ubsan_ptr_check_stmt): Likewise.
6779 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6780 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6781 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6783 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6786 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6787 aarch64_simd_valid_immediate on CONST_VECTORs.
6788 (aarch64_reg_or_bic_imm): Likewise.
6790 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6792 PR rtl-optimization/82396
6793 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6795 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6797 * graphite-dependences.c (scop_get_reads): Move code to...
6798 (scop_get_must_writes): Move code to...
6799 (scop_get_may_writes): Move code to...
6800 (scop_get_reads_and_writes): ... here.
6801 (scop_get_dependences): Call scop_get_reads_and_writes.
6803 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6805 PR tree-optimization/82434
6806 * fold-const.h (can_native_encode_type_p,
6807 can_native_encode_string_p): Remove.
6808 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6809 don't encode anything, just return what would be otherwise returned.
6810 (native_encode_fixed, native_encode_complex, native_encode_vector):
6812 (native_encode_string): Likewise. Inline by hand
6813 can_native_encode_string_p.
6814 (can_native_encode_type_p): Remove.
6815 (can_native_encode_string_p): Remove.
6816 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6817 STRING_CSTs using can_native_encode_string_p, test all
6818 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6819 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6820 argument from native_encode_expr.
6821 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6822 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6825 2017-10-06 Richard Biener <rguenther@suse.de>
6827 PR tree-optimization/82397
6828 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6829 operand_equal_p but rely on data_ref_compare_tree for detecting
6831 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6832 to match up with dr_group_sort_cmp.
6834 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6837 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6839 * config/s390/s390-builtin-types.def: Regenerate.
6841 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6844 * config/s390/s390-builtin-types.def: Regenerate.
6845 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6846 Change flag from B_VXE to B_VX.
6847 (s390_vec_min_dbl): Remove B_VXE flag.
6849 2017-10-06 Richard Biener <rguenther@suse.de>
6851 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6852 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6853 translate_isl_ast_to_gimple::is_valid_rename,
6854 translate_isl_ast_to_gimple::get_rename,
6855 translate_isl_ast_to_gimple::get_def_bb_for_const,
6856 translate_isl_ast_to_gimple::get_new_name,
6857 translate_isl_ast_to_gimple::collect_all_ssa_names,
6858 translate_isl_ast_to_gimple::copy_loop_phi_args,
6859 translate_isl_ast_to_gimple::collect_all_ssa_names,
6860 translate_isl_ast_to_gimple::copy_loop_phi_args,
6861 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6862 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6863 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6864 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6865 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6866 translate_isl_ast_to_gimple::copy_cond_phi_args,
6867 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6868 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6869 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6870 translate_isl_ast_to_gimple::rename_uses,
6871 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6872 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6873 (set_rename_for_each_def): Likewise.
6874 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6875 here. Handle rewriting SCEV analyzable uses here.
6876 (copy_bb_and_scalar_dependences): Generate code for PHI
6878 (graphite_regenerate_ast_isl): Adjust.
6879 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6880 (add_write, add_read): New functions.
6881 (build_cross_bb_scalars_def): Use it and simplify.
6882 (build_cross_bb_scalars_use): Likewise.
6883 (graphite_find_cross_bb_scalar_vars): Inline into...
6884 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6885 simulating out-of-SSA. Compute liveout and add dependencies.
6886 (build_scops): Force an empty entry block.
6887 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6889 (sese_build_liveouts): Declare.
6890 (sese_trivially_empty_bb_p): Likewise.
6891 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6892 compute liveout and debug_liveout.
6893 (sese_bad_liveouts_use): Remove.
6894 (sese_reset_debug_liveouts_bb): Likewise.
6895 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6896 (sese_build_liveouts): Build liveout and debug_liveout and store
6898 (new_sese_info): Adjust.
6899 (free_sese_info): Likewise.
6900 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6901 do not build liveout here.
6902 (move_sese_in_condition): Adjust region entry.
6903 (scev_analyzable_p): Match up with chrec_apply requirements.
6904 (sese_trivially_empty_bb_p): New.
6905 * tree-into-ssa.c (get_reaching_def): Properly support generating
6906 default-defs for incremental rewrite of anonymous names.
6908 2017-10-06 Richard Biener <rguenther@suse.de>
6910 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6911 precision do not perform modulo reduction.
6913 2017-10-06 Richard Biener <rguenther@suse.de>
6915 PR tree-optimization/82436
6916 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6917 conservatively choose the vectorization factor when checking
6918 whether we can perform the required load permutation.
6919 (vect_transform_slp_perm_load): Assert when we may not fail.
6921 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6923 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6924 message for incompatible -msdata=* and -mcall-* options.
6926 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6928 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6929 rate for post-reload scheduling.
6931 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6933 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6935 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6937 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6938 to improve monte carlo in scimark.
6940 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6942 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6943 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6944 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6945 pentium4_cost, nocona_cost): Set reassociation width to 1.
6946 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6947 width to 2 for fp operations and 1 otherwise.
6948 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6950 (atom_cost): Set reassociation width to 2.
6951 (slm_cost, generic_cost): Set fp reassociation width
6952 to 2 and 1 otherwise.
6953 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6954 (core_cost): Set fp reassociation width to 4 and vector to 2.
6955 (ix86_reassociation_width): Rewrite using cost table; special case
6956 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6957 and TARGET_AVX128_OPTIMAL.
6958 * config/i386/i386.h (processor_costs): Add
6959 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6960 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6961 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6962 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6963 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6964 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6966 2017-10-05 Nathan Sidwell <nathan@acm.org>
6968 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6970 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6972 * config/arm/arm.c (arm_test_fpu_data): New.
6973 (arm_run_selftests): Call arm_test_fpu_data.
6975 2017-10-04 Nathan Sidwell <nathan@acm.org>
6977 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6979 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6981 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6983 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6984 check type for aarch64_simd_valid_immediate.
6985 (aarch64_output_simd_mov_immediate): Update prototype.
6986 (aarch64_simd_valid_immediate): Update prototype.
6987 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6988 support for ORR-immediate.
6989 (and<mode>3): modified pattern to add support for BIC-immediate.
6990 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6991 now checks for valid immediate for BIC and ORR based on new enum
6993 (aarch64_output_simd_mov_immediate): Function now used to output
6994 BIC/ORR imm as well based on new enum argument.
6995 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6997 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6998 (aarch64_reg_or_bic_imm): Likewise.
7000 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7002 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7003 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7005 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7009 PR rtl-optimization/82396
7010 * haifa-sched.c (autopref_multipass_init): Simplify
7012 (autopref_rank_data): Simplify sort order.
7013 * sched-int.h (autopref_multipass_data_): Remove
7014 multi_mem_insn_p, min_offset and max_offset.
7016 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7018 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7020 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7022 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7025 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7027 * doc/sourcebuild.texi: Document vect_long_mult.
7029 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7031 PR tree-optimization/82413
7032 * fold-const.c (build_range_check): Use widest_int when comparing
7033 the maximum ETYPE value with HIGH.
7035 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7037 PR rtl-optimization/82396
7038 * haifa-sched.c (autopref_multipass_init): Simplify
7040 (autopref_rank_data): Simplify sort order.
7041 * sched-int.h (autopref_multipass_data_): Remove
7042 multi_mem_insn_p, min_offset and max_offset.
7044 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7046 PR tree-optimization/82381
7047 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7048 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7051 PR tree-optimization/82374
7052 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7053 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7054 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7055 current_function_decl to the new decl.
7057 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7059 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7060 helper macro for IEEE float128 hardware built-in functions.
7061 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7063 (TRUNCF128_ODD): Likewise.
7064 (ADDF128_ODD): Likewise.
7065 (SUBF128_ODD): Likewise.
7066 (MULF128_ODD): Likewise.
7067 (DIVF128_ODD): Likewise.
7068 (FMAF128_ODD): Likewise.
7069 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7070 UNSPEC_TRUNC_ROUND_TO_ODD.
7071 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7072 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7073 floating point round to odd instructions.
7074 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7075 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7076 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7077 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7078 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7079 (trunc<mode>sf2_hw): Change the truncate with round to odd
7080 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7081 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7082 to odd hardware instructions.
7083 (sub<mode>3_odd): Likewise.
7084 (mul<mode>3_odd): Likewise.
7085 (div<mode>3_odd): Likewise.
7086 (sqrt<mode>2_odd): Likewise.
7087 (fma<mode>4_odd): Likewise.
7088 (fms<mode>4_odd): Likewise.
7089 (nfma<mode>4_odd): Likewise.
7090 (nfms<mode>4_odd): Likewise.
7091 (trunc<mode>df2_odd): Change the truncate with round to odd
7092 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7094 * doc/extend.texi (PowerPC built-in functions): Update documentation
7095 for existing IEEE float128-bit built-in functions. Add built-in
7096 functions that generate the IEEE 128-bit floating point round to
7099 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7101 PR rtl-optimization/77729
7102 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7103 to (X&(C1&~C2))|C2 transformations.
7105 2017-10-03 Martin Jambor <mjambor@suse.cz>
7107 PR tree-optimization/82363
7108 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7109 mismatch, mark lacc written regardless of racc.
7111 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7113 PR tree-optimization/82381
7114 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7115 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7116 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7117 fallthrough into reassoc_stmt_dominates_stmt_p.
7120 * combine.c (combine_instructions): Don't combine in unreachable
7123 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7126 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7127 function to not use the have_cpu variable. Do not set cpu_index,
7128 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7130 (rs6000_valid_attribute_p): Remove duplicate initializations of
7131 old_optimize and func_optimize.
7132 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7133 (rs6000_activate_target_options): Make global.
7134 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7137 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7139 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7140 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7141 Set *poffset to 0 and *psize and *pmax_size to -1 if
7142 *poffset + *psize overflows in HOST_WIDE_INT.
7144 PR tree-optimization/82387
7145 PR tree-optimization/82388
7146 PR tree-optimization/82389
7147 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7148 instead of live_bytes non-NULL.
7150 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7153 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7156 2017-10-02 Richard Biener <rguenther@suse.de>
7158 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7159 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7160 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7162 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7164 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7165 requested precision matches the type's.
7166 * calls.c (alloc_max_size): Calculate the new candidate size as
7167 a widest_int and use wi::to_widest when comparing it with the
7168 current candidate size.
7169 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7170 zero rather than integer_zero_node.
7171 * match.pd: Check for a no-op conversion before using wi::add
7172 rather than after. Use tree_to_uhwi when summing small shift
7173 counts into an unsigned int.
7175 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7176 Alan Hayward <alan.hayward@arm.com>
7177 David Sherwood <david.sherwood@arm.com>
7180 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7181 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7182 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7183 POINTER_AND_FP_REGS.
7185 2017-10-02 Richard Biener <rguenther@suse.de>
7187 PR tree-optimization/82355
7188 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7189 a mapping for the enclosing loop but avoid generating one for
7191 (copy_bb_and_scalar_dependences): Remove premature codegen
7192 error on PHIs in blocks duplicated into multiple places.
7193 * graphite-scop-detection.c
7194 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7195 in the region use it as loop and nest to analyze the DR in.
7196 (try_generate_gimple_bb): Likewise.
7197 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7198 (add_loop_constraints): For blocks in a loop not in the region
7199 create a dimension with a single iteration.
7200 * sese.h (gbb_loop_at_index): Remove assert.
7202 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7204 * omp-expand.c (adjust_context_scope): New function.
7205 (expand_parallel_call): Call adjust_context_scope.
7207 2017-10-01 Jeff Law <law@redhat.com>
7209 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7210 dom_opt_dom_walker class with direct access to private members.
7211 Add comments. Call test_for_singularity.
7212 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7213 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7214 m_dummy_cond anymore.
7215 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7217 (pass_dominator:execute): Build the dummy_cond here and pass it
7218 to the dom_opt_dom_walker ctor.
7219 (test_for_singularity): New function.
7221 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7222 Maya Rashish <coypu@sdf.org>
7224 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7225 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7226 (alpha*-*-netbsd*) Use nbsd_tm_file.
7227 (arm*-*-netbsdelf*) Likewise.
7228 (i[34567]86-*-netbsdelf*) Likewise.
7229 (x86_64-*-netbsd*) Likewise.
7230 (mips*-*-netbsd*) Likewise.
7231 (powerpc-*-netbsd*) Likewise.
7232 (sh*-*-netbsd*) Likewise.
7233 (sparc-*-netbsdelf*) Likewise.
7234 (sparc64-*-netbsd*) Likewise.
7235 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7237 (vax-*-netbsdelf*) Likewise.
7238 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7239 (UINT_FAST8_TYPE) Likewise.
7240 (INT_FAST16_TYPE) Check CHAR_FAST16.
7241 (UINT_FAST16_TYPE) Likewise.
7243 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7246 * config/i386/i386.md
7247 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7248 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7249 *divmodsi4_zext_2): New define_insn_and_split.
7250 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7251 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7252 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7253 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7254 New define_insn_and_split.
7255 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7256 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7257 operands[1] having DImode when mode is SImode.
7259 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7260 always SImode for DIV and MOD in REG_EQUAL notes.
7262 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7265 * match.pd: Fix handling of NaNs in pattern.
7267 2017-09-29 Jeff Law <law@redhat.com>
7269 * sbitmap.c (bitmap_bit_in_range_p): New function.
7270 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7271 * tree-ssa-dse.c (live_bytes_read): New function.
7272 (dse_classify_store): Ignore reads of dead bytes.
7274 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7275 typos and whitespace errors.
7276 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7277 * config/s390/s390.c (s390_emit_prologue): Likewise.
7279 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7282 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7283 with a symbol for LRA.
7285 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7287 PR rtl-optimization/82338
7288 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7290 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7292 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7293 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7294 (qsort_chk): Declare.
7295 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7296 (qsort_chk): New function.
7298 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7300 PR tree-optimization/82337
7301 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7302 phi definition if the PHI result appears in an abnormal PHI.
7303 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7304 basis appears in an abnormal PHI.
7306 2017-09-29 Richard Biener <rguenther@suse.de>
7308 * graphite-isl-ast-to-gimple.c
7309 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7310 (binary_op_to_tree): Use it.
7311 (get_rename_from_scev): Likewise.
7312 (copy_loop_phi_nodes): Likewise.
7313 (copy_bb_and_scalar_dependences): Likewise.
7314 (translate_pending_phi_nodes): Likewise.
7316 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7319 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7320 for movabsq $(i32 << shift), r64.
7322 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7324 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7325 index when encoding %esp as %rsp to avoid 0x67 prefix.
7327 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7329 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7330 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7332 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7334 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7335 Extensions with more than 16 double VFP registers.
7336 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7337 to_clear_mask and all code related to it. Replace the remaining
7338 entry by a sbitmap and adapt code accordingly.
7340 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7342 * brig-builtins.def: Change pure attributes to const.
7344 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7346 * config.gcc (default_gnu_indirect_function): Default to yes for
7347 sparc*-*-linux* with glibc.
7349 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7351 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7352 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7353 when creating .init_array and .fini_array sections with priority
7356 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7359 * config/aarch64/aarch64.c
7360 (aarch64_builtin_support_vector_misalignment): Always return false
7361 when misalignment is unknown.
7363 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7365 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7366 this function to return false if the definition used by the swap
7367 instruction is artificial, or if the memory address from which the
7368 constant value is loaded is not represented by a base address held
7369 in a register or if the base address register is a frame or stack
7370 pointer. Additionally, return false if the base address of the
7371 loaded constant is a SYMBOL_REF but is not considered to be a
7373 (replace_swapped_load_constant): New function.
7374 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7375 loaded constant vector with a load of a swapped constant vector.
7377 2017-09-27 Carl Love <cel@us.ibm.com>
7379 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7380 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7381 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7384 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7386 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7387 first, always call autopref_rank_data otherwise.
7389 2017-09-27 Richard Biener <rguenther@suse.de>
7391 * graphite-scop-detection.c (find_scop_parameters): Move
7392 loop bound handling ...
7393 (gather_bbs::before_dom_children): ... here, avoiding the need
7394 to build scop_info->loop_nest.
7395 (record_loop_in_sese): Remove.
7396 * sese.h (sese_info_t::loop_nest): Remove.
7397 * sese.c (new_sese_info): Do not allocate loop_nest.
7398 (free_sese_info): Do not free loop_nest.
7400 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7403 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7404 lhs from calls if the lhs has addressable type.
7406 2017-09-27 Richard Biener <rguenther@suse.de>
7408 * graphite.h (scop::max_alias_set): New member.
7409 * graphite-scop-detection.c: Remove references to non-existing
7410 --param in comments.
7411 (build_alias_sets): Record the maximum alias set used for drs.
7412 (build_scops): Support zero as unlimited for
7413 --param graphite-max-arrays-per-scop.
7414 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7416 (build_poly_sr_1): ... here. Compute alias set based on the
7417 maximum alias set used for drs rather than
7418 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7420 2017-09-27 Richard Biener <rguenther@suse.de>
7422 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7423 --param loop-block-tile-size=0 to disable tiling.
7425 2017-09-27 Richard Biener <rguenther@suse.de>
7427 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7428 (graphite-max-nb-scop-params): Document special value zero.
7429 * domwalk.h (dom_walker::STOP): New symbolical constant.
7430 (dom_walker::dom_walker): Add optional parameter for bb to
7432 (dom_walker::~dom_walker): Declare.
7433 (dom_walker::before_dom_children): Document STOP return value.
7434 (dom_walker::m_user_bb_to_rpo): New member.
7435 (dom_walker::m_bb_to_rpo): Likewise.
7436 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7437 mapping here if not provided by the user.
7438 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7439 provided by the user.
7440 (dom_walker::STOP): Define.
7441 (dom_walker::walk): Do not compute bb to RPO mapping here.
7442 Support STOP return value from before_dom_children to stop
7444 * graphite-optimize-isl.c (optimize_isl): If the schedule
7445 is the same still generate code if -fgraphite-identity
7446 or -floop-parallelize-all are given.
7447 * graphite-scop-detection.c: Include cfganal.h.
7448 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7450 (gather_bbs::before_dom_children): Return STOP for BBs
7452 (build_scops): Compute bb to RPO mapping and pass it to
7453 the domwalk. Treat --param graphite-max-nb-scop-params=0
7454 as not limiting the number of params.
7455 * graphite.c (graphite_initialize): Remove limit on the
7456 number of basic-blocks in a function.
7457 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7458 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7459 default value of 10.
7461 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7463 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7464 Adjust code to eliminate needing to do the shift right 32-bits
7465 operation after XSCVDPSPN.
7467 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7469 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7470 ((X / Y) != 0 -> X >= Y): Likewise.
7472 2017-09-26 Carl Love <cel@us.ibm.com>
7474 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7475 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7476 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7477 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7478 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7479 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7480 definitions and overloading.
7481 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7482 statement for P9V_BUILTIN_XST_LEN_R.
7483 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7484 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7485 define_expand and define_insn for the instructions and builtins.
7486 * doc/extend.texi: Update the built-in documentation file for the new
7488 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7489 define_insn for the instructions
7491 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7494 * gcc/config/netbsd-protos.h: New file.
7495 * gcc/config/netbsd.c: New file.
7496 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7497 * gcc/config/t-netbsd: New file.
7498 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7499 (tmake_file) Add t-netbsd.
7500 (extra_objs) Add netbsd.o.
7502 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7506 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7508 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7510 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7511 sign extension from a vector register to a GPR by doing a 32-bit
7512 direct move and then an EXTSW.
7513 (extendsi<mode>2 splitter): Likewise.
7514 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7515 right or vector extract after doing XSCVDPSPN. Use
7516 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7518 (movdi_from_sf_zero_ext): Likewise.
7519 (reload_gpr_from_vsxsf): Likewise.
7520 (p8_mfvsrd_4_disf): Delete, no longer used.
7521 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7522 then needing to move the SFmode to a GPR to use the XSCVDPSP
7523 instruction instead of FRSP and XSCVDPSPN.
7524 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7525 it is adjacent to the other XSCVSPDP insns.
7526 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7527 SFmode to be in traditional Altivec registers.
7528 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7529 (vsx_xscvspdpn): Likewise.
7530 (vsx_xscvspdpn_scalar): Likewise.
7532 2017-09-26 Martin Jambor <mjambor@suse.cz>
7534 * tree-sra.c (compare_access_positions): Put integral types first,
7535 stabilize sorting of integral types, remove conditions putting
7536 non-full-precision integers last.
7537 (sort_and_splice_var_accesses): Disable scalarization if a
7538 non-integert would be represented by a non-full-precision integer.
7540 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7542 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7543 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7544 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7545 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7546 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7547 conditionals inside the function instead of around it. Call
7548 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7549 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7551 2017-09-26 Richard Biener <rguenther@suse.de>
7553 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7555 (scop_detection::build_scop_breadth): ... this. Removed.
7556 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7557 (scop_detection::harmful_stmt_in_bb): Likewise.
7558 (scop_detection::graphite_can_represent_stmt): Likewise.
7559 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7560 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7561 (scop_detection::can_represent_loop_1): ... this. Removed.
7562 (scop_detection::harmful_loop_in_region): Simplify after inlining
7563 the above and remove more quadraticness.
7564 (build_scops): Adjust.
7565 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7568 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7571 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7572 REGNO (base) == SP_REG if base is a REG.
7575 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7576 if it is different SSA_NAME.
7577 (optimize_range_tests_cmp_bitwise): New function.
7578 (optimize_range_tests): Call it.
7580 2017-09-26 Richard Biener <rguenther@suse.de>
7582 PR tree-optimization/82321
7583 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7584 for the def being inside the loop.
7586 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7588 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7590 * config/s390/s390-builtins.def: Fix constraint on op4.
7592 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7594 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7595 independent expanders.
7596 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7597 ("vec_ordered", "vec_unordered"): New expanders.
7599 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7601 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7604 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7606 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7607 vec_unpacks_lo_v16qi.
7608 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7610 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7612 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7613 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7614 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7616 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7618 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7620 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7622 ("*vec_slb<mode>"): New insn pattern.
7623 ("vec_shr_<mode>"): New expander.
7624 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7625 and force the shift count operand to V16QImode.
7626 ("vec_srb<mode>"): Set shift count mode to V16QI.
7628 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7630 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7631 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7632 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7634 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7637 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7639 2017-09-26 Richard Biener <rguenther@suse.de>
7641 PR tree-optimization/82320
7642 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7645 2017-09-25 Jeff Law <law@redhat.com>
7647 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7648 prototype for new argument.
7649 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7650 mostly extracted from rs6000_emit_allocate_stack.
7651 (rs6000_emit_probe_stack_range_stack_clash): New function.
7652 (rs6000_emit_allocate_stack): Call
7653 rs6000_emit_probe_stack_range_stack_clash as needed.
7654 (rs6000_emit_probe_stack_range): Add additional argument
7655 to call to gen_probe_stack_range{si,di}.
7656 (output_probe_stack_range): New.
7657 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7658 (output_probe_stack_range_stack_clash): New.
7659 (rs6000_emit_prologue): Emit notes into dump file as requested.
7660 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7661 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7662 Add additional operand and pass it to output_probe_stack_range.
7664 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7666 PR tree-optimization/82163
7667 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7668 (checking_verify_loop_closed_ssa): New parameter.
7669 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7670 (check_loop_closed_ssa_stmt): Delete.
7671 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7672 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7673 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7676 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7678 * brig-builtins.def: Treat HSAIL barrier builtins as
7679 setjmp/longjump style functions.
7681 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7683 * target.def (constant_alignment): New hook.
7684 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7685 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7686 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7687 * doc/tm.texi: Regenerate.
7688 * targhooks.h (default_constant_alignment): Declare.
7689 (constant_alignment_word_strings): Likewise.
7690 * targhooks.c (default_constant_alignment): New function.
7691 (constant_alignment_word_strings): Likewise.
7692 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7693 instead of CONSTANT_ALIGNMENT.
7694 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7695 (force_const_mem): Likewise.
7696 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7697 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7698 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7699 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7700 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7702 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7703 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7704 constant_alignment_word_strings.
7705 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7706 (CONSTANT_ALIGNMENT): Likewise.
7707 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7708 (arm_constant_alignment): New function.
7709 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7710 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7711 constant_alignment_word_strings.
7712 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7713 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7714 constant_alignment_word_strings.
7715 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7716 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7717 (cris_constant_alignment): New function.
7718 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7719 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7720 (epiphany_constant_alignment): New function.
7721 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7722 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7723 constant_alignment_word_strings.
7724 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7725 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7726 constant_alignment_word_strings.
7727 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7728 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7729 constant_alignment_word_strings.
7730 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7731 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7732 * config/i386/i386.c (ix86_constant_alignment): Make static.
7733 Use the same interface as the target hook.
7734 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7735 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7736 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7737 constant_alignment_word_strings.
7738 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7739 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7740 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7741 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7742 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7743 constant_alignment_word_strings.
7744 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7745 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7746 constant_alignment_word_strings.
7747 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7748 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7749 constant_alignment_word_strings.
7750 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7751 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7753 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7754 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7755 * config/mips/mips.c (mips_constant_alignment): New function.
7756 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7757 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7758 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7759 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7760 (mmix_constant_alignment): Make static. Use the same interface
7762 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7763 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7764 constant_alignment_word_strings.
7765 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7766 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7767 constant_alignment_word_strings.
7768 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7769 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7770 constant_alignment_word_strings.
7771 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7772 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7773 (rs6000_constant_alignment): New function.
7774 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7775 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7776 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7777 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7778 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7779 (rs6000_constant_alignment): New function.
7780 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7781 * config/s390/s390.c (s390_constant_alignment): New function.
7782 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7783 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7784 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7785 constant_alignment_word_strings.
7786 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7787 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7788 (sparc_constant_alignment): New function.
7789 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7790 * config/spu/spu.c (spu_constant_alignment): New function.
7791 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7792 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7793 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7794 constant_alignment_word_strings.
7795 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7796 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7797 constant_alignment_word_strings.
7798 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7799 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7800 constant_alignment_word_strings.
7801 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7802 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7803 (visium_constant_alignment): New function.
7804 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7805 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7806 (xtensa_constant_alignment): New function.
7807 * system.h (CONSTANT_ALIGNMENT): Poison.
7809 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7811 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7812 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7813 (rs6000_builtin_valid_without_lhs): New helper function.
7814 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7815 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7817 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7819 * target.h (vec_perm_indices): Use unsigned short rather than
7821 (auto_vec_perm_indices): Likewise.
7822 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7823 Use unsigned int rather than unsigned char.
7824 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7826 2017-09-25 Richard Biener <rguenther@suse.de>
7828 * cfgloop.h (sort_sibling_loops): Declare.
7829 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7830 (sort_sibling_loops): New function sorting the sibling loop list
7832 * graphite.c (graphite_transform_loops): Sort sibling loops.
7834 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7836 * target.def (vec_perm_const_ok): Change sel parameter to
7838 * optabs-query.c (can_vec_perm_p): Update accordingly.
7839 * doc/tm.texi: Regenerate.
7840 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7841 auto_vec_perm_indices and remove separate nelt field.
7842 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7843 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7844 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7845 (aarch64_expand_vec_perm_const): Update accordingly.
7846 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7847 to vec_perm_indices.
7848 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7849 auto_vec_perm_indices and remove separate nelt field.
7850 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7851 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7852 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7854 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7855 to vec_perm_indices.
7856 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7857 sel to vec_perm_indices.
7858 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7859 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7860 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7862 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7865 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7868 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7869 on the FUNCTION_DECL function context if it has a DIE that is a
7872 2017-09-25 Richard Biener <rguenther@suse.de>
7874 PR tree-optimization/82285
7875 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7878 2017-09-25 Tom de Vries <tom@codesourcery.com>
7882 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7885 2017-09-25 Richard Biener <rguenther@suse.de>
7887 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7888 ISL errors other than isl_error_quota happen. Dump if the
7889 schedule is the same.
7890 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7891 errors instead of aborting inside ISL.
7893 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7896 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7897 of libgcc_eh for m64.
7898 * config/i386/darwin64.h: Likewise.
7900 2017-09-25 Richard Biener <rguenther@suse.de>
7903 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7904 attribute for incomplete types nor twice for complete ones.
7906 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7909 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7910 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7913 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7916 * config/i386/i386.opt (mprefer-avx256): Use
7917 ix86_target_flags variable.
7918 * config/i386/i386.c (ix86_target_string): Move
7919 -mprefer-avx256 to flag2_opts.
7921 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7924 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7925 and x != -1 | y != -1 into (x & y) != -1.
7927 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7929 * config.gcc: Add new case statement to set
7930 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7931 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7932 s390x-*-linux* case statements. Added aarch64 to the list of
7933 supported architectures.
7935 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7937 PR tree-optimization/82289
7938 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7939 STMT_VINFO_RELEVANT_P.
7941 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7942 Alan Hayward <alan.hayward@arm.com>
7943 David Sherwood <david.sherwood@arm.com>
7945 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7946 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7947 (extract_range_from_binary_expr_1): Don't call
7948 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7950 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7951 Alan Hayward <alan.hayward@arm.com>
7952 David Sherwood <david.sherwood@arm.com>
7954 * target.def (preferred_vector_alignment): New hook.
7955 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7957 * doc/tm.texi: Regenerate.
7958 * targhooks.h (default_preferred_vector_alignment): Declare.
7959 * targhooks.c (default_preferred_vector_alignment): New function.
7960 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7962 (DR_TARGET_ALIGNMENT): New macro.
7963 (aligned_access_p): Update commentary.
7964 (vect_known_alignment_in_bytes): New function.
7965 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7967 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7968 Calculate the misalignment based on the target alignment rather than
7970 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7971 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7972 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7973 the target alignment, rather than masking the element misalignment
7974 with the number of elements in a vector. Also use the target
7975 alignment when calculating the maximum number of peels.
7976 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7977 instead of TYPE_ALIGN_UNIT.
7978 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7979 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7980 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7981 (vect_create_data_ref_ptr): Likewise.
7982 (vect_setup_realignment): Realign by ANDing with
7983 -DR_TARGET_MISALIGNMENT.
7984 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7985 the number of peels based on DR_TARGET_ALIGNMENT.
7986 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7987 with the guaranteed alignment boundary when deciding whether
7989 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7990 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7991 (ensure_base_align): Remove stmt_info parameter. Get the
7992 target base alignment from DR_TARGET_ALIGNMENT.
7993 (vectorizable_store): Update call accordingly. Interpret
7994 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7996 (vectorizable_load): Likewise.
7998 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7999 Alan Hayward <alan.hayward@arm.com>
8000 David Sherwood <david.sherwood@arm.com>
8002 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8003 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8004 (vect_enhance_data_refs_alignment): Likewise.
8006 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8008 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8009 error. Only quit immediately if parsing is complete.
8010 (BEGIN): Initialize fatal_err and parse_done.
8011 (begin fpu, end fpu): Check number of arguments.
8012 (begin arch, end arch): Likewise.
8013 (begin cpu, end cpu): Likewise.
8014 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8015 (optalias): Likewise.
8017 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8019 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8020 * config/arm/arm-isa.h: Delete. Move definitions to ...
8021 * arm-cpus.in: ... here. Use new feature and fgroup values.
8022 * config/arm/arm.c (arm_option_override): Use lower case for feature
8024 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8025 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8026 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8028 (print_isa_bits_for): New function.
8029 (gen_isa): New function.
8030 (gen_comm_data): Use print_isa_bits_for.
8031 (define feature): New keyword.
8032 (define fgroup): New keyword.
8033 * config/arm/t-arm (TM_H): Remove.
8034 (GTM_H): Add arm-isa.h.
8035 (arm-isa.h): Add rule to generate file.
8036 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8037 case for feature bit names.
8039 2017-09-22 Richard Biener <rguenther@suse.de>
8041 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8043 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8044 print of no dependency loops ...
8045 * graphite.c (graphite_transform_loops): ... here.
8046 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8048 (same_close_phi_node, remove_duplicate_close_phi,
8049 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8050 (canonicalize_loop_closed_ssa): ... here and simplify.
8051 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8052 (optimize_isl): Use dump_printf_loc to tell when we stopped
8053 optimizing because of an ISL timeout.
8055 2017-09-22 Richard Biener <rguenther@suse.de>
8057 PR tree-optimization/82291
8058 * tree-if-conv.c (predicate_mem_writes): Make sure to
8059 remove writes in blocks predicated with false.
8061 2017-09-22 Richard Biener <rguenther@suse.de>
8063 * sese.c: Include cfganal.h.
8064 (if_region_set_false_region): Remove.
8065 (create_if_region_on_edge): Likewise.
8066 (move_sese_in_condition): Re-implement without destroying
8069 2017-09-22 Richard Biener <rguenther@suse.de>
8071 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8072 Verify both BBs contain loop PHI nodes before dispatching to
8074 (graphite_regenerate_ast_isl): Do not recompute dominators,
8075 do not verify three times. Restructure for clarity.
8076 * graphite-scop-detection.c (same_close_phi_node,
8077 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8078 defined_in_loop_p, canonicalize_loop_closed_ssa,
8079 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8080 checking and SSA rewrite, move to ...
8081 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8082 (graphite_initialize): Do not pass in ctx, do not reset the
8083 SCEV cache, compute only dominators.
8084 (graphite_transform_loops): Allocate ISL ctx after
8085 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8086 Maintain post-dominators only around build_scops.
8087 * sese.c (if_region_set_false_region): Make static. Free
8088 and recompute dominators.
8089 (move_sese_in_condition): Assert we don't get called with
8090 post-dominators computed.
8091 * sese.h (if_region_set_false_region): Remove.
8093 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8095 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8096 mode attribute for TARGET_AVX512VL.
8098 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8100 * config/i386/i386.opt (mprefer-avx256): New option.
8101 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8103 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8104 for TARGET_PREFER_AVX256.
8105 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8107 2017-09-21 Jeff Law <law@redhat.com>
8109 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8110 Fix dump output if the only stack space is for pushed registers.
8112 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8114 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8117 2017-09-21 Martin Sebor <msebor@redhat.com>
8120 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8121 code (in C++) or code that triggers warnings.
8123 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8125 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8127 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8129 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8130 * sched-rgn.c: Ditto.
8131 * sel-sched-ir.c: Ditto.
8133 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8135 * toplev.h (set_random_seed): Adjust return type.
8136 * toplev.c (init_local_tick): Move eager initialization of random_seed
8137 to get_random_seed. Adjust comment.
8138 (init_random_seed): Inline to get_random_seed, delete.
8139 (get_random_seed): Initialize random_seed lazily.
8140 (set_random_seed): Do not return previous value.
8141 (print_switch_value): Do not call get_random_seed.
8143 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8145 * cgraph.c (delete_function_version): New, broken out from...
8146 (cgraph_node::delete_function_version): ...here. Rename to
8147 cgraph_node::delete_function_version_by_decl. Update all uses.
8148 (cgraph_node::remove): Call delete_function_version.
8150 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8153 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8154 VAR_DECLs to which addressable non-volatile parameters are mapped
8155 and for id->retvar after the return value assignment. Clear
8156 id->retval and id->retbnd after inlining.
8158 2017-09-21 Richard Biener <rguenther@suse.de>
8160 PR tree-optimization/82276
8161 PR tree-optimization/82244
8162 * tree-vrp.c (build_assert_expr_for): Set
8163 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8165 (remove_range_assertions): Revert earlier change.
8167 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8170 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8172 2017-09-21 Richard Biener <rguenther@suse.de>
8174 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8175 Restore valid IL after code generation errors.
8176 * graphite.c (graphite_transform_loops): Diagnose code
8177 generation issues as MSG_MISSED_OPTIMIZATION and continue
8178 with processing SCOPs.
8180 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8181 Alan Hayward <alan.hayward@arm.com>
8182 David Sherwood <david.sherwood@arm.com>
8184 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8185 rather than choosing between plus_constant and gen_rtx_<CODE>.
8186 * expr.c (emit_push_insn): Likewise.
8187 (expand_expr_real_2): Likewise.
8189 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8190 Alan Hayward <alan.hayward@arm.com>
8191 David Sherwood <david.sherwood@arm.com>
8193 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8195 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8196 Alan Hayward <alan.hayward@arm.com>
8197 David Sherwood <david.sherwood@arm.com>
8199 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8200 calling tree_to_uhwi.
8202 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8203 Alan Hayward <alan.hayward@arm.com>
8204 David Sherwood <david.sherwood@arm.com>
8206 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8207 INTEGER_CST rather than a negative test for ADDR_EXPR.
8209 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8210 Alan Hayward <alan.hayward@arm.com>
8211 David Sherwood <david.sherwood@arm.com>
8213 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8214 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8216 2017-09-21 Richard Biener <rguenther@suse.de>
8218 PR tree-optimization/71351
8219 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8220 graphite_create_new_loop_guard): Remove, fold remaining parts
8222 (translate_isl_ast_node_for): ... here and simplify.
8224 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8227 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8228 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8229 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8230 alternative always use QI mode, for -Os imov (=R,R) alternative
8231 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8234 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8235 Jeff Law <law@redhat.com>
8237 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8238 (allocate_stack_space): New function, partially extracted from
8240 (s390_emit_prologue): Track offset to most recent stack probe.
8241 Code to allocate space moved into allocate_stack_space.
8242 Dump actions when no stack is allocated.
8243 (s390_prologue_plus_offset): New function.
8244 (s390_emit_stack_probe): Likewise.
8246 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8248 * common.opt (Wa, Wl, Wp, g, gz=): Add
8250 (gno-column-info): Remove.
8251 (gcolumn-info): Drop RejectNegative.
8253 (gno-record-gcc-switches): Remove.
8254 (grecord-gcc-switches): Drop RejectNegative.
8255 (gno-split-dwarf): Remove.
8256 (gsplit-dwarf): Drop RejectNegative.
8257 (gno-strict-dwarf): Remove.
8258 (gstrict-dwarf): Drop RejectNegative.
8259 * config/darwin.opt (gfull, gused): Add RejectNegative.
8260 * dwarf2out.c (gen_producer_string): Drop
8261 gno-record-gcc-switches handler.
8262 * optc-gen.awk: Add g to prefixes with negative forms.
8263 * opts-common.c (remapping_prefix_p): New.
8264 (find_opt): Check it.
8265 (generate_canonical_option): Test g prefix.
8266 (option_map): Add -gno- mapping.
8267 (add_misspelling_candidates): Check remapping_prefix_p.
8269 2017-09-20 Jeff Law <law@redhat.com>
8271 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8272 thinko in stack clash protection support.
8274 * explow.c (compute_stack_clash_protection_loop_data): Use
8275 CONST_INT_P instead of explicit test. Verify object is a
8276 CONST_INT_P before looking at INTVAL.
8277 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8278 instead of explicit test.
8280 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8283 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8284 address instead of to r1 and r11.
8286 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8288 * config.gcc: Support "knm".
8289 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8290 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8292 * config/i386/i386.c (m_KNM): Define.
8293 (processor_target_table): Add "knm".
8295 (ix86_option_override_internal): Add "knm".
8296 (ix86_issue_rate): Add PROCESSOR_KNM.
8297 (ix86_adjust_cost): Ditto.
8298 (ia32_multipass_dfa_lookahead): Ditto.
8299 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8300 (fold_builtin_cpu): Add M_INTEL_KNM.
8301 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8302 (processor_type): Add PROCESSOR_KNM.
8303 * config/i386/x86-tune.def: Add m_KNM.
8304 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8306 2017-09-20 Richard Biener <rguenther@suse.de>
8308 PR tree-optimization/80213
8309 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8310 are allowed in empty BBs as well.
8311 (canonicalize_loop_closed_ssa): Also look for other complex
8313 (scop_detection::get_sese): Include the loop-closed PHI block
8315 (scop_detection::merge_sese): Remove code adding extra blocks.
8316 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8317 (build_scops): Assert the final returned scop is invalid.
8319 2017-09-20 Richard Biener <rguenther@suse.de>
8321 PR tree-optimization/82264
8322 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8324 (vn_phi_lookup): Likewise.
8325 (vn_phi_insert): Likewise.
8327 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8329 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8330 that fits into uhwi or shwi, add DW_AT_const_value regardless
8331 of early_dwarf without going through RTL, using add_AT_unsigned
8334 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8335 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8336 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8337 (DEBUG_MACRO_SECTION): Likewise.
8338 (DEBUG_LTO_MACRO_SECTION): Likewise.
8339 (DEBUG_STR_DWO_SECTION): Likewise.
8340 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8341 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8342 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8343 (DEBUG_LTO_LINE_STR_SECTION): Define.
8344 (init_sections_and_labels): Initialize debug_line_str_section
8345 variable. Initialize debug_loc_section for -gdwarf-5 to
8346 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8348 2017-09-20 Richard Biener <rguenther@suse.de>
8350 * graphite-sese-to-poly.c (extract_affine): Properly handle
8351 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8353 2017-09-20 Richard Biener <rguenther@suse.de>
8355 PR tree-optimization/81373
8356 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8357 Force SESE live-out defs to be handled even if they are
8360 2017-09-19 Jeff Law <law@redhat.com>
8362 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8363 nothing for stack adjustments with REG_STACK_CHECK.
8364 * sched-deps.c (parse_add_or_inc): Reject insns with
8365 REG_STACK_CHECK from dependency breaking.
8366 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8367 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8368 * reg-notes.def (STACK_CHECK): New note.
8370 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8371 (ix86_expand_prologue): Dump stack clash info as needed.
8372 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8374 * function.c (dump_stack_clash_frame_info): New function.
8375 * function.h (dump_stack_clash_frame_info): Prototype.
8376 (enum stack_clash_probes): New enum.
8378 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8379 flag_stack_clash_protection.
8380 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8381 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8382 (arm_frame_pointer_required): Likewise.
8383 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8384 (ia64_expand_prologue): Likewise.
8385 * config/mips/mips.c (mips_expand_prologue): Likewise.
8386 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8387 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8388 (sparc_flat_expand_prologue): Likewise.
8389 * config/spu/spu.c (spu_expand_prologue): Likewise.
8391 * explow.c: Include "params.h".
8392 (anti_adjust_stack_and_probe_stack_clash): New function.
8393 (get_stack_check_protect): Likewise.
8394 (compute_stack_clash_protection_loop_data): Likewise.
8395 (emit_stack_clash_protection_loop_start): Likewise.
8396 (emit_stack_clash_protection_loop_end): Likewise.
8397 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8398 Use anti_adjust_stack_and_probe_stack_clash.
8399 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8400 (emit_stack_clash_protection_loop_start): Likewise.
8401 (emit_stack_clash_protection_loop_end): Likewise.
8402 * rtl.h (get_stack_check_protect): Prototype.
8403 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8404 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8405 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8407 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8409 * doc/tm.texi: Rebuilt.
8410 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8411 get_stack_check_protect.
8412 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8413 * config/arm/arm.c (arm_expand_prologue): Likewise.
8414 (arm_frame_pointer_required): Likewise.
8415 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8416 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8417 * config/mips/mips.c (mips_expand_prologue): Likewise.
8418 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8419 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8420 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8421 (sparc_flat_expand_prologue): Likewise.
8423 * common.opt (-fstack-clash-protection): New option.
8424 * flag-types.h (enum stack_check_type): Note difference between
8425 -fstack-check= and -fstack-clash-protection.
8426 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8427 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8428 * toplev.c (process_options): Issue warnings/errors for cases
8429 not handled with -fstack-clash-protection.
8430 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8431 (-fstack-check): Note additional problem with -fstack-check=generic.
8432 Note that -fstack-check is primarily for Ada and refer users
8433 to -fstack-clash-protection for stack-clash-protection.
8434 Document new params for stack clash protection.
8436 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8438 * config/i386/i386.c (ix86_split_long_move): Do not handle
8439 address used for LEA in a special way.
8441 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8443 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8445 2017-09-19 Martin Sebor <msebor@redhat.com>
8448 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8449 of incompatible types.
8451 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8453 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8454 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8455 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8456 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8458 2017-09-19 Richard Biener <rguenther@suse.de>
8460 PR tree-optimization/82244
8461 * tree-vrp.c (remove_range_assertions): Do not propagate
8462 a constant to abnormals but replace the assert with a copy.
8464 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8466 PR rtl-optimization/57878
8467 PR rtl-optimization/68988
8468 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8469 avoidance test involving non_reload_pseudos. Move frequency test
8470 below the general fragmentation avoidance test.
8472 2017-09-19 Richard Biener <rguenther@suse.de>
8474 PR tree-optimization/69728
8475 * graphite-sese-to-poly.c (schedule_error): New global.
8476 (add_loop_schedule): Handle empty domain by failing the
8478 (build_original_schedule): Handle schedule_error.
8480 2017-09-19 Richard Biener <rguenther@suse.de>
8482 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8483 Do not iterate to sibling loops but only to siblings of inner
8486 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8489 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8490 that the registers are different.
8492 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8494 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8495 to processor_model and "amdfam17h" to arch_names_table.
8496 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8498 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8501 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8503 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8504 Alan Hayward <alan.hayward@arm.com>
8505 David Sherwood <david.sherwood@arm.com>
8507 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8509 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8511 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8512 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8513 vect_schedule_slp_instance.
8514 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8515 Update call to vect_slp_analyze_node_operations. Simplify return
8517 (vect_slp_analyze_bb_1): Update call accordingly.
8518 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8519 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8520 (vect_schedule_slp): Update call accordingly.
8522 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8523 Alan Hayward <alan.hayward@arm.com>
8524 David Sherwood <david.sherwood@arm.com>
8526 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8527 with types that aren't in fact scalar.
8529 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8531 * tree-vect-slp.c (vect_record_max_nunits): New function,
8533 (vect_build_slp_tree_1): ...here.
8534 (vect_build_slp_tree_2): Call it for phis too.
8536 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8538 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8539 to vect_get_vec_def_for_operand when getting the mask operand.
8541 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8542 Alan Hayward <alan.hayward@arm.com>
8543 David Sherwood <david.sherwood@arm.com>
8545 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8548 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8549 Alan Hayward <alan.hayward@arm.com>
8550 David Sherwood <david.sherwood@arm.com>
8552 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8553 calculation for vector booleans.
8555 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8556 Alan Hayward <alan.hayward@arm.com>
8557 David Sherwood <david.sherwood@arm.com>
8559 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8561 (vect_transform_stmt): ...here.
8562 (vect_analyze_stmt): Use it instead of calling
8563 vectorizable_live_operation directly.
8565 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8567 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8568 non-SIMT targets in acc vector loops.
8570 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8572 * configure.ac: Add arc and check if assembler supports gdwarf2.
8573 * configure: Regenerate.
8575 2017-09-18 Richard Biener <rguenther@suse.de>
8577 PR tree-optimization/82220
8578 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8579 epilogue niters from the min_profitable_iters compute.
8581 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8584 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8585 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8586 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8588 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8591 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8592 switching to a new text section.
8594 2017-09-18 Richard Biener <rguenther@suse.de>
8596 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8598 (build_alias_set): Reject aliases with no access function.
8600 2017-09-18 Richard Biener <rguenther@suse.de>
8602 PR tree-optimization/79622
8603 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8605 (build_cross_bb_scalars_use): Likewise.
8607 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8609 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8611 2017-09-18 Alan Modra <amodra@gmail.com>
8614 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8615 stack_pointer_rtx for count 0. Update comments. Break up
8616 large rtl expression.
8618 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8620 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8621 Increase to 20 bytes.
8622 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8623 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8624 or avx version of the stub.
8626 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8629 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8630 compute the minimum stack alignment. Also update preferred stack
8631 boundary for leaf functions.
8633 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8635 PR tree-optimization/82228
8636 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8639 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8641 * common/config/nds32/nds32-common.c
8642 (nds32_option_optimization_table): Refine formatting.
8643 (nds32_option_optimization_table): Use -fsched-pressure and
8644 -fomit-frame-pointer for specific optimization level.
8646 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8648 * config/nds32/nds32.c: Refine formatting and comments.
8649 * config/nds32/nds32.h: Likewise.
8650 * config/nds32/nds32.md: Likewise.
8651 * config/nds32/nds32-cost.c: Likewise.
8652 * config/nds32/nds32-isr.c: Likewise.
8653 * config/nds32/nds32-md-auxiliary.c: Likewise.
8654 * config/nds32/nds32-multiple.md: Likewise.
8655 * config/nds32/nds32-predicates.c: Likewise.
8657 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8658 Jakub Jelinek <jakub@redhat.com>
8660 Add support for -std=c++2a.
8661 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8663 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8665 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8668 * doc/extend.texi (Common Function Attributes): Add
8669 references to ARM, AArch64, and S/390 specific attributes.
8670 (Function Specific Option Pragmas): Add AArch64 and S/390
8671 to list of back ends that support the target pragma.
8673 2017-09-15 Nathan Sidwell <nathan@acm.org>
8675 * doc/standards.texi: Fix C++17 description. Update URLs for
8678 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8680 * common.opt (Wcast-align=strict): New warning option.
8681 * doc/invoke.texi: Document -Wcast-align=strict.
8683 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8685 * cgraph.h (cgraph_thunk_info): Add comments.
8686 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8687 assert for VIRTUAL_* arguments stricter.
8689 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8691 PR tree-optimization/71026
8692 * match.pd: Move RDIV patterns from fold-const.c
8693 * fold-const.c (distribute_real_division): Removed.
8694 (fold_binary_loc): Remove calls to distribute_real_divison.
8696 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8698 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8699 c++1z and gnu++1z as deprecated. Change other references to
8700 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8701 Change -Wc++1z-compat to -Wc++17-compat.
8702 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8703 * dwarf2out.c (highest_c_language): Handle C++17.
8704 (gen_compile_unit_die): Likewise.
8706 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8708 PR rtl-optimization/82192
8709 * combine.c (make_extraction): Don't look through non-paradoxical
8710 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8713 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8714 Alan Hayward <alan.hayward@arm.com>
8715 David Sherwood <david.sherwood@arm.com>
8717 * target.def (function_arg_offset): New hook.
8718 * targhooks.h (default_function_arg_offset): Declare.
8719 * targhooks.c (default_function_arg_offset): New function.
8720 * function.c (locate_and_pad_parm): Use
8721 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8722 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8723 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8724 * doc/tm.texi: Regenerate.
8725 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8726 * config/spu/spu.c (spu_function_arg_offset): New function.
8727 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8728 * system.h (FUNCTION_ARG_OFFSET): Poison.
8730 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8731 Alan Hayard <alan.hayward@arm.com>
8732 David Sherwood <david.sherwood@arm.com>
8734 * target.def (truly_noop_truncation): New hook.
8735 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8736 than TRULY_NOOP_TRUNCATION.
8737 * hooks.h (hook_bool_uint_uint_true): Declare.
8738 * hooks.c (hook_bool_uint_uint_true): New function.
8739 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8740 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8741 * doc/tm.texi: Regenerate.
8742 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8743 rather than TRULY_NOOP_TRUNCATION in comments.
8744 (simplify_comparison): Likewise.
8745 (record_truncated_value): Likewise.
8746 * expmed.c (extract_bit_field_1): Likewise.
8747 (extract_split_bit_field): Likewise.
8748 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8749 instead of TRULY_NOOP_TRUNCATION.
8750 * function.c (assign_parm_setup_block): Likewise.
8751 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8752 * rtlhooks.c: Include target.h.
8753 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8754 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8755 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8756 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8757 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8758 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8759 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8760 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8761 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8762 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8763 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8764 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8765 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8766 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8767 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8768 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8769 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8770 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8771 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8772 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8773 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8774 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8775 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8776 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8777 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8778 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8779 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8780 rather than TRULY_NOOP_TRUNCATION in comments.
8781 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8782 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8783 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8784 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8785 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8786 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8787 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8788 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8789 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8790 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8791 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8792 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8793 rather than TRULY_NOOP_TRUNCATION in comments.
8794 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8795 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8796 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8797 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8798 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8799 TRULY_NOOP_TRUNCATION condition.
8800 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8801 (TRULY_NOOP_TRUNCATION): Delete.
8802 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8803 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8804 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8805 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8806 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8807 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8808 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8809 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8810 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8811 rather than TRULY_NOOP_TRUNCATION in comments.
8812 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8813 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8814 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8815 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8816 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8817 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8819 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8822 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8823 (*cmp_ior): Likewise.
8824 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8825 (*ior_scc_scc_cmp): Likewise.
8826 (*and_scc_scc): Likewise.
8827 (*and_scc_scc_cmp): Likewise.
8829 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8830 Alan Hayard <alan.hayward@arm.com>
8831 David Sherwood <david.sherwood@arm.com>
8833 * target.def (can_change_mode_class): New hook.
8834 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8835 (hard_regno_nregs): Likewise.
8836 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8837 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8838 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8839 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8840 (LOAD_EXTEND_OP): Update accordingly.
8841 * doc/tm.texi: Regenerate.
8842 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8843 CANNOT_CHANGE_MODE_CLASS.
8844 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8845 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8846 * combine.c (simplify_set): Update accordingly.
8847 * emit-rtl.c (validate_subreg): Likewise.
8848 * recog.c (general_operand): Likewise.
8849 * regcprop.c (mode_change_ok): Likewise.
8850 * reload1.c (choose_reload_regs): Likewise.
8851 (inherit_piecemeal_p): Likewise.
8852 * rtlanal.c (simplify_subreg_regno): Likewise.
8853 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8854 instead of CANNOT_CHANGE_MODE_CLASS.
8855 (reload_cse_simplify_operands): Likewise.
8856 * reload.c (push_reload): Use targetm.can_change_mode_class
8857 instead of CANNOT_CHANGE_MODE_CLASS.
8858 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8859 REG_CANNOT_CHANGE_MODE_P.
8860 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8861 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8862 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8863 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8864 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8865 (arm_can_change_mode_class): New function.
8866 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8867 than CANNOT_CHANGE_MODE_CLASS in comments.
8868 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8869 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8870 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8871 (ix86_can_change_mode_class): ...this new function, inverting the
8872 sense of the return value.
8873 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8874 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8875 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8876 (ia64_can_change_mode_class): New function.
8877 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8878 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8879 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8880 (m32c_can_change_mode_class): ...this new function, inverting the
8881 sense of the return value.
8882 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8883 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8884 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8885 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8886 (mips_can_change_mode_class): ...this new function, inverting the
8887 sense of the return value.
8888 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8889 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8890 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8891 (msp430_can_change_mode_class): New function.
8892 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8893 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8894 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8895 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8896 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8897 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8898 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8899 (pa_cannot_change_mode_class): Replace with...
8900 (pa_can_change_mode_class): ...this new function, inverting the
8901 sense of the return value.
8902 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8903 than CANNOT_CHANGE_MODE_CLASS in comments.
8904 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8905 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8906 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8907 (pdp11_cannot_change_mode_class): Replace with...
8908 (pdp11_can_change_mode_class): ...this new function, inverting the
8909 sense of the return value.
8910 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8911 * config/powerpcspe/powerpcspe-protos.h
8912 (rs6000_cannot_change_mode_class_ptr): Delete.
8913 * config/powerpcspe/powerpcspe.c
8914 (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/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8925 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8926 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8927 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8928 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8930 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8931 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8932 (rs6000_option_override_internal): Assign to
8933 targetm.can_change_mode_class instead of
8934 rs6000_cannot_change_mode_class_ptr.
8935 (rs6000_cannot_change_mode_class): Replace with...
8936 (rs6000_can_change_mode_class): ...this new function, inverting the
8937 sense of the return value.
8938 (rs6000_debug_cannot_change_mode_class): Replace with...
8939 (rs6000_debug_can_change_mode_class): ...this new function.
8940 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8941 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8942 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8943 (s390_can_change_mode_class): ...this new function, inverting the
8944 sense of the return value.
8945 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8946 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8947 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8948 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8949 (sh_cannot_change_mode_class): Replace with...
8950 (sh_can_change_mode_class): ...this new function, inverting the
8951 sense of the return value.
8952 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8953 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8954 (sparc_can_change_mode_class): New function.
8955 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8956 * config/spu/spu.c (spu_can_change_mode_class): New function.
8957 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8958 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8959 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8960 (visium_can_change_mode_class): New function.
8961 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8963 2017-09-15 Richard Biener <rguenther@suse.de>
8965 PR tree-optimization/82217
8966 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8967 but not undefined case.
8969 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8972 * postreload.c (reload_cse_simplify_operands): Skip
8973 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8975 2017-09-15 Richard Biener <rguenther@suse.de>
8977 PR tree-optimization/68823
8978 * graphite-scop-detection.c (build_alias_set): If we have a
8979 possible dependence check whether we can handle them by just
8980 looking at the DRs DR_ACCESS_FNs.
8981 (build_scops): If build_alias_set fails, fail the SCOP.
8983 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8985 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8986 to support float128 built-in functions that require the ISA 3.0
8988 (BU_FLOAT128_3_HW): Likewise.
8989 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8991 (FMAF128): Likewise.
8993 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8994 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8995 floating point instructions.
8996 (rs6000_invalid_builtin): Likewise.
8997 (rs6000_builtin_mask_names): Likewise.
8998 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8999 (RS6000_BTM_FLOAT128_HW): Likewise.
9000 (RS6000_BTM_COMMON): Likewise.
9001 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9003 * doc/extend.texi (RS/6000 built-in functions): Document the
9004 IEEE 128-bit floating point square root and fused multiply-add
9007 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9009 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9010 reg (r2) isn't in the set of registers defined in the prologue.
9012 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9013 Alan Hayward <alan.hayward@arm.com>
9014 David Sherwood <david.sherwood@arm.com>
9016 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9017 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9018 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9019 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9020 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9022 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9023 max_vectorization_factor.
9024 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9026 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9027 Alan Hayward <alan.hayward@arm.com>
9028 David Sherwood <david.sherwood@arm.com>
9030 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9031 (vect_worthwhile_without_simd_p): Declare.
9032 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9033 (vectorizable_reduction): Use it.
9034 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9035 (vectorizable_operation): Likewise.
9037 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9038 Alan Hayward <alan.hayward@arm.com>
9039 David Sherwood <david.sherwood@arm.com>
9041 * tree-vectorizer.h (vect_get_num_copies): New function.
9042 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9043 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9044 (vectorizable_induction): Likewise.
9045 (vectorizable_live_operation): Likewise.
9046 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9047 (vectorizable_bswap): Likewise.
9048 (vectorizable_call): Likewise.
9049 (vectorizable_conversion): Likewise.
9050 (vectorizable_assignment): Likewise.
9051 (vectorizable_shift): Likewise.
9052 (vectorizable_operation): Likewise.
9053 (vectorizable_store): Likewise.
9054 (vectorizable_load): Likewise.
9055 (vectorizable_condition): Likewise.
9056 (vectorizable_comparison): Likewise.
9057 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9059 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9060 Alan Hayward <alan.hayward@arm.com>
9061 David Sherwood <david.sherwood@arm.com>
9063 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9064 of vect_init_vector.
9066 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9067 Alan Hayward <alan.hayward@arm.com>
9068 David Sherwood <david.sherwood@arm.com>
9070 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9071 an inline wrapper that provides a location.
9072 (gimple_build_vector): Likewise.
9073 * gimple-fold.c (gimple_build_vector_from_val): New function.
9074 (gimple_build_vector): Likewise.
9075 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9076 functions to build the initial value. Always return a gimple value.
9077 (get_initial_defs_for_reduction): Likewise. Only compute
9079 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9080 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9081 (vectorizable_induction): Use gimple_build_vector rather than
9084 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9085 Alan Hayward <alan.hayward@arm.com>
9086 David Sherwood <david.sherwood@arm.com>
9088 * target.h (vec_perm_indices): New typedef.
9089 (auto_vec_perm_indices): Likewise.
9090 * optabs-query.h: Include target.h
9091 (can_vec_perm_p): Take a vec_perm_indices *.
9092 * optabs-query.c (can_vec_perm_p): Likewise.
9093 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9094 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9095 * tree-vect-generic.c (lower_vec_perm): Likewise.
9096 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9097 (vect_grouped_load_supported): Likewise.
9098 (vect_shift_permute_load_chain): Likewise.
9099 (vect_permute_store_chain): Use auto_vec_perm_indices.
9100 (vect_permute_load_chain): Likewise.
9101 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9102 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9103 Update uses of can_vec_perm_p.
9104 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9105 mode with a number of elements. Take a vec_perm_indices *.
9106 (vect_create_epilog_for_reduction): Update accordingly.
9107 Use auto_vec_perm_indices.
9108 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9109 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9110 (vect_transform_slp_perm_load): Likewise.
9111 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9112 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9113 (vect_gen_perm_mask_checked): Likewise.
9114 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9115 (vect_gen_perm_mask_checked): Likewise.
9116 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9117 (vectorizable_store): Likewise.
9118 (vectorizable_load): Likewise.
9119 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9120 (vectorizable_bswap): Likewise.
9122 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9123 Alan Hayward <alan.hayward@arm.com>
9124 David Sherwood <david.sherwood@arm.com>
9126 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9127 * tree.c (build_vector): Likewise.
9128 (build_vector_from_ctor): Update accordingly.
9129 (build_vector_from_val): Likewise.
9130 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9131 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9132 * tree-vect-generic.c (add_rshift): Likewise.
9133 (expand_vector_divmod): Likewise.
9134 (optimize_vector_constructor): Likewise.
9135 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9136 (vect_transform_slp_perm_load): Likewise.
9137 (vect_schedule_slp_instance): Likewise.
9138 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9139 (vectorizable_call): Likewise.
9140 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9141 * expmed.c (make_tree): Likewise.
9142 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9143 a vector passed to build_vector.
9144 (fold_convert_const): Likewise.
9145 (exact_inverse): Likewise.
9146 (fold_ternary_loc): Likewise.
9147 (fold_relational_const): Likewise.
9148 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9149 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9150 (const_unop): Likewise. Store the reduction accumulator in a
9151 variable rather than an array.
9152 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9153 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9154 the new vector, rather than constructing it after the input arrays.
9155 (native_interpret_vector): Use auto_vec<tree> when building
9156 a vector passed to build_vector. Add elements in order.
9157 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9158 auto_vec<tree> when building a vector passed to build_vector.
9159 (vect_create_epilog_for_reduction): Likewise.
9160 (vectorizable_induction): Likewise.
9161 (get_initial_def_for_reduction): Likewise. Fix indentation of
9163 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9165 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9166 passed to build_vector.
9168 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9169 Alan Hayward <alan.hayward@arm.com>
9170 David Sherwood <david.sherwood@arm.com>
9172 * tree-core.h (tree_base::u): Add an "nelts" field.
9173 (tree_vector): Use VECTOR_CST_NELTS as the length.
9174 * tree.c (tree_size): Likewise.
9175 (make_vector): Initialize VECTOR_CST_NELTS.
9176 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9177 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9178 TYPE_VECTOR_SUBPARTS.
9179 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9180 as the number of units, setting it from VECTOR_CST_NELTS.
9181 (const_vector_from_tree): Likewise.
9182 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9183 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9184 (fold_negate_expr_1): Likewise.
9185 (fold_convert_const): Likewise.
9186 (const_binop): Likewise. Differentiate the number of output and
9188 (const_unop): Likewise.
9189 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9190 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9191 in cases that did the opposite.
9193 2017-09-14 Richard Biener <rguenther@suse.de>
9195 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9198 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9200 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9202 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9205 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9206 if and where to split a bb, except for splitting before debug insn
9207 sequences followed by non-label real insn. Delete debug insns
9208 in between basic blocks.
9210 * combine.c (make_compound_operation_int): Formatting fixes.
9212 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9213 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9214 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9215 * config/sol2.h (LINK_EH_SPEC): Likewise.
9216 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9217 * config/s390/linux.h (LINK_SPEC): Likewise.
9218 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9219 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9220 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9221 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9222 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9223 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9224 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9225 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9227 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9229 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9231 (ENDFILE_LINUX_SPEC): Likewise.
9232 (LINK_EH_SPEC): Likewise.
9233 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9234 (LINK_OS_LINUX_SPEC32): Likewise.
9235 (LINK_OS_LINUX_SPEC64): Likewise.
9236 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9237 (LINK_OS_LINUX_SPEC): Likewise.
9239 2017-09-13 Martin Liska <mliska@suse.cz>
9242 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9243 CASE_HIGH is NULL_TREE.
9245 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9246 Alan Hayward <alan.hayward@arm.com>
9247 David Sherwood <david.sherwood@arm.com>
9249 * target.def (secondary_memory_needed): New hook.
9250 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9251 instead of SECONDARY_MEMORY_NEEDED.
9252 (secondary_memory_needed_mode): Likewise.
9253 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9254 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9255 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9256 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9257 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9258 * doc/tm.texi: Regenerate.
9259 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9260 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9261 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9262 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9263 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9264 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9265 mode argument first and change the reg_class arguments to reg_class_t.
9266 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9267 Make static. Update the call to inline_secondary_memory_needed.
9268 (ix86_register_move_cost): Update the call to
9269 inline_secondary_memory_needed.
9270 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9271 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9273 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9274 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9276 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9277 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9278 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9279 and match hook interface. Add comment from mips.h.
9280 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9281 * config/mmix/mmix.md (truncdfsf2): Refer to
9282 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9284 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9285 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9286 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9287 (pa_secondary_memory_needed): New function.
9288 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9289 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9290 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9291 (pdp11_secondary_memory_needed): Make static and match hook interface.
9292 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9293 * config/powerpcspe/powerpcspe-protos.h
9294 (rs6000_secondary_memory_needed_ptr): Delete.
9295 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9297 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9298 (rs6000_option_override_internal): Assign to
9299 targetm.secondary_memory_needed rather than
9300 rs6000_secondary_memory_needed_ptr.
9301 (rs6000_secondary_memory_needed): Match hook interface.
9302 (rs6000_debug_secondary_memory_needed): Likewise.
9303 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9304 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9305 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9306 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9307 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9308 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9310 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9311 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9312 (rs6000_option_override_internal): Assign to
9313 targetm.secondary_memory_needed rather than
9314 rs6000_secondary_memory_needed_ptr.
9315 (rs6000_secondary_memory_needed): Match hook interface.
9316 (rs6000_debug_secondary_memory_needed): Likewise.
9317 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9318 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9319 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9320 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9321 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9322 (sparc_secondary_memory_needed): New function.
9323 * lra-constraints.c (check_and_process_move): Refer to
9324 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9326 (curr_insn_transform): Likewise.
9327 (process_alt_operands): Use targetm.secondary_memory_needed
9328 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9329 (check_secondary_memory_needed_p): Likewise.
9330 (choose_split_class): Likewise.
9331 * reload.c: Unconditionally include code that was previously
9332 conditional on SECONDARY_MEMORY_NEEDED.
9333 (push_secondary_reload): Use targetm.secondary_memory_needed
9334 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9335 (push_reload): Likewise.
9336 * reload1.c: Unconditionally include code that was previously
9337 conditional on SECONDARY_MEMORY_NEEDED.
9338 (choose_reload_regs): Use targetm.secondary_memory_needed
9339 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9340 (gen_reload): Likewise.
9341 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9343 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9344 Alan Hayward <alan.hayward@arm.com>
9345 David Sherwood <david.sherwood@arm.com>
9347 * target.def (secondary_memory_needed_mode): New hook:
9348 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9349 * targhooks.h (default_secondary_memory_needed_mode): New function.
9350 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9351 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9352 * doc/tm.texi: Regenerate.
9353 * lra-constraints.c (check_and_process_move): Use
9354 targetm.secondary_memory_needed_mode instead of
9355 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9356 (curr_insn_transform): Likewise.
9357 * reload.c (get_secondary_mem): Likewise.
9358 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9359 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9361 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9362 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9363 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9364 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9365 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9367 * config/powerpcspe/powerpcspe-protos.h
9368 (rs6000_secondary_memory_needed_mode): Delete.
9369 * config/powerpcspe/powerpcspe.c
9370 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9371 (rs6000_secondary_memory_needed_mode): Make static.
9372 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9373 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9375 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9377 (rs6000_secondary_memory_needed_mode): Make static.
9378 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9379 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9380 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9381 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9382 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9384 (sparc_secondary_memory_needed_mode): New function.
9385 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9387 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9389 * config/aarch64/constraints.md (Umq): New constraint.
9390 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9392 (mov<mode>): Update condition.
9394 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9396 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9397 when bitposition is the same.
9399 2017-09-13 Richard Biener <rguenther@suse.de>
9401 * dwarf2out.c (output_die_symbol): Remove.
9402 (output_die): Do not output a DIEs symbol.
9404 2017-09-13 Richard Biener <rguenther@suse.de>
9407 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9408 default-def to avoid breaking iterator update with the weird
9409 interaction with cgraph_update_edges_for_call_stmt_node.
9411 2017-09-13 Richard Biener <rguenther@suse.de>
9413 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9414 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9415 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9416 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9418 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9420 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9421 Disable pc relative literal load irrespective of
9422 TARGET_FIX_ERR_A53_84341 for default.
9424 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9426 * config/sparc/sparc.c (output_return): Output the source location of
9427 the insn in the delay slot, if any.
9428 (output_sibcall): Likewise.
9430 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9433 * common.opt (-static-pie): New alias.
9434 (shared): Negate static-pie.
9435 (-no-pie): Update help text.
9437 (static-pie): New option.
9438 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9439 -static-pie support.
9440 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9441 (LINK_EH_SPEC): Likewise.
9442 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9443 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9444 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9445 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9446 (init_gcc_specs): Likewise.
9447 (init_spec): Likewise.
9448 (display_help): Update help message for -pie.
9449 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9452 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9454 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9455 (movdi_aarch64): Likewise.
9456 (movti_aarch64): Likewise.
9458 2017-09-12 Simon Wright <simon@pushface.org>
9461 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9462 calculation of the minor version, always output as 0.
9464 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9467 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9468 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9469 on it early, rather than manual conversion late. For
9470 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9471 instead of performing manual conversion.
9473 2017-09-12 Carl Love <cel@us.ibm.com>
9475 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9476 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9478 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9479 VMULOSW): Add definitions.
9480 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9481 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9482 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9483 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9484 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9486 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9488 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9490 (movti_aarch64): Likewise.
9491 (movdf_aarch64): Likewise.
9492 (movtf_aarch64): Likewise.
9493 (load_pairdi): Likewise.
9494 (store_pairdi): Likewise.
9495 (load_pairdf): Likewise.
9496 (store_pairdf): Likewise.
9497 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9498 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9499 (ldr_got_small_<mode>): Likewise.
9500 (ldr_got_small_28k_<mode>): Likewise.
9501 (ldr_got_tiny): Likewise.
9502 * config/aarch64/iterators.md (ldst_sz): New.
9503 (ldpstp_sz): Likewise.
9504 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9506 (thunderx_load): Split load_8 to load_16.
9507 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9509 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9510 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9511 (xgene1_store_pair): Split store_8 to store_16.
9512 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9513 (falkor_st_0_st_sd): Split store_8 to store_16.
9515 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9517 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9518 and store1/2/3/4 to store_4/8/12/16.
9519 * config/aarch64/aarch64.md: Update for rename.
9520 * config/arm/arm.md: Likewise.: Likewise.
9521 * config/arm/arm.c: Likewise.
9522 * config/arm/thumb1.md: Likewise.
9523 * config/arm/thumb2.md: Likewise.
9524 * config/arm/vfp.md: Likewise.
9525 * config/arm/arm-generic.md: Likewise.
9526 * config/arm/arm1020e.md: Likewise.
9527 * config/arm/arm1026ejs.md: Likewise.
9528 * config/arm/arm1136jfs.md: Likewise.
9529 * config/arm/arm926ejs.md: Likewise.
9530 * config/arm/cortex-a15.md: Likewise.
9531 * config/arm/cortex-a17.md: Likewise.
9532 * config/arm/cortex-a5.md: Likewise.
9533 * config/arm/cortex-a53.md: Likewise.
9534 * config/arm/cortex-a57.md: Likewise.
9535 * config/arm/cortex-a7.md: Likewise.
9536 * config/arm/cortex-a8.md: Likewise.
9537 * config/arm/cortex-a9.md: Likewise.
9538 * config/arm/cortex-m4.md: Likewise.
9539 * config/arm/cortex-m7.md: Likewise.
9540 * config/arm/cortex-r4.md: Likewise.
9541 * config/arm/exynos-m1.md: Likewise.
9542 * config/arm/fa526.md: Likewise.
9543 * config/arm/fa606te.md: Likewise.
9544 * config/arm/fa626te.md: Likewise.
9545 * config/arm/fa726te.md: Likewise.
9546 * config/arm/fmp626.md: Likewise.
9547 * config/arm/iwmmxt.md: Likewise.
9548 * config/arm/ldmstm.md: Likewise.
9549 * config/arm/marvell-pj4.md: Likewise.
9550 * config/arm/xgene1.md: Likewise.
9551 * config/aarch64/thunderx.md: Likewise.
9552 * config/aarch64/thunderx2t99.md: Likewise.
9553 * config/aarch64/falkor.md: Likewise.
9555 2017-09-12 Martin Liska <mliska@suse.cz>
9557 * attribs.c (private_lookup_attribute): New function.
9558 * attribs.h (private_lookup_attribute): Declared here.
9559 (lookup_attribute): Called from this place.
9561 2017-09-12 Richard Biener <rguenther@suse.de>
9563 PR tree-optimization/82157
9564 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9565 stmts with side-effects.
9567 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9568 Alan Hayward <alan.hayward@arm.com>
9569 David Sherwood <david.sherwood@arm.com>
9571 * target.def (hard_regno_nregs): New hook.
9572 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9573 * targhooks.h (default_hard_regno_nregs): Declare.
9574 * targhooks.c (default_hard_regno_nregs): New function.
9575 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9576 (TARGET_HARD_REGNO_NREGS): ...this hook.
9577 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9578 (CLASS_MAX_NREGS): Likewise.
9579 * doc/tm.texi: Regenerate.
9580 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9581 instead of HARD_REGNO_NREGS.
9582 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9583 HARD_REGNO_NREGS in the comment.
9584 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9585 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9586 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9587 Return an unsigned int.
9588 (TARGET_HARD_REGNO_NREGS): Redefine.
9589 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9590 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9591 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9592 (arc_hard_regno_nregs): New function.
9593 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9594 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9595 (arm_hard_regno_nregs): New function.
9596 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9597 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9598 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9599 (TARGET_HARD_REGNO_NREGS): Redefine.
9600 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9601 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9602 (HARD_REGNO_NREGS): Delete.
9603 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9604 (cr16_hard_regno_nregs): New function.
9605 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9606 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9607 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9608 (cris_hard_regno_nregs): New function.
9609 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9610 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9611 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9612 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9613 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9614 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9615 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9616 (frv_hard_regno_nregs): Make static. Take and return an
9618 (frv_class_max_nregs): Remove outdated copy of documentation.
9619 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9620 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9621 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9622 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9623 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9624 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9625 (TARGET_HARD_REGNO_NREGS): Redefine.
9626 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9627 (CLASS_MAX_NREGS): Update comment.
9628 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9629 (ia64_hard_regno_nregs): New function.
9630 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9631 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9632 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9633 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9634 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9636 (m32c_hard_regno_nregs): Likewise. Make static.
9637 (TARGET_HARD_REGNO_NREGS): Redefine.
9638 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9639 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9640 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9641 (m68k_hard_regno_nregs): New function.
9642 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9643 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9644 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9645 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9646 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9647 Take and return an unsigned int.
9648 (TARGET_HARD_REGNO_NREGS): Redefine.
9649 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9650 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9651 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9652 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9653 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9654 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9655 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9656 (msp430_hard_regno_nregs): Make static. Take and return an
9658 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9659 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9660 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9661 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9662 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9663 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9664 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9665 (TARGET_HARD_REGNO_NREGS): Redefine.
9666 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9667 (PA_HARD_REGNO_NREGS): ...this.
9668 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9669 (PA_HARD_REGNO_NREGS): ...this.
9670 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9671 (pa_hard_regno_nregs): New function.
9672 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9673 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9674 (pdp11_hard_regno_nregs): New function.
9675 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9676 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9677 (rs6000_hard_regno_nregs_hook): New function.
9678 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9679 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9680 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9681 Take and return an unsigned int. Move earlier in file.
9682 (TARGET_HARD_REGNO_NREGS): Redefine.
9683 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9684 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9685 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9686 (rl78_hard_regno_nregs): Make static. Take and return an
9688 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9689 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9690 (rs6000_hard_regno_nregs_hook): New function.
9691 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9692 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9693 (TARGET_HARD_REGNO_NREGS): Redefine.
9694 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9695 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9696 instead of HARD_REGNO_NREGS.
9697 (s390_hard_regno_nregs): New function.
9698 (s390_hard_regno_mode_ok): Add comment from s390.h.
9699 (TARGET_HARD_REGNO_NREGS): Redefine.
9700 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9701 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9702 (sh_hard_regno_nregs): New function.
9703 (sh_pass_in_reg_p): Use it.
9704 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9705 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9706 (sparc_hard_regno_nregs): New function.
9707 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9708 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9709 (spu_function_arg_advance): Use it, supplying a valid register number.
9710 (TARGET_HARD_REGNO_NREGS): Redefine.
9711 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9712 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9713 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9714 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9715 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9716 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9717 (CLASS_MAX_NREGS): Remove copy of old documentation.
9718 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9719 (visium_hard_regno_nregs): New function.
9720 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9721 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9722 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9723 xtensa_hard_regno_nregs): New function.
9724 * system.h (HARD_REGNO_NREGS): Poison.
9726 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9728 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9729 hard_regno_nregs instead of HARD_REGNO_NREGS.
9730 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9731 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9732 (c6x_expand_epilogue): Likewise.
9733 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9734 (frv_read_iacc_argument): Likewise.
9735 * config/sh/sh.c: Include regs.h.
9736 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9737 (regs_used): Likewise.
9738 (output_stack_adjust): Likewise.
9739 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9740 * expmed.c: Include regs.h.
9741 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9742 * ree.c: Include regs.h.
9743 (combine_reaching_defs): Use hard_regno_nregs instead of
9745 (add_removable_extension): Likewise.
9747 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9749 * regs.h (hard_regno_nregs): Turn into a function.
9750 (end_hard_regno): Update accordingly.
9751 * caller-save.c (setup_save_areas): Likewise.
9752 (save_call_clobbered_regs): Likewise.
9753 (replace_reg_with_saved_mem): Likewise.
9754 (insert_restore): Likewise.
9755 (insert_save): Likewise.
9756 * combine.c (can_change_dest_mode): Likewise.
9757 (move_deaths): Likewise.
9758 (distribute_notes): Likewise.
9759 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9760 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9761 (rs6000_split_multireg_move): Likewise.
9762 (rs6000_register_move_cost): Likewise.
9763 (rs6000_memory_move_cost): Likewise.
9764 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9765 (rs6000_split_multireg_move): Likewise.
9766 (rs6000_register_move_cost): Likewise.
9767 (rs6000_memory_move_cost): Likewise.
9768 * cselib.c (cselib_reset_table): Likewise.
9769 (cselib_lookup_1): Likewise.
9770 * emit-rtl.c (set_mode_and_regno): Likewise.
9771 * function.c (aggregate_value_p): Likewise.
9772 * ira-color.c (setup_profitable_hard_regs): Likewise.
9773 (check_hard_reg_p): Likewise.
9774 (calculate_saved_nregs): Likewise.
9775 (assign_hard_reg): Likewise.
9776 (improve_allocation): Likewise.
9777 (calculate_spill_cost): Likewise.
9778 * ira-emit.c (modify_move_list): Likewise.
9779 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9780 (ira_hard_reg_in_set_p): Likewise.
9781 * ira.c (setup_reg_mode_hard_regset): Likewise.
9782 (clarify_prohibited_class_mode_regs): Likewise.
9783 (check_allocation): Likewise.
9784 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9785 (lra_setup_reg_renumber): Likewise.
9786 (setup_try_hard_regno_pseudos): Likewise.
9787 (spill_for): Likewise.
9788 (assign_hard_regno): Likewise.
9789 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9790 * lra-constraints.c (in_class_p): Likewise.
9791 (lra_constraint_offset): Likewise.
9792 (simplify_operand_subreg): Likewise.
9793 (lra_constraints): Likewise.
9794 (split_reg): Likewise.
9795 (split_if_necessary): Likewise.
9796 (invariant_p): Likewise.
9797 (inherit_in_ebb): Likewise.
9798 * lra-lives.c (process_bb_lives): Likewise.
9799 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9800 (get_hard_regs): Likewise.
9801 (do_remat): Likewise.
9802 * lra-spills.c (assign_spill_hard_regs): Likewise.
9803 * mode-switching.c (create_pre_exit): Likewise.
9804 * postreload.c (reload_combine_recognize_pattern): Likewise.
9805 * recog.c (peep2_find_free_register): Likewise.
9806 * regcprop.c (kill_value_regno): Likewise.
9807 (set_value_regno): Likewise.
9808 (copy_value): Likewise.
9809 (maybe_mode_change): Likewise.
9810 (find_oldest_value_reg): Likewise.
9811 (copyprop_hardreg_forward_1): Likewise.
9812 * regrename.c (check_new_reg_p): Likewise.
9813 (regrename_do_replace): Likewise.
9814 * reload.c (push_reload): Likewise.
9815 (combine_reloads): Likewise.
9816 (find_dummy_reload): Likewise.
9817 (operands_match_p): Likewise.
9818 (find_reloads): Likewise.
9819 (find_equiv_reg): Likewise.
9820 (reload_adjust_reg_for_mode): Likewise.
9821 * reload1.c (count_pseudo): Likewise.
9822 (count_spilled_pseudo): Likewise.
9823 (find_reg): Likewise.
9824 (clear_reload_reg_in_use): Likewise.
9825 (free_for_value_p): Likewise.
9826 (allocate_reload_reg): Likewise.
9827 (choose_reload_regs): Likewise.
9828 (reload_adjust_reg_for_temp): Likewise.
9829 (emit_reload_insns): Likewise.
9830 (delete_output_reload): Likewise.
9831 * rtlanal.c (subreg_get_info): Likewise.
9832 * sched-deps.c (sched_analyze_reg): Likewise.
9833 * sel-sched.c (init_regs_for_mode): Likewise.
9834 (mark_unavailable_hard_regs): Likewise.
9835 (choose_best_reg_1): Likewise.
9836 (verify_target_availability): Likewise.
9837 * valtrack.c (dead_debug_insert_temp): Likewise.
9838 * var-tracking.c (track_loc_p): Likewise.
9839 (emit_note_insn_var_location): Likewise.
9840 * varasm.c (make_decl_rtl): Likewise.
9841 * reginfo.c (choose_hard_reg_mode): Likewise.
9842 (init_reg_modes_target): Refer directly to
9843 this_target_regs->x_hard_regno_nregs.
9845 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9847 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9848 instead of hard_regno_nregs.
9850 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9852 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9853 end_hard_regno instead of hard_regno_nregs.
9854 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9855 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9856 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9857 * ira-color.c (improve_allocation): Likewise.
9858 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9859 * lra-lives.c (mark_regno_live): Likewise.
9860 (mark_regno_dead): Likewise.
9861 * lra-remat.c (operand_to_remat): Likewise.
9862 * lra.c (collect_non_operand_hard_regs): Likewise.
9863 * postreload.c (reload_combine_note_store): Likewise.
9864 (move2add_valid_value_p): Likewise.
9865 * reload.c (regno_clobbered_p): Likewise.
9867 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9869 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9871 * config/v850/v850.c (v850_reorg): Likewise.
9872 * reload.c (refers_to_regno_for_reload_p): Likewise.
9873 (find_equiv_reg): Likewise.
9874 * reload1.c (reload_reg_reaches_end_p): Likewise.
9876 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9878 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9880 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9881 * config/arm/arm.c (output_move_neon): Likewise.
9882 (arm_attr_length_move_neon): Likewise.
9883 (neon_split_vcombine): Likewise.
9884 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9885 (c6x_mark_reg_written): Likewise.
9886 (c6x_dwarf_register_span): Likewise.
9887 * config/i386/i386.c (ix86_save_reg): Likewise.
9888 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9889 (rws_access_reg): Likewise.
9890 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9891 * mode-switching.c (create_pre_exit): Likewise.
9892 * ree.c (combine_reaching_defs): Likewise.
9893 (add_removable_extension): Likewise.
9894 * regcprop.c (find_oldest_value_reg): Likewise.
9895 (copyprop_hardreg_forward_1): Likewise.
9896 * reload.c (reload_inner_reg_of_subreg): Likewise.
9897 (push_reload): Likewise.
9898 (combine_reloads): Likewise.
9899 (find_dummy_reload): Likewise.
9900 (reload_adjust_reg_for_mode): Likewise.
9901 * reload1.c (find_reload_regs): Likewise.
9902 (forget_old_reloads_1): Likewise.
9903 (reload_reg_free_for_value_p): Likewise.
9904 (reload_adjust_reg_for_temp): Likewise.
9905 (emit_reload_insns): Likewise.
9906 (delete_output_reload): Likewise.
9907 * sel-sched.c (choose_best_reg_1): Likewise.
9908 (choose_best_pseudo_reg): Likewise.
9910 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9911 Alan Hayward <alan.hayward@arm.com>
9912 David Sherwood <david.sherwood@arm.com>
9914 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9915 * target.def (slow_unaligned_access): New hook.
9916 * targhooks.h (default_slow_unaligned_access): Declare.
9917 * targhooks.c (default_slow_unaligned_access): New function.
9918 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9919 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9920 * doc/tm.texi: Regenerate.
9921 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9922 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9923 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9925 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9926 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9928 (rs6000_slow_unaligned_access): New function.
9929 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9930 (expand_block_compare): Likewise.
9931 (expand_strn_compare): Likewise.
9932 (rs6000_rtx_costs): Likewise.
9933 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9934 (riscv_slow_unaligned_access): Likewise.
9935 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9936 (riscv_slow_unaligned_access_p): ...this and make static.
9937 (riscv_option_override): Update accordingly.
9938 (riscv_slow_unaligned_access): New function.
9939 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9940 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9941 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9942 (rs6000_slow_unaligned_access): New function.
9943 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9944 (rs6000_rtx_costs): Likewise.
9945 * config/rs6000/rs6000-string.c (expand_block_compare)
9946 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9947 of SLOW_UNALIGNED_ACCESS.
9948 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9949 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9950 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9951 of SLOW_UNALIGNED_ACCESS.
9952 * expmed.c (simple_mem_bitfield_p): Likewise.
9953 * expr.c (alignment_for_piecewise_move): Likewise.
9954 (emit_group_load_1): Likewise.
9955 (emit_group_store): Likewise.
9956 (copy_blkmode_from_reg): Likewise.
9957 (emit_push_insn): Likewise.
9958 (expand_assignment): Likewise.
9959 (store_field): Likewise.
9960 (expand_expr_real_1): Likewise.
9961 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9962 * lra-constraints.c (simplify_operand_subreg): Likewise.
9963 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9964 * gimple-ssa-store-merging.c: Likewise in block comment at start
9966 * tree-ssa-strlen.c: Include target.h.
9967 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9968 of SLOW_UNALIGNED_ACCESS.
9969 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9971 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9973 PR rtl-optimization/82185
9974 * expmed.c (emit_store_flag_int): Only test tem if it has been
9977 2017-09-12 Richard Biener <rguenther@suse.de>
9980 * match.pd ((FTYPE) N CMP CST): Fix typo.
9982 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9984 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9986 (mips_near_type_p): Add 'short_call' attribute as a synonym
9988 * doc/extend.texi (short_call): Document new function attribute.
9990 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9993 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9994 assertion check that in the condition.
9995 (get_atomic_generic_size): Likewise. Before testing if parameter
9996 has pointer type, if it has array type, call for C++
9997 default_conversion to perform array-to-pointer conversion.
9999 2017-09-12 Richard Biener <rguenther@suse.de>
10001 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10002 for operations we cannot scalarize.
10004 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10006 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10007 vectors heap vectors. Clean up comments.
10008 Make visited_bbs a reference.
10009 (profitable_jump_thread_path): Make GC
10010 vectors heap vectors. Clean up comments.
10012 (convert_and_register_jump_thread_path): Make GC vectors heap
10014 (check_subpath_and_update_thread_path): Same. Clean up comments.
10015 Make visited_bbs a reference.
10016 (handle_phi): Abstract common code to to
10017 register_jump_thread_path_if_profitable.
10018 Rename VAR_BB to DEF_BB.
10020 Make GC vectors heap vectors.
10021 Make visited_bbs a reference.
10022 (handle_assignment): Same.
10023 (register_jump_thread_path_if_profitable): New.
10024 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10026 Make GC vectors heap vectors. Clean up comments.
10027 Make visited_bbs a reference.
10028 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10029 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10032 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10035 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10036 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10038 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10040 Revert r251800 and r251799.
10042 2017-09-11 Martin Jambor <mjambor@suse.cz>
10045 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10046 arguments in advance.
10047 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10048 use it to find predecessor edges.
10049 (naive_outof_ssa): Collect vector of predecessors.
10051 2017-09-08 Jason Merrill <jason@redhat.com>
10053 PR c++/70029 - ICE with ref-qualifier and -flto
10054 * langhooks.h (struct lang_hooks_for_types): Add
10055 copy_lang_qualifiers.
10056 * attribs.c (build_type_attribute_qual_variant): Use it.
10057 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10059 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10060 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10062 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10065 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10066 (*mulsi3_sp64): New instruction.
10067 (mulsi3): New expander.
10069 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10071 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10073 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10075 * sancov.c: Include memmodel.h.
10077 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10080 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10083 2017-09-07 Carl Love <cel@us.ibm.com>
10085 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10086 the sldi instruction.
10088 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10090 * sancov.c: Include tm_p.h.
10092 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10095 * output.h (switch_to_other_text_partition): New declaration.
10096 * varasm.c (switch_to_other_text_partition): New function.
10097 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10098 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10099 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10100 to the other text partition before emitting LCL label and switch back
10101 after emitting the word after it.
10103 2017-09-07 Richard Biener <rguenther@suse.de>
10105 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10106 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10107 critical edges here, after loop init.
10108 (pass_data_pre): Remove PROP_no_crit_edges flags.
10109 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10110 for valueization of call args to avoid leaking VN_TOP.
10111 (visit_use): Assert we do not visit default defs.
10112 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10113 Use error_mark_node to more easily detect leaking VN_TOP.
10114 All default-defs are varying, not VN_TOP. Mark them visited.
10115 (run_scc_vn): Make code match comment.
10117 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10119 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10120 OPTION_MASK_FLOAT128_KEYWORD.
10121 (POWERPC_MASKS): Likewise.
10122 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10123 support for the -mfloat128-type option, and make -mfloat128
10124 default on PowerPC Linux systems. Define or undefine
10125 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10126 Define __float128 to be __ieee128 if IEEE 128-bit support is
10127 enabled, or undefine it.
10128 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10129 Delete defining __FLOAT128_TYPE__.
10130 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10131 -mfloat128-type option and make -mfloat128 default on PowerPC
10133 (TARGET_FLOAT128_TYPE): Likewise.
10134 (-mfloat128-type): Likewise.
10135 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10136 Delete the -mfloat128-type option and make -mfloat128 default on
10137 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10138 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10139 128-bit floating point is enabled. Change tests from using
10140 -mfloat128-type to -mfloat128.
10141 (rs6000_mangle_type): Use the correct mangling for the __float128
10142 type even if normal long double is restricted to 64-bits.
10143 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10145 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10146 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10147 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10149 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10150 documentation for -mfloat128.
10152 2017-09-06 Olivier Hainque <hainque@adacore.com>
10154 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10156 2017-09-06 Wish Wu <wishwu007@gmail.com>
10157 Jakub Jelinek <jakub@redhat.com>
10159 * asan.c (initialize_sanitizer_builtins): Add
10160 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10161 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10162 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10163 BT_FN_VOID_UINT64_PTR variables.
10164 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10165 (BT_FN_VOID_UINT16_UINT16): Likewise.
10166 (BT_FN_VOID_UINT32_UINT32): Likewise.
10167 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10168 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10169 (BT_FN_VOID_UINT64_PTR): Likewise.
10170 * common.opt (flag_sanitize_coverage): New variable.
10171 (fsanitize-coverage=trace-pc): Remove.
10172 (fsanitize-coverage=): Add.
10173 * flag-types.h (enum sanitize_coverage_code): New enum.
10174 * fold-const.c (fold_range_test): Disable non-short-circuit
10175 optimization if flag_sanitize_coverage.
10176 (fold_truth_andor): Likewise.
10177 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10178 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10179 (coverage_sanitizer_opts): New array.
10180 (get_closest_sanitizer_option): Add OPTS argument, handle also
10181 OPT_fsanitize_coverage_.
10182 (parse_sanitizer_options): Adjusted to also handle
10183 OPT_fsanitize_coverage_.
10184 (common_handle_option): Add OPT_fsanitize_coverage_.
10185 * sancov.c (instrument_comparison, instrument_switch): New function.
10186 (sancov_pass): Add trace-cmp support.
10187 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10188 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10189 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10190 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10191 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10192 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10193 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10194 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10195 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10196 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10198 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10200 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10201 error. Only quit immediately if parsing is complete.
10202 (BEGIN): Initialize fatal_err and parse_done.
10203 (begin fpu, end fpu): Check number of arguments.
10204 (begin arch, end arch): Likewise.
10205 (begin cpu, end cpu): Likewise.
10206 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10207 (optalias): Likewise.
10209 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10211 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10212 * config/arm/arm-isa.h: Delete. Move definitions to ...
10213 * arm-cpus.in: ... here. Use new feature and fgroup values.
10214 * config/arm/arm.c (arm_option_override): Use lower case for feature
10216 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10217 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10218 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10220 (print_isa_bits_for): New function.
10221 (gen_isa): New function.
10222 (gen_comm_data): Use print_isa_bits_for.
10223 (define feature): New keyword.
10224 (define fgroup): New keyword.
10225 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10226 (arm-isa.h): Add rule to generate file.
10227 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10228 case for feature bit names.
10230 2017-09-06 Richard Biener <rguenther@suse.de>
10232 * tree-ssa-pre.c (NECESSARY): Remove.
10233 (create_expression_by_pieces): Do not touch pass-local flags.
10234 (insert_into_preds_of_block): Likewise.
10235 (do_pre_regular_insertion): Likewise.
10236 (eliminate_insert): Likewise.
10237 (eliminate_dom_walker::before_dom_children): Likewise.
10238 (fini_eliminate): Do not look at inserted_exprs.
10239 (mark_operand_necessary): Remove.
10240 (remove_dead_inserted_code): Replace with simple work-list
10241 algorithm based on inserted_exprs and SSA uses.
10242 (pass_pre::execute): Re-order fini_eliminate and
10243 remove_dead_inserted_code.
10245 2017-09-06 Olivier Hainque <hainque@adacore.com>
10247 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10248 for VxWorks 7. Adjust surrounding comments.
10250 2017-09-06 Richard Biener <rguenther@suse.de>
10252 * gimple-ssa-strength-reduction.c
10253 (find_candidates_dom_walker::before_dom_children): Also allow
10256 2017-09-06 Richard Biener <rguenther@suse.de>
10258 PR tree-optimization/82108
10259 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10260 for gap in the non-permutation SLP case.
10262 2017-09-06 Martin Jambor <mjambor@suse.cz>
10264 PR tree-optimization/82078
10265 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10266 add_access_to_work_queue...
10267 (build_accesses_from_assign): ...here.
10268 (propagate_all_subaccesses): Make sure racc is the group
10269 representative, if there is one.
10271 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10273 PR middle-end/82095
10274 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10277 2017-09-06 Richard Biener <rguenther@suse.de>
10279 * gimple-ssa-strength-reduction.c
10280 (find_candidates_dom_walker::before_doom_children): Use a
10281 type and not a mode check.
10283 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10286 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10288 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10290 2017-09-05 Jeff Law <law@redhat.com>
10292 PR tree-optimization/64910
10293 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10294 cases where we have 3 or more operands.
10296 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10298 PR middle-end/81768
10299 * omp-low.c (lower_omp_for): Recompute tree invariant if
10300 gimple_omp_for_initial/final is ADDR_EXPR.
10302 PR middle-end/81768
10303 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10304 into gimple val before gimplification fo the COND_EXPR.
10306 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10308 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10309 REGION_COPY argument.
10310 (thread_through_all_blocks): Remove unused argument to
10311 duplicate_thread_path.
10313 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10314 Alan Hayward <alan.hayward@arm.com>
10315 David Sherwood <david.sherwood@arm.com>
10317 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10318 Take a scalar_mode rather than a machine_mode.
10319 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10320 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10321 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10322 (aarch64_gen_adjusted_ldpstp): Likewise.
10323 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10325 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10326 Alan Hayward <alan.hayward@arm.com>
10327 David Sherwood <david.sherwood@arm.com>
10329 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10330 Take a scalar_int_mode instead of a machine_mode.
10331 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10332 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10333 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10334 (aarch64_simd_attr_length_rglist): Delete.
10335 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10336 a scalar_int_mode instead of a machine_mode.
10337 (aarch64_add_offset): Likewise.
10338 (aarch64_internal_mov_immediate): Likewise
10339 (aarch64_add_constant_internal): Likewise.
10340 (aarch64_add_constant): Likewise.
10341 (aarch64_movw_imm): Likewise.
10342 (aarch64_rtx_arith_op_extract_p): Likewise.
10343 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10344 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10345 Remove assert that the mode isn't a vector.
10346 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10347 (aarch64_expand_mov_immediate): Update calls after above changes.
10348 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10349 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10350 (aarch64_move_imm): Likewise.
10351 (aarch64_can_const_movi_rtx_p): Likewise.
10352 (aarch64_strip_extend): Likewise.
10353 (aarch64_extr_rtx_p): Likewise.
10354 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10355 a CONST_INT when the mode parameter is VOIDmode.
10356 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10358 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10360 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10361 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10362 (bitwise_type_for_mode): Update accordingly.
10364 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10366 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10367 * stor-layout.c (mode_for_size_tree): Likewise.
10368 (mode_for_array): Update accordingly.
10369 (layout_decl): Likewise.
10370 (compute_record_mode): Likewise. Only set the mode once.
10372 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10374 * target.def (get_mask_mode): Change return type to opt_mode.
10376 * doc/tm.texi: Regenerate.
10377 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10378 * targhooks.c (default_get_mask_mode): Likewise.
10379 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10380 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10381 targetm.get_mask_mode.
10382 * tree.c (build_truth_vector_type): Likewise.
10384 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10386 * machmode.h (mode_for_vector): Return an opt_mode.
10387 * stor-layout.c (mode_for_vector): Likewise.
10388 (mode_for_int_vector): Update accordingly.
10389 (layout_type): Likewise.
10390 * config/i386/i386.c (emit_memmov): Likewise.
10391 (ix86_expand_set_or_movmem): Likewise.
10392 (ix86_expand_vector_init): Likewise.
10393 (ix86_get_mask_mode): Likewise.
10394 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10396 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10397 * expmed.c (extract_bit_field_1): Likewise.
10398 * expr.c (expand_expr_real_2): Likewise.
10399 * optabs-query.c (can_vec_perm_p): Likewise.
10400 (can_vec_mask_load_store_p): Likewise.
10401 * optabs.c (expand_vec_perm): Likewise.
10402 * targhooks.c (default_get_mask_mode): Likewise.
10403 * tree-vect-stmts.c (vectorizable_store): Likewise.
10404 (vectorizable_load): Likewise.
10405 (get_vectype_for_scalar_type_and_size): Likewise.
10407 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10409 * machmode.h (mode_for_int_vector): New function.
10410 * stor-layout.c (mode_for_int_vector): Likewise.
10411 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10412 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10413 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10414 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10415 (s390_expand_vcond): Likewise.
10417 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10419 * machmode.h (opt_machine_mode): New type.
10420 (opt_mode<T>): Allow construction from anything that can be
10422 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10423 (mode_for_size): Return an opt_machine_mode.
10424 * stor-layout.c (mode_for_size): Likewise.
10425 (mode_for_size_tree): Update call accordingly.
10426 (bitwise_mode_for_mode): Likewise.
10427 (make_fract_type): Likewise.
10428 (make_accum_type): Likewise.
10429 * caller-save.c (replace_reg_with_saved_mem): Update call
10431 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10432 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10433 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10434 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10435 * expmed.c (extract_bit_field_1): Likewise.
10436 * reload.c (get_secondary_mem): Likewise.
10437 * varasm.c (assemble_integer): Likewise.
10438 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10441 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10443 * machmode.h (decimal_float_mode_for_size): New function.
10444 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10445 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10446 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10447 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10448 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10449 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10451 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10453 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10454 (get_builtin_sync_mode): Likewise.
10455 (expand_ifn_atomic_compare_exchange): Likewise.
10456 (expand_builtin_atomic_clear): Likewise.
10457 (expand_builtin_atomic_test_and_set): Likewise.
10458 (fold_builtin_atomic_always_lock_free): Likewise.
10459 * calls.c (compute_argument_addresses): Likewise.
10460 (emit_library_call_value_1): Likewise.
10461 (store_one_arg): Likewise.
10462 * combine.c (combine_instructions): Likewise.
10463 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10464 * config/arm/arm.c (arm_function_value): Likewise.
10465 (aapcs_allocate_return_reg): Likewise.
10466 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10467 * config/i386/i386.c (construct_container): Likewise.
10468 (ix86_gimplify_va_arg): Likewise.
10469 (ix86_expand_sse_cmp): Likewise.
10470 (emit_memmov): Likewise.
10471 (emit_memset): Likewise.
10472 (expand_small_movmem_or_setmem): Likewise.
10473 (ix86_expand_pextr): Likewise.
10474 (ix86_expand_pinsr): Likewise.
10475 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10476 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10478 * config/mips/mips.c (mips_function_value_1) Likewise.
10479 (mips_block_move_straight): Likewise.
10480 (mips_expand_ins_as_unaligned_store): Likewise.
10481 * config/powerpcspe/powerpcspe.c
10482 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10483 (rs6000_darwin64_record_arg_flush): Likewise.
10484 * config/rs6000/rs6000.c
10485 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10486 (rs6000_darwin64_record_arg_flush): Likewise.
10487 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10488 (sparc_function_value_1): Likewise.
10489 * config/spu/spu.c (adjust_operand): Likewise.
10490 (spu_emit_branch_or_set): Likewise.
10491 (arith_immediate_p): Likewise.
10492 * emit-rtl.c (gen_lowpart_common): Likewise.
10493 * expr.c (expand_expr_real_1): Likewise.
10494 * function.c (assign_parm_setup_block): Likewise.
10495 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10496 * reload1.c (alter_reg): Likewise.
10497 * stor-layout.c (mode_for_vector): Likewise.
10498 (layout_type): Likewise.
10500 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10502 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10503 (spu_convert_move): Likewise.
10504 * lower-subreg.c (resolve_simple_move): Likewise.
10506 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10509 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10510 define_insn to a define_expand.
10511 (altivec_vsum2sws_direct): New define_insn.
10512 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10514 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10516 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10519 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10524 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10525 realign stack if stack alignment needed is less than incoming
10528 2017-09-05 Marek Polacek <polacek@redhat.com>
10531 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10534 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10536 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10538 2017-09-05 Richard Biener <rguenther@suse.de>
10540 PR tree-optimization/82084
10541 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10543 2017-09-05 Richard Biener <rguenther@suse.de>
10545 PR tree-optimization/82102
10546 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10548 2017-09-05 Martin Liska <mliska@suse.cz>
10550 PR tree-optimization/82032
10551 * tree-cfg.c (generate_range_test): New function.
10552 * tree-cfg.h (generate_range_test): Declared here.
10553 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10554 (cleanup_control_expr_graph): Use it.
10555 * tree-switch-conversion.c (try_switch_expansion): Remove
10557 (emit_case_nodes): Use generate_range_test.
10559 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10562 * config/i386/i386.md (*<btsc><mode>_mask): Add
10563 TARGET_USE_BT to insn constraint.
10564 (*btr<mode>_mask): Ditto.
10566 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10568 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10569 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10571 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10574 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10575 TARGET_NEON and TARGET_IWMMXT.
10576 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10577 TARGET_NEON and TARGET_IWMMXT.
10578 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10580 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10582 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10583 (ix86_rewrite_tls_address): Ditto.
10584 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10585 (ix86_rewrite_tls_address_1): Ditto.
10586 (ix86_rewrite_tls_address): Ditto.
10587 * config/i386/predicates.md (tls_address_pattern): New predicate.
10588 * config/i386/i386.md (TLS address splitter): New splitter.
10590 2017-09-04 Richard Biener <rguenther@suse.de>
10592 PR tree-optimization/82084
10593 * fold-const.h (can_native_encode_string_p): Declare.
10594 * fold-const.c (can_native_encode_string_p): Factor out from ...
10595 (native_encode_string): ... here.
10596 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10597 vectorizing stores from constants we later cannot handle.
10599 2017-09-04 Marek Polacek <polacek@redhat.com>
10602 * doc/invoke.texi: Update -Wtautological-compare documentation.
10604 2017-09-04 Jeff Law <law@redhat.com>
10606 PR tree-optimization/64910
10607 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10608 swap the first and last operand if the last is a constant.
10610 2017-09-04 Marek Polacek <polacek@redhat.com>
10613 * convert.c (do_narrow): When sanitizing signed integer overflows,
10614 bail out for signed types.
10615 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10617 2017-09-04 Richard Biener <rguenther@suse.de>
10619 PR tree-optimization/82060
10620 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10621 Move devirtualization after stmt folding and before EH/AB/noreturn
10622 cleanup to get the stmt refs canonicalized. Use a bool instead
10623 of gimple_modified_p since that doesn't work for NOPs. Schedule
10624 NOPs generated by folding for removal.
10626 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10627 Alan Hayward <alan.hayward@arm.com>
10628 David Sherwood <david.sherwood@arm.com>
10630 * coretypes.h (pad_direction): New enum.
10631 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10632 (FUNCTION_ARG_PADDING): Likewise.
10633 * target.def (function_arg_padding): New hook.
10634 * targhooks.h (default_function_arg_padding): Declare.
10635 * targhooks.c (default_function_arg_padding): New function.
10636 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10637 (TARGET_FUNCTION_ARG_PADDING): ...this.
10638 * doc/tm.texi: Regenerate.
10639 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10640 instead of direction.
10641 (compute_argument_addresses): Likewise.
10642 (load_register_parameters): Likewise.
10643 (emit_library_call_value_1): Likewise.
10644 (store_one_arg): Use targetm.calls.function_arg_padding instead
10645 of FUNCTION_ARG_PADDING.
10646 (must_pass_in_stack_var_size_or_pad): Likewise.
10647 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10648 (emit_group_store): Likewise.
10649 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10650 instead of FUNCTION_ARG_PADDING.
10651 (emit_push_insn): Likewise, and propagate enum change throughout
10653 * function.h (direction): Delete.
10654 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10656 * function.c (assign_parm_find_stack_rtl): Likewise.
10657 (assign_parm_setup_block_p): Likewise.
10658 (assign_parm_setup_block): Likewise.
10659 (gimplify_parameters): Likewise.
10660 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10661 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10663 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10664 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10665 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10666 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10667 (aarch64_function_arg_padding): ...this new function.
10668 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10669 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10670 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10671 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10672 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10673 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10674 (arm_pad_arg_upward): Replace with...
10675 (arm_function_arg_padding): ...this new function.
10676 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10678 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10679 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10680 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10681 (ia64_hpux_function_arg_padding): Replace with...
10682 (ia64_function_arg_padding): ...this new function. Use pad_direction
10683 instead of direction. Check for TARGET_HPUX.
10684 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10685 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10686 (iq2000_function_arg_padding): New function.
10687 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10688 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10689 (mips_function_arg_padding): ...this new function.
10690 (mips_pad_reg_upward): Update accordingly.
10691 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10692 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10693 targetm.calls.function_arg_padding.
10694 (FUNCTION_ARG_PADDING): Delete.
10695 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10696 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10697 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10698 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10699 (nios2_block_reg_padding): Return pad_direction instead of direction.
10700 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10701 instead of direction.
10702 (nios2_function_arg_padding): Likewise. Make static.
10703 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10704 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10705 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10706 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10707 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10708 (pa_function_arg_padding): Make static. Return pad_direction instead
10710 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10711 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10712 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10713 instead of direction. Use targetm.calls.function_arg_padding.
10714 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10715 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10716 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10717 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10718 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10720 (function_arg_padding): Rename to...
10721 (rs6000_function_arg_padding): ...this. Make static. Return
10722 pad_direction instead of direction.
10723 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10724 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10725 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10726 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10727 instead of direction. Use targetm.calls.function_arg_padding.
10728 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10729 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10730 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10731 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10732 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10733 (function_arg_padding): Rename to...
10734 (rs6000_function_arg_padding): ...this. Make static. Return
10735 pad_direction instead of direction.
10736 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10737 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10738 * config/s390/s390.c (s390_function_arg_padding): New function.
10739 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10740 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10741 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10742 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10743 (function_arg_padding): Rename to...
10744 (sparc_function_arg_padding): ...this. Make static. Return
10745 pad_direction instead of direction.
10746 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10747 * config/spu/spu.c (spu_function_arg_padding): New function.
10748 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10749 * system.h (FUNCTION_ARG_PADDING): Poison.
10751 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10752 Alan Hayward <alan.hayward@arm.com>
10753 David Sherwood <david.sherwood@arm.com>
10755 * target.def (modes_tieable_p): New hook.
10756 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10757 (TARGET_MODES_TIEABLE_P): ...this.
10758 * doc/tm.texi.in: Regenerate.
10759 * hooks.h (hook_bool_mode_mode_true): Declare.
10760 * hooks.c (hook_bool_mode_mode_true): New function.
10761 * combine.c (subst): Use targetm.modes_tieable_p instead of
10763 * dse.c (find_shift_sequence): Likewise.
10764 * expmed.c (extract_low_bits): Likewise.
10765 * lower-subreg.c: Include target.h.
10766 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10768 * rtlanal.c (rtx_cost): Likewise.
10769 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10770 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10771 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10772 (TARGET_MODES_TIEABLE_P): Redefine.
10773 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10774 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10775 (TARGET_MODES_TIEABLE_P): Redefine.
10776 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10777 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10778 (arc_modes_tieable_p): New function.
10779 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10780 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10781 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10782 (arm_modes_tieable_p): Make static.
10783 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10784 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10785 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10786 (TARGET_MODES_TIEABLE_P): Redefine.
10787 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10788 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10789 (TARGET_MODES_TIEABLE_P): Redefine.
10790 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10791 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10792 (cr16_modes_tieable_p): New function.
10793 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10794 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10795 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10796 (TRULY_NOOP_TRUNCATION): Update comment.
10797 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10798 (TRULY_NOOP_TRUNCATION): Update comment.
10799 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10800 (frv_modes_tieable_p): New function.
10801 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10802 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10803 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10804 (TARGET_MODES_TIEABLE_P): Redefine.
10805 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10806 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10807 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10808 (TARGET_MODES_TIEABLE_P): Redefine.
10809 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10810 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10811 (ia64_modes_tieable_p): New function.
10812 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10813 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10814 (iq2000_modes_tieable_p): New function.
10815 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10816 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10817 (lm32_modes_tieable_p): New function.
10818 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10819 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10820 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10821 (TARGET_MODES_TIEABLE_P): Redefine.
10822 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10823 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10824 (m32r_modes_tieable_p): New function.
10825 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10826 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10827 (m68k_modes_tieable_p): New function.
10828 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10829 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10830 (mcore_modes_tieable_p): New function.
10831 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10832 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10834 (TARGET_MODES_TIEABLE_P): Redefine.
10835 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10836 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10837 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10838 (TARGET_MODES_TIEABLE_P): Redefine.
10839 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10840 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10841 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10842 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10843 (mn10300_modes_tieable_p): ...this and make static.
10844 (TARGET_MODES_TIEABLE_P): Redefine.
10845 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10846 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10847 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10848 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10849 (msp430_modes_tieable_p): Make static.
10850 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10851 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10852 (TARGET_MODES_TIEABLE_P): Redefine.
10853 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10854 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10855 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10856 (TARGET_MODES_TIEABLE_P): Redefine.
10857 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10858 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10859 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10860 (TARGET_MODES_TIEABLE_P): Redefine.
10861 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10862 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10863 (pdp11_modes_tieable_p): New function.
10864 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10865 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10866 (rs6000_modes_tieable_p): New function.
10867 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10868 * config/powerpcspe/powerpcspe.md: Update comment.
10869 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10870 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10871 (TARGET_MODES_TIEABLE_P): Redefine.
10872 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10873 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10874 (rl78_modes_tieable_p): New function.
10875 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10876 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10877 (rs6000_modes_tieable_p): New function.
10878 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10879 * config/rs6000/rs6000.md: Update comment.
10880 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10881 * config/rx/rx.c (rx_modes_tieable_p): New function.
10882 (TARGET_MODES_TIEABLE_P): Redefine.
10883 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10884 * config/s390/s390.c (s390_modes_tieable_p): New function.
10885 (TARGET_MODES_TIEABLE_P): Redefine.
10886 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10887 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10888 (sh_modes_tieable_p): New function.
10889 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10890 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10891 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10892 (sparc_modes_tieable_p): Make static.
10893 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10894 * config/spu/spu.c (spu_modes_tieable_p): New function.
10895 (TARGET_MODES_TIEABLE_P): Redefine.
10896 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10897 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10898 (TARGET_MODES_TIEABLE_P): Redefine.
10899 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10900 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10901 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10902 * config/v850/v850.c (v850_modes_tieable_p): New function.
10903 (TARGET_MODES_TIEABLE_P): Redefine.
10904 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10905 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10906 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10907 (visium_modes_tieable_p): New function.
10908 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10909 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10910 (xtensa_modes_tieable_p): New function.
10911 * system.h (MODES_TIEABLE_P): Poison.
10913 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10914 Alan Hayward <alan.hayward@arm.com>
10915 David Sherwood <david.sherwood@arm.com>
10917 * target.def (hard_regno_mode_ok): New hook.
10918 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10919 (TARGET_HARD_REGNO_MODE_OK): ...this.
10920 * doc/tm.texi.in: Regenerate.
10921 * hooks.h (hook_bool_uint_mode_true): Declare.
10922 * hooks.c (hook_bool_uint_mode_true): New function.
10923 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10924 HARD_REGNO_MODE_OK.
10925 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10926 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10927 instead of HARD_REGNO_MODE_OK.
10928 * caller-save.c: Include target.h.
10929 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10930 HARD_REGNO_MODE_OK.
10931 * combine.c (can_combine_p): Likewise.
10932 (combinable_i3pat): Likewise.
10933 (can_change_dest_mode): Likewise.
10934 * expr.c (init_expr_target): Likewise.
10935 (convert_move): Likewise.
10936 (convert_modes): Likewise.
10937 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10938 (setup_prohibited_mode_move_regs): Likewise.
10939 * ira.h (target_ira): Likewise.
10940 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10941 * lra-constraints.c (process_alt_operands): Likewise.
10942 (split_reg): Likewise.
10943 * recog.c (peep2_find_free_register): Likewise.
10944 * ree.c (combine_reaching_defs): Likewise.
10945 * regcprop.c (maybe_mode_change): Likewise.
10946 * reginfo.c (init_reg_sets_1): Likewise.
10947 (choose_hard_reg_mode): Likewise.
10948 (simplifiable_subregs): Likewise.
10949 * regrename.c (check_new_reg_p): Likewise.
10950 * reload.c (find_valid_class): Likewise.
10951 (find_valid_class_1): Likewise.
10952 (reload_inner_reg_of_subreg): Likewise.
10953 (push_reload): Likewise.
10954 (combine_reloads): Likewise.
10955 (find_dummy_reload): Likewise.
10956 (find_reloads): Likewise.
10957 * reload1.c (find_reg): Likewise.
10958 (set_reload_reg): Likewise.
10959 (allocate_reload_reg): Likewise.
10960 (choose_reload_regs): Likewise.
10961 (reload_adjust_reg_for_temp): Likewise.
10962 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10963 (simplify_subreg_regno): Likewise.
10964 * sel-sched.c (init_regs_for_mode): Likewise.
10965 * varasm.c (make_decl_rtl): Likewise.
10966 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10967 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10968 HARD_REGNO_MODE_OK.
10969 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10970 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10971 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10972 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10973 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10974 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10975 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10976 (arc_mode_class): Delete.
10977 (HARD_REGNO_MODE_OK): Delete.
10978 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10979 (arc_hard_regno_mode_ok): Rename old array to...
10980 (arc_hard_regno_mode_ok_modes): ...this.
10981 (arc_conditional_register_usage): Update accordingly.
10982 (arc_mode_class): Make static.
10983 (arc_hard_regno_mode_ok): New function.
10984 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10985 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10986 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10987 (arm_hard_regno_mode_ok): Make static.
10988 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10989 HARD_REGNO_MODE_OK.
10990 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10991 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10992 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10994 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10995 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10996 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10997 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10998 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10999 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11000 * config/bfin/predicates.md (valid_reg_operand): Use
11001 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11002 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11003 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11004 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11005 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11006 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11007 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11008 (cr16_hard_regno_mode_ok): Make static and return a bool.
11009 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11010 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11011 (cris_hard_regno_mode_ok): New function.
11012 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11013 (epiphany_mode_class): Delete.
11014 (HARD_REGNO_MODE_OK): Delete.
11015 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11016 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11017 (hard_regno_mode_ok): Rename to...
11018 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11019 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11020 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11021 HARD_REGNO_MODE_OK.
11022 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11023 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11024 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11025 (frv_hard_regno_mode_ok): Make static and return a bool.
11026 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11027 HARD_REGNO_MODE_OK.
11028 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11029 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11030 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11031 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11033 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11034 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11035 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11036 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11038 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11039 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11040 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11041 (ia64_hard_regno_mode_ok): New function.
11042 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11043 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11044 (iq2000_hard_regno_mode_ok): New function.
11045 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11046 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11047 (lm32_hard_regno_mode_ok): New function.
11048 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11049 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11050 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11051 instead of HARD_REGNO_MODE_OK.
11052 (m32c_hard_regno_ok): Rename to...
11053 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11054 (m32c_cannot_change_mode_class): Update accordingly.
11055 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11056 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11057 (m32r_mode_class): Delete.
11058 (HARD_REGNO_MODE_OK): Delete.
11059 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11060 (m32r_hard_regno_mode_ok): Rename to...
11061 (m32r_hard_regno_modes): ...this.
11062 (m32r_mode_class): Make static.
11063 (m32r_hard_regno_mode_ok): New function.
11064 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11065 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11066 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11067 (m68k_hard_regno_mode_ok): Make static.
11068 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11069 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11070 (mcore_hard_regno_mode_ok): New function.
11071 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11072 (HARD_REGNO_MODE_OK): Delete.
11073 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11075 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11076 (microblaze_hard_regno_mode_ok): New function.
11077 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11078 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11079 (mips_hard_regno_mode_ok): Delete.
11080 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11081 (mips_hard_regno_mode_ok_p): ...this and make static.
11082 (mips_hard_regno_mode_ok_p): Rename to...
11083 (mips_hard_regno_mode_ok_uncached): ...this.
11084 (mips_hard_regno_mode_ok): New function.
11085 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11086 of HARD_REGNO_MODE_OK.
11087 (mips_option_override): Update after above name changes.
11088 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11089 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11090 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11091 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11092 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11093 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11094 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11095 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11096 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11097 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11098 (msp430_hard_regno_mode_ok): Make static and return a bool.
11099 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11100 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11101 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11103 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11104 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11105 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11106 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11107 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11108 (PA_HARD_REGNO_MODE_OK): ...this
11109 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11110 (PA_HARD_REGNO_MODE_OK): ...this.
11111 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11112 (pa_hard_regno_mode_ok): New function.
11113 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11114 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11115 (pdp11_hard_regno_mode_ok): New function.
11116 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11117 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11119 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11121 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11122 (rs6000_hard_regno_mode_ok): Rename to...
11123 (rs6000_hard_regno_mode_ok_uncached): ...this.
11124 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11125 (rs6000_hard_regno_mode_ok): New function.
11126 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11127 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11128 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11129 (riscv_hard_regno_mode_ok): ...this and make static.
11130 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11131 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11132 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11133 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11134 (rl78_hard_regno_mode_ok): Make static and return bool.
11135 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11136 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11138 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11139 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11140 (rs6000_hard_regno_mode_ok): Rename to...
11141 (rs6000_hard_regno_mode_ok_uncached): ...this.
11142 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11143 (rs6000_hard_regno_mode_ok): New function.
11144 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11145 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11146 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11147 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11148 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11149 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11150 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11151 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11152 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11153 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11154 (sh_hard_regno_mode_ok): Make static.
11155 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11156 instead of HARD_REGNO_MODE_OK.
11157 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11158 (sparc_mode_class): Delete.
11159 (HARD_REGNO_MODE_OK): Delete.
11160 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11161 (hard_regno_mode_classes): Make static.
11162 (sparc_mode_class): Likewise.
11163 (sparc_hard_regno_mode_ok): New function.
11164 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11165 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11166 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11168 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11169 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11170 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11171 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11172 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11173 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11174 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11175 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11176 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11177 (visium_hard_regno_mode_ok): New function.
11178 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11179 instead of HARD_REGNO_MODE_OK.
11180 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11181 (HARD_REGNO_MODE_OK): Delete.
11182 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11183 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11184 (xtensa_option_override): Update accordingly.
11185 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11186 (xtensa_hard_regno_mode_ok): New function.
11187 * system.h (HARD_REGNO_MODE_OK): Poison.
11189 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11190 Alan Hayward <alan.hayward@arm.com>
11191 David Sherwood <david.sherwood@arm.com>
11193 * target.def (hard_regno_call_part_clobbered): New hook.
11194 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11195 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11196 * doc/tm.texi: Regenerate.
11197 * hooks.h (hook_bool_uint_mode_false): Declare.
11198 * hooks.c (hook_bool_uint_mode_false): New function.
11199 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11200 * cselib.c (cselib_process_insn): Use
11201 targetm.hard_regno_call_part_clobbered instead of
11202 HARD_REGNO_CALL_PART_CLOBBERED.
11203 * ira-conflicts.c (ira_build_conflicts): Likewise.
11204 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11205 * lra-constraints.c (need_for_call_save_p): Likewise.
11206 * lra-lives.c: Include target.h.
11207 (check_pseudos_live_through_calls): Use
11208 targetm.hard_regno_call_part_clobbered instead of
11209 HARD_REGNO_CALL_PART_CLOBBERED.
11210 * regcprop.c: Include target.h.
11211 (copyprop_hardreg_forward_1): Use
11212 targetm.hard_regno_call_part_clobbered instead of
11213 HARD_REGNO_CALL_PART_CLOBBERED.
11214 * reginfo.c (choose_hard_reg_mode): Likewise.
11215 * regrename.c (check_new_reg_p): Likewise.
11216 * reload.c (find_equiv_reg): Likewise.
11217 * reload1.c (emit_reload_insns): Likewise.
11218 * sched-deps.c (deps_analyze_insn): Likewise.
11219 * sel-sched.c (init_regs_for_mode): Likewise.
11220 (mark_unavailable_hard_regs): Likewise.
11221 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11222 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11223 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11225 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11226 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11227 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11229 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11231 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11232 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11233 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11235 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11236 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11237 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11239 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11240 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11242 * config/powerpcspe/powerpcspe.c
11243 (rs6000_hard_regno_call_part_clobbered): New function.
11244 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11245 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11246 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11248 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11249 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11250 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11252 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11253 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11254 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11256 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11257 Alan Hayward <alan.hayward@arm.com>
11258 David Sherwood <david.sherwood@arm.com>
11260 * rtl.h (subreg_memory_offset): Declare.
11261 * emit-rtl.c (subreg_memory_offset): New function.
11262 * expmed.c (store_bit_field_1): Use it.
11263 * expr.c (undefined_operand_subword_p): Likewise.
11264 * simplify-rtx.c (simplify_subreg): Likewise.
11266 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11268 PR rtl-optimization/57448
11271 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11272 using atomic_load pattern.
11273 (expand_atomic_store): Likewise.
11275 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11278 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11279 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11280 handling. Use replace_call_with_value with NULL instead of
11281 gsi_replace, unlink_stmt_vdef and release_defs.
11283 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11286 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11288 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11291 * rtl.h (emit_library_call_value_1): Declare.
11292 (emit_library_call): Replace declaration with a series of overloads.
11293 Remove the parameter count argument.
11294 (emit_library_call_value): Likewise.
11295 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11296 with an "rtx_mode_t *".
11297 (emit_library_call_value): Delete.
11298 (emit_library_call): Likewise.
11299 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11300 (asan_emit_allocas_unpoison): Likewise.
11301 * builtins.c (expand_builtin_powi): Likewise.
11302 (expand_asan_emit_allocas_unpoison): Likewise.
11303 * cfgexpand.c (expand_main_function): Likewise.
11304 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11305 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11306 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11307 * config/arm/arm.c (arm_trampoline_init): Likewise.
11308 (arm_call_tls_get_addr): Likewise.
11309 (arm_expand_divmod_libfunc): Likewise.
11310 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11311 (smulsi3_highpart): Likewise.
11312 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11313 (c6x_expand_compare): Likewise.
11314 (c6x_expand_movmem): Likewise.
11315 * config/frv/frv.c (frv_trampoline_init): Likewise.
11316 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11317 (ix86_expand_divmod_libfunc): Likewise.
11318 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11319 (ia64_expand_compare): Likewise.
11320 (ia64_profile_hook): Likewise.
11321 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11322 (nonlocal_goto): Likewise.
11323 (restore_stack_nonlocal): Likewise.
11324 * config/m32r/m32r.c (block_move_call): Likewise.
11325 (m32r_trampoline_init): Likewise.
11326 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11327 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11328 (m68k_call_m68k_read_tp): Likewise.
11329 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11330 (microblaze_expand_divide): Likewise.
11331 * config/mips/mips.h (mips_args): Likewise.
11332 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11333 (MIPS_ICACHE_SYNC): Likewise.
11334 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11335 (nios2_trampoline_init): Likewise.
11336 * config/pa/pa.c (hppa_tls_call): Likewise.
11337 (pa_trampoline_init): Likewise.
11338 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11339 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11340 (expand_strn_compare): Likewise.
11341 (rs6000_generate_compare): Likewise.
11342 (rs6000_expand_float128_convert): Likewise.
11343 (output_profile_hook): Likewise.
11344 (rs6000_trampoline_init): Likewise.
11345 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11346 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11347 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11348 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11349 (rs6000_generate_compare): Likewise.
11350 (rs6000_expand_float128_convert): Likewise.
11351 (output_profile_hook): Likewise.
11352 (rs6000_trampoline_init): Likewise.
11353 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11354 * config/sh/sh.c (sh_trampoline_init): Likewise.
11355 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11356 (sparc_emit_float_lib_cmp): Likewise.
11357 (sparc32_initialize_trampoline): Likewise.
11358 (sparc64_initialize_trampoline): Likewise.
11359 (sparc_profile_hook): Likewise.
11360 * config/spu/spu.c (ea_load_store): Likewise.
11361 * config/spu/spu.md (floatunssidf2): Likewise.
11362 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11363 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11364 * config/visium/visium.c (expand_block_move_4): Likewise.
11365 (expand_block_move_2): Likewise.
11366 (expand_block_move_1): Likewise.
11367 (expand_block_set_4): Likewise.
11368 (expand_block_set_2): Likewise.
11369 (expand_block_set_1): Likewise.
11370 (visium_trampoline_init): Likewise.
11371 (visium_profile_hook): Likewise.
11372 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11373 (xtensa_setup_frame_addresses): Likewise.
11374 (xtensa_trampoline_init): Likewise.
11375 * except.c (sjlj_emit_function_enter): Likewise.
11376 (sjlj_emit_function_exit): Likewise.
11377 * explow.c (allocate_dynamic_stack_space): Likewise.
11378 (probe_stack_range): Likewise.
11379 * expr.c (convert_mode_scalar): Likewise.
11380 * optabs.c (expand_binop): Likewise.
11381 (expand_twoval_binop_libfunc): Likewise.
11382 (expand_unop): Likewise.
11383 (prepare_cmp_insn): Likewise.
11384 (prepare_float_lib_cmp): Likewise.
11385 (expand_float): Likewise.
11386 (expand_fix): Likewise.
11387 (expand_fixed_convert): Likewise.
11388 (maybe_emit_sync_lock_test_and_set): Likewise.
11389 (expand_atomic_compare_and_swap): Likewise.
11390 (expand_mem_thread_fence): Likewise.
11391 (expand_atomic_fetch_op): Likewise.
11393 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11395 * doc/generic.texi (OpenACC): Adjust URL.
11396 * doc/invoke.texi (C Dialect Options): Ditto.
11398 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11400 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11401 predicate for operand 1. Add (m,<S>) constraint.
11402 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11403 Prevent memory operand 1 with register operand 2.
11405 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11407 PR rtl-optimization/82024
11408 * combine.c (try_combine): If the combination result is a PARALLEL,
11409 and we only need to retain the SET in there that would be placed
11410 at I2, check that we can place that at I3 instead, before doing so.
11412 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11415 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11417 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11418 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11421 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11422 Jeff Law <law@redhat.com>
11424 * varasm.c (bss_initializer_p): Do not put constants into .bss
11425 (categorize_decl_for_section): Handle bss_initializer_p returning
11426 false when DECL_INITIAL is NULL.
11428 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11431 * config/s390/s390.c (s390_can_inline_p): New function.
11433 2017-09-01 Jeff Law <law@redhat.com>
11435 PR tree-optimization/82052
11436 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11437 Always initialize the returned slot after a hash table miss
11438 when INSERT is true.
11440 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11442 * config/s390/s390.md (mem_signal_fence): Remove.
11443 * doc/md.texi (mem_signal_fence): Remove.
11444 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11446 * target-insns.def (mem_signal_fence): Remove.
11448 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11451 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11454 * asan.c (create_odr_indicator): Strip name encoding from assembler
11455 name before appending it after __odr_asan_.
11457 2017-09-01 Martin Liska <mliska@suse.cz>
11459 PR tree-optimization/82059
11460 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11461 frequency only when an edge is redirected.
11463 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11465 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11466 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11467 (arc_conditional_register_usage): Remove ARC600 lp_count
11469 (arc_file_start): Emit Tag_ARC_CPU_variation.
11470 (arc_can_use_doloop_p): New conditions to use ZOLs.
11471 (hwloop_fail): New function.
11472 (hwloop_optimize): Likewise.
11473 (hwloop_pattern_reg): Likewise.
11474 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11475 (arc_reorg_loops): New function, calls reorg_loops.
11476 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11477 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11479 (arc_loop_hazard): Remove function, functionality moved into
11481 (arc_hazard): Remove arc_loop_hazard call.
11482 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11483 into hwloop_optimize.
11484 (arc_label_align): Remove ZOL handling.
11485 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11486 * config/arc/arc.md (doloop_begin): Remove pattern.
11487 (doloop_begin_i): Likewise.
11488 (doloop_end_i): Likewise.
11489 (doloop_fallback): Likewise.
11490 (doloop_fallback_m): Likewise.
11491 (doloop_end): Reimplement expand.
11492 (arc_lp): New pattern for LP instruction.
11493 (loop_end): New pattern.
11494 (loop_fail): Likewise.
11495 (decrement_and_branch_until_zero): Likewise.
11496 * config/arc/arc.opt (mlpc-width): New option.
11497 * doc/invoke.texi (mlpc-width): Document option.
11499 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11501 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11502 (arc_ccfsm_advance): Fix checking for delay slots.
11503 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11505 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11507 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11509 (movhi_insn): Update store instruction constraint which are saving
11510 6-bit short immediates.
11511 (movsi_insn): Consider also short scaled load operations.
11512 (zero_extendhisi2_i): Use Usd constraint instead of T.
11513 (extendhisi2_i): Add q constraint.
11514 (arc_clzsi2): Add type and length attributes.
11515 (arc_ctzsi2): Likewise.
11516 * config/arc/constraints.md (Usc): Update constraint, the
11517 assembler can parse two relocations for a single instruction.
11519 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11521 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11522 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11524 2017-08-31 Olivier Hainque <hainque@adacore.com>
11526 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11527 match as powerpc-wrs-vxworks*.
11529 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11531 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11532 register constraint for by-element operand.
11533 (aarch64_mls_elt_merge<mode>): Likewise.
11535 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11537 * config/arc/arc.c (arc_can_follow_jump): Check for short
11540 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11542 * config.gcc: Use g.opt for arc.
11543 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11544 functionality moved to ...
11545 (legitimate_scaled_address_p): New function, ...here.
11546 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11547 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11548 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11550 (arc_override_options): Handle G option.
11551 (arc_output_pic_addr_const): Correct function definition.
11552 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11553 (arc_decl_anon_ns_mem_p): Delete.
11554 (arc_in_small_data_p): Overhaul this function to take into
11555 consideration the value given via G option.
11556 (arc_rewrite_small_data_1): Renamed and corrected old
11557 arc_rewrite_small_data function.
11558 (arc_rewrite_small_data): New function.
11559 (small_data_pattern): Don't use pic_offset_table_rtx.
11560 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11561 * config/arc/simdext.md (movmisalignv2hi): Use
11562 prepare_move_operands function.
11564 (movmisalign*): Likewise.
11565 * doc/invoke.texi (ARC options): Document -G option.
11567 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11569 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11571 * config/arc/arc.c (arc_print_operand): Output scalled address for
11572 sdata whenever is possible.
11573 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11574 load/stores are available.
11575 (compact_sda_memory_operand): Check for the alignment required by
11576 code density instructions.
11577 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11579 * config/arc/constraints.md (Usd): Update constraint.
11580 (Us0): New constraint.
11581 (Usc): Update constraint.
11583 2017-08-31 Richard Biener <rguenther@suse.de>
11585 PR middle-end/82054
11586 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11587 function only once.
11589 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11591 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11592 Resize type_signature.
11594 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11595 Alan Hayward <alan.hayward@arm.com>
11596 David Sherwood <david.sherwood@arm.com>
11598 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11599 subregs whose inner modes can be stored in GPRs.
11600 (aarch64_classify_index): Likewise.
11602 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11603 Alan Hayward <alan.hayward@arm.com>
11604 David Sherwood <david.sherwood@arm.com>
11606 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11607 (V_INT_EQUIV): ...this.
11608 (v_cmp_result): Rename to...
11609 (v_int_equiv): ...this.
11610 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11611 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11612 (copysign<mode>3): Likewise.
11613 (aarch64_simd_bsl<mode>_internal): Likewise.
11614 (aarch64_simd_bsl<mode>): Likewise.
11615 (vec_cmp<mode><mode>): Likewise.
11616 (vcond<mode><mode>): Likewise.
11617 (vcond<v_cmp_mixed><mode>): Likewise.
11618 (vcondu<mode><v_cmp_mixed>): Likewise.
11619 (aarch64_cm<optab><mode>): Likewise.
11620 (aarch64_cmtst<mode>): Likewise.
11621 (aarch64_fac<optab><mode>): Likewise.
11622 (vec_perm_const<mode>): Likewise.
11623 (vcond_mask_<mode><v_cmp_result>): Rename to...
11624 (vcond_mask_<mode><v_int_equiv>): ...this.
11625 (vec_cmp<mode><v_cmp_result>): Rename to...
11626 (vec_cmp<mode><v_int_equiv>): ...this.
11628 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11629 Alan Hayward <alan.hayward@arm.com>
11630 David Sherwood <david.sherwood@arm.com>
11632 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11634 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11635 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11636 (UNSPEC_LD4_DREG): New unspecs.
11637 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11638 (aarch64_ld2<mode>_dreg_be): Replace with...
11639 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11641 (aarch64_ld3<mode>_dreg_le)
11642 (aarch64_ld3<mode>_dreg_be): Replace with...
11643 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11645 (aarch64_ld4<mode>_dreg_le)
11646 (aarch64_ld4<mode>_dreg_be): Replace with...
11647 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11650 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11652 PR tree-optimization/81987
11653 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11654 insert an initializer in a location not dominated by the stride
11657 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11659 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11660 on the entire header of the finally block in the fallthru case.
11662 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11664 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11666 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11668 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11669 rs6000_emit_move_from_cr and call renamed function.
11670 (rs6000_emit_prologue): Call renamed functions.
11671 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11672 movesi_from_cr, remove volatile CRs.
11674 2017-08-30 Jon Beniston <jon@beniston.com>
11675 Richard Biener <rguenther@suse.de>
11677 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11678 of VECTOR_MODE_P check.
11679 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11680 element vector types.
11682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11684 * df.h (df_read_modify_subreg_p): Remove in favor of...
11685 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11686 const_rtx instead of an rtx.
11687 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11688 * df-problems.c (df_word_lr_mark_ref): Likewise.
11689 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11690 (mark_pseudo_reg_dead): Likewise.
11691 (mark_ref_dead): Likewise.
11692 * reginfo.c (init_subregs_of_mode): Likewise.
11693 * sched-deps.c (sched_analyze_1): Likewise.
11694 * df-scan.c (df_def_record_1): Likewise.
11695 (df_uses_record): Likewise.
11696 (df_read_modify_subreg_p): Remove in favor of...
11697 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11698 const_rtx instead of an rtx.
11700 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11701 Alan Hayward <alan.hayward@arm.com>
11702 David Sherwood <david.sherwood@arm.com>
11704 * rtl.h (partial_subreg_p): New function.
11705 * caller-save.c (save_call_clobbered_regs): Use it.
11706 * calls.c (expand_call): Likewise.
11707 * combine.c (combinable_i3pat): Likewise.
11708 (simplify_set): Likewise.
11709 (make_extraction): Likewise.
11710 (make_compound_operation_int): Likewise.
11711 (gen_lowpart_or_truncate): Likewise.
11712 (force_to_mode): Likewise.
11713 (make_field_assignment): Likewise.
11714 (reg_truncated_to_mode): Likewise.
11715 (record_truncated_value): Likewise.
11716 (move_deaths): Likewise.
11717 * cse.c (record_jump_cond): Likewise.
11718 (cse_insn): Likewise.
11719 * cselib.c (cselib_lookup_1): Likewise.
11720 * expmed.c (extract_bit_field_using_extv): Likewise.
11721 * function.c (assign_parm_setup_reg): Likewise.
11722 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11723 * ira-build.c (create_insn_allocnos): Likewise.
11724 * lra-coalesce.c (merge_pseudos): Likewise.
11725 * lra-constraints.c (match_reload): Likewise.
11726 (simplify_operand_subreg): Likewise.
11727 (curr_insn_transform): Likewise.
11728 * lra-lives.c (process_bb_lives): Likewise.
11729 * lra.c (new_insn_reg): Likewise.
11730 (lra_substitute_pseudo): Likewise.
11731 * regcprop.c (mode_change_ok): Likewise.
11732 (maybe_mode_change): Likewise.
11733 (copyprop_hardreg_forward_1): Likewise.
11734 * reload.c (push_reload): Likewise.
11735 (find_reloads): Likewise.
11736 (find_reloads_subreg_address): Likewise.
11737 * reload1.c (alter_reg): Likewise.
11738 (eliminate_regs_1): Likewise.
11739 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11741 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11743 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11744 back to if statements, including unpack.
11746 2017-08-30 Martin Liska <mliska@suse.cz>
11748 PR inline-asm/82001
11749 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11751 (func_checker::compare_asm_inputs_outputs): ... this function.
11752 (func_checker::compare_gimple_asm): Use the function to compare
11753 also ASM constrains.
11754 * ipa-icf-gimple.h: Rename the function.
11756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11757 Alan Hayward <alan.hayward@arm.com>
11758 David Sherwood <david.sherwood@arm.com>
11760 * coretypes.h (complex_mode): New type.
11761 * gdbhooks.py (build_pretty_printer): Handle it.
11762 * machmode.h (complex_mode): New class.
11763 (complex_mode::includes_p): New function.
11764 (is_complex_int_mode): Likewise.
11765 (is_complex_float_mode): Likewise.
11766 * genmodes.c (get_mode_class): Handle complex mode classes.
11767 * function.c (expand_function_end): Use is_complex_int_mode.
11769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11770 Alan Hayward <alan.hayward@arm.com>
11771 David Sherwood <david.sherwood@arm.com>
11773 * coretypes.h (scalar_mode_pod): New typedef.
11774 * gdbhooks.py (build_pretty_printer): Handle it.
11775 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11776 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11777 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11778 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11779 as_a <scalar_mode>.
11781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11782 Alan Hayward <alan.hayward@arm.com>
11783 David Sherwood <david.sherwood@arm.com>
11785 * machmode.h (mode_for_vector): Take a scalar_mode instead
11787 * stor-layout.c (mode_for_vector): Likewise.
11788 * explow.c (promote_mode): Use as_a <scalar_mode>.
11789 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11792 Alan Hayward <alan.hayward@arm.com>
11793 David Sherwood <david.sherwood@arm.com>
11795 * target.def (preferred_simd_mode): Take a scalar_mode
11796 instead of a machine_mode.
11797 * targhooks.h (default_preferred_simd_mode): Likewise.
11798 * targhooks.c (default_preferred_simd_mode): Likewise.
11799 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11800 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11801 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11802 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11803 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11804 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11805 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11806 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11808 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11809 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11810 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11811 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11812 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11813 * doc/tm.texi: Regenerate.
11814 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11818 Alan Hayward <alan.hayward@arm.com>
11819 David Sherwood <david.sherwood@arm.com>
11821 * target.def (scalar_mode_supported_p): Take a scalar_mode
11822 instead of a machine_mode.
11823 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11824 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11825 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11826 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11827 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11828 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11829 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11830 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11831 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11832 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11833 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11834 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11835 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11836 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11838 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11839 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11840 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11841 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11842 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11844 * doc/tm.texi: Regenerate.
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 * coretypes.h (opt_scalar_mode): New typedef.
11851 * gdbhooks.py (build_pretty_printers): Handle it.
11852 * machmode.h (mode_iterator::get_2xwider): Add overload for
11854 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11856 * expr.c (convert_mode_scalar): Likewise.
11857 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11858 * optabs.c (expand_float): Likewise.
11859 (expand_fix): Likewise.
11860 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
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 * optabs.c (expand_float): Explicitly check for scalars before
11867 using a branching expansion.
11868 (expand_fix): 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 * expr.c (convert_mode): Split scalar handling out into...
11875 (convert_mode_scalar): ...this new function. Treat the modes
11878 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11879 Alan Hayward <alan.hayward@arm.com>
11880 David Sherwood <david.sherwood@arm.com>
11882 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11884 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11886 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11887 Alan Hayward <alan.hayward@arm.com>
11888 David Sherwood <david.sherwood@arm.com>
11890 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11891 rather than a machine_mode.
11892 (fixed_from_string): Likewise.
11893 (fixed_convert): Likewise.
11894 (fixed_convert_from_int): Likewise.
11895 (fixed_convert_from_real): Likewise.
11896 (real_convert_from_fixed): Likewise.
11897 * fixed-value.c (fixed_from_double_int): Likewise.
11898 (fixed_from_string): Likewise.
11899 (fixed_convert): Likewise.
11900 (fixed_convert_from_int): Likewise.
11901 (fixed_convert_from_real): Likewise.
11902 (real_convert_from_fixed): Likewise.
11903 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
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 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11910 of separate mode class checks. Do not allow vector modes here.
11911 (immed_wide_int_const): Use as_a <scalar_mode>.
11912 * explow.c (trunc_int_for_mode): Likewise.
11913 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11914 (wi::shwi): Likewise.
11915 (wi::min_value): Likewise.
11916 (wi::max_value): Likewise.
11917 * dwarf2out.c (loc_descriptor): Likewise.
11918 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11919 for CONST_WIDE_INT.
11921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11922 Alan Hayward <alan.hayward@arm.com>
11923 David Sherwood <david.sherwood@arm.com>
11925 * tree.h (SCALAR_TYPE_MODE): New macro.
11926 * expr.c (expand_expr_addr_expr_1): Use it.
11927 (expand_expr_real_2): Likewise.
11928 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11929 (fold_convert_const_fixed_from_int): Likewise.
11930 (fold_convert_const_fixed_from_real): Likewise.
11931 (native_encode_fixed): Likewise
11932 (native_encode_complex): Likewise
11933 (native_encode_vector): Likewise.
11934 (native_interpret_fixed): Likewise.
11935 (native_interpret_real): Likewise.
11936 (native_interpret_complex): Likewise.
11937 (native_interpret_vector): Likewise.
11938 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11939 (simd_clone_adjust_argument_types): Likewise.
11940 (simd_clone_init_simd_arrays): Likewise.
11941 (simd_clone_adjust): Likewise.
11942 * stor-layout.c (layout_type): Likewise.
11943 * tree.c (build_minus_one_cst): Likewise.
11944 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11945 * tree-inline.c (estimate_move_cost): Likewise.
11946 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11947 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11948 (vectorizable_reduction): Likewise.
11949 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11950 (vect_recog_mixed_size_cond_pattern): Likewise.
11951 (check_bool_pattern): Likewise.
11952 (adjust_bool_pattern): Likewise.
11953 (search_type_for_mask_1): Likewise.
11954 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11955 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11956 (vectorizable_load): Likewise.
11957 (vectorizable_store): Likewise.
11958 * ubsan.c (ubsan_encode_value): Likewise.
11959 * varasm.c (output_constant): Likewise.
11961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11962 Alan Hayward <alan.hayward@arm.com>
11963 David Sherwood <david.sherwood@arm.com>
11965 * coretypes.h (scalar_mode): New class.
11966 * machmode.h (scalar_mode): Likewise.
11967 (scalar_mode::includes_p): New function.
11968 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11969 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11970 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11971 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11972 * expmed.c (store_bit_field_1): Likewise.
11973 (extract_bit_field_1): Likewise.
11974 * expr.c (write_complex_part): Likewise.
11975 (read_complex_part): Likewise.
11976 (emit_move_complex_push): Likewise.
11977 (expand_expr_real_2): Likewise.
11978 * function.c (assign_parm_setup_reg): Likewise.
11979 (assign_parms_unsplit_complex): Likewise.
11980 * optabs.c (expand_binop): Likewise.
11981 * rtlanal.c (subreg_get_info): Likewise.
11982 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11983 * varasm.c (output_constant_pool_2): Likewise.
11985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11986 Alan Hayward <alan.hayward@arm.com>
11987 David Sherwood <david.sherwood@arm.com>
11989 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11991 (expmed_mult_highpart_optab): Likewise.
11992 (expmed_mult_highpart): Likewise.
11994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11995 Alan Hayward <alan.hayward@arm.com>
11996 David Sherwood <david.sherwood@arm.com>
11998 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11999 instead of a machine_mode.
12000 (builtin_memset_read_str): Likewise.
12001 * builtins.c (c_readstr): Likewise.
12002 (builtin_memcpy_read_str): Likewise.
12003 (builtin_strncpy_read_str): Likewise.
12004 (builtin_memset_read_str): Likewise.
12005 (builtin_memset_gen_str): Likewise.
12006 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12007 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12008 instead of a machine_mode.
12009 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12011 (make_extraction): Likewise.
12012 (try_widen_shift_mode): Take and return scalar_int_modes instead
12014 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12015 a scalar_int_mode instead of a machine_mode.
12016 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12017 (avr_addr_space_pointer_mode): Likewise.
12018 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12019 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12020 (msp430_unwind_word_mode): Likewise.
12021 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12022 (spu_addr_space_pointer_mode): Likewise.
12023 (spu_addr_space_address_mode): Likewise.
12024 (spu_libgcc_cmp_return_mode): Likewise.
12025 (spu_libgcc_shift_count_mode): Likewise.
12026 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12027 (rl78_addr_space_pointer_mode): Likewise.
12028 (fl78_unwind_word_mode): Likewise.
12029 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12031 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12032 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12033 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12034 (mips_valid_pointer_mode): Likewise.
12035 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12036 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12037 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12039 (ft32_addr_space_address_mode): Likewise.
12040 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12041 scalar_int_mode instead of a machine_mode.
12042 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12044 (m32c_addr_space_address_mode): Likewise.
12045 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12046 (rs6000_eh_return_filter_mode): Likewise.
12047 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12048 (rs6000_eh_return_filter_mode): Likewise.
12049 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12050 (s390_libgcc_shift_count_mode): Likewise.
12051 (s390_unwind_word_mode): Likewise.
12052 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12054 * target.def (mode_rep_extended): Likewise.
12055 (valid_pointer_mode): Likewise.
12056 (addr_space.valid_pointer_mode): Likewise.
12057 (eh_return_filter_mode): Return a scalar_int_mode rather than
12059 (libgcc_cmp_return_mode): Likewise.
12060 (libgcc_shift_count_mode): Likewise.
12061 (unwind_word_mode): Likewise.
12062 (addr_space.pointer_mode): Likewise.
12063 (addr_space.address_mode): Likewise.
12064 * doc/tm.texi: Regenerate.
12065 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12067 (do_jump): Use scalar_int_mode for local variables.
12068 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12069 rather than a machine_mode.
12070 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12071 (scompare_loc_descriptor_wide): Likewise.
12072 (scompare_loc_descriptor_narrow): Likewise.
12073 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12075 * except.c (sjlj_emit_dispatch_table): Likewise.
12076 (expand_builtin_eh_copy_values): Likewise.
12077 * explow.c (convert_memory_address_addr_space_1): Likewise.
12078 Take a scalar_int_mode rather than a machine_mode.
12079 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12080 than a machine_mode.
12081 (memory_address_addr_space): Use scalar_int_mode for local variables.
12082 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12083 rather than a machine_mode.
12084 * expmed.c (mask_rtx): Likewise.
12085 (init_expmed_one_conv): Likewise.
12086 (expand_mult_highpart_adjust): Likewise.
12087 (extract_high_half): Likewise.
12088 (expmed_mult_highpart_optab): Likewise.
12089 (expmed_mult_highpart): Likewise.
12090 (expand_smod_pow2): Likewise.
12091 (expand_sdiv_pow2): Likewise.
12092 (emit_store_flag_int): Likewise.
12093 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12095 (extract_low_bits): Likewise.
12096 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12098 * expr.c (pieces_addr::adjust): Likewise.
12099 (can_store_by_pieces): Likewise.
12100 (store_by_pieces): Likewise.
12101 (clear_by_pieces_1): Likewise.
12102 (expand_expr_addr_expr_1): Likewise.
12103 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12104 (expand_expr_real_1): Likewise.
12105 (try_casesi): Likewise.
12106 * final.c (shorten_branches): Likewise.
12107 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12108 type of "mode" to machine_mode.
12109 * internal-fn.c (expand_arith_overflow_result_store): Take a
12110 scalar_int_mode rather than a machine_mode.
12111 (expand_mul_overflow): Use scalar_int_mode for local variables.
12112 * loop-doloop.c (doloop_modify): Likewise.
12113 (doloop_optimize): Likewise.
12114 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12115 than a machine_mode.
12116 (expand_doubleword_shift_condmove): Likewise.
12117 (expand_doubleword_shift): Likewise.
12118 (expand_doubleword_clz): Likewise.
12119 (expand_doubleword_popcount): Likewise.
12120 (expand_doubleword_parity): Likewise.
12121 (expand_absneg_bit): Use scalar_int_mode for local variables.
12122 (prepare_float_lib_cmp): Likewise.
12123 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12124 rather than a machine_mode.
12125 (convert_memory_address_addr_space): Likewise.
12126 (get_mode_bounds): Likewise.
12127 (get_address_mode): Return a scalar_int_mode rather than a
12129 * rtlanal.c (get_address_mode): Likewise.
12130 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12131 than a machine_mode.
12132 * targhooks.c (default_mode_rep_extended): Likewise.
12133 (default_valid_pointer_mode): Likewise.
12134 (default_addr_space_valid_pointer_mode): Likewise.
12135 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12136 than a machine_mode.
12137 (default_libgcc_cmp_return_mode): Likewise.
12138 (default_libgcc_shift_count_mode): Likewise.
12139 (default_unwind_word_mode): Likewise.
12140 (default_addr_space_pointer_mode): Likewise.
12141 (default_addr_space_address_mode): Likewise.
12142 * targhooks.h (default_eh_return_filter_mode): Likewise.
12143 (default_libgcc_cmp_return_mode): Likewise.
12144 (default_libgcc_shift_count_mode): Likewise.
12145 (default_unwind_word_mode): Likewise.
12146 (default_addr_space_pointer_mode): Likewise.
12147 (default_addr_space_address_mode): Likewise.
12148 (default_mode_rep_extended): Take a scalar_int_mode rather than
12150 (default_valid_pointer_mode): Likewise.
12151 (default_addr_space_valid_pointer_mode): Likewise.
12152 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12154 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12155 rather than a machine_mode.
12156 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12157 for local variables.
12158 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12159 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12160 than a machine_mode.
12162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12163 Alan Hayward <alan.hayward@arm.com>
12164 David Sherwood <david.sherwood@arm.com>
12166 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12167 the mode argument to scalar_int_mode.
12168 (do_jump_by_parts_zero_rtx): Likewise.
12169 (do_jump_by_parts_equality_rtx): Likewise.
12170 (do_jump_by_parts_greater): Take a mode argument.
12171 (do_jump_by_parts_equality): Likewise.
12172 (do_jump_1): Update calls accordingly.
12174 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12175 Alan Hayward <alan.hayward@arm.com>
12176 David Sherwood <david.sherwood@arm.com>
12178 * is-a.h (safe_dyn_cast): New function.
12179 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12180 (jump_table_for_label): Likewise.
12181 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12182 instead of an rtx_insn *.
12183 (shorten_branches): Use dyn_cast instead of LABEL_P and
12184 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12185 rtx_jump_table_data::get_data_mode.
12186 (final_scan_insn): Likewise.
12188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12189 Alan Hayward <alan.hayward@arm.com>
12190 David Sherwood <david.sherwood@arm.com>
12192 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12193 trying to combine a full-register integer set with a subreg
12196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12197 Alan Hayward <alan.hayward@arm.com>
12198 David Sherwood <david.sherwood@arm.com>
12200 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12201 that is always either address_mode or pointer_mode.
12203 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12204 Alan Hayward <alan.hayward@arm.com>
12205 David Sherwood <david.sherwood@arm.com>
12207 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12208 when the two are known to be equal.
12210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12211 Alan Hayward <alan.hayward@arm.com>
12212 David Sherwood <david.sherwood@arm.com>
12214 * simplify-rtx.c (simplify_const_unary_operation): Use
12215 is_a <scalar_int_mode> instead of checking for a nonzero
12216 precision. Forcibly convert op_mode to a scalar_int_mode
12217 in that case. More clearly differentiate the operand and
12218 result modes and use the former when deciding what the value
12219 of a count-bits operation should be. Use is_int_mode instead
12220 of checking for a MODE_INT. Remove redundant check for whether
12221 this mode has a zero precision.
12223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12224 Alan Hayward <alan.hayward@arm.com>
12225 David Sherwood <david.sherwood@arm.com>
12227 * optabs.c (widen_leading): Change the type of the mode argument
12228 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12229 (widen_bswap): Likewise.
12230 (expand_parity): Likewise.
12231 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12232 (expand_ffs): Likewise.
12233 (epand_unop): Check for scalar integer modes before calling the
12236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12237 Alan Hayward <alan.hayward@arm.com>
12238 David Sherwood <david.sherwood@arm.com>
12240 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12242 (expand_expr_real_1): Update call accordingly.
12244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12245 Alan Hayward <alan.hayward@arm.com>
12246 David Sherwood <david.sherwood@arm.com>
12248 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12249 value_mode arguments. Use scalar_int_mode internally.
12250 (store_bit_field_1): Rename the new integer mode from imode
12251 to op0_mode and use it instead of GET_MODE (op0). Update calls
12252 to store_split_bit_field, store_bit_field_using_insv and
12253 store_fixed_bit_field.
12254 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12255 Use scalar_int_mode internally. Use a bit count rather than a mode
12256 when calculating the largest bit size for get_best_mode.
12257 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12258 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12259 Remove assertion that OP0 has a scalar integer mode.
12260 (store_split_bit_field): Add op0_mode and value_mode arguments.
12261 Update calls to extract_fixed_bit_field.
12262 (extract_bit_field_using_extv): Add an op0_mode argument.
12263 Use scalar_int_mode internally.
12264 (extract_bit_field_1): Rename the new integer mode from imode to
12265 op0_mode and use it instead of GET_MODE (op0). Update calls to
12266 extract_split_bit_field, extract_bit_field_using_extv and
12267 extract_fixed_bit_field.
12268 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12269 to extract_split_bit_field and extract_fixed_bit_field_1.
12270 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12271 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12272 on the target mode.
12273 (extract_split_bit_field): Add an op0_mode argument. Update call
12274 to extract_fixed_bit_field.
12276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12277 Alan Hayward <alan.hayward@arm.com>
12278 David Sherwood <david.sherwood@arm.com>
12280 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12281 * explow.c (hard_function_value): Likewise.
12282 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12283 convert_to_mode call outside the loop.
12284 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12285 for the mode iterator. Require the mode specified by max_pieces
12287 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12289 (copy_blkmode_to_reg): Likewise.
12290 (set_storage_via_setmem): Likewise.
12291 * optabs.c (prepare_cmp_insn): Likewise.
12292 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12293 * stor-layout.c (finish_bitfield_representative): Likewise.
12295 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12296 Alan Hayward <alan.hayward@arm.com>
12297 David Sherwood <david.sherwood@arm.com>
12299 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12300 * expr.c (convert_move): Use them.
12301 (convert_modes): Likewise.
12302 (store_expr_with_bounds): Likewise.
12304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12305 Alan Hayward <alan.hayward@arm.com>
12306 David Sherwood <david.sherwood@arm.com>
12308 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12309 parameter for the mode of "x". Remove the "known_x", "known_mode"
12310 and "known_ret" arguments. Change the type of the mode argument
12311 to scalar_int_mode.
12312 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12313 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12314 (reg_num_sign_bit_copies_for_combine): Likewise.
12315 * rtlanal.c (nonzero_bits1): Likewise.
12316 (num_sign_bit_copies1): Likewise.
12317 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12318 (reg_num_sign_bit_copies_general): Likewise.
12319 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12320 (reg_nonzero_bits_general): Likewise.
12322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12323 Alan Hayward <alan.hayward@arm.com>
12324 David Sherwood <david.sherwood@arm.com>
12326 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12327 than in subroutines. Return 1 for non-integer modes.
12328 (cached_num_sign_bit_copies): Change the type of the mode parameter
12329 to scalar_int_mode.
12330 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12331 classes. Handle CONST_INT_P first and then check whether X also
12332 has a scalar integer mode. Check the same thing for inner registers
12333 of a SUBREG and for values that are being extended or truncated.
12335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12336 Alan Hayward <alan.hayward@arm.com>
12337 David Sherwood <david.sherwood@arm.com>
12339 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12340 in subroutines. Return the mode mask for non-integer modes.
12341 (cached_nonzero_bits): Change the type of the mode parameter
12342 to scalar_int_mode.
12343 (nonzero_bits1): Likewise. Remove early exit for other mode
12344 classes. Handle CONST_INT_P first and then check whether X
12345 also has a scalar integer mode.
12347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12348 Alan Hayward <alan.hayward@arm.com>
12349 David Sherwood <david.sherwood@arm.com>
12351 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12352 Return a scalar_int_mode and assert that the size is greater than
12354 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12355 (op_by_pieces_d::op_by_pieces_d): Likewise.
12356 (op_by_pieces_d::run): Likewise.
12357 (can_store_by_pieces): Likewise.
12359 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12360 Alan Hayward <alan.hayward@arm.com>
12361 David Sherwood <david.sherwood@arm.com>
12363 * combine.c (extract_left_shift): Add a mode argument and update
12365 (make_compound_operation_int): Change the type of the mode parameter
12366 to scalar_int_mode and update the call to extract_left_shift.
12368 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12369 Alan Hayward <alan.hayward@arm.com>
12370 David Sherwood <david.sherwood@arm.com>
12372 * combine.c (simplify_and_const_int): Change the type of the mode
12373 parameter to scalar_int_mode.
12374 (simplify_and_const_int_1): Likewise. Update recursive call.
12376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12377 Alan Hayward <alan.hayward@arm.com>
12378 David Sherwood <david.sherwood@arm.com>
12380 * combine.c (simplify_compare_const): Check that the mode is a
12381 scalar_int_mode (rather than VOIDmode) before testing its
12383 (simplify_comparison): Move COMPARISON_P handling out of the
12384 loop and restrict the latter part of the loop to scalar_int_modes.
12385 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12386 and when considering SUBREG_REGs. Use is_int_mode instead of
12387 checking GET_MODE_CLASS against MODE_INT.
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 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12394 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12395 shift_unit_mode and for modes involved in scalar shifts.
12397 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12398 Alan Hayward <alan.hayward@arm.com>
12399 David Sherwood <david.sherwood@arm.com>
12401 * combine.c (force_int_to_mode): New function, split out from...
12402 (force_to_mode): ...here. Keep xmode up-to-date and use it
12403 instead of GET_MODE (x).
12405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12406 Alan Hayward <alan.hayward@arm.com>
12407 David Sherwood <david.sherwood@arm.com>
12409 * optabs-query.h (extraction_insn::struct_mode): Change type to
12410 opt_scalar_int_mode and update comment.
12411 (extraction_insn::field_mode): Change type to scalar_int_mode.
12412 (extraction_insn::pos_mode): Likewise.
12413 * combine.c (make_extraction): Update accordingly.
12414 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12415 (get_optab_extraction_insn): Likewise.
12416 * recog.c (simplify_while_replacing): Likewise.
12417 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12418 parameter to opt_scalar_int_mode.
12420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12421 Alan Hayward <alan.hayward@arm.com>
12422 David Sherwood <david.sherwood@arm.com>
12424 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12425 to a scalar_int_mode instead of a machine_mode.
12426 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12427 (get_best_mode): Return a boolean and use a pointer argument to store
12428 the selected mode. Replace the limit mode parameter with a bit limit.
12429 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12430 for the values returned by bit_field_mode_iterator::next_mode.
12431 (store_bit_field): Update call to get_best_mode.
12432 (store_fixed_bit_field): Likewise.
12433 (extract_fixed_bit_field): Likewise.
12434 * expr.c (optimize_bitfield_assignment_op): Likewise.
12435 * fold-const.c (optimize_bit_field_compare): Likewise.
12436 (fold_truth_andor_1): Likewise.
12437 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12438 Update for new type of m_mode.
12439 (get_best_mode): As above.
12441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12442 Alan Hayward <alan.hayward@arm.com>
12443 David Sherwood <david.sherwood@arm.com>
12445 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12446 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12447 (store_bit_field): Check is_a <scalar_int_mode> before calling
12448 strict_volatile_bitfield_p.
12449 (extract_bit_field): Likewise.
12451 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12452 Alan Hayward <alan.hayward@arm.com>
12453 David Sherwood <david.sherwood@arm.com>
12455 * target.def (cstore_mode): Return a scalar_int_mode.
12456 * doc/tm.texi: Regenerate.
12457 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12458 * targhooks.h (default_cstore_mode): Likewise.
12459 * targhooks.c (default_cstore_mode): Likewise, using a forced
12461 * expmed.c (emit_cstore): Expect the target of the cstore to be
12464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12465 Alan Hayward <alan.hayward@arm.com>
12466 David Sherwood <david.sherwood@arm.com>
12468 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12470 (niter_desc): Likewise mode.
12471 (iv_analyze): Add a mode parameter.
12473 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12474 and change its type to scalar_int_mode.
12475 * loop-iv.c: Update commentary at head of file.
12476 (iv_constant): Pass the mode paraeter before the rtx it describes
12477 and change its type to scalar_int_mode. Remove VOIDmode handling.
12478 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12479 (iv_extend): Likewise.
12480 (shorten_into_mode): Likewise.
12481 (iv_add): Use scalar_int_mode.
12482 (iv_mult): Likewise.
12483 (iv_shift): Likewise.
12484 (canonicalize_iv_subregs): Likewise.
12485 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12486 it describes and change its mode to scalar_int_mode. Also change
12487 the type of the returned inner_mode to scalar_int_mode.
12488 (get_biv_step): Likewise, turning outer_mode from a pointer
12489 into a direct parameter. Update call to get_biv_step_1.
12490 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12491 iv_constant and get_biv_step.
12492 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12493 and change its type to scalar_int_mode. Don't initialise iv->mode
12494 to VOIDmode and remove later checks for its still being VOIDmode.
12495 Update calls to iv_analyze_op and iv_analyze_expr. Check
12496 is_a <scalar_int_mode> when changing the mode under consideration.
12497 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12498 Update call to iv_analyze_expr.
12499 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12500 inner register is not also a scalar_int_mode. Update call to
12502 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12503 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12504 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12505 separate mode class checks. Update calls to iv_analyze. Remove
12506 fix-up of VOIDmodes after iv_analyze_biv.
12507 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12508 don't have a scalar_int_mode. Update call to biv_p.
12510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12511 Alan Hayward <alan.hayward@arm.com>
12512 David Sherwood <david.sherwood@arm.com>
12514 * cfgexpand.c (convert_debug_memory_address): Use
12515 as_a <scalar_int_mode>.
12516 * combine.c (expand_compound_operation): Likewise.
12517 (make_extraction): Likewise.
12518 (change_zero_ext): Likewise.
12519 (simplify_comparison): Likewise.
12520 * cse.c (cse_insn): Likewise.
12521 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12522 (mem_loc_descriptor): Likewise.
12523 (loc_descriptor): Likewise.
12524 * expmed.c (init_expmed_one_mode): Likewise.
12525 (synth_mult): Likewise.
12526 (emit_store_flag_1): Likewise.
12527 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12528 of a comparison with size.
12529 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12530 (reduce_to_bit_field_precision): Likewise.
12531 * function.c (expand_function_end): Likewise.
12532 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12533 * loop-doloop.c (doloop_modify): Likewise.
12534 * optabs.c (expand_binop): Likewise.
12535 (expand_unop): Likewise.
12536 (expand_copysign_absneg): Likewise.
12537 (prepare_cmp_insn): Likewise.
12538 (maybe_legitimize_operand): Likewise.
12539 * recog.c (const_scalar_int_operand): Likewise.
12540 * rtlanal.c (get_address_mode): Likewise.
12541 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12542 (simplify_cond_clz_ctz): Likewise.
12543 * tree-nested.c (get_nl_goto_field): Likewise.
12544 * tree.c (build_vector_type_for_mode): Likewise.
12545 * var-tracking.c (use_narrower_mode): Likewise.
12547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12548 Alan Hayward <alan.hayward@arm.com>
12549 David Sherwood <david.sherwood@arm.com>
12551 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12552 * builtins.c (expand_builtin_signbit): Use it.
12553 * cfgexpand.c (expand_debug_expr): Likewise.
12554 * dojump.c (do_jump): Likewise.
12555 (do_compare_and_jump): Likewise.
12556 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12557 * expmed.c (make_tree): Likewise.
12558 * expr.c (expand_expr_real_2): Likewise.
12559 (expand_expr_real_1): Likewise.
12560 (try_casesi): Likewise.
12561 * fold-const-call.c (fold_const_call_ss): Likewise.
12562 * fold-const.c (unextend): Likewise.
12563 (extract_muldiv_1): Likewise.
12564 (fold_single_bit_test): Likewise.
12565 (native_encode_int): Likewise.
12566 (native_encode_string): Likewise.
12567 (native_interpret_int): Likewise.
12568 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12569 * internal-fn.c (expand_addsub_overflow): Likewise.
12570 (expand_neg_overflow): Likewise.
12571 (expand_mul_overflow): Likewise.
12572 (expand_arith_overflow): Likewise.
12573 * match.pd: Likewise.
12574 * stor-layout.c (layout_type): Likewise.
12575 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12576 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12577 * tree-ssanames.c (get_range_info): Likewise.
12578 * tree-switch-conversion.c (array_value_type) Likewise.
12579 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12580 (vect_recog_divmod_pattern): Likewise.
12581 (vect_recog_mixed_size_cond_pattern): Likewise.
12582 * tree-vrp.c (extract_range_basic): Likewise.
12583 (simplify_float_conversion_using_ranges): Likewise.
12584 * tree.c (int_fits_type_p): Likewise.
12585 * ubsan.c (instrument_bool_enum_load): Likewise.
12586 * varasm.c (mergeable_string_section): Likewise.
12587 (narrowing_initializer_constant_valid_p): Likewise.
12588 (output_constant): Likewise.
12590 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12591 Alan Hayward <alan.hayward@arm.com>
12592 David Sherwood <david.sherwood@arm.com>
12594 * machmode.h (NARROWEST_INT_MODE): New macro.
12595 * expr.c (alignment_for_piecewise_move): Use it instead of
12596 GET_CLASS_NARROWEST_MODE (MODE_INT).
12597 (push_block): Likewise.
12598 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12600 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12603 Alan Hayward <alan.hayward@arm.com>
12604 David Sherwood <david.sherwood@arm.com>
12606 * postreload.c (move2add_valid_value_p): Change the type of the
12607 mode parameter to scalar_int_mode.
12608 (move2add_use_add2_insn): Add a mode parameter and use it instead
12610 (move2add_use_add3_insn): Likewise.
12611 (reload_cse_move2add): Update accordingly.
12613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12614 Alan Hayward <alan.hayward@arm.com>
12615 David Sherwood <david.sherwood@arm.com>
12617 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12619 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12620 * optabs.c (expand_unop): Likewise.
12622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12623 Alan Hayward <alan.hayward@arm.com>
12624 David Sherwood <david.sherwood@arm.com>
12626 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12627 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12628 (popcount_loc_descriptor): Likewise.
12629 (bswap_loc_descriptor): Likewise.
12630 (rotate_loc_descriptor): Likewise.
12631 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12632 calling the functions above.
12634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12635 Alan Hayward <alan.hayward@arm.com>
12636 David Sherwood <david.sherwood@arm.com>
12638 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12640 (try_combine): Likewise.
12641 (simplify_if_then_else): Likewise.
12642 * cse.c (cse_insn): Likewise.
12643 * dwarf2out.c (mem_loc_descriptor): Likewise.
12644 * emit-rtl.c (gen_lowpart_common): Likewise.
12645 * simplify-rtx.c (simplify_truncation): Likewise.
12646 (simplify_binary_operation_1): Likewise.
12647 (simplify_const_relational_operation): Likewise.
12648 (simplify_ternary_operation): Likewise.
12649 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12652 Alan Hayward <alan.hayward@arm.com>
12653 David Sherwood <david.sherwood@arm.com>
12655 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12656 * reload.c (push_reload): Likewise.
12657 (find_reloads): Likewise.
12659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12660 Alan Hayward <alan.hayward@arm.com>
12661 David Sherwood <david.sherwood@arm.com>
12663 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12664 (make_compound_operation_int): Likewise.
12665 (change_zero_ext): Likewise.
12666 * expr.c (convert_move): Likewise.
12667 (convert_modes): Likewise.
12668 * fwprop.c (forward_propagate_subreg): Likewise.
12669 * loop-iv.c (get_biv_step_1): Likewise.
12670 * optabs.c (widen_operand): Likewise.
12671 * postreload.c (move2add_valid_value_p): Likewise.
12672 * recog.c (simplify_while_replacing): Likewise.
12673 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12674 (simplify_binary_operation_1): Likewise. Remove redundant
12675 mode equality check.
12677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12678 Alan Hayward <alan.hayward@arm.com>
12679 David Sherwood <david.sherwood@arm.com>
12681 * combine.c (combine_simplify_rtx): Add checks for
12682 is_a <scalar_int_mode>.
12683 (simplify_if_then_else): Likewise.
12684 (make_field_assignment): Likewise.
12685 (simplify_comparison): Likewise.
12686 * ifcvt.c (noce_try_bitop): Likewise.
12687 * loop-invariant.c (canonicalize_address_mult): Likewise.
12688 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12691 Alan Hayward <alan.hayward@arm.com>
12692 David Sherwood <david.sherwood@arm.com>
12694 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12695 is_a <scalar_int_mode> instead of != BLKmode.
12697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12698 Alan Hayward <alan.hayward@arm.com>
12699 David Sherwood <david.sherwood@arm.com>
12701 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12702 instead of != VOIDmode.
12703 * combine.c (if_then_else_cond): Likewise.
12704 (change_zero_ext): Likewise.
12705 * dwarf2out.c (mem_loc_descriptor): Likewise.
12706 (loc_descriptor): Likewise.
12707 * rtlanal.c (canonicalize_condition): Likewise.
12708 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12711 Alan Hayward <alan.hayward@arm.com>
12712 David Sherwood <david.sherwood@arm.com>
12714 * simplify-rtx.c (simplify_binary_operation_1): Use
12715 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12718 Alan Hayward <alan.hayward@arm.com>
12719 David Sherwood <david.sherwood@arm.com>
12721 * wide-int.h (int_traits<unsigned char>) New class.
12722 (int_traits<unsigned short>) Likewise.
12723 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12724 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12726 * combine.c (set_nonzero_bits_and_sign_copies): Use
12727 is_a <scalar_int_mode>.
12728 (find_split_point): Likewise.
12729 (combine_simplify_rtx): Likewise.
12730 (simplify_logical): Likewise.
12731 (expand_compound_operation): Likewise.
12732 (expand_field_assignment): Likewise.
12733 (make_compound_operation): Likewise.
12734 (extended_count): Likewise.
12735 (change_zero_ext): Likewise.
12736 (simplify_comparison): Likewise.
12737 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12738 (ucompare_loc_descriptor): Likewise.
12739 (minmax_loc_descriptor): Likewise.
12740 (mem_loc_descriptor): Likewise.
12741 (loc_descriptor): Likewise.
12742 * expmed.c (init_expmed_one_mode): Likewise.
12743 * lra-constraints.c (lra_constraint_offset): Likewise.
12744 * optabs.c (prepare_libcall_arg): Likewise.
12745 * postreload.c (move2add_note_store): Likewise.
12746 * reload.c (operands_match_p): Likewise.
12747 * rtl.h (load_extend_op): Likewise.
12748 * rtlhooks.c (gen_lowpart_general): Likewise.
12749 * simplify-rtx.c (simplify_truncation): Likewise.
12750 (simplify_unary_operation_1): Likewise.
12751 (simplify_binary_operation_1): Likewise.
12752 (simplify_const_binary_operation): Likewise.
12753 (simplify_const_relational_operation): Likewise.
12754 (simplify_subreg): Likewise.
12755 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12756 * var-tracking.c (adjust_mems): Likewise.
12757 (prepare_call_arguments): Likewise.
12759 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12760 Alan Hayward <alan.hayward@arm.com>
12761 David Sherwood <david.sherwood@arm.com>
12763 * machmode.h (is_int_mode): New fuction.
12764 * combine.c (find_split_point): Use it.
12765 (combine_simplify_rtx): Likewise.
12766 (simplify_if_then_else): Likewise.
12767 (simplify_set): Likewise.
12768 (simplify_shift_const_1): Likewise.
12769 (simplify_comparison): Likewise.
12770 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12771 * cse.c (notreg_cost): Likewise.
12772 (cse_insn): Likewise.
12773 * cselib.c (cselib_lookup_1): Likewise.
12774 * dojump.c (do_jump_1): Likewise.
12775 (do_compare_rtx_and_jump): Likewise.
12776 * dse.c (get_call_args): Likewise.
12777 * dwarf2out.c (rtl_for_decl_init): Likewise.
12778 (native_encode_initializer): Likewise.
12779 * expmed.c (emit_store_flag_1): Likewise.
12780 (emit_store_flag): Likewise.
12781 * expr.c (convert_modes): Likewise.
12782 (store_field): Likewise.
12783 (expand_expr_real_1): Likewise.
12784 * fold-const.c (fold_read_from_constant_string): Likewise.
12785 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12786 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12787 * optabs.c (expand_binop): Likewise.
12788 (expand_unop): Likewise.
12789 (expand_abs_nojump): Likewise.
12790 (expand_one_cmpl_abs_nojump): Likewise.
12791 * simplify-rtx.c (mode_signbit_p): Likewise.
12792 (val_signbit_p): Likewise.
12793 (val_signbit_known_set_p): Likewise.
12794 (val_signbit_known_clear_p): Likewise.
12795 (simplify_relational_operation_1): Likewise.
12796 * tree.c (vector_type_mode): Likewise.
12798 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12799 Alan Hayward <alan.hayward@arm.com>
12800 David Sherwood <david.sherwood@arm.com>
12802 * machmode.h (smallest_mode_for_size): Fix formatting.
12803 (smallest_int_mode_for_size): New function.
12804 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12805 instead of smallest_mode_for_size.
12806 * combine.c (make_extraction): Likewise.
12807 * config/arc/arc.c (arc_expand_movmem): Likewise.
12808 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12809 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12810 * config/s390/s390.c (s390_expand_insv): Likewise.
12811 * config/sparc/sparc.c (assign_int_registers): Likewise.
12812 * config/spu/spu.c (spu_function_value): Likewise.
12813 (spu_function_arg): Likewise.
12814 * coverage.c (get_gcov_type): Likewise.
12815 (get_gcov_unsigned_t): Likewise.
12816 * dse.c (find_shift_sequence): Likewise.
12817 * expmed.c (store_bit_field_1): Likewise.
12818 * expr.c (convert_move): Likewise.
12819 (store_field): Likewise.
12820 * internal-fn.c (expand_arith_overflow): Likewise.
12821 * optabs-query.c (get_best_extraction_insn): Likewise.
12822 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12823 * stor-layout.c (layout_type): Likewise.
12824 (initialize_sizetypes): Likewise.
12825 * targhooks.c (default_get_mask_mode): Likewise.
12826 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12829 Alan Hayward <alan.hayward@arm.com>
12830 David Sherwood <david.sherwood@arm.com>
12832 * machmode.h (opt_mode::else_blk): New function.
12833 (int_mode_for_mode): Declare.
12834 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12835 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12837 * cfgexpand.c (expand_debug_expr): Likewise.
12838 * combine.c (gen_lowpart_or_truncate): Likewise.
12839 (gen_lowpart_for_combine): Likewise.
12840 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12841 * config/avr/avr.c (avr_to_int_mode): Likewise.
12842 (avr_out_plus_1): Likewise.
12843 (avr_out_plus): Likewise.
12844 (avr_out_round): Likewise.
12845 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12846 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12847 (s390_expand_vcond): Likewise.
12848 * config/spu/spu.c (spu_split_immediate): Likewise.
12849 (spu_expand_mov): Likewise.
12850 * dse.c (get_stored_val): Likewise.
12851 * expmed.c (store_bit_field_1): Likewise.
12852 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12854 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12855 (extract_low_bits): Likewise.
12856 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12857 handling rather than repeating the check.
12858 (emit_group_store): Likewise.
12859 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12860 * optabs.c (expand_absneg_bit): Likewise.
12861 (expand_copysign_absneg): Likewise.
12862 (expand_copysign_bit): Likewise.
12863 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12864 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12865 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12866 * var-tracking.c (prepare_call_arguments): Likewise.
12867 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12868 int_mode_for_mode instead of mode_for_size.
12869 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12872 Alan Hayward <alan.hayward@arm.com>
12873 David Sherwood <david.sherwood@arm.com>
12875 * machmode.h (int_mode_for_size): New function.
12876 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12877 instead of mode_for_size.
12878 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12880 * combine.c (expand_field_assignment): Use int_mode_for_size
12881 instead of mode_for_size.
12882 (make_extraction): Likewise.
12883 (simplify_shift_const_1): Likewise.
12884 (simplify_comparison): Likewise.
12885 * dojump.c (do_jump): Likewise.
12886 * dwarf2out.c (mem_loc_descriptor): Likewise.
12887 * emit-rtl.c (init_derived_machine_modes): Likewise.
12888 * expmed.c (flip_storage_order): Likewise.
12889 (convert_extracted_bit_field): Likewise.
12890 * expr.c (copy_blkmode_from_reg): Likewise.
12891 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12892 * internal-fn.c (expand_mul_overflow): Likewise.
12893 * lower-subreg.c (simple_move): Likewise.
12894 * optabs-libfuncs.c (init_optabs): Likewise.
12895 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12896 * tree.c (vector_type_mode): Likewise.
12897 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12898 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12899 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12900 * tree-vect-stmts.c (vectorizable_load): Likewise.
12901 (vectorizable_store): Likewise.
12903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12904 Alan Hayward <alan.hayward@arm.com>
12905 David Sherwood <david.sherwood@arm.com>
12907 * coretypes.h (pod_mode): New type.
12908 (scalar_int_mode_pod): New typedef.
12909 * machmode.h (pod_mode): New class.
12910 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12911 * genmodes.c (emit_mode_int_n): Update accordingly.
12912 * lower-subreg.h (target_lower_subreg): Change type to
12913 scalar_int_mode_pod.
12914 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12915 scalar_int_mode_pod.
12917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12918 Alan Hayward <alan.hayward@arm.com>
12919 David Sherwood <david.sherwood@arm.com>
12921 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12922 machine_mode to scalar_int_mode.
12923 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12924 (rs6000_option_override_internal): Remove cast to int.
12925 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12926 machine_mode to scalar_int_mode.
12927 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12928 (rs6000_option_override_internal): Remove cast to int.
12929 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12930 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12932 * config/s390/s390.c (s390_expand_builtin): Likewise.
12933 * coretypes.h (scalar_int_mode): New type.
12934 (opt_scalar_int_mode): New typedef.
12935 * machmode.h (scalar_int_mode): New class.
12936 (scalar_int_mode::includes_p): New function.
12937 (byte_mode): Change type to scalar_int_mode.
12938 (word_mode): Likewise.
12939 (ptr_mode): Likewise.
12940 * emit-rtl.c (byte_mode): Likewise.
12941 (word_mode): Likewise.
12942 (ptr_mode): Likewise.
12943 (init_derived_machine_modes): Update accordingly.
12944 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12945 and MODE_PARTIAL_INT.
12946 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12947 opt_scalar_int_mode.
12949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12950 Alan Hayward <alan.hayward@arm.com>
12951 David Sherwood <david.sherwood@arm.com>
12953 * target.def (libgcc_floating_mode_supported_p): Take a
12955 * doc/tm.texi: Regenerate.
12956 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12958 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12959 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12963 Alan Hayward <alan.hayward@arm.com>
12964 David Sherwood <david.sherwood@arm.com>
12966 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12967 * doc/tm.texi: Regenerate.
12968 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12969 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12970 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12971 * targhooks.h (default_floatn_mode): Likewise.
12972 * targhooks.c (default_floatn_mode): Likewise.
12973 * tree.c (build_common_tree_nodes): Update accordingly.
12975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12976 Alan Hayward <alan.hayward@arm.com>
12977 David Sherwood <david.sherwood@arm.com>
12979 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12980 (mode_iterator::iterate_p): Likewise.
12981 (mode_iterator::get_wider): Likewise.
12982 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12985 Alan Hayward <alan.hayward@arm.com>
12986 David Sherwood <david.sherwood@arm.com>
12988 * coretypes.h (opt_scalar_float_mode): New typedef.
12989 * machmode.h (float_mode_for_size): New function.
12990 * emit-rtl.c (double_mode): Delete.
12991 (init_emit_once): Use float_mode_for_size.
12992 * stor-layout.c (layout_type): Likewise.
12993 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12996 Alan Hayward <alan.hayward@arm.com>
12997 David Sherwood <david.sherwood@arm.com>
12999 * output.h (assemble_real): Take a scalar_float_mode.
13000 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13001 * config/arm/arm.md (consttable_4): Likewise.
13002 (consttable_8): Likewise.
13003 (consttable_16): Likewise.
13004 * config/mips/mips.md (consttable_float): Likewise.
13005 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13006 * varasm.c (assemble_real): Take a scalar_float_mode.
13007 (output_constant_pool_2): Update accordingly.
13008 (output_constant): Likewise.
13010 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13011 Alan Hayward <alan.hayward@arm.com>
13012 David Sherwood <david.sherwood@arm.com>
13014 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13015 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13016 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13017 (native_encode_real): Likewise.
13018 (native_interpret_real): Likewise.
13019 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13020 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13023 Alan Hayward <alan.hayward@arm.com>
13024 David Sherwood <david.sherwood@arm.com>
13026 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13027 <scalar_float_mode>. Simplify.
13028 (gen_extend_conv_libfunc): Likewise.
13030 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13031 Alan Hayward <alan.hayward@arm.com>
13032 David Sherwood <david.sherwood@arm.com>
13034 * coretypes.h (scalar_float_mode): New type.
13035 * machmode.h (mode_traits::from_int): Use machine_mode if
13036 USE_ENUM_MODES is defined.
13037 (is_a): New function.
13039 (dyn_cast): Likewise.
13040 (scalar_float_mode): New class.
13041 (scalar_float_mode::includes_p): New function.
13042 (is_float_mode): Likewise.
13043 * gdbhooks.py (MachineModePrinter): New class.
13044 (build_pretty_printer): Use it for scalar_float_mode.
13045 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13046 (format_helper::format_helper): Turn into a template.
13047 * genmodes.c (get_mode_class): New function.
13048 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13049 or machine_mode if none.
13050 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13051 as_a <scalar_float_mode>.
13052 * dwarf2out.c (mem_loc_descriptor): Likewise.
13053 (insert_float): Likewise.
13054 (add_const_value_attribute): Likewise.
13055 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13056 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13057 (expand_unop): Update accordingly.
13058 (expand_abs_nojump): Likewise.
13059 (expand_copysign_absneg): Take a scalar_float_mode.
13060 (expand_copysign_bit): Likewise.
13061 (expand_copysign): Update accordingly.
13063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13064 Alan Hayward <alan.hayward@arm.com>
13065 David Sherwood <david.sherwood@arm.com>
13067 * coretypes.h (opt_mode): New class.
13068 * machmode.h (opt_mode): Likewise.
13069 (opt_mode::else_void): New function.
13070 (opt_mode::require): Likewise.
13071 (opt_mode::exists): Likewise.
13072 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13073 (GET_MODE_2XWIDER_MODE): Likewise.
13074 (mode_iterator::get_wider): Update accordingly.
13075 (mode_iterator::get_2xwider): Likewise.
13076 (mode_iterator::get_known_wider): Likewise, turning into a template.
13077 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13078 forcing a wider mode to exist.
13079 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13080 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13081 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13082 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13083 * lower-subreg.c (init_lower_subreg): Likewise.
13084 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13085 on the final iteration.
13086 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13087 a wider mode exists before asking for a move pattern.
13088 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13089 forcing a wider mode to exist.
13090 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13091 returning false if no such mode exists.
13092 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13093 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13094 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13095 Avoid checking for a MODE_INT if we already know the mode is not a
13097 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13098 forcing a wider mode to exist.
13099 (expmed_mult_highpart_optab): Likewise.
13100 (expmed_mult_highpart): Likewise.
13101 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13103 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13104 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13105 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13106 * internal-fn.c (expand_mul_overflow): Update use of
13107 GET_MODE_2XWIDER_MODE.
13108 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13109 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13110 GET_MODE_WIDER_MODE.
13111 (convert_plusminus_to_widen): Likewise.
13112 * tree-switch-conversion.c (array_value_type): Likewise.
13113 * var-tracking.c (emit_note_insn_var_location): Likewise.
13114 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13115 Return false inside rather than outside the loop if no wider mode
13117 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13118 and GET_MODE_2XWIDER_MODE
13119 (can_compare_p): Use else_void.
13120 * gdbhooks.py (OptMachineModePrinter): New class.
13121 (build_pretty_printer): Use it for opt_mode.
13123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13124 Alan Hayward <alan.hayward@arm.com>
13125 David Sherwood <david.sherwood@arm.com>
13127 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13128 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13131 Alan Hayward <alan.hayward@arm.com>
13132 David Sherwood <david.sherwood@arm.com>
13134 * machmode.h (mode_traits): New structure.
13135 (get_narrowest_mode): New function.
13136 (mode_iterator::start): Likewise.
13137 (mode_iterator::iterate_p): Likewise.
13138 (mode_iterator::get_wider): Likewise.
13139 (mode_iterator::get_known_wider): Likewise.
13140 (mode_iterator::get_2xwider): Likewise.
13141 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13142 (FOR_EACH_MODE): Likewise.
13143 (FOR_EACH_MODE_FROM): Likewise.
13144 (FOR_EACH_MODE_UNTIL): Likewise.
13145 (FOR_EACH_WIDER_MODE): Likewise.
13146 (FOR_EACH_2XWIDER_MODE): Likewise.
13147 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13148 * combine.c (simplify_comparison): Likewise
13149 * config/i386/i386.c (type_natural_mode): Likewise.
13150 * cse.c (cse_insn): Likewise.
13151 * dse.c (find_shift_sequence): Likewise.
13152 * emit-rtl.c (init_derived_machine_modes): Likewise.
13153 (init_emit_once): Likewise.
13154 * explow.c (hard_function_value): Likewise.
13155 * expmed.c (extract_fixed_bit_field_1): Likewise.
13156 (extract_bit_field_1): Likewise.
13157 (expand_divmod): Likewise.
13158 (emit_store_flag_1): Likewise.
13159 * expr.c (init_expr_target): Likewise.
13160 (convert_move): Likewise.
13161 (alignment_for_piecewise_move): Likewise.
13162 (widest_int_mode_for_size): Likewise.
13163 (emit_block_move_via_movmem): Likewise.
13164 (copy_blkmode_to_reg): Likewise.
13165 (set_storage_via_setmem): Likewise.
13166 (compress_float_constant): Likewise.
13167 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13168 * optabs-query.c (get_best_extraction_insn): Likewise.
13169 * optabs.c (expand_binop): Likewise.
13170 (expand_twoval_unop): Likewise.
13171 (expand_twoval_binop): Likewise.
13172 (widen_leading): Likewise.
13173 (widen_bswap): Likewise.
13174 (expand_parity): Likewise.
13175 (expand_unop): Likewise.
13176 (prepare_cmp_insn): Likewise.
13177 (prepare_float_lib_cmp): Likewise.
13178 (expand_float): Likewise.
13179 (expand_fix): Likewise.
13180 (expand_sfix_optab): Likewise.
13181 * postreload.c (move2add_use_add2_insn): Likewise.
13182 * reg-stack.c (reg_to_stack): Likewise.
13183 * reginfo.c (choose_hard_reg_mode): Likewise.
13184 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13185 * stor-layout.c (mode_for_size): Likewise.
13186 (smallest_mode_for_size): Likewise.
13187 (mode_for_vector): Likewise.
13188 (finish_bitfield_representative): Likewise.
13189 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13190 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13191 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13192 * var-tracking.c (prepare_call_arguments): Likewise.
13194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13195 Alan Hayward <alan.hayward@arm.com>
13196 David Sherwood <david.sherwood@arm.com>
13198 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13199 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13200 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13201 * machmode.h (mode_size): Move earlier in file.
13202 (mode_precision): Likewise.
13203 (mode_inner): Likewise.
13204 (mode_nunits): Likewise.
13205 (mode_unit_size): Likewise.
13206 (unit_unit_precision): Likewise.
13207 (mode_wider): Likewise.
13208 (mode_2xwider): Likewise.
13209 (machine_mode): New class.
13210 (mode_to_bytes): New function.
13211 (mode_to_bits): Likewise.
13212 (mode_to_precision): Likewise.
13213 (mode_to_inner): Likewise.
13214 (mode_to_unit_size): Likewise.
13215 (mode_to_unit_precision): Likewise.
13216 (mode_to_nunits): Likewise.
13217 (GET_MODE_SIZE): Use mode_to_bytes.
13218 (GET_MODE_BITSIZE): Use mode_to_bits.
13219 (GET_MODE_PRECISION): Use mode_to_precision.
13220 (GET_MODE_INNER): Use mode_to_inner.
13221 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13222 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13223 (GET_MODE_NUNITS): Use mode_to_nunits.
13224 * system.h (ALWAYS_INLINE): New macro.
13225 * config/powerpcspe/powerpcspe-c.c
13226 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13227 int for arg1_mode and arg2_mode.
13229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13230 Alan Hayward <alan.hayward@arm.com>
13231 David Sherwood <david.sherwood@arm.com>
13233 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13234 Prefix mode names with E_ in case statements.
13235 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13236 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13237 (aarch64_split_simd_move): Likewise.
13238 (aarch64_gen_storewb_pair): Likewise.
13239 (aarch64_gen_loadwb_pair): Likewise.
13240 (aarch64_gen_store_pair): Likewise.
13241 (aarch64_gen_load_pair): Likewise.
13242 (aarch64_get_condition_code_1): Likewise.
13243 (aarch64_constant_pool_reload_icode): Likewise.
13244 (get_rsqrte_type): Likewise.
13245 (get_rsqrts_type): Likewise.
13246 (get_recpe_type): Likewise.
13247 (get_recps_type): Likewise.
13248 (aarch64_gimplify_va_arg_expr): Likewise.
13249 (aarch64_simd_container_mode): Likewise.
13250 (aarch64_emit_load_exclusive): Likewise.
13251 (aarch64_emit_store_exclusive): Likewise.
13252 (aarch64_expand_compare_and_swap): Likewise.
13253 (aarch64_gen_atomic_cas): Likewise.
13254 (aarch64_emit_bic): Likewise.
13255 (aarch64_emit_atomic_swap): Likewise.
13256 (aarch64_emit_atomic_load_op): Likewise.
13257 (aarch64_evpc_trn): Likewise.
13258 (aarch64_evpc_uzp): Likewise.
13259 (aarch64_evpc_zip): Likewise.
13260 (aarch64_evpc_ext): Likewise.
13261 (aarch64_evpc_rev): Likewise.
13262 (aarch64_evpc_dup): Likewise.
13263 (aarch64_gen_ccmp_first): Likewise.
13264 (aarch64_gen_ccmp_next): Likewise.
13265 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13266 (alpha_emit_xfloating_libcall): Likewise.
13267 (emit_insxl): Likewise.
13268 (alpha_arg_type): Likewise.
13269 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13270 (arc_preferred_simd_mode): Likewise.
13271 (arc_secondary_reload): Likewise.
13272 (get_arc_condition_code): Likewise.
13273 (arc_print_operand): Likewise.
13274 (arc_legitimate_constant_p): Likewise.
13275 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13276 * config/arc/arc.md (casesi_load): Likewise.
13277 (casesi_compact_jump): Likewise.
13278 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13279 (cc_use_register): Likewise.
13280 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13281 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13282 (arm_init_iwmmxt_builtins): Likewise.
13283 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13284 (neon_expand_vector_init): Likewise.
13285 (arm_attr_length_move_neon): Likewise.
13286 (maybe_get_arm_condition_code): Likewise.
13287 (arm_emit_vector_const): Likewise.
13288 (arm_preferred_simd_mode): Likewise.
13289 (arm_output_iwmmxt_tinsr): Likewise.
13290 (thumb1_output_casesi): Likewise.
13291 (thumb2_output_casesi): Likewise.
13292 (arm_emit_load_exclusive): Likewise.
13293 (arm_emit_store_exclusive): Likewise.
13294 (arm_expand_compare_and_swap): Likewise.
13295 (arm_evpc_neon_vuzp): Likewise.
13296 (arm_evpc_neon_vzip): Likewise.
13297 (arm_evpc_neon_vrev): Likewise.
13298 (arm_evpc_neon_vtrn): Likewise.
13299 (arm_evpc_neon_vext): Likewise.
13300 (arm_validize_comparison): Likewise.
13301 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13302 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13303 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13304 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13305 (c6x_preferred_simd_mode): Likewise.
13306 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13307 (epiphany_rtx_costs): Likewise.
13308 * config/epiphany/predicates.md (proper_comparison_operator):
13310 * config/frv/frv.c (condexec_memory_operand): Likewise.
13311 (frv_emit_move): Likewise.
13312 (output_move_single): Likewise.
13313 (output_condmove_single): Likewise.
13314 (frv_hard_regno_mode_ok): Likewise.
13315 (frv_matching_accg_mode): Likewise.
13316 * config/h8300/h8300.c (split_adds_subs): Likewise.
13317 (h8300_rtx_costs): Likewise.
13318 (h8300_print_operand): Likewise.
13319 (compute_mov_length): Likewise.
13320 (output_logical_op): Likewise.
13321 (compute_logical_op_length): Likewise.
13322 (compute_logical_op_cc): Likewise.
13323 (h8300_shift_needs_scratch_p): Likewise.
13324 (output_a_shift): Likewise.
13325 (compute_a_shift_length): Likewise.
13326 (compute_a_shift_cc): Likewise.
13327 (expand_a_rotate): Likewise.
13328 (output_a_rotate): Likewise.
13329 * config/i386/i386.c (classify_argument): Likewise.
13330 (function_arg_advance_32): Likewise.
13331 (function_arg_32): Likewise.
13332 (function_arg_64): Likewise.
13333 (function_value_64): Likewise.
13334 (ix86_gimplify_va_arg): Likewise.
13335 (ix86_legitimate_constant_p): Likewise.
13336 (put_condition_code): Likewise.
13337 (split_double_mode): Likewise.
13338 (ix86_avx256_split_vector_move_misalign): Likewise.
13339 (ix86_expand_vector_logical_operator): Likewise.
13340 (ix86_split_idivmod): Likewise.
13341 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13342 (ix86_build_const_vector): Likewise.
13343 (ix86_build_signbit_mask): Likewise.
13344 (ix86_match_ccmode): Likewise.
13345 (ix86_cc_modes_compatible): Likewise.
13346 (ix86_expand_branch): Likewise.
13347 (ix86_expand_sse_cmp): Likewise.
13348 (ix86_expand_sse_movcc): Likewise.
13349 (ix86_expand_int_sse_cmp): Likewise.
13350 (ix86_expand_vec_perm_vpermi2): Likewise.
13351 (ix86_expand_vec_perm): Likewise.
13352 (ix86_expand_sse_unpack): Likewise.
13353 (ix86_expand_int_addcc): Likewise.
13354 (ix86_split_to_parts): Likewise.
13355 (ix86_vectorize_builtin_gather): Likewise.
13356 (ix86_vectorize_builtin_scatter): Likewise.
13357 (avx_vpermilp_parallel): Likewise.
13358 (inline_memory_move_cost): Likewise.
13359 (ix86_tieable_integer_mode_p): Likewise.
13360 (x86_maybe_negate_const_int): Likewise.
13361 (ix86_expand_vector_init_duplicate): Likewise.
13362 (ix86_expand_vector_init_one_nonzero): Likewise.
13363 (ix86_expand_vector_init_one_var): Likewise.
13364 (ix86_expand_vector_init_concat): Likewise.
13365 (ix86_expand_vector_init_interleave): Likewise.
13366 (ix86_expand_vector_init_general): Likewise.
13367 (ix86_expand_vector_set): Likewise.
13368 (ix86_expand_vector_extract): Likewise.
13369 (emit_reduc_half): Likewise.
13370 (ix86_emit_i387_round): Likewise.
13371 (ix86_mangle_type): Likewise.
13372 (ix86_expand_round_sse4): Likewise.
13373 (expand_vec_perm_blend): Likewise.
13374 (canonicalize_vector_int_perm): Likewise.
13375 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13376 (expand_vec_perm_1): Likewise.
13377 (expand_vec_perm_interleave3): Likewise.
13378 (expand_vec_perm_even_odd_pack): Likewise.
13379 (expand_vec_perm_even_odd_1): Likewise.
13380 (expand_vec_perm_broadcast_1): Likewise.
13381 (ix86_vectorize_vec_perm_const_ok): Likewise.
13382 (ix86_expand_vecop_qihi): Likewise.
13383 (ix86_expand_mul_widen_hilo): Likewise.
13384 (ix86_expand_sse2_abs): Likewise.
13385 (ix86_expand_pextr): Likewise.
13386 (ix86_expand_pinsr): Likewise.
13387 (ix86_preferred_simd_mode): Likewise.
13388 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13389 * config/i386/sse.md (*andnot<mode>3): Likewise.
13390 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13391 (*<code><mode>3): Likewise.
13392 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13393 (ia64_expand_atomic_op): Likewise.
13394 (ia64_arg_type): Likewise.
13395 (ia64_mode_to_int): Likewise.
13396 (ia64_scalar_mode_supported_p): Likewise.
13397 (ia64_vector_mode_supported_p): Likewise.
13398 (expand_vec_perm_broadcast): Likewise.
13399 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13400 (iq2000_function_arg_advance): Likewise.
13401 (iq2000_function_arg): Likewise.
13402 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13403 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13404 (m68k_libcall_value): Likewise.
13405 (m68k_function_value): Likewise.
13406 (sched_attr_op_type): Likewise.
13407 * config/mcore/mcore.c (mcore_output_move): Likewise.
13408 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13410 (microblaze_function_arg): Likewise.
13411 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13412 (mips_print_operand): Likewise.
13413 (mips_mode_ok_for_mov_fmt_p): Likewise.
13414 (mips_vector_mode_supported_p): Likewise.
13415 (mips_preferred_simd_mode): Likewise.
13416 (mips_expand_vpc_loongson_even_odd): Likewise.
13417 (mips_expand_vec_unpack): Likewise.
13418 (mips_expand_vi_broadcast): Likewise.
13419 (mips_expand_vector_init): Likewise.
13420 (mips_expand_vec_reduc): Likewise.
13421 (mips_expand_msa_cmp): Likewise.
13422 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13423 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13424 (cc_flags_for_mode): Likewise.
13425 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13426 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13427 (nds32_output_casesi_pc_relative): Likewise.
13428 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13429 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13430 (nvptx_gen_unpack): Likewise.
13431 (nvptx_gen_pack): Likewise.
13432 (nvptx_gen_shuffle): Likewise.
13433 (nvptx_gen_wcast): Likewise.
13434 (nvptx_preferred_simd_mode): Likewise.
13435 * config/pa/pa.c (pa_secondary_reload): Likewise.
13436 * config/pa/predicates.md (base14_operand): Likewise.
13437 * config/powerpcspe/powerpcspe-c.c
13438 (altivec_resolve_overloaded_builtin): Likewise.
13439 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13441 (rs6000_preferred_simd_mode): Likewise.
13442 (output_vec_const_move): Likewise.
13443 (rs6000_expand_vector_extract): Likewise.
13444 (rs6000_split_vec_extract_var): Likewise.
13445 (reg_offset_addressing_ok_p): Likewise.
13446 (rs6000_legitimate_offset_address_p): Likewise.
13447 (rs6000_legitimize_address): Likewise.
13448 (rs6000_emit_set_const): Likewise.
13449 (rs6000_const_vec): Likewise.
13450 (rs6000_emit_move): Likewise.
13451 (spe_build_register_parallel): Likewise.
13452 (rs6000_darwin64_record_arg_recurse): Likewise.
13453 (swap_selector_for_mode): Likewise.
13454 (spe_init_builtins): Likewise.
13455 (paired_init_builtins): Likewise.
13456 (altivec_init_builtins): Likewise.
13457 (do_load_for_compare): Likewise.
13458 (rs6000_generate_compare): Likewise.
13459 (rs6000_expand_float128_convert): Likewise.
13460 (emit_load_locked): Likewise.
13461 (emit_store_conditional): Likewise.
13462 (rs6000_output_function_epilogue): Likewise.
13463 (rs6000_handle_altivec_attribute): Likewise.
13464 (rs6000_function_value): Likewise.
13465 (emit_fusion_gpr_load): Likewise.
13466 (emit_fusion_p9_load): Likewise.
13467 (emit_fusion_p9_store): Likewise.
13468 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13469 (fusion_gpr_mem_load): Likewise.
13470 (fusion_addis_mem_combo_load): Likewise.
13471 (fusion_addis_mem_combo_store): Likewise.
13472 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13473 (fusion_gpr_mem_load): Likewise.
13474 (fusion_addis_mem_combo_load): Likewise.
13475 (fusion_addis_mem_combo_store): Likewise.
13476 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13478 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13479 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13480 (rs6000_preferred_simd_mode): Likewise.
13481 (output_vec_const_move): Likewise.
13482 (rs6000_expand_vector_extract): Likewise.
13483 (rs6000_split_vec_extract_var): Likewise.
13484 (reg_offset_addressing_ok_p): Likewise.
13485 (rs6000_legitimate_offset_address_p): Likewise.
13486 (rs6000_legitimize_address): Likewise.
13487 (rs6000_emit_set_const): Likewise.
13488 (rs6000_const_vec): Likewise.
13489 (rs6000_emit_move): Likewise.
13490 (rs6000_darwin64_record_arg_recurse): Likewise.
13491 (swap_selector_for_mode): Likewise.
13492 (paired_init_builtins): Likewise.
13493 (altivec_init_builtins): Likewise.
13494 (rs6000_expand_float128_convert): Likewise.
13495 (emit_load_locked): Likewise.
13496 (emit_store_conditional): Likewise.
13497 (rs6000_output_function_epilogue): Likewise.
13498 (rs6000_handle_altivec_attribute): Likewise.
13499 (rs6000_function_value): Likewise.
13500 (emit_fusion_gpr_load): Likewise.
13501 (emit_fusion_p9_load): Likewise.
13502 (emit_fusion_p9_store): Likewise.
13503 * config/rx/rx.c (rx_gen_move_template): Likewise.
13504 (flags_from_mode): Likewise.
13505 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13506 (s390_slb_comparison): Likewise.
13507 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13508 (s390_vector_mode_supported_p): Likewise.
13509 (s390_cc_modes_compatible): Likewise.
13510 (s390_match_ccmode_set): Likewise.
13511 (s390_canonicalize_comparison): Likewise.
13512 (s390_emit_compare_and_swap): Likewise.
13513 (s390_branch_condition_mask): Likewise.
13514 (s390_rtx_costs): Likewise.
13515 (s390_secondary_reload): Likewise.
13516 (__SECONDARY_RELOAD_CASE): Likewise.
13517 (s390_expand_cs): Likewise.
13518 (s390_preferred_simd_mode): Likewise.
13519 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13520 * config/sh/sh.c (sh_print_operand): Likewise.
13521 (dump_table): Likewise.
13522 (sh_secondary_reload): Likewise.
13523 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13524 * config/sh/sh.md (casesi_worker_1): Likewise.
13525 (casesi_worker_2): Likewise.
13526 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13527 (fcc_comparison_operator): Likewise.
13528 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13529 (emit_soft_tfmode_cvt): Likewise.
13530 (sparc_preferred_simd_mode): Likewise.
13531 (output_cbranch): Likewise.
13532 (sparc_print_operand): Likewise.
13533 (sparc_expand_vec_perm_bmask): Likewise.
13534 (vector_init_bshuffle): Likewise.
13535 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13536 (spu_vector_mode_supported_p): Likewise.
13537 (spu_expand_insv): Likewise.
13538 (spu_emit_branch_or_set): Likewise.
13539 (spu_handle_vector_attribute): Likewise.
13540 (spu_builtin_splats): Likewise.
13541 (spu_builtin_extract): Likewise.
13542 (spu_builtin_promote): Likewise.
13543 (spu_expand_sign_extend): Likewise.
13544 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13545 (tilegx_simd_int): Likewise.
13546 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13547 (tilepro_simd_int): Likewise.
13548 * config/v850/v850.c (const_double_split): Likewise.
13549 (v850_print_operand): Likewise.
13550 (ep_memory_offset): Likewise.
13551 * config/vax/vax.c (vax_rtx_costs): Likewise.
13552 (vax_output_int_move): Likewise.
13553 (vax_output_int_add): Likewise.
13554 (vax_output_int_subtract): Likewise.
13555 * config/visium/predicates.md (visium_branch_operator): Likewise.
13556 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13557 (visium_print_operand_address): Likewise.
13558 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13559 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13560 (xtensa_expand_conditional_branch): Likewise.
13561 (xtensa_copy_incoming_a7): Likewise.
13562 (xtensa_output_literal): Likewise.
13563 * dfp.c (decimal_real_maxval): Likewise.
13564 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13567 Alan Hayward <alan.hayward@arm.com>
13568 David Sherwood <david.sherwood@arm.com>
13570 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13571 (mode_nunits_inline): Likewise.
13572 (mode_inner_inline): Likewise.
13573 (mode_unit_size_inline): Likewise.
13574 (mode_unit_precision_inline): Likewise.
13575 (emit_insn_modes_h): Likewise. Also emit a #define of the
13577 (emit_mode_wider): Add an E_ prefix to mode names.
13578 (emit_mode_complex): Likewise.
13579 (emit_mode_inner): Likewise.
13580 (emit_mode_adjustments): Likewise.
13581 (emit_mode_int_n): Likewise.
13582 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13583 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13584 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13585 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13586 (CRC32_BUILTIN, ENTRY): Likewise.
13587 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13588 (aarch64_pop_regs): Likewise.
13589 (aarch64_process_components): Likewise.
13590 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13591 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13592 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13593 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13594 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13595 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13597 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13598 (pcmpistr): Likewise.
13599 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13600 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13601 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13603 * config/rl78/rl78.c (mduc_regs): Likewise.
13604 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13605 (htm_expand_builtin): Likewise.
13606 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13607 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13608 * config/xtensa/xtensa.c (print_operand): Likewise.
13609 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13610 (NUM_MODE_VECTOR_INT): Likewise.
13611 * genoutput.c (null_operand): Likewise.
13612 (output_operand_data): Likewise.
13613 * genrecog.c (print_parameter_value): Likewise.
13614 * lra.c (debug_operand_data): Likewise.
13616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13618 * dwarf2out.c (scompare_loc_descriptor_wide)
13619 (scompare_loc_descriptor_narrow): New functions, split out from...
13620 (scompare_loc_descriptor): ...here.
13621 * expmed.c (emit_store_flag_int): New function, split out from...
13622 (emit_store_flag): ...here.
13624 2017-08-30 Richard Biener <rguenther@suse.de>
13626 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13627 (dwarf2out_early_finish): Move setting of AT_pubnames from
13628 early debug output to early finish.
13630 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13632 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13633 and -mdata-region to the assembler.
13635 2017-08-30 Richard Biener <rguenther@suse.de>
13637 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13639 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13640 (dwarf2out_early_global_decl): Only generate a DIE for the
13641 abstract origin if it doesn't already exist or is a declaration DIE.
13642 (resolve_addr): Do not add the linkage name twice when
13643 generating a stub DIE for the DW_TAG_GNU_call_site target.
13645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13647 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13648 Use machine_mode rather than int for arg1_mode.
13650 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13653 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13654 that the second argument of the built-in functions to unpack
13655 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13656 switch statement instead a lot of if statements.
13657 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13658 Allow 64-bit values to be in Altivec registers as well as
13659 traditional floating point registers.
13660 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13662 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13664 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13665 MAX_REGS_PER_ADDRESS == 1.
13667 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13669 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13670 * config/i386/i386.c (ix86_option_override_internal): Simplify
13671 setting of opts->x_flag_entry.
13673 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13674 Jakub Jelinek <jakub@redhat.com>
13675 Richard Biener <rguenther@suse.de>
13677 PR tree-optimization/81503
13678 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13679 folded constant fits in the target type; reorder tests for clarity.
13681 2017-08-29 Martin Liska <mliska@suse.cz>
13683 * passes.def: Include pass_lower_switch.
13684 * stmt.c (dump_case_nodes): Remove and move to
13685 tree-switch-conversion.
13686 (case_values_threshold): Likewise.
13687 (expand_switch_as_decision_tree_p): Likewise.
13688 (emit_case_decision_tree): Likewise.
13689 (expand_case): Likewise.
13690 (balance_case_nodes): Likewise.
13691 (node_has_low_bound): Likewise.
13692 (node_has_high_bound): Likewise.
13693 (node_is_bounded): Likewise.
13694 (emit_case_nodes): Likewise.
13695 (struct simple_case_node): New struct.
13696 (add_case_node): Remove.
13697 (emit_case_dispatch_table): Use vector instead of case_list.
13698 (reset_out_edges_aux): Remove.
13699 (compute_cases_per_edge): Likewise.
13700 (expand_case): Build list of simple_case_node.
13701 (expand_sjlj_dispatch_table): Use it.
13702 * tree-switch-conversion.c (struct case_node): Moved from
13703 stmt.c and adjusted.
13704 (emit_case_nodes): Likewise.
13705 (node_has_low_bound): Likewise.
13706 (node_has_high_bound): Likewise.
13707 (node_is_bounded): Likewise.
13708 (case_values_threshold): Likewise.
13709 (reset_out_edges_aux): Likewise.
13710 (compute_cases_per_edge): Likewise.
13711 (add_case_node): Likewise.
13712 (dump_case_nodes): Likewise.
13713 (balance_case_nodes): Likewise.
13714 (expand_switch_as_decision_tree_p): Likewise.
13715 (emit_jump): Likewise.
13716 (emit_case_decision_tree): Likewise.
13717 (try_switch_expansion): Likewise.
13718 (do_jump_if_equal): Likewise.
13719 (emit_cmp_and_jump_insns): Likewise.
13720 (fix_phi_operands_for_edge): New function.
13721 (record_phi_operand_mapping): Likewise.
13722 (class pass_lower_switch): New pass.
13723 (pass_lower_switch::execute): New function.
13724 (make_pass_lower_switch): Likewise.
13725 (conditional_probability):
13726 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13727 * tree-pass.h: Add make_pass_lower_switch.
13729 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13732 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13735 2017-08-29 Richard Biener <rguenther@suse.de>
13737 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13738 we do not add a DW_AT_inline attribute twice.
13739 (gen_subprogram_die): Remove code setting DW_AT_inline on
13740 DECL_ABSTRACT_P nodes.
13742 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13744 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13745 calls to internal functions.
13746 (gimplify_modify_expr): Likewise.
13747 * tree-call-cdce.c (use_internal_fn): Likewise.
13748 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13749 (convert_to_divmod): Set the nothrow flag.
13750 * tree-if-conv.c (predicate_mem_writes): Likewise.
13751 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13752 (vectorizable_call): Likewise.
13753 (vectorizable_store): Likewise.
13754 (vectorizable_load): Likewise.
13755 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13756 (vect_recog_mask_conversion_pattern): Likewise.
13758 2017-08-29 Martin Liska <mliska@suse.cz>
13761 * gcc.c (driver_handle_option): Add new argument.
13762 * opts-common.c (handle_option): Pass
13763 target_option_override_hook.
13764 * opts-global.c (lang_handle_option): Add new option.
13765 (set_default_handlers): Add new argument.
13766 (decode_options): Likewise.
13767 * opts.c (target_handle_option): Likewise.
13768 (common_handle_option): Call target_option_override_hook.
13769 * opts.h (struct cl_option_handler_func): Add hook for
13770 target option override.
13771 (struct cl_option_handlers): Likewise.
13772 (set_default_handlers): Add new argument.
13773 (decode_options): Likewise.
13774 (common_handle_option): Likewise.
13775 (target_handle_option): Likewise.
13776 * toplev.c (toplev::main): Pass targetm.target_option.override
13779 2017-08-29 Richard Biener <rguenther@suse.de>
13780 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13782 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13783 life to the active subtree.
13785 2017-08-28 Jeff Law <law@redhat.com>
13787 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13788 derive_equivalences.
13789 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13790 Code moved into....
13791 (edge_info::derive_equivalences): New private member function
13793 * tree-ssa-dom.c (class edge_info): Changed from a struct
13794 to a class. Add ctor/dtor, methods and data members.
13795 (edge_info::edge_info): Renamed from allocate_edge_info.
13796 Initialize additional members.
13797 (edge_info::~edge_info): New.
13798 (free_dom_edge_info): Delete the edge info.
13799 (record_edge_info): Use new class & associated member functions.
13800 Tighten forms for testing for edge equivalences.
13801 (record_temporary_equivalences): Iterate over the simple
13802 equivalences rather than assuming there's only one per edge.
13803 (cprop_into_successor_phis): Iterate over the simple
13804 equivalences rather than assuming there's only one per edge.
13805 (optimize_stmt): Use operand_equal_p rather than pointer
13806 equality for mini-DSE code.
13808 2017-08-28 Nathan Sidwell <nathan@acm.org>
13810 * gcc.c (execute): Fold SIGPIPE handling into switch
13811 statement. Adjust internal error message.
13813 2017-08-28 Richard Biener <rguenther@suse.de>
13816 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13817 Do nothing for removed DIEs.
13819 2017-08-28 Richard Biener <rguenther@suse.de>
13821 PR tree-optimization/81977
13822 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13825 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13828 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13829 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13830 using targetm.gen_mem_thread_fence.
13832 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13835 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13836 predicate to register_operand. Reorder operands.
13837 (*btr<mode>): Ditto.
13838 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13839 (*btr<mode>_mask): Ditto.
13841 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13843 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13844 * config/rs6000/xmmintrin.h: New file.
13845 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13847 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13850 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13851 parameter and_insn and return it.
13852 (recombine_lvx_pattern): Insert a copy to ensure availability of
13853 the base register of the copied masking operation at the point of
13854 the instruction replacement.
13855 (recombine_stvx_pattern): Likewise.
13857 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13859 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13860 undocumented switches.
13861 (-mpower9-dform-vector): Likewise.
13862 (-mpower9-dform): Likewise.
13863 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13864 comments to delete references to -mpower9-dform* switches.
13865 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13866 Delete reference to -mpower9-dform* switches, test for
13867 -mpower9-vector instead.
13868 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13869 (OTHER_P9_VECTOR_MASKS): Likewise.
13870 (POWERPC_MASKS): Likewise.
13871 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13872 tests against -mpower9-dform* to -mpower9-vector. Delete code
13873 that checked for -mpower9-dform* consistancy with other options.
13874 Add test for -mpower9-misc to enable other power9 switches.
13875 (rs6000_init_hard_regno_mode_ok): Likewise.
13876 (rs6000_option_override_internal): Likewise.
13877 (rs6000_emit_prologue): Likewise.
13878 (rs6000_emit_epilogue): Likewise.
13879 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13880 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13881 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13883 (emit_fusion_p9_store): Likewise.
13884 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13885 resetting these macros if the assembler does not support ISA 3.0
13887 (TARGET_P9_DFORM_VECTOR): Likewise.
13888 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13889 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13891 2017-08-25 Alan Modra <amodra@gmail.com>
13893 PR rtl-optimization/81747
13894 * cse.c (cse_extended_basic_block): Don't attempt to record
13895 equivalences for degenerate conditional branches that branch
13896 to their fall-through.
13898 2017-08-24 Martin Sebor <msebor@redhat.com>
13900 PR middle-end/81908
13901 * gimple-fold.c (size_must_be_zero_p): New function.
13902 (gimple_fold_builtin_memory_op): Call it.
13904 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13906 * config/rs6000/mm_malloc.h: New file.
13908 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13910 PR tree-optimization/81913
13911 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13912 analysis when either IVs in condition can wrap.
13914 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13916 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13917 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13919 2017-08-24 Richard Biener <rguenther@suse.de>
13922 * targhooks.c (default_target_can_inline_p): Properly
13923 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13924 is present and always compare.
13925 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13926 infer -mfpmath=sse from TARGET_SSE_P.
13927 (ix86_can_inline_p): Properly use target_option_default_node when
13928 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13930 2017-08-24 Richard Biener <rguenther@suse.de>
13933 * dwarf2out.c (output_die): Handle flag_generate_offload like
13935 (output_comp_unit): Likewise.
13936 (gen_array_type_die): Likewise.
13937 (dwarf2out_early_finish): Likewise.
13938 (note_variable_value_in_expr): Likewise.
13939 (dwarf2out_finish): Likewise. Adjust assert.
13940 * cgraphunit.c (symbol_table::compile): Move setting of
13941 flag_generate_offload earlier ...
13942 (symbol_table::finalize_compilation_unit): ... here, before
13943 early debug finalization.
13945 2017-08-24 Richard Biener <rguenther@suse.de>
13947 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13948 and ipa-fnsummary.h.
13949 (ix86_can_inline_p): When ix86_fpmath flags do not match
13950 check whether the callee uses FP math at all.
13952 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13954 PR middle-end/81931
13955 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13956 instead of TYPE_PRECISION.
13958 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13959 Alan Hayward <alan.hayward@arm.com>
13960 David Sherwood <david.sherwood@arm.com>
13962 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13964 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13966 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13968 (ix86_frame::stack_realign_allocate): New field.
13969 (struct machine_frame_state): Modify comments.
13970 (machine_frame_state::sp_realigned_fp_end): New field.
13971 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13972 layout calculation.
13973 (sp_valid_at): Add assertion to assure no attempt to access invalid
13974 offset of a realigned stack.
13975 (fp_valid_at): Likewise.
13976 (choose_baseaddr): Modify comments.
13977 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13978 ix86_expand_prologue.
13979 (ix86_expand_prologue): Modify stack realignment and allocation.
13980 (ix86_expand_epilogue): Modify comments.
13981 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13982 avx2_runtime, avx512f, and avx512f_runtime.
13984 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13986 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13987 (mstackrealign): Do not init to -1.
13988 * config/i386/i386.c (ix86_option_override_internal):
13989 Check opts_set, not opts when setting default value of
13990 opts->x_ix86_force_align_arg_pointer.
13992 2017-08-23 Richard Biener <rguenther@suse.de>
13994 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13995 lang_hooks.decl_printable_name.
13996 * print-rtl-function.c (print_rtx_function): Likewise.
13997 * tree-pretty-print.c (dump_function_header): Likewise.
13999 2017-08-23 Richard Biener <rguenther@suse.de>
14002 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14003 -g0 at compile-time.
14005 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14007 PR middle-end/19706
14008 * doc/sourcebuild.texi (Other hardware attributes):
14011 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14013 PR middle-end/19706
14014 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14015 Fix single-use check.
14017 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14019 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14021 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14023 * doc/install.texi: Modify to add more details on running selected
14026 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14028 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14029 is combined with -mabi=ms.
14030 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14033 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14035 PR tree-optimization/81488
14036 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14037 and cached_basis fields.
14038 (MAX_SPREAD): New constant.
14039 (alloc_cand_and_find_basis): Initialize new fields.
14040 (clear_visited): New function.
14041 (create_phi_basis_1): Rename from create_phi_basis, set visited
14042 and cached_basis fields.
14043 (create_phi_basis): New wrapper function.
14044 (phi_add_costs_1): Rename from phi_add_costs, add spread
14045 parameter, set visited field, short-circuit when limits reached.
14046 (phi_add_costs): New wrapper function.
14047 (record_phi_increments_1): Rename from record_phi_increments, set
14049 (record_phi_increments): New wrapper function.
14050 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14051 (phi_incr_cost): New wrapper function.
14052 (all_phi_incrs_profitable_1): Rename from
14053 all_phi_incrs_profitable, set visited field.
14054 (all_phi_incrs_profitable): New wrapper function.
14056 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14057 Alan Hayward <alan.hayward@arm.com>
14058 David Sherwood <david.sherwood@arm.com>
14060 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14061 that takes the outer and inner modes.
14062 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14063 comparison as the canonical test for a paradoxical subreg.
14064 * combine.c (simplify_set): Use paradoxical_subreg_p.
14065 (make_extraction): Likewise.
14066 (force_to_mode): Likewise.
14067 (rtx_equal_for_field_assignment_p): Likewise.
14068 (gen_lowpart_for_combine): Likewise.
14069 (simplify_comparison): Likewise.
14070 * cse.c (equiv_constant): Likewise.
14071 * expmed.c (store_bit_field_1): Likewise.
14072 * final.c (alter_subreg): Likewise.
14073 * fwprop.c (propagate_rtx): Likewise.
14074 (forward_propagate_subreg): Likewise.
14075 * ira-conflicts.c (ira_build_conflicts): Likewise.
14076 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14077 * lra-constraints.c (curr_insn_transform): Likewise.
14078 (split_reg): Likewise.
14079 * lra-eliminations.c (move_plus_up): Likewise.
14080 (lra_eliminate_regs_1): Likewise.
14081 * recog.c (general_operand): Likewise.
14082 * ree.c (combine_reaching_defs): Likewise.
14083 * reload.c (push_reload): Likewise.
14084 (find_reloads): Likewise.
14085 * reload1.c (elimination_effects): Likewise.
14086 (compute_reload_subreg_offset): Likewise.
14087 (choose_reload_regs): Likewise.
14088 * rtlanal.c (subreg_lsb_1): Likewise.
14089 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14090 (simplify_subreg): Likewise.
14091 * var-tracking.c (track_loc_p): Likewise.
14092 * emit-rtl.c (byte_lowpart_offset): Likewise.
14093 (paradoxical_subreg_p): Delete out-of-line definition.
14095 2017-08-22 Jeff Law <law@redhat.com>
14097 PR tree-optimization/81741
14098 PR tree-optimization/71947
14099 * tree-ssa-dom.c: Include tree-inline.h.
14100 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14101 equivalences if one is more expensive to compute than the other.
14102 * tree-ssa-scopedtables.h (class const_or_copies): Make
14103 record_const_or_copy_raw method private.
14104 (class avail_exprs_stack): New method simplify_binary_operation.
14105 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14106 avail_exprs_stack::simplify_binary_operation as needed.
14107 (avail_exprs_stack::simplify_binary_operation): New function.
14109 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14111 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14112 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14113 (DOT_SYMBOLS): Likewise.
14114 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14115 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14116 (RS6000_ABI_NAME): Likewise.
14117 (TARGET_CMODEL): Likewise.
14118 (TOC_SECTION_ASM_OP): Likewise.
14119 (SET_CMODEL): New macro.
14120 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14122 2017-08-22 Richard Biener <rguenther@suse.de>
14124 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14125 to allow for free-lang-data replacements similar to verify_type_variant.
14127 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14128 Alan Hayward <alan.hayward@arm.com>
14129 David Sherwood <david.sherwood@arm.com>
14131 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14132 VOIDmode for the LABEL_REF.
14134 2017-08-22 Richard Biener <rguenther@suse.de>
14136 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14138 2017-08-22 Richard Biener <rguenther@suse.de>
14140 * common.opt (feliminate-dwarf2-dups): Ignore.
14141 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14142 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14143 same_die_p_wrap, compute_section_prefix,
14144 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14145 (comdat_symbol_id, comdat_symbol_number): Likewise.
14146 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14148 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14149 (output_die): Mark unreachable path unreachable.
14150 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14151 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14152 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14153 (dwarf2out_early_finish): Likewise.
14155 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14157 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14159 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14162 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14163 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14164 (avr_attribute_table) <io, io_low, address>: Initialize
14165 .decl_required with true.
14167 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14169 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14170 undocumented debugging options.
14171 (-mvsx-scalar-double): Likewise.
14172 (-mallow-df-permute): Likewise.
14173 (-mvectorize-builtins): Likewise.
14174 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14175 (rs6000_builtin_vectorized_function): Likewise.
14176 (rs6000_builtin_md_vectorized_function): Likewise.
14177 (rs6000_opt_vars): Likewise.
14179 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14182 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14183 (*btrq_imm): Rename from *btrq.
14184 (*btcq_imm): Rename from *btcq.
14185 (btsc): New code attribute.
14186 (*<btsc><mode>): New insn pattern.
14187 (*btr<mode>): Ditto.
14188 (*<btsc><mode>_mask): New insn_and_split pattern.
14189 (*btr<mode>_mask): Ditto.
14191 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14192 Alan Hayward <alan.hayward@arm.com>
14193 David Sherwood <david.sherwood@arm.com>
14195 * function.c (pad_below): Simplify padding calculation.
14197 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14198 Alan Hayward <alan.hayward@arm.com>
14199 David Sherwood <david.sherwood@arm.com>
14201 * target.def (function_prologue): Remove frame size argument.
14202 (function_epilogue): Likewise.
14203 * doc/tm.texi: Regenerate.
14204 * final.c (final_start_function): Update call to function_prologue.
14205 (final_end_function): Update call to function_epilogue.
14206 (default_function_pro_epilogue): Remove frame size argument.
14207 * output.h (default_function_pro_epilogue): Likewise.
14208 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14209 (arm_output_function_prologue): Likewise.
14210 * config/frv/frv.c (frv_function_prologue): Likewise.
14211 (frv_function_epilogue): Likewise.
14212 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14213 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14214 (ia64_output_function_epilogue): Likewise.
14215 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14216 (m32r_output_function_epilogue): Likewise.
14217 * config/microblaze/microblaze.c (microblaze_function_prologue)
14218 (microblaze_function_epilogue): Likewise.
14219 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14220 (mips_output_function_epilogue): Likewise.
14221 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14222 (mmix_target_asm_function_epilogue): Likewise.
14223 * config/msp430/msp430.c (msp430_start_function): Likewise.
14224 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14225 (nds32_asm_function_epilogue): Likewise.
14226 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14227 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14228 (pa_output_function_epilogue): Likewise.
14229 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14230 (rs6000_output_function_epilogue): Likewise.
14231 * config/rl78/rl78.c (rl78_start_function): Likewise.
14232 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14233 (rs6000_output_function_epilogue): Likewise.
14234 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14235 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14236 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14237 (sparc_asm_function_epilogue): Likewise.
14239 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14241 * tree.h (type_has_mode_precision_p): New function.
14242 * convert.c (convert_to_integer_1): Use it.
14243 * expr.c (expand_expr_real_2): Likewise.
14244 (expand_expr_real_1): Likewise.
14245 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14246 * match.pd: Likewise.
14247 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14248 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14249 * tree-tailcall.c (process_assignment): Likewise.
14250 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14251 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14252 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14253 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14254 (vectorizable_assignment): Likewise.
14255 (vectorizable_shift): Likewise.
14256 (vectorizable_operation): Likewise.
14257 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14259 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14261 * match.pd: Add pow (C, x) simplification.
14263 2017-08-21 Richard Biener <rguenther@suse.de>
14265 PR tree-optimization/81900
14266 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14267 for blocks with abnormal predecessors.
14268 (compute_antic): Do not set visited flag prematurely.
14270 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14273 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14275 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14277 * stor-layout.h (vector_type_mode): Move to...
14278 * tree.h (vector_type_mode): ...here.
14279 * stor-layout.c (vector_type_mode): Move to...
14280 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14282 2017-08-21 Richard Biener <rguenther@suse.de>
14284 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14285 register_external_die hooks.
14286 (debug_false_tree_charstarstar_uhwistar): Declare.
14287 (debug_nothing_tree_charstar_uhwi): Likewise.
14288 * debug.c (do_nothing_debug_hooks): Adjust.
14289 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14290 (debug_nothing_tree_charstar_uhwi): Likewise.
14291 * dbxout.c (dbx_debug_hooks): Adjust.
14292 (xcoff_debug_hooks): Likewise.
14293 * sdbout.c (sdb_debug_hooks): Likewise.
14294 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14295 * dwarf2out.c (macinfo_label_base): New global.
14296 (dwarf2out_register_external_die): New function for the
14297 register_external_die hook.
14298 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14299 (dwarf2_debug_hooks): Use them.
14300 (dwarf2_lineno_debug_hooks): Adjust.
14301 (struct die_struct): Add with_offset flag.
14302 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14303 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14304 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14305 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14306 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14307 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14308 defining section names for the early LTO debug variants.
14309 (reset_indirect_string): New helper.
14310 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14311 (print_dw_val): Add support for offsetted symbol references.
14312 (get_ultimate_context): Split out from is_cxx.
14313 (is_cxx): Use get_ultimate_context.
14314 (is_fortran): Add decl overload.
14315 (compute_comp_unit_symbol): Split out worker from
14316 compute_section_prefix.
14317 (compute_section_prefix): Call compute_comp_unit_symbol and
14318 set comdat_type_p here.
14319 (output_die): Skip DIE symbol output for the LTO added one.
14320 Handle DIE symbol references with offset.
14321 (output_comp_unit): Guard section name mangling properly.
14322 For LTO debug sections emit a symbol at the section beginning
14323 which we use to refer to its DIEs.
14324 (add_abstract_origin_attribute): For DIEs registered via
14325 dwarf2out_register_external_die directly refer to the early
14326 DIE rather than indirectly through the shadow one we created.
14327 Remove obsolete call to dwarf2out_abstract_function for
14328 non-function/block origins.
14329 (gen_array_type_die): When generating early LTO debug do
14330 not emit DW_AT_string_length.
14331 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14332 late when in LTO. As suggested place a gcc_unreachable for
14333 the DECL_ABSTRACT_P case.
14334 (gen_subprogram_die): Avoid another specification DIE
14335 for early built declarations/definitions for the late LTO case.
14336 (gen_variable_die): Add type references for late duplicated VLA dies
14338 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14339 we have the abstract instance already.
14340 (process_scope_var): Adjust decl DIE contexts in LTO which
14341 first puts them in limbo.
14342 (gen_decl_die): Do not generate type DIEs late apart from
14343 types for VLAs or for decls we do not yet have a DIE. Do not
14344 call dwarf2out_abstract_function late.
14345 (dwarf2out_early_global_decl): Make sure to create DIEs
14346 for abstract instances of a decl first.
14347 (dwarf2out_late_global_decl): Adjust comment.
14348 (output_macinfo_op): With multiple macro sections use
14349 macinfo_label_base to distinguish labels.
14350 (output_macinfo): Likewise. Update macinfo_label_base.
14351 Pass in the line info label.
14352 (note_variable_value_in_expr): When generating LTO resolve
14353 all variable values here by generating DIEs as needed.
14354 (init_sections_and_labels): Add early LTO debug flag parameter
14355 and generate different sections and names if set. Add generation
14356 counter for the labels so we can have multiple of them.
14357 (reset_dies): Helper to allow DIEs to be output multiple times.
14358 (dwarf2out_finish): When outputting DIEs to the fat part of an
14359 LTO object first reset DIEs.
14360 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14361 (modified_type_die): Check for decl_ultimate_origin being self
14363 (gen_type_die_with_usage): Likewise.
14364 (gen_typedef_die): Allow decl_ultimate_origin being self.
14365 (set_decl_abstract_flags): Remove.
14366 (set_block_abstract_flags): Likewise.
14367 (dwarf2out_abstract_function): Treat the early generated DIEs
14368 as the abstract copy and only add DW_AT_inline and
14369 DW_AT_artificial here and call set_decl_origin_self.
14370 If the DIE has an abstract origin don't do anything.
14371 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14372 if we have none yet (Go fails to build one, PR78628).
14373 (variably_modified_type_p): Prevent endless recursion for Ada
14374 cyclic pointer types.
14375 * lto-streamer-in.c: Include debug.h.
14376 (dref_queue): New global.
14377 (lto_read_tree_1): Stream in DIE references.
14378 (lto_input_tree): Register DIE references.
14379 (input_function): Stream DECL_DEBUG_ARGS.
14380 * lto-streamer-out.c: Include debug.h.
14381 (lto_write_tree_1): Output DIE references.
14382 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14383 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14384 (output_function): Stream DECL_DEBUG_ARGS.
14385 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14386 Stream DECL_ABSTRACT_ORIGIN.
14387 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14388 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14389 DECL_CONTEXT for file-scope decls.
14390 * lto-streamer.h (struct dref_entry): Declare.
14391 (dref_queue): Likewise.
14392 * cfgexpand.c (pass_expand::execute): Do not call the
14393 outlining_inline_function hook here.
14394 * lto-wrapper.c (debug_obj): New global.
14395 (tool_cleanup): Unlink it if required.
14396 (debug_objcopy): New function.
14397 (run_gcc): Handle early debug sections in the IL files by
14398 extracting them to separate files, partially linkin them and
14399 feeding the result back as result to the linker.
14400 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14401 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14402 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14403 sections into a separate segment.
14404 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14406 (darwin_asm_dwarf_section): Likewise.
14407 (darwin_asm_output_dwarf_offset): Likewise.
14408 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14410 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14411 Alan Hayward <alan.hayward@arm.com>
14412 David Sherwood <david.sherwood@arm.com>
14414 * read-md.h (md_reader::record_potential_iterator_use): Replace
14415 pointer argument with an rtx and an index.
14416 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14417 (apply_mode_iterator): Likewise.
14418 (apply_code_iterator): Likewise.
14419 (apply_int_iterator): Likewise.
14420 (apply_subst_iterator): Likewise.
14421 (record_iterator_use): Likewise.
14422 (record_attribute_use): Likewise.
14423 (md_reader::record_potential_iterator_use): Likewise. Update calls
14424 to record_iterator_use and apply_iterator.
14425 (iterator_use): Replace ptr with x and index.
14426 (attribute_use): Likewise.
14427 (apply_attribute_uses): Update calls to apply_iterator.
14428 (apply_iterators): Likewise. Update initialization of iterator_use.
14429 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14430 and record_potential_iterator_use.
14431 (rtx_reader::read_rtx_operand): Likewise.
14433 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14434 Alan Hayward <alan.hayward@arm.com>
14435 David Sherwood <david.sherwood@arm.com>
14437 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14440 2017-08-21 Richard Biener <rguenther@suse.de>
14442 PR middle-end/81884
14443 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14444 at struct end conservatively when comparing common bases.
14446 2017-08-21 Richard Biener <rguenther@suse.de>
14448 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14449 (mem_ref_in_stmt): Remove.
14450 (determine_max_movement): Use ref index to get at the reference.
14451 (invariantness_dom_walker::before_dom_children): Deal with
14452 lim data already initialized.
14453 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14455 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14457 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14458 (TARGET_ISA_ROUND): Ditto.
14459 (TARGET_ROUND): Ditto.
14460 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14461 * config/i386/i386.md: Ditto.
14462 * config/i386/sse.md: Ditto.
14463 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14464 with OPTION_MASK_ISA_SSE4_1.
14466 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14469 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14470 __builtin_ia32_lzcnt_u16.
14472 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14475 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14476 (rs6000_set_current_function): Rewrite function to use it.
14478 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14481 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14482 and TYPE_WARN_IF_NOT_ALIGN.
14483 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14484 (handle_warn_if_not_align): New.
14485 (place_union_field): Call handle_warn_if_not_align.
14486 (place_field): Call handle_warn_if_not_align.
14487 Copy TYPE_WARN_IF_NOT_ALIGN.
14488 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14489 (layout_type): Likewise.
14490 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14492 (tree_decl_common): Add warn_if_not_align.
14493 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14494 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14495 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14496 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14497 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14498 * doc/extend.texi: Document warn_if_not_aligned attribute.
14499 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14501 2017-08-17 Martin Liska <mliska@suse.cz>
14504 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14505 (get_data_dependence): Use it as pointer type.
14506 (distribute_loop): Likewise.
14508 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14510 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14511 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14512 (p8_vmrgow_<mode>_direct): New define_insn.
14513 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14514 handle endianness for vmrgew and vmrgow permute patterns.
14516 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14518 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14519 * config/rs6000/rs6000-cpus.def: Remove comment.
14520 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14521 (POWERPC_MASKS): Likewise.
14522 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14523 use of TARGET_VSX_TIMODE.
14524 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14525 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14526 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14527 (rs6000_option_override_internal): Remove dead code.
14528 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14529 (rs6000_legitimize_reload_address): Likewise.
14530 (rs6000_legitimate_address_p): Likewise.
14531 (rs6000_opt_masks): Delete "vsx-timode".
14532 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14533 from function comment.
14534 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14535 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14536 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14538 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14539 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14540 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14541 (VSX_TI): Likewise.
14543 (define_peephole2): Likewise.
14545 2017-08-17 Martin Sebor <msebor@redhat.com>
14548 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14549 past the end of an array.
14550 (test_pp_format): Add test cases.
14552 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14554 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14555 missing ECF_NOTHROW flags.
14557 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14560 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14561 operands residing in integer registers.
14562 (*vsx_le_perm_load_<mode>): Likewise.
14563 (*vsx_le_perm_store_<mode>): Likewise.
14564 (define_peephole2): Add peepholes to optimize the above.
14566 2017-08-17 Marek Polacek <polacek@redhat.com>
14568 PR middle-end/81814
14569 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14570 to mimic what shorten_compare did. Change the return type to bool.
14571 (fold_cond_expr_with_comparison): Update call to
14572 operand_equal_for_comparison_p.
14573 (fold_ternary_loc): Likewise.
14575 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14577 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14579 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14581 2017-08-17 Richard Biener <rguenther@suse.de>
14583 * tree-ssa-structalias.c (solve_graph): When propagating
14584 to successors update the graphs succ edges and avoid duplicate work.
14586 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14589 * config/i386/i386.c (ix86_option_override_internal): Save target
14590 specific options after ix86_stack_protector_guard_reg was changed.
14592 2017-08-17 Richard Biener <rguenther@suse.de>
14594 PR tree-optimization/81827
14595 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14596 (new_var_info): Initialize it conservatively.
14597 (get_call_vi): Mark register vars.
14598 (new_scalar_tmp_constraint_exp): Likewise.
14599 (handle_rhs_call): Likewise.
14600 (handle_const_call): Likewise.
14601 (create_function_info_for): Likewise.
14602 (solve_constraints): Sort varinfos to separate register from
14603 non-register vars to pack points-to solution bitmaps during
14606 2017-08-17 Marek Polacek <polacek@redhat.com>
14608 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14610 2017-08-17 Richard Biener <rguenther@suse.de>
14612 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14613 to true when overflow is undefined and we saturated the result.
14615 2017-08-17 Alan Modra <amodra@gmail.com>
14618 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14619 Don't use store multiple if only one reg needs saving.
14620 (interesting_frame_related_regno): New function.
14621 (rs6000_frame_related): Don't emit frame info for regs that
14623 (rs6000_emit_epilogue): Likewise.
14625 2017-08-16 Nathan Sidwell <nathan@acm.org>
14627 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14628 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14629 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14630 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14631 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14632 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14633 (verify_type): Adjust for TYPE_BINFO move.
14634 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14635 process TYPE_BINFO directly.
14636 (hash_tree): Likewise.
14637 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14639 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14642 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14644 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14646 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14649 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14650 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14651 Add ix86_binary_operator_ok to insn constraint.
14652 (*iordi_1_bts): Ditto.
14653 (*xordi_1_btc): Ditto.
14654 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14655 Update corresponding peephole2 pattern.
14659 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14661 PR tree-optimization/81832
14662 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14663 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14665 2017-08-16 Marek Polacek <polacek@redhat.com>
14668 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14669 regarding size_zero_node.
14671 2017-08-16 Martin Liska <mliska@suse.cz>
14674 * config.gcc: Respect previously set extra_objs in case
14677 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14679 PR tree-optimization/81835
14680 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14681 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14682 not depend on the phi.
14684 2017-08-16 Alan Modra <amodra@gmail.com>
14686 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14689 2017-08-16 Alan Modra <amodra@gmail.com>
14691 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14692 (save_reg_p): ..into this. Update all callers.
14693 (first_reg_to_save): Simplify.
14695 2017-08-16 Alan Modra <amodra@gmail.com>
14697 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14700 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14704 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14705 constant count if that count is less than 32.
14707 2017-08-15 Nathan Sidwell <nathan@acm.org>
14709 * gcc.c (execute): Emit friendlier message if inferior is killed
14710 by an external cause.
14712 2017-08-15 Richard Biener <rguenther@suse.de>
14714 PR tree-optimization/81790
14715 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14716 CONSTRUCTORs from simplifying and VN.
14718 2017-08-14 Martin Sebor <msebor@redhat.com>
14720 * builtin-attrs.def: Add comments.
14722 2017-08-14 Martin Sebor <msebor@redhat.com>
14725 * doc/extend.texi (attribute nonstring): Document new attribute.
14727 2017-08-14 Martin Sebor <msebor@redhat.com>
14730 * tree-diagnostic.c (default_tree_printer): Handle %G.
14731 * gimple-pretty-print.h (percent_G_format): Declare new function.
14732 * gimple-pretty-print.c (percent_G_format): Define.
14733 * tree-pretty-print.c (percent_K_format): Add argument.
14735 2017-08-14 Martin Sebor <msebor@redhat.com>
14737 PR translation/79998
14738 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14739 Remove a stray space.
14741 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14744 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14745 (*iordi_1_bts): Ditto.
14746 (*xordi_1_btc): Ditto.
14748 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14751 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14752 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14754 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14755 quoted strings, and make more translator-friendly.
14756 (darwin_rs6000_override_options): Likewise.
14757 (rs6000_option_override_internal): Likewise.
14758 (rs6000_return_in_memory): Fix overlong line.
14759 (init_cmulative_args): Use quoted strings, and make more
14760 translator-friendly.
14761 (rs6000_pass_by_reference): Fix overlong line.
14762 (def_builtin): Use quoted strings.
14763 (altivec_expand_predicate_builtin): Use quoted strings, and make
14764 more translator-friendly.
14765 (htm_expand_builtin): Use quoted strings.
14766 (cpu_expand_builtin): Use quoted strings, and make more
14767 translator-friendly.
14768 (altivec_expand_builtin): Likewise.
14769 (paired_expand_predicate_builtin): Likewise.
14770 (rs6000_invalid_builtin): Likewise.
14771 (builtin_function_type): Use quoted strings.
14772 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14773 more translator-friendly.
14774 (rs6000_trampoline_init): Likewise.
14775 (rs6000_handle_altivec_attribute): Likewise.
14776 (rs6000_inner_target_options): Use quoted strings.
14777 (rs6000_disable_incompatible_switches): Likewise.
14778 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14779 strings, and make more translator-friendly.
14780 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14782 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14784 PR tree-optimization/81799
14785 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14786 cond_expr to simple gimple operand.
14788 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14790 PR middle-end/46932
14791 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14793 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14797 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14798 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14799 TARGET_GASISR_PROLOGUES.
14800 * config/avr/avr.c (avr_option_override): Same.
14801 (avr_pass_pre_proep::execute): Same.
14803 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14806 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14807 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14809 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14811 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14812 define_insn_and_split. Split to a memory load from 0 in
14813 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14814 using PTR mode iterator.
14815 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14816 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14817 (*add_tp_<mode>): Redefine as define_insn_and_split.
14818 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14819 address space. Merge with *add_tp_x32 using PTR mode iterator.
14820 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14821 Split to an add with a memory load from 0 in
14822 DEFAULT_TLS_SEG_REG address space.
14824 2017-08-12 Andrew Pinski <apinski@cavium.com>
14826 * config/aarch64/aarch64-option-extensions.def (rdma):
14827 Fix feature string to what Linux prints out in /proc/cpuinfo.
14829 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14832 * dwarf2out.c (modified_type_die): For C typedef types that have
14833 an ultimate origin, process the ultimate origin instead of the
14835 (gen_typedef_die): Assert that input DECLs have no ultimate
14837 (gen_type_die_with_usage): For typedef variants that have an
14838 ultimate origin, just call gen_decl_die on the original DECL.
14839 (process_scope_var): Avoid creating DIEs for local typedefs and
14840 concrete static variables.
14842 2017-08-12 Alan Modra <amodra@gmail.com>
14846 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14847 match gnu-user.h startfile.
14848 (ENDFILE_LINUX_SPEC): Similarly.
14850 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14853 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14855 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14857 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14858 * config/aarch64/aarch64.md (mov<mode>): Change.
14859 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14860 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14861 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14863 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14865 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14866 for storage order barriers.
14868 2017-08-11 Martin Liska <mliska@suse.cz>
14871 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14872 variables of void type.
14874 2017-08-11 Martin Liska <mliska@suse.cz>
14876 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14877 TARGET_SUPPORTS_ALIASES.
14878 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14879 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14880 (optimize_weakref): Likewise.
14881 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14882 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14883 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14885 2017-08-11 Martin Liska <mliska@suse.cz>
14888 * config/i386/i386.c (make_resolver_func): Do complete
14889 refactoring of the function.
14891 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14894 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14895 * config/i386/i386.c (ix86_stack_protect_guard): Use
14896 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14897 * doc/invoke.texi (x86 Options): Document
14898 -mstack-protector-guard-symbol= option.
14900 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14902 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14903 * config/i386/i386.c (ix86_split_stack_guard): New function.
14904 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14905 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14906 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14907 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14908 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14909 (split_stack_space_check): Call ix86_split_stack_guard.
14911 2017-08-10 Martin Sebor <msebor@redhat.com>
14913 * print-tree.c (print_node): Print location using the established
14915 Replace spaces with colons.
14916 (debug_raw, debug): Ditto.
14918 2017-08-10 Martin Sebor <msebor@redhat.com>
14921 * pretty-print.c (pp_format): Correct the handling of %s precision.
14923 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14926 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14928 (ix86_finalize_stack_frame_flags): This. Also clear
14929 frame_pointer_needed if -fno-omit-frame-pointer is used without
14931 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14932 with ix86_finalize_stack_frame_flags.
14933 (ix86_expand_epilogue): Likewise.
14934 (ix86_expand_split_stack_prologue): Likewise.
14935 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14937 2017-08-10 Martin Liska <mliska@suse.cz>
14940 * c-attribs.c (handle_target_attribute):
14941 Report warning for an empty string argument of target attribute.
14943 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14946 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14948 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14949 or DECL_NONLOCAL labels.
14950 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14951 or DECL_NONLOCAL labels here.
14953 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14955 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14956 to indicate when early gimple folding has been disabled.
14957 (rs6000_gimple_fold_builtin): Add debug content.
14958 (rs6000_invalid_builtin): Fix whitespace.
14959 (rs6000_expand_builtin): Fix whitespace.
14960 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14962 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14965 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14966 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14968 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14970 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14971 (qdf24xx): Likewise.
14972 * config/aarch64/aarch64.md: Include falkor.md.
14973 * config/aarch64/falkor.md: New.
14975 2017-08-09 Marek Polacek <polacek@redhat.com>
14978 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14979 * diagnostic.c (emit_diagnostic): Add a comment.
14980 (emit_diagnostic_valist): New function.
14982 2017-08-09 Marek Polacek <polacek@redhat.com>
14985 * input.c (make_location): New overload.
14986 * input.h (make_location): Declare.
14988 2017-08-08 Alan Modra <amodra@gmail.com>
14989 H.J. Lu <hongjiu.lu@intel.com>
14992 * gcc.c (NO_PIE_SPEC): Delete.
14993 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14995 (LINK_PIE_SPEC): ..to here.
14996 (LINK_COMMAND_SPEC): Support -no-pie.
14997 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14998 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15000 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15001 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15002 (ENDFILE_CRTEND_SPEC): Similarly.
15004 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15007 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15008 (mstack-protector-guard-offset=): Ditto.
15009 * config/i386/i386.c (ix86_option_override): Handle
15010 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15012 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15013 ix86_stack_protect_guard_offset variables.
15014 (TARGET_STACK_PROTECT_GUARD): Always define.
15015 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15016 and -mstack-protector-guard-offset= options.
15018 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15020 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15021 boundary case for the last candidate.
15023 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15025 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15026 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15028 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15030 PR middle-end/19706
15031 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15032 * config/aarch64/aarch64-builtins.c
15033 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15034 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15035 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15037 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15038 Andrew Pinski <pinskia@gmail.com>
15040 PR middle-end/19706
15041 * internal-fn.def (XORSIGN): New.
15042 * optabs.def (xorsign_optab): New.
15043 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15044 (convert_expand_mult_copysign): New.
15045 (pass_optimize_widening_mul::execute): Call
15046 convert_expand_mult_copysign.
15048 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15050 PR tree-optimization/81354
15051 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15052 Insert on edges rather than explicitly creating landing pads.
15053 (analyze_candidates_and_replace): Commit edge inserts.
15055 2017-08-08 Richard Biener <rguenther@suse.de>
15057 PR middle-end/81719
15058 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15059 (expand_simple_operations): Also look through ADDR_EXPRs with
15060 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15062 2017-08-08 Richard Biener <rguenther@suse.de>
15064 PR tree-optimization/81723
15065 * tree-vect-slp.c (struct bst_traits): New hash traits.
15066 (bst_fail): New global.
15067 (vect_build_slp_tree_2): New worker, split out from ...
15068 (vect_build_slp_tree): ... this now wrapping it with using
15069 bst_fail set to cache SLP tree build fails. Properly handle
15071 (vect_analyze_slp_instance): Allocate and free bst_fail.
15073 2017-08-08 Martin Liska <mliska@suse.cz>
15076 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15077 LABEL_DECLs that can be from a different function.
15079 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15081 PR tree-optimization/81744
15082 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15083 loop's number of iterations.
15085 2017-08-08 Martin Liska <mliska@suse.cz>
15087 * asan.c: Include header files.
15088 * attribs.c (build_decl_attribute_variant): New function moved
15090 (build_type_attribute_qual_variant): Likewise.
15091 (cmp_attrib_identifiers): Likewise.
15092 (simple_cst_list_equal): Likewise.
15093 (omp_declare_simd_clauses_equal): Likewise.
15094 (attribute_value_equal): Likewise.
15095 (comp_type_attributes): Likewise.
15096 (build_type_attribute_variant): Likewise.
15097 (lookup_ident_attribute): Likewise.
15098 (remove_attribute): Likewise.
15099 (merge_attributes): Likewise.
15100 (merge_type_attributes): Likewise.
15101 (merge_decl_attributes): Likewise.
15102 (merge_dllimport_decl_attributes): Likewise.
15103 (handle_dll_attribute): Likewise.
15104 (attribute_list_equal): Likewise.
15105 (attribute_list_contained): Likewise.
15106 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15107 (lookup_attribute_by_prefix): Likewise.
15108 * bb-reorder.c: Include header files.
15109 * builtins.c: Likewise.
15110 * calls.c: Likewise.
15111 * cfgexpand.c: Likewise.
15112 * cgraph.c: Likewise.
15113 * cgraphunit.c: Likewise.
15114 * convert.c: Likewise.
15115 * dwarf2out.c: Likewise.
15116 * final.c: Likewise.
15117 * fold-const.c: Likewise.
15118 * function.c: Likewise.
15119 * gimple-expr.c: Likewise.
15120 * gimple-fold.c: Likewise.
15121 * gimple-pretty-print.c: Likewise.
15122 * gimple.c: Likewise.
15123 * gimplify.c: Likewise.
15124 * hsa-common.c: Likewise.
15125 * hsa-gen.c: Likewise.
15126 * internal-fn.c: Likewise.
15127 * ipa-chkp.c: Likewise.
15128 * ipa-cp.c: Likewise.
15129 * ipa-devirt.c: Likewise.
15130 * ipa-fnsummary.c: Likewise.
15131 * ipa-inline.c: Likewise.
15132 * ipa-visibility.c: Likewise.
15134 * lto-cgraph.c: Likewise.
15135 * omp-expand.c: Likewise.
15136 * omp-general.c: Likewise.
15137 * omp-low.c: Likewise.
15138 * omp-offload.c: Likewise.
15139 * omp-simd-clone.c: Likewise.
15140 * opts-global.c: Likewise.
15141 * passes.c: Likewise.
15142 * predict.c: Likewise.
15143 * sancov.c: Likewise.
15144 * sanopt.c: Likewise.
15145 * symtab.c: Likewise.
15146 * toplev.c: Likewise.
15147 * trans-mem.c: Likewise.
15148 * tree-chkp.c: Likewise.
15149 * tree-eh.c: Likewise.
15150 * tree-into-ssa.c: Likewise.
15151 * tree-object-size.c: Likewise.
15152 * tree-parloops.c: Likewise.
15153 * tree-profile.c: Likewise.
15154 * tree-ssa-ccp.c: Likewise.
15155 * tree-ssa-live.c: Likewise.
15156 * tree-ssa-loop.c: Likewise.
15157 * tree-ssa-sccvn.c: Likewise.
15158 * tree-ssa-structalias.c: Likewise.
15159 * tree-ssa.c: Likewise.
15160 * tree-streamer-in.c: Likewise.
15161 * tree-vectorizer.c: Likewise.
15162 * tree-vrp.c: Likewise.
15163 * tsan.c: Likewise.
15164 * ubsan.c: Likewise.
15165 * varasm.c: Likewise.
15166 * varpool.c: Likewise.
15167 * tree.c: Remove functions moved to attribs.[ch].
15168 * tree.h: Likewise.
15169 * config/aarch64/aarch64.c: Add attrs.h header file.
15170 * config/alpha/alpha.c: Likewise.
15171 * config/arc/arc.c: Likewise.
15172 * config/arm/arm.c: Likewise.
15173 * config/avr/avr.c: Likewise.
15174 * config/bfin/bfin.c: Likewise.
15175 * config/c6x/c6x.c: Likewise.
15176 * config/cr16/cr16.c: Likewise.
15177 * config/cris/cris.c: Likewise.
15178 * config/darwin.c: Likewise.
15179 * config/epiphany/epiphany.c: Likewise.
15180 * config/fr30/fr30.c: Likewise.
15181 * config/frv/frv.c: Likewise.
15182 * config/ft32/ft32.c: Likewise.
15183 * config/h8300/h8300.c: Likewise.
15184 * config/i386/winnt.c: Likewise.
15185 * config/ia64/ia64.c: Likewise.
15186 * config/iq2000/iq2000.c: Likewise.
15187 * config/lm32/lm32.c: Likewise.
15188 * config/m32c/m32c.c: Likewise.
15189 * config/m32r/m32r.c: Likewise.
15190 * config/m68k/m68k.c: Likewise.
15191 * config/mcore/mcore.c: Likewise.
15192 * config/microblaze/microblaze.c: Likewise.
15193 * config/mips/mips.c: Likewise.
15194 * config/mmix/mmix.c: Likewise.
15195 * config/mn10300/mn10300.c: Likewise.
15196 * config/moxie/moxie.c: Likewise.
15197 * config/msp430/msp430.c: Likewise.
15198 * config/nds32/nds32-isr.c: Likewise.
15199 * config/nds32/nds32.c: Likewise.
15200 * config/nios2/nios2.c: Likewise.
15201 * config/nvptx/nvptx.c: Likewise.
15202 * config/pa/pa.c: Likewise.
15203 * config/pdp11/pdp11.c: Likewise.
15204 * config/powerpcspe/powerpcspe.c: Likewise.
15205 * config/riscv/riscv.c: Likewise.
15206 * config/rl78/rl78.c: Likewise.
15207 * config/rx/rx.c: Likewise.
15208 * config/s390/s390.c: Likewise.
15209 * config/sh/sh.c: Likewise.
15210 * config/sol2.c: Likewise.
15211 * config/sparc/sparc.c: Likewise.
15212 * config/spu/spu.c: Likewise.
15213 * config/stormy16/stormy16.c: Likewise.
15214 * config/tilegx/tilegx.c: Likewise.
15215 * config/tilepro/tilepro.c: Likewise.
15216 * config/v850/v850.c: Likewise.
15217 * config/vax/vax.c: Likewise.
15218 * config/visium/visium.c: Likewise.
15219 * config/xtensa/xtensa.c: Likewise.
15221 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15224 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15225 constraints since the -mupper-regs-* switches have been
15227 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15228 into a vector from a double word element that was extracted from
15229 another vector, and eliminate extra XXPERMDI instructions.
15230 (vsx_concat_<mode>_2): Likewise.
15231 (vsx_concat_<mode>_3): Likewise.
15232 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15233 concat to allow optimizing inserts from previous extracts.
15235 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15237 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15238 memory reference to a SSP offset in TLS address space.
15239 (ix86_print_operand) <case '@'>: Remove.
15240 (ix86_print_operand_punct_valid_p): Remove '@' code.
15241 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15242 UNSPEC_SP_TLS_TEST.
15243 (stack_tls_protect_set_<mode>): Remove.
15244 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15245 (stack_tls_protect_test_<mode>): Remove.
15246 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15248 2017-08-07 Olivier Hainque <hainque@adacore.com>
15251 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15253 2017-08-07 Douglas Rupp <rupp@adacore.com>
15255 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15256 variable was referenced as multidir in command.
15258 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15261 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15264 2017-08-07 Martin Liska <mliska@suse.cz>
15266 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15267 * config/rl78/rl78.c: Add include of attribs.h.
15268 * config/sh/sh.c: Likewise.
15269 * config/v850/v850.c: Likewise.
15271 2017-08-07 Tom de Vries <tom@codesourcery.com>
15273 PR middle-end/78266
15274 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15276 2017-08-07 Martin Liska <mliska@suse.cz>
15278 * config/mips/mips.c: Include attribs.h.
15280 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15283 * doc/invoke.texi: Document change in behvaior for -Ofast for
15286 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15288 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15290 (aarch64_pop_regs): Likewise.
15291 (aarch64_gen_load_pair): Likewise.
15292 (aarch64_save_callee_saves): Likewise.
15293 (aarch64_restore_callee_saves): Likewise.
15295 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15297 * config/i386/i386.c: Revert the last change.
15299 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15302 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15304 (ix86_finalize_stack_frame_flags): This. Also clear
15305 frame_pointer_needed if -fno-omit-frame-pointer is used without
15307 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15308 with ix86_finalize_stack_frame_flags.
15309 (ix86_expand_epilogue): Likewise.
15310 (ix86_expand_split_stack_prologue): Likewise.
15312 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15315 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15316 to P_AES for Westmere.
15318 2017-08-07 Jonathan Yong <10walls@gmail.com>
15320 * config/i386/mingw.opt (fset-stack-executable): Removed.
15321 * config/i386/cygming.opt (fset-stack-executable): Moved
15323 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15325 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15327 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15329 2017-08-07 Marek Polacek <polacek@redhat.com>
15331 PR middle-end/81737
15332 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15334 2017-08-07 Martin Liska <mliska@suse.cz>
15336 * attribs.h (canonicalize_attr_name): New function.
15337 (cmp_attribs): Move from c-format.c and adjusted.
15338 (is_attribute_p): Moved from tree.h.
15339 * tree-inline.c: Add new includes.
15340 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15341 (private_is_attribute_p): Remove.
15342 (private_lookup_attribute): Likewise.
15343 (private_lookup_attribute_by_prefix): Simplify.
15344 (remove_attribute): Use is_attribute_p.
15345 * tree.h: Remove removed declarations.
15347 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15349 PR middle-end/81698
15350 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15351 instead of computing it in the function. Formatting fix.
15352 (expand_case): Don't rely on default_edge being the first edge,
15353 clear it if removing it, pass default_edge to
15354 emit_case_dispatch_table.
15355 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15358 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15360 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15361 insn in the function, emit NOP after the insn.
15363 2017-08-06 Tom de Vries <tom@codesourcery.com>
15365 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15368 2017-08-06 Tom de Vries <tom@codesourcery.com>
15370 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15373 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15375 PR tree-optimization/57371
15376 * match.pd: New pattern.
15378 2017-08-04 Marek Polacek <polacek@redhat.com>
15380 PR middle-end/81695
15381 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15382 perform the computation in offset_int.
15384 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15386 PR tree-optimization/81136
15387 * tree-vectorizer.h: Include tree-hash-traits.h.
15388 (vec_base_alignments): New typedef.
15389 (vec_info): Add a base_alignments field.
15390 (vect_record_base_alignments): Declare.
15391 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15393 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15394 (create_data_ref): Add an is_conditional_in_stmt argument.
15395 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15396 the is_conditional_in_stmt field.
15397 (data_ref_loc): Add an is_conditional_in_stmt field.
15398 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15399 (find_data_references_in_stmt): Update call to create_data_ref.
15400 (graphite_find_data_references_in_stmt): Likewise.
15401 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15402 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15403 (vect_record_base_alignment): New function.
15404 (vect_record_base_alignments): Likewise.
15405 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15406 for nested statements even if we fail to compute a misalignment.
15407 Use pooled base alignments for unconditional references.
15408 (vect_find_same_alignment_drs): Compare base addresses instead
15410 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15411 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15413 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15415 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15416 Add an explicit name for the enum. Use auto_vec for slp_instances
15417 and grouped_stores.
15418 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15420 (_bb_vec_info): Add a constructor and destructor.
15421 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15422 (destroy_loop_vec_info): Delete.
15423 (vect_destroy_datarefs): Likewise.
15424 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15425 (vec_info::vec_info): New function.
15426 (vec_info::~vec_info): Likewise.
15427 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15428 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15429 destroy_loop_vec_info.
15430 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15431 (_loop_vec_info::_loop_vec_info): ...this.
15432 (destroy_loop_vec_info): Replace with...
15433 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15434 the stmt_vec_infos. Leave handling of vec_info information to its
15435 destructor. Remove explicit vector releases.
15436 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15437 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15438 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15439 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15440 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15441 (destroy_bb_vec_info): Replace with...
15442 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15443 information to its destructor.
15444 (vect_slp_analyze_bb_1): Use new and delete instead of
15445 new_bb_vec_info and destroy_bb_vec_info.
15446 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15449 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15451 * tree-data-ref.h (subscript): Add access_fn field.
15452 (data_dependence_relation): Add could_be_independent_p.
15453 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15454 (same_access_functions): Move to tree-data-ref.c.
15455 * tree-data-ref.c (ref_contains_union_access_p): New function.
15456 (access_fn_component_p): Likewise.
15457 (access_fn_components_comparable_p): Likewise.
15458 (dr_analyze_indices): Add a reference to access_fn_component_p.
15459 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15461 (constant_access_functions): Likewise.
15462 (add_other_self_distances): Likewise.
15463 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15464 (initialize_data_dependence_relation): Use XCNEW and remove
15465 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15466 of access functions that have the same type. Allow the
15467 subsequence to end with different bases in some circumstances.
15468 Record the chosen access functions in SUB_ACCESS_FN.
15469 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15470 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15471 (subscript_dependence_tester_1): Likewise dra and drb.
15472 (build_classic_dist_vector): Update calls accordingly.
15473 (subscript_dependence_tester): Likewise.
15474 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15475 DDR_COULD_BE_INDEPENDENT_P.
15476 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15477 comp_alias_ddrs instead of may_alias_ddrs.
15478 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15480 (vect_analyze_data_ref_dependence): Use it if
15481 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15482 distance vectors if that fails.
15483 (dependence_distance_ge_vf): New function.
15484 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15485 LOOP_VINFO_MAY_ALIAS_DDRS.
15487 2017-08-04 Richard Biener <rguenther@suse.de>
15489 PR middle-end/81705
15490 * fold-const.c (fold_binary_loc): Properly restrict
15491 minus_var0 && minus_var1 case when associating undefined overflow
15494 2017-08-04 Tom de Vries <tom@codesourcery.com>
15496 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15498 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15500 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15501 Don't start diagnostic messages with a capital letter.
15502 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15504 (rs6000_invalid_builtin): Likewise.
15505 (rs6000_trampoline_init): Likewise.
15507 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15510 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15511 after setting changeable df flags.
15513 2017-08-03 Richard Biener <rguenther@suse.de>
15515 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15516 up if the use is in USE - X.
15518 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15520 * toplev.c (dumpfile.h): New include.
15521 (internal_error_reentered): New static function. Use it...
15522 (internal_error_function): ...here to handle reentered internal_error.
15524 2017-08-03 Richard Biener <rguenther@suse.de>
15526 PR middle-end/81148
15527 * fold-const.c (split_tree): Add minus_var and minus_con
15528 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15529 here but always use minus_*.
15530 (associate_trees): Assert we never associate with MINUS_EXPR
15531 and NULL first operand. Do not recurse for PLUS_EXPR operands
15532 when associating as MINUS_EXPR either.
15533 (fold_binary_loc): Track minus_var and minus_con.
15535 2017-08-03 Tom de Vries <tom@codesourcery.com>
15538 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15539 ACCEL_COMPILER, apply finish_options on
15540 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15542 2017-08-03 Tom de Vries <tom@codesourcery.com>
15545 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15546 function_entry_patch_area_size > 0.
15548 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15551 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15552 instead of 10??LU, perform unit multiplication in wide_int,
15553 don't change alloc_object_size_limit if the limit is larger
15556 PR tree-optimization/81655
15557 PR tree-optimization/81588
15558 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15559 the case when ranges[i].low and high are 1 for unsigned type with
15562 PR middle-end/81052
15563 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15564 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15566 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15568 * tree-vrp.h: Add include guard.
15570 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15573 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15574 (ud2): New insn pattern.
15575 * config/i386/i386.c (ix86_expand_epilogue):
15576 For naked functions, generate ud2 instead of trap insn.
15578 2017-08-02 Marek Polacek <polacek@redhat.com>
15581 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15583 2017-08-02 Tom de Vries <tom@codesourcery.com>
15584 Cesar Philippidis <cesar@codesourcery.com>
15586 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15587 Add missing edge probabilities.
15589 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15591 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15592 Correct endianness.
15594 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15596 PR middle-end/79499
15597 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15598 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15599 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15600 prologue_seq sequences - if any.
15602 2017-08-02 Richard Biener <rguenther@suse.de>
15604 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15605 via vectors if supported, integer extracts via punning if supported
15606 or otherwise vector extracts.
15608 2017-08-02 Richard Biener <rguenther@suse.de>
15610 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15612 (bitmap_insert_into_set): ... this.
15614 2017-08-02 Richard Biener <rguenther@suse.de>
15616 PR tree-optimization/81633
15618 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15620 PR tree-optimization/71752
15621 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15623 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15625 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15626 (machine_function::call_ms2sysv_pad_out): Remove field.
15627 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15628 (ix86_compute_frame_layout): Likewise.
15630 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15633 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15634 attribute with interrupt attribute.
15636 2017-08-01 Andrew Pinski <apinski@cavium.com>
15638 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15639 BIT_INSERT_EXPR's operand 1
15640 to see if the types precision matches.
15642 2017-08-01 Martin Liska <mliska@suse.cz>
15644 PR middle-end/70140
15645 * builtins.c (expand_builtin_memcpy_args): Remove.
15646 (expand_builtin_memcpy): Call newly added function
15647 expand_builtin_memory_copy_args.
15648 (expand_builtin_memcpy_with_bounds): Likewise.
15649 (expand_builtin_mempcpy): Remove last argument.
15650 (expand_builtin_mempcpy_with_bounds): Likewise.
15651 (expand_builtin_memory_copy_args): New function created from
15652 expand_builtin_mempcpy_args with small modifications.
15653 (expand_builtin_mempcpy_args): Remove.
15654 (expand_builtin_stpcpy): Remove unused argument.
15655 (expand_builtin): Likewise.
15656 (expand_builtin_with_bounds): Likewise.
15658 2017-08-01 Martin Liska <mliska@suse.cz>
15661 Make mempcpy more optimal (PR middle-end/70140).
15663 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15666 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15667 __builtin_vec_cmpne verify both arguments are compatible vectors
15668 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15669 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15670 move computation of aligned to after checking the argument types.
15674 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15675 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15678 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15679 Doug Rupp <rupp@adacore.com>
15680 Olivier Hainque <hainque@adacore.com>
15682 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15683 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15685 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15686 for TARGET_OS_CPP_BUILTIN.
15687 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15688 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15690 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15691 passing required abi options to the assembler for EABI configurations.
15692 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15693 of .text.hot and .text.unlikely sections for kernel modules when
15694 using ARM style exceptions.
15695 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15696 options. Add EXTRA_CC1_SPEC.
15697 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15699 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15701 (ARM_TARGET2_DWARF_FORMAT): Define.
15702 * config/arm/t-vxworks: Adjust multilib control to removal of the
15703 Diab command line options.
15705 2017-08-01 Martin Liska <mliska@suse.cz>
15707 PR gcov-profile/81561
15708 * gcov.c (unblock): Make unblocking safe as we need to preserve
15709 index correspondence of blocks and block_lists.
15711 2017-08-01 Richard Biener <rguenther@suse.de>
15713 PR tree-optimization/81181
15714 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15715 (compute_antic): ... end of iteration here.
15717 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15719 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15720 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15721 (ftree-slp-vectorize): Likewise.
15722 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15723 can no longer be set independent of flag_tree_loop_vectorize.
15724 * omp-general.c (emp_max_vf): Likewise.
15725 * opts.c (enable_fdo_optimizations): Remove references to
15726 flag_tree_vectorize, these are now implicit.
15727 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15728 and leave it for the options machinery.
15730 2017-08-01 Martin Liska <mliska@suse.cz>
15732 PR middle-end/70140
15733 * builtins.c (expand_builtin_memcpy_args): Remove.
15734 (expand_builtin_memcpy): Call newly added function
15735 expand_builtin_memory_copy_args.
15736 (expand_builtin_memcpy_with_bounds): Likewise.
15737 (expand_builtin_mempcpy): Remove last argument.
15738 (expand_builtin_mempcpy_with_bounds): Likewise.
15739 (expand_builtin_memory_copy_args): New function created from
15740 expand_builtin_mempcpy_args with small modifications.
15741 (expand_builtin_mempcpy_args): Remove.
15742 (expand_builtin_stpcpy): Remove unused argument.
15743 (expand_builtin): Likewise.
15744 (expand_builtin_with_bounds): Likewise.
15746 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15749 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15750 print "ds:" only for immediates in generic address space.
15752 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15755 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15756 (ix86_function_ok_for_sibcall): Return false for naked functions.
15758 2017-08-01 Richard Biener <rguenther@suse.de>
15760 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15761 (compute_antic): Seed worklist with exit block predecessors.
15762 * cfganal.c (dfs_find_deadend): For a cycle return the source
15763 of the edge closing it.
15765 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15767 * config/aarch64/aarch64.c
15768 (aarch64_can_const_movi_rtx_p): Move 0 check.
15770 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15772 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15773 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15775 * match.pd: Ditto in address comparison pattern.
15777 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15779 PR tree-optimization/81627
15780 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15781 closed ssa form for store-store chain.
15783 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15785 PR tree-optimization/81620
15786 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15787 for store-store chain.
15789 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15791 PR tree-optimization/81588
15792 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15793 ranges[i].in_p, invert comparison code ccode. For >/>=,
15794 swap rhs1 and rhs2 and comparison code unconditionally,
15795 for </<= don't do that. Don't swap rhs1/rhs2 again if
15796 ranges[i].in_p, instead invert comparison code ccode if
15797 opcode or oe->rank is BIT_IOR_EXPR.
15800 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15801 a direct optab to conversion optab.
15802 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15803 with GET_MODE_INNER as last argument instead of optab_handler.
15804 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15805 vector extraction if possible and optab is available.
15806 * expr.c (store_constructor): Use convert_optab_handler instead
15807 of optab_handler. Use vector initialization from smaller
15808 vectors if possible and optab is available.
15809 * tree-vect-stmts.c (vectorizable_load): Likewise.
15810 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15811 now have two modes.
15812 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15813 of vec_init from half-sized vectors with the same element mode.
15814 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15815 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15816 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15817 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15818 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15819 after mode in gen_vec_extract* calls.
15820 (vec_extract<mode>): Renamed to ...
15821 (vec_extract<mode><ssescalarmodelower>): ... this.
15822 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15823 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15824 element mode after mode in gen_vec_init* calls.
15825 (VEC_INIT_HALF_MODE): New mode iterator.
15826 (vec_init<mode>): Renamed to ...
15827 (vec_init<mode><ssescalarmodelower>): ... this.
15828 (vec_init<mode><ssehalfvecmodelower>): New expander.
15829 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15830 (vec_extractv2sfsf): ... this.
15831 (vec_initv2sf): Renamed to ...
15832 (vec_initv2sfsf): ... this.
15833 (vec_extractv2si): Renamed to ...
15834 (vec_extractv2sisi): ... this.
15835 (vec_initv2si): Renamed to ...
15836 (vec_initv2sisi): ... this.
15837 (vec_extractv4hi): Renamed to ...
15838 (vec_extractv4hihi): ... this.
15839 (vec_initv4hi): Renamed to ...
15840 (vec_initv4hihi): ... this.
15841 (vec_extractv8qi): Renamed to ...
15842 (vec_extractv8qiqi): ... this.
15843 (vec_initv8qi): Renamed to ...
15844 (vec_initv8qiqi): ... this.
15845 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15846 (vec_init<mode>): Renamed to ...
15847 (vec_init<mode><VEC_base_l>): ... this.
15848 (vec_extract<mode>): Renamed to ...
15849 (vec_extract<mode><VEC_base_l>): ... this.
15850 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15851 (vec_initv2sfsf): ... this.
15852 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15853 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15854 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15855 element mode after mode in gen_vec_init* calls.
15856 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15857 (vec_init<mode><Vel>): ... this.
15858 (vec_extract<mode>): Renamed to ...
15859 (vec_extract<mode><Vel>): ... this.
15860 * config/aarch64/iterators.md (Vel): New mode attribute.
15861 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15862 Add element mode after mode in gen_vec_extract* calls.
15863 * config/s390/vector.md (non_vec_l): New mode attribute.
15864 (vec_extract<mode>): Renamed to ...
15865 (vec_extract<mode><non_vec_l>): ... this.
15866 (vec_init<mode>): Renamed to ...
15867 (vec_init<mode><non_vec_l>): ... this.
15868 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15869 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15871 * config/arm/iterators.md (V_elem_l): New mode attribute.
15872 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15873 (vec_extract<mode><V_elem_l>): ... this.
15874 (vec_extractv2di): Renamed to ...
15875 (vec_extractv2didi): ... this.
15876 (vec_init<mode>): Renamed to ...
15877 (vec_init<mode><V_elem_l>): ... this.
15878 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15879 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15880 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15881 Add element mode after gen_vec_extract* calls.
15882 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15883 (vec_init<mode><unitmode>): ... this.
15884 (vec_extract<mode>): Renamed to ...
15885 (vec_extract<mode><unitmode>): ... this.
15886 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15887 (vec_init<mode><unitmode>): ... this.
15888 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15889 (vec_initv2sfsf): ... this.
15890 (vec_extractv2sf): Renamed to ...
15891 (vec_extractv2sfsf): ... this.
15892 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15893 Add element mode after gen_vec_extract* calls.
15894 * config/mips/mips.md (unitmode): New mode iterator.
15895 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15896 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15897 * config/spu/spu.md (inner_l): New mode attribute.
15898 (vec_init<mode>): Renamed to ...
15899 (vec_init<mode><inner_l>): ... this.
15900 (vec_extract<mode>): Renamed to ...
15901 (vec_extract<mode><inner_l>): ... this.
15902 * config/sparc/sparc.md (veltmode): New mode iterator.
15903 (vec_init<VMALL:mode>): Renamed to ...
15904 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15905 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15906 (vec_initv2sisi): ... this.
15907 (vec_initv2sf): Renamed to ...
15908 (vec_initv2sfsf): ... this.
15909 (vec_extractv2sf): Renamed to ...
15910 (vec_extractv2sfsf): ... this.
15911 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15912 (vec_init<mode>): Renamed to ...
15913 (vec_init<mode><VEC_base_l>): ... this.
15914 (vec_extract<mode>): Renamed to ...
15915 (vec_extract<mode><VEC_base_l>): ... this.
15916 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15917 (vec_initv2sfsf): ... this.
15918 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15919 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15920 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15921 gen_vec_init* calls.
15923 2017-08-01 Richard Biener <rguenther@suse.de>
15925 PR tree-optimization/81297
15926 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15927 TREE_OVERFLOW from INTEGER_CSTs.
15929 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15931 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15933 2017-07-31 Carl Love <cel@us.ibm.com>
15935 * config/rs6000/rs6000-c: Add support for built-in functions
15936 vector signed char vec_xl_be (signed long long, signed char *);
15937 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15938 vector signed int vec_xl_be (signed long long, signed int *);
15939 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15940 vector signed long long vec_xl_be (signed long long, signed long long *);
15941 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15942 vector signed short vec_xl_be (signed long long, signed short *);
15943 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15944 vector double vec_xl_be (signed long long, double *);
15945 vector float vec_xl_be (signed long long, float *);
15946 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15947 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15948 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15950 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15951 (altivec_expand_builtin): Add switch statement to call
15952 altivec_expand_xl_be for each builtin.
15953 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15954 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15955 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15956 __builtin_vsx_le_be_v16qi.
15957 * doc/extend.texi: Update the built-in documentation file for the
15958 new built-in functions.
15960 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15963 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15965 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15967 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15969 * config.gcc: Add z14.
15970 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15971 CPU model numbers for z13s and z14.
15972 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15974 * config/s390/s390-opts.h (enum processor_type): Rename
15975 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15976 * config/s390/s390.c (processor_table): Add field for CPU name to
15977 be passed to Binutils.
15978 (s390_asm_output_machine_for_arch): Use the new field in
15979 processor_table for Binutils.
15980 (s390_expand_builtin): Replace arch12 with z14.
15981 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15982 (s390_get_sched_attrmask): Likewise.
15983 (s390_get_unit_mask): Likewise.
15984 * config/s390/s390.opt: Add z14 to processor_type enum.
15986 2017-07-31 Martin Jambor <mjambor@suse.cz>
15989 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15990 regardless of optimization level.
15992 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15993 Martin Liska <mliska@suse.cz>
15995 * predict.def: Remove old comment and adjust probability.
15996 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15997 PREDICT statements.
15999 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16002 * config/i386/i386.c (ix86_function_naked): New function.
16003 (ix86_can_use_return_insn_p): Return false for naked functions.
16004 (ix86_expand_prologue): Skip prologue for naked functions.
16005 (ix86_expand_epilogue): Skip epilogue for naked functions
16006 and emit trap instruction.
16007 (ix86_warn_func_return): New function.
16008 (ix86_attribute_table): Add "naked" attribute specification.
16009 (TARGET_WARN_FUNC_RETURN): Define.
16010 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16012 2017-07-31 Martin Liska <mliska@suse.cz>
16014 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16015 (dump_gimple_bb_header): Always dump BB info.
16016 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16018 2017-07-31 Martin Liska <mliska@suse.cz>
16021 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16022 also with current_function_decl non-null equality.
16024 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16027 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16028 change type to the element type, instead add eltype variable and
16029 use it where we are interested in the element type.
16031 PR tree-optimization/81603
16032 * ipa-polymorphic-call.c
16033 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16034 offset arithmetic in offset_int, bail out if the resulting bit offset
16035 doesn't fit into shwi.
16037 2017-07-31 Martin Liska <mliska@suse.cz>
16039 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16040 (gimplify_save_expr): Fix comment.
16042 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16045 * config/i386/i386.c (ix86_function_arg): Update arguments for
16047 (ix86_compute_frame_layout): Set the initial stack offset to
16048 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16049 INCOMING_FRAME_SP_OFFSET.
16050 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16051 stack before exception handler returns.
16052 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16053 the 'ERROR_CODE' for exception handler.
16055 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16057 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16058 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16059 (ASM_OUTPUT_REG_POP): Ditto.
16060 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16061 instead of asm_fprintf to output pure string.
16063 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16065 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16066 to imported_module_or_decl hook.
16067 (debug_nothing_tree_tree_tree_bool): Remove.
16068 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16069 * debug.c (do_nothing_debug_hooks): Use
16070 debug_nothing_tree_tree_tree_bool_bool instead of
16071 debug_nothing_tree_tree_tree_bool.
16072 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16073 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16074 * sdbout.c (sdb_debug_hooks): Likewise.
16075 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16076 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16078 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16079 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16080 attribute, don't add anything.
16082 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16084 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16085 (fold_build2_stat_loc): Likewise.
16086 (fold_build3_stat_loc): Likewise.
16087 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16088 (fold_build1_loc): Remove macro.
16089 (fold_build2_loc): Likewise.
16090 (fold_build3_loc): Likewise.
16092 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16094 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16095 (gimple_build_debug_bind_source_stat): Likewise.
16096 * gimple.h (gimple_build_debug_bind): Remove macro.
16097 (gimple_build_debug_bind_source): Likewise.
16099 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16101 * bitmap.c (bitmap_alloc): Adjust.
16102 (bitmap_gc_alloc): Likewise.
16103 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16105 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16107 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16108 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16109 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16110 (bitmap_gc_alloc_stat): Likewise.
16111 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16113 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16115 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16116 * rtl.h (shallow_copy_rtx): Remove macro.
16118 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16120 * emit-rtl.c (gen_raw_REG): Adjust.
16121 * gengenrtl.c (gendef): Likewise.
16122 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16123 * rtl.h (rtx_alloc): Remove macro.
16125 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16127 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16128 (build_tree_list_stat): Likewise.
16129 * tree.h (build_tree_list): Remove macro.
16130 (build_tree_list_vec): Likewise.
16132 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16134 * tree.c (make_vector_stat): Remove _stat from name.
16135 (build_vector_stat): Likewise.
16136 * tree.h (make_vector_stat): Remove macro.
16137 (build_vector_stat): Likewise.
16139 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16141 * tree.h (build_var_debug_value): Remove prototype.
16143 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16145 * tree.c (tree_cons_stat): Remove _stat from name.
16146 * tree.h (tree_cons): Remove macro.
16148 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16150 * tree.c (build_vl_exp_stat): Remove _stat from name.
16151 * tree.h (build_vl_exp): Remove macro.
16153 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16155 * tree.c (build_decl_stat): Remove _stat from name.
16156 * tree.h (build_decl): Remove macro.
16158 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16160 * gimple.c (gimple_build_with_ops_stat): Adjust.
16161 (gimple_alloc_stat): Remove _stat from name.
16162 * gimple.h (gimple_alloc): Remove macro.
16164 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16166 * tree.c (make_tree_vec_stat): Remove _stat from name.
16167 (grow_tree_vec_stat): Likewise.
16168 * tree.h (make_tree_vec_stat): Adjust prototype.
16169 (grow_tree_vec_stat): Likewise.
16170 (make_tree_vec): Remove macro.
16171 (grow_tree_vec): Likewise.
16173 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16175 * fold-const.c (fold_build1_stat_loc): Adjust.
16176 (fold_build2_stat_loc): Likewise.
16177 (fold_build3_stat_loc): Likewise.
16178 * tree.c (build0_stat): Remove _stat from name.
16179 (build1_stat): Likewise.
16180 (build2_stat): Likewise.
16181 (build3_stat): Likewise.
16182 (build4_stat): Likewise.
16183 (build5_stat): Likewise.
16184 * tree.h (build1_loc): Remove macro, and rename _stat function
16186 (build2_loc): Likewise.
16187 (build3_loc): Likewise.
16188 (build4_loc): Likewise.
16189 (build5_loc): Likewise.
16191 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16193 * tree.c (make_int_cst_stat): Remove _stat from name.
16194 * tree.h (make_int_cst_stat): Adjust prototype.
16195 (make_int_cst): Remove macro.
16197 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16199 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16200 * tree.h (make_tree_binfo_stat): Adjust prototype.
16201 (make_tree_binfo): Remove.
16203 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16205 * tree.c (copy_node_stat): Rename to copy_node.
16206 (build_distinct_type_copy): Adjust.
16207 * tree.h (copy_node_stat): Adjust prototype.
16208 (copy_node): Remove macro.
16210 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16212 * tree.c (make_node_stat): rename to make_node.
16213 (build_tree_list_stat): Adjust.
16214 (build0_stat): Likewise.
16215 (build2_stat): Likewise.
16216 (build3_stat): Likewise.
16217 (build4_stat): Likewise.
16218 (build5_stat): Likewise.
16219 (build_decl_stat): Likewise.
16220 * tree.h (make_node_stat): Adjust prototype.
16221 (make_node): remove macro.
16223 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16225 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16226 (PPC_FEATURE2_SCV): Likewise.
16227 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16229 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16231 * config/aarch64/aarch64.c
16232 (aarch64_internal_mov_immediate): Add new special pattern.
16233 * config/aarch64/aarch64.md (*movdi_aarch64):
16234 Add reg/32bit const mov case.
16236 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16237 Richard Sandiford <richard.sandiford@linaro.org>
16239 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16240 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16241 Add integer and movi cases.
16242 (movi-split-hf-df-sf split, fp16): New.
16243 (enabled): Added TARGET_FP_F16INST.
16244 * config/aarch64/iterators.md (GPF_HF): New.
16245 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16247 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16249 * config/aarch64/aarch64.c
16250 (aarch64_simd_container_mode): Add prototype.
16251 (aarch64_expand_mov_immediate): Add HI support.
16252 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16253 (aarch64_can_const_movi_rtx_p): New.
16254 (aarch64_preferred_reload_class):
16255 Remove restrictions of using FP registers for certain SIMD operations.
16256 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16257 (aarch64_valid_floating_const): Add integer move validation.
16258 (aarch64_simd_imm_scalar_p): Remove.
16259 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16260 (aarch64_legitimate_constant_p): Expand list of supported cases.
16261 * config/aarch64/aarch64-protos.h
16262 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16263 (aarch64_reinterpret_float_as_int): New.
16264 (aarch64_simd_imm_scalar_p): Remove.
16265 * config/aarch64/constraints.md (Uvi): New.
16266 (Dd): Split into Ds and new Dd.
16267 * config/aarch64/aarch64.md (*movsi_aarch64):
16269 (*movdi_aarch64): Add SIMD mov case.
16271 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16273 * tree-predcom.c: (struct chain): Handle store-store chain in which
16274 stores for elimination only store loop invariant values.
16275 (execute_pred_commoning_chain): Ditto.
16276 (prepare_initializers_chain_store_elim): Ditto.
16277 (prepare_finalizers): Ditto.
16278 (is_inv_store_elimination_chain): New function.
16279 (initialize_root_vars_store_elim_1): New function.
16281 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16283 * tree-predcom.c: Revise general description of the pass.
16284 (enum chain_type): New enum type for store elimination.
16285 (struct chain): New field supporting store elimination.
16286 (struct component): Ditto.
16287 (dump_chain): Dump store-stores chain.
16288 (release_chain): Release resources.
16289 (split_data_refs_to_components): Compute and create component
16290 contains only stores for elimination.
16291 (get_chain_last_ref_at): New function.
16292 (make_invariant_chain): Initialization.
16293 (make_rooted_chain): Specify chain type in parameter and record it.
16294 (add_looparound_copies): Skip for store-stores chain.
16295 (determine_roots_comp): Compute type of chain and pass it to
16297 (initialize_root_vars_store_elim_2): New function.
16298 (finalize_eliminated_stores): New function.
16299 (remove_stmt): Handle store for elimination.
16300 (execute_pred_commoning_chain): Execute predictive commoning on
16301 store-store chains.
16302 (determine_unroll_factor): Skip unroll for store-stores chain.
16303 (prepare_initializers_chain_store_elim): New function.
16304 (prepare_initializers_chain): Hanlde store-store chain.
16305 (prepare_finalizers_chain, prepare_finalizers): New function.
16306 (tree_predictive_commoning_loop): Return integer value indicating
16307 if loop is unrolled or lcssa form is corrupted.
16308 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16310 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16312 * tree-predcom.c (initialize_root): Delete.
16313 (execute_pred_commoning_chain): Initialize root vars and replace
16314 reference of non-combined chain directly, rather than call above
16317 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16319 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16320 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16322 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16324 * tree-predcom.c (struct chain): New field init_seq.
16325 (release_chain): Release init_seq.
16326 (prepare_initializers_chain): Record intialization stmts in above
16328 (insert_init_seqs): New function.
16329 (tree_predictive_commoning_loop): Call insert_init_seqs.
16331 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16333 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16335 2017-07-28 Richard Biener <rguenther@suse.de>
16337 * match.pd: Remove superfluous :c.
16338 * genmatch.c (simplify::id): Add member.
16339 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16341 (current_id): New global.
16342 (dt_node::parent): Move from ...
16343 (dt_operand::parent): ... here. Add for_id member.
16344 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16345 (decision_tree::find_node): Relax order requirement when
16346 merging DT_TRUE nodes to ones inbetween the current simplify
16347 and the one we try to merge with. Add diagnostic whenever
16348 we need to enforce pattern order by not merging.
16349 (decision_tree::insert): Set current_id.
16350 (decision_tree::print_node): Dump parent node and for_id.
16351 (parser::last_id): Add member.
16352 (parser::push_simplify): Assign unique id.
16353 (parser::parser): Initialize last_id.
16355 2017-07-28 Martin Liska <mliska@suse.cz>
16358 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16359 gimple_build_debug_bind.
16361 2017-07-28 Richard Biener <rguenther@suse.de>
16363 PR tree-optimization/81502
16364 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16366 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16368 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16369 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16370 for BIT_FIELD_REF args.
16371 * fold-const.c (make_bit_field_ref): Likewise.
16372 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16374 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16377 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16378 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16379 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16380 Or it into SANITIZER_UNDEFINED.
16381 * ubsan.c: Include gimple-fold.h and varasm.h.
16382 (ubsan_expand_ptr_ifn): New function.
16383 (instrument_pointer_overflow): New function.
16384 (maybe_instrument_pointer_overflow): New function.
16385 (instrument_object_size): Formatting fix.
16386 (pass_ubsan::execute): Call instrument_pointer_overflow
16387 and maybe_instrument_pointer_overflow.
16388 * internal-fn.c (expand_UBSAN_PTR): New function.
16389 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16390 * sanitizer.def (__ubsan_handle_pointer_overflow,
16391 __ubsan_handle_pointer_overflow_abort): New builtins.
16392 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16393 * internal-fn.def (UBSAN_PTR): New internal function.
16394 * opts.c (sanitizer_opts): Add pointer-overflow.
16395 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16396 * fold-const.c (build_range_check): Compute pointer range check in
16397 integral type if pointer arithmetics would be needed. Formatting
16400 2017-07-28 Martin Liska <mliska@suse.cz>
16403 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16404 parameters that are of a variable-length.
16406 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16408 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16410 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16411 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16412 (CRT_CALL_STATIC_FUNCTION): Likewise.
16413 (ASM_DEFAULT_SPEC): New define.
16414 (ASM_SPEC32): Likewise.
16415 (ASM_SPEC64): Likewise.
16416 (ASM_SPEC_COMMON): Likewise.
16417 (ASM_SPEC): Likewise.
16418 (INVALID_64BIT): Likewise.
16419 (LINK_OS_DEFAULT_SPEC): Likewise.
16420 (LINK_OS_SPEC32): Likewise.
16421 (LINK_OS_SPEC64): Likewise.
16422 (POWERPC_LINUX): Likewise.
16423 (PTRDIFF_TYPE): Likewise.
16424 (RESTORE_FP_PREFIX): Likewise.
16425 (RESTORE_FP_SUFFIX): Likewise.
16426 (SAVE_FP_PREFIX): Likewise.
16427 (SAVE_FP_SUFFIX): Likewise.
16428 (SIZE_TYPE): Likewise.
16429 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16430 (TARGET_64BIT): Likewise.
16431 (TARGET_64BIT): Likewise.
16432 (TARGET_AIX): Likewise.
16433 (WCHAR_TYPE_SIZE): Likewise.
16434 (WCHAR_TYPE): Undefine.
16435 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16436 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16437 (CPP_OS_RTEMS_SPEC): Delete.
16438 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16439 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16441 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16443 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16445 PR tree-optimization/81578
16446 * tree-parloops.c (build_new_reduction): Bail out if
16447 reduction_code isn't one of the standard OpenMP reductions.
16448 Move the details printing after that decision.
16450 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16452 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16453 related to reload_in_progress.
16454 (splat_input_operand): Likewise.
16455 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16457 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16459 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16460 (TARGET_INSTANTIATE_DECLS): Likewise.
16461 (legitimate_indexed_address_p): Delete reload_in_progress code.
16462 (rs6000_debug_legitimate_address_p): Likewise.
16463 (rs6000_eliminate_indexed_memrefs): Likewise.
16464 (rs6000_emit_le_vsx_store): Likewise.
16465 (rs6000_emit_move_si_sf_subreg): Likewise.
16466 (rs6000_emit_move): Likewise.
16467 (register_to_reg_type): Likewise.
16468 (rs6000_pre_atomic_barrier): Likewise.
16469 (rs6000_machopic_legitimize_pic_address): Likewise.
16470 (rs6000_allocate_stack_temp): Likewise.
16471 (rs6000_address_for_fpconvert): Likewise.
16472 (rs6000_address_for_altivec): Likewise.
16473 (rs6000_secondary_memory_needed_rtx): Delete function.
16474 (rs6000_check_sdmode): Likewise.
16475 (rs6000_alloc_sdmode_stack_slot): Likewise.
16476 (rs6000_instantiate_decls): Likewise.
16477 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16478 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16479 Delete reload_in_progress.
16480 (*vec_reload_and_plus_<mptrsize>): Likewise.
16481 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16482 (vsx_div_v2di): Likewise.
16483 (vsx_udiv_v2di): Likewise.
16485 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16487 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16488 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16489 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16490 (rs6000_debug_reg_global): Delete print of LRA status.
16491 (rs6000_option_override_internal): Delete dead LRA related code.
16492 (rs6000_lra_p): Delete function.
16493 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16495 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16497 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16498 * config/riscv/rtems.h: New file.
16500 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16501 Sudakshina Das <sudi.das@arm.com>
16503 * config/aarch64/aarch64.md
16504 (define_split for and<mode>3nr_compare): Move
16505 non aarch64_logical_operand to a register.
16506 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16507 register immediate operand to a register.
16508 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16510 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16512 PR middle-end/81564
16513 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16515 2017-07-27 Richard Biener <rguenther@suse.de>
16517 PR tree-optimization/81573
16518 PR tree-optimization/81494
16519 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16520 multi defuse cycle case.
16522 2017-07-27 Richard Biener <rguenther@suse.de>
16524 PR tree-optimization/81571
16525 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16528 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16530 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16531 earlier and only if MASK_FPU is set. Adjust formatting.
16533 2017-07-27 Martin Liska <mliska@suse.cz>
16535 * opt-functions.awk: Add validation of value of Init.
16536 * optc-gen.awk: Pass new argument.
16538 2017-07-27 Martin Liska <mliska@suse.cz>
16540 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16541 Fix wrong condition.
16543 2017-07-27 Martin Liska <mliska@suse.cz>
16545 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16546 BBs and edges seen by autoFDO.
16548 2017-07-27 Richard Biener <rguenther@suse.de>
16550 PR tree-optimization/81502
16551 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16552 with incompatible but same sized type.
16553 (execute_update_addresses_taken): Likewise.
16555 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16557 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16558 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16560 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16563 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16564 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16565 Change s_operand to memory_operand.
16567 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16569 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16570 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16571 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16572 Emit instructions rather than returning an expression. Handle TFmode
16573 and KFmode by casting to TImode.
16574 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16575 (rs6000_emit_le_vsx_store): Likewise.
16576 * config/rs6000/vsx.md (VSX_TI): New iterator.
16577 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16578 (*vsx_le_undo_permute_<mode>): Likewise.
16579 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16580 emit the split sequence.
16581 (*vsx_le_perm_store_<mode>): Likewise.
16583 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16585 PR tree-optimization/81555
16586 PR tree-optimization/81556
16587 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16588 if true, force CHANGED for the recursive invocation.
16589 (reassociate_bb): Remember original length of ops array, pass
16590 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16592 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16593 attributes for noipa attribute. For naked attribute use
16594 lookup_attribute first before lookup_attribute_spec.
16595 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16597 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16598 for functions with noipa attribute.
16599 (cgraph_externally_visible_p): Return true for functions with noipa
16601 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16602 for functions with noipa attribute.
16603 * doc/extend.texi: Document noipa function attribute.
16604 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16605 also for functions with noipa attribute.
16606 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16608 2017-07-26 Andrew Pinski <apinski@cavium.com>
16610 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16611 vec_unalign_load_cost and vec_unalign_store_cost.
16613 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16615 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16616 -mvsx-small-integer option.
16617 (ISA_3_0_MASKS_IEEE): Likewise.
16618 (OTHER_VSX_VECTOR_MASKS): Likewise.
16619 (POWERPC_MASKS): Likewise.
16620 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16621 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16622 code, only testing for DImode being allowed in non-VSX floating
16624 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16625 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16627 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16628 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16629 TARGET_P8_VECTOR test.
16630 (rs6000_secondary_reload_simple_move): Likewise.
16631 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16632 since TARGET_P9_VECTOR was already tested.
16633 (rs6000_opt_masks): Remove -mvsx-small-integer.
16634 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16635 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16637 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16638 test for TARGET_VEXTRACTUB was used, and that uses
16640 (p9 extract splitter): Likewise.
16641 (vsx_extract_<mode>_di_p9): Likewise.
16642 (vsx_extract_<mode>_store_p9): Likewise.
16643 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16644 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16645 the elimination of TARGET_VSX_SMALL_INTEGER.
16646 (vsx_extract_<mode>_p8): Likewise.
16647 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16648 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16649 (vsx_set_<mode>_p9): Likewise.
16650 (vsx_set_v4sf_p9): Likewise.
16651 (vsx_set_v4sf_p9_zero): Likewise.
16652 (vsx_insert_extract_v4sf_p9): Likewise.
16653 (vsx_insert_extract_v4sf_p9_2): Likewise.
16654 * config/rs6000/rs6000.md (sign extend splitter): Change
16655 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16656 (floatsi<mode>2_lfiwax_mem): Likewise.
16657 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16658 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16659 since a test for TARGET_P9_VECTOR was used.
16660 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16661 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16662 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16663 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16664 TARGET_P8_VECTOR test.
16665 (fix_trunc<mode>si2_stfiwx): Likewise.
16666 (fix_trunc<mode>si2_internal): Likewise.
16667 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16668 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16670 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16671 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16672 TARGET_P8_VECTOR test.
16673 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16674 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16675 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16677 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16678 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16679 since a test for TARGET_P9_VECTOR was used.
16680 (splitter for loading small constants): Likewise.
16682 2017-07-26 Andrew Pinski <apinski@cavium.com>
16684 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16687 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16690 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16691 (fp_valid_at): Likewise.
16693 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16695 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16696 (qdf24xx_addrcost_table): Likewise.
16697 (cortexa57_tunings): Update to use generic_branch_cost.
16698 (cortexa72_tunings): Likewise.
16699 (cortexa73_tunings): Likewise.
16700 (qdf24xx_tunings): Likewise.
16702 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16704 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16705 (thunderx2t99_branch_cost): Likewise.
16706 (cortexa35_tunings): Update to use generic_branch_cost.
16707 (cortexa53_tunings): Likewise.
16708 (cortexa57_tunings): Likewise.
16709 (cortexa72_tunings): Likewise.
16710 (cortexa73_tunings): Likewise.
16711 (thunderx2t99_tunings): Likewise.
16713 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16715 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16716 (sparc_option_override): Honour MASK_FSMULD.
16717 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16718 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16719 * config/sparc/sparc.opt (mfsmuld): New option.
16720 * doc/invoke.texi (mfsmuld): Document option.
16722 2017-07-26 Marek Polacek <polacek@redhat.com>
16724 PR middle-end/70992
16725 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16727 2017-07-26 Richard Biener <rguenther@suse.de>
16729 * gimple-match-head.c (do_valueize): Return OP if valueize
16731 (get_def): New helper to get at the def stmt of a SSA name
16732 if valueize allows.
16733 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16734 do_valueize to get at the def stmt.
16735 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16737 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16739 PR middle-end/46932
16740 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16742 2017-07-26 Martin Liska <mliska@suse.cz>
16745 * function.c (expand_function_start): Make expansion of
16746 nonlocal_goto_save_area after parm_birth_insn.
16748 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16750 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16751 from all CPU target flags enable members.
16753 2017-07-26 Richard Biener <rguenther@suse.de>
16755 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16756 (decision_tree::gen): Make 'type' const.
16757 (write_predicate): Likewise.
16759 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16761 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16762 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16763 (rs6000_option_override_internal): Likewise.
16764 (rs6000_expand_vector_set): Likewise.
16765 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16766 (TARGET_UPPER_REGS_SF): Likewise.
16767 (TARGET_UPPER_REGS_DI): Likewise.
16768 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16769 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16770 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16771 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16772 (Splitters for DI constants in Altivec registers): Likewise.
16773 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16774 (vsx_set_v4sf_p9): Likewise.
16775 (vsx_set_v4sf_p9_zero): Likewise.
16776 (vsx_insert_extract_v4sf_p9): Likewise.
16777 (vsx_insert_extract_v4sf_p9_2): Likewise.
16779 2017-07-25 Carl Love <cel@us.ibm.com>
16781 * doc/extend.texi: Update the built-in documentation file for the
16782 existing built-in functions
16783 vector signed char vec_cnttz (vector signed char);
16784 vector unsigned char vec_cnttz (vector unsigned char);
16785 vector signed short vec_cnttz (vector signed short);
16786 vector unsigned short vec_cnttz (vector unsigned short);
16787 vector signed int vec_cnttz (vector signed int);
16788 vector unsigned int vec_cnttz (vector unsigned int);
16789 vector signed long long vec_cnttz (vector signed long long);
16790 vector unsigned long long vec_cnttz (vector unsigned long long);
16792 2017-07-25 Andrew Pinski <apinski@cavium.com>
16794 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16795 accesses where the use is for the first operand of a BIT_INSERT.
16797 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16800 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16801 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16803 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16805 * config/i386/gstabs.h: Delete.
16806 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16808 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16810 * config/i386/i386.c (ix86_decompose_address): Do not check for
16811 register RTX when looking at index_reg or base_reg.
16812 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16814 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16816 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16817 to update EH info here.
16819 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16821 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16823 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16825 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16827 2017-07-25 Torsten Duwe <duwe@suse.de>
16829 * common.opt: Introduce -fpatchable-function-entry
16830 command line option, and its variables function_entry_patch_area_size
16831 and function_entry_patch_area_start.
16832 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16833 including a two-value parser.
16834 * target.def (print_patchable_function_entry): New target hook.
16835 * targhooks.h (default_print_patchable_function_entry): New function.
16836 * targhooks.c (default_print_patchable_function_entry): Likewise.
16837 * toplev.c (process_options): Switch off IPA-RA if
16838 patchable function entries are being generated.
16839 * varasm.c (assemble_start_function): Look at the
16840 patchable-function-entry command line switch and current
16841 function attributes and maybe generate NOP instructions by
16842 calling the print_patchable_function_entry hook.
16843 * doc/extend.texi: Document patchable_function_entry attribute.
16844 * doc/invoke.texi: Document -fpatchable_function_entry
16845 command line option.
16846 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16848 * doc/tm.texi: Re-generate.
16850 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16853 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16854 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16856 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16858 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16860 2017-07-25 Richard Biener <rguenther@suse.de>
16862 PR tree-optimization/81455
16863 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16864 not walk in cycles when looking for guards.
16866 2017-07-25 Richard Biener <rguenther@suse.de>
16868 PR tree-optimization/81529
16869 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16870 when optimizing backedge uses.
16872 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16874 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16876 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16877 to dl_section_ref. On AIX, append an expression to subtract
16878 the size of the section length to dl_section_ref.
16880 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16882 * configure.ac: If any of the config.* scripts fail, exit 1.
16883 * configure: Regenerate.
16885 2017-07-25 Richard Biener <rguenther@suse.de>
16887 PR middle-end/81546
16888 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16889 of immediate uses, be more verbose on errors.
16891 2017-07-25 Richard Biener <rguenther@suse.de>
16893 PR tree-optimization/81510
16894 * tree-vect-loop.c (vect_is_simple_reduction): When the
16895 reduction stmt is not inside the loop bail out.
16897 2017-07-25 Richard Biener <rguenther@suse.de>
16899 PR tree-optimization/81303
16900 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16901 profitability check against LOOP_VINFO_NITERSM1.
16903 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16905 * domwalk.c (cmp_bb_postorder): Simplify.
16906 (sort_bbs_postorder): New function. Use it...
16907 (dom_walker::walk): ...here to optimize common cases.
16909 2017-07-25 Martin Liska <mliska@suse.cz>
16912 * ipa-visibility.c (function_and_variable_visibility): Make the
16913 redirection just on target that supports aliasing.
16914 Fix GNU coding style.
16916 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16919 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16921 * config/arm/rtems.h: Likewise.
16922 * config/bfin/rtems.h: Likewise.
16923 * config/i386/rtemself.h: Likewise.
16924 * config/lm32/rtems.h: Likewise.
16925 * config/m32c/rtems.h: Likewise.
16926 * config/m68k/rtemself.h: Likewise.
16927 * config/microblaze/rtems.h: Likewise.
16928 * config/mips/rtems.h: Likewise.
16929 * config/moxie/rtems.h: Likewise.
16930 * config/nios2/rtems.h: Likewise.
16931 * config/powerpcspe/rtems.h: Likewise.
16932 * config/rs6000/rtems.h: Likewise.
16933 * config/rtems.h: Likewise.
16934 * config/sh/rtems.h: Likewise.
16935 * config/sh/rtemself.h: Likewise.
16936 * config/sparc/rtemself.h: Likewise.
16938 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16941 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16942 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16943 * tree-ssa-structalias.c (alias_get_name): Same.
16945 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16948 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16949 instructions if no du chain is found.
16951 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16953 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16955 2017-07-25 Richard Biener <rguenther@suse.de>
16957 PR middle-end/81505
16958 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16961 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16963 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16964 upper-regs options.
16965 (ISA_2_7_MASKS_SERVER): Likewise.
16966 (ISA_3_0_MASKS_IEEE): Likewise.
16967 (OTHER_P8_VECTOR_MASKS): Likewise.
16968 (OTHER_VSX_VECTOR_MASKS): Likewise.
16969 (POWERPC_MASKS): Likewise.
16970 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16971 duplicate list of options.
16972 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16973 explicit -mupper-regs options.
16974 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16975 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16976 alias for -mupper-regs-df.
16977 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16978 (rs6000_init_hard_regno_mode_ok): Likewise.
16979 (rs6000_option_override_internal): Likewise.
16980 (rs6000_opt_masks): Likewise.
16981 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16982 options in terms of whether -mvsx or -mpower8-vector was used.
16983 (TARGET_UPPER_REGS_DI): Likewise.
16984 (TARGET_UPPER_REGS_SF): Likewise.
16985 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16986 -mupper-regs-* options.
16988 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16990 * passes.c (emergency_dump_function): Print some empty lines and a
16991 header before the RTL dump.
16993 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16995 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16997 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17000 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17001 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17003 2017-07-24 Carl Love <cel@us.ibm.com>
17005 * config/rs6000/rs6000-c.c: Add support for built-in functions
17006 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17007 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17008 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17009 vec_extract_fp_from_shortl): Add defines for the two builtins.
17010 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17011 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17013 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17014 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17015 * doc/extend.texi: Update the built-in documentation file for the
17016 new built-in function.
17018 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17021 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17023 * doc/generic.texi: Likewise.
17024 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17025 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17027 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17029 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17030 (aarch64_mls_elt_merge<mode>): Likewise.
17032 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17034 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17035 having __cxa_atexit.
17037 2017-07-23 Michael Collison <michael.collison@arm.com>
17039 * config/arm/arm.c (arm_option_override): Deprecate
17040 use of -mstructure-size-boundary.
17041 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17042 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17044 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17047 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17048 Reduce cost estimate for direct moves.
17050 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17053 * config/i386/i386.c (ix86_option_override_internal): Disable
17054 BMI, BMI2 and TBM instructions for -m16.
17056 2017-07-21 Carl Love <cel@us.ibm.com>
17058 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17059 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17060 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17061 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17062 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17063 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17064 VMULOSW): New enum "unspec" values.
17065 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17066 altivec_vmulosw): New patterns.
17067 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17068 VMULOSW): Add definitions.
17070 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17072 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17073 (qdf24xx): Likewise.
17074 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17075 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17076 (AARCH64_FL_V8_1): Renumber.
17077 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17078 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17079 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17080 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17081 rdma to feature modifiers list.
17083 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17085 PR middle-end/56727
17086 * ipa-visibility (function_and_variable_visibility): Convert
17087 recursive PLT call to direct call if appropriate.
17089 2017-07-21 Andrew Pinski <apinski@cavium.com>
17091 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17092 operand 1 to see if the types precision matches.
17093 * fold-const.c (operand_equal_p): Likewise.
17095 2017-07-21 Richard Biener <rguenther@suse.de>
17097 PR tree-optimization/81303
17098 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17099 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17100 (vect_peeling_hash_get_lowest_cost): Adjust.
17101 (vect_enhance_data_refs_alignment): Likewise. Use
17102 vect_get_peeling_costs_all_drs to compute the penalty for no
17103 peeling to match up costs.
17105 2017-07-21 Richard Biener <rguenther@suse.de>
17107 PR tree-optimization/81500
17108 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17109 we didn't identify a reduction path.
17111 2017-07-21 Tom de Vries <tom@codesourcery.com>
17112 Cesar Philippidis <cesar@codesourcery.com>
17114 PR gcov-profile/81442
17115 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17118 2017-07-21 Tom de Vries <tom@codesourcery.com>
17121 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17123 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17124 nvptx_override_options_after_change.
17126 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17128 * dwarf2out.c (output_file_names): Avoid double testing for
17129 dwarf_version >= 5.
17131 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17133 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17135 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17137 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17139 (try_crossjump_to_edge): Do not punt on partitioned functions.
17141 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17143 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17144 Put all BBs reachable only via paths crossing cold region to cold
17146 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17148 2016-07-21 Richard Biener <rguenther@suse.de>
17150 PR tree-optimization/81303
17151 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17152 into account prologue and epilogue iterations when raising
17153 min_profitable_iters to sth at least covering one vector iteration.
17155 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17157 * config/arm/arm.c (arm_test_cpu_arch_dat):
17160 2017-07-20 Nathan Sidwell <nathan@acm.org>
17162 Remove TYPE_METHODS.
17163 * tree.h (TYPE_METHODS): Delete.
17164 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17165 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17166 (dbxout_type_methods): Scan TYPE_FIELDS.
17167 (dbxout_type): Don't check TYPE_METHODS here.
17168 * function.c (use_register_for_decl): Always ignore register for
17169 class types when not optimizing.
17170 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17171 * tree.c (free_lang_data_in_type): Stitch out member functions and
17172 templates from TYPE_FIELDS.
17173 (build_distinct_type_copy, verify_type_variant,
17174 verify_type): Member fns are on TYPE_FIELDS.
17175 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17176 * tree-pretty-print.c (dump_generic_node): Likewise.
17178 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17181 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17182 V2TImode and V4TImode.
17183 (ix86_expand_vector_extract): Likewise.
17184 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17185 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17186 (ssescalarmode): Handle V4TImode and V2TImode.
17187 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17188 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17189 (VEXTRACTI128_MODE): New mode iterator.
17190 (splitter for *vec_extractv?ti first element): New.
17191 (VEC_INIT_MODE): New mode iterator.
17192 (vec_init<mode>): Consolidate 3 expanders into one using
17193 VEC_INIT_MODE mode iterator.
17195 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17197 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17198 non_spilled_static_chain_regno_p.
17200 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17202 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17204 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17206 * bb-reorder.c (connect_traces): Allow copying of blocks within
17209 2017-07-20 Richard Biener <rguenther@suse.de>
17211 * gimple.h (gimple_phi_result): Add gphi * overload.
17212 (gimple_phi_result_ptr): Likewise.
17213 (gimple_phi_arg): Likewise. Adjust index assert to only
17214 allow actual argument accesses rather than all slots available
17216 (gimple_phi_arg_def): Add gphi * overload.
17217 * tree-phinodes.c (make_phi_node): Initialize only actual
17219 (resize_phi_node): Clear memory not covered by old node,
17220 do not initialize excess argument slots.
17221 (reserve_phi_args_for_new_edge): Initialize new argument slot
17224 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17226 PR tree-optimization/81388
17228 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17230 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17231 by removing computation of may_be_zero.
17233 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17234 Tom de Vries <tom@codesourcery.com>
17236 PR middle-end/81030
17237 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17238 when gimple level profile disagrees with what RTL expander did.
17240 2017-07-20 Richard Biener <rguenther@suse.de>
17242 PR tree-optimization/61171
17243 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17244 (vect_analyze_stmt): Add slp instance parameter.
17245 (vectorizable_reduction): Likewise.
17246 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17247 (vect_is_simple_reduction): Deal with chains not detected
17248 as SLP reduction chain, specifically not properly associated
17249 chains containing a mix of plus/minus.
17250 (get_reduction_op): Remove.
17251 (get_initial_defs_for_reduction): Simplify, pass in whether
17252 this is a reduction chain, pass in the SLP node for the PHIs.
17253 (vect_create_epilog_for_reduction): Get the SLP instance as
17255 (vectorizable_reduction): Get the SLP instance as arg.
17256 During analysis remember the SLP node with the PHIs in the
17257 instance. Simplify getting at the vectorized reduction PHIs.
17258 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17259 through SLP instance.
17260 (vect_slp_analyze_operations): Likewise.
17261 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17262 (vect_transform_stmt): Likewise.
17264 2017-07-20 Tom de Vries <tom@codesourcery.com>
17266 PR tree-optimization/81489
17267 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17268 read of phi arg location to before loop that modifies phi.
17270 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17272 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17275 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17277 PR middle-end/81331
17278 * except.c (execute): Fix ordering issue.
17280 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17282 PR rtl-optimization/81423
17283 * combine.c (make_compound_operation_int): Don't try to optimize
17284 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17286 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17288 PR rtl-optimization/81423
17289 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17290 with a constant that is -1 in the truncated to mode.
17292 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17294 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17295 (determine_unlikely_bbs): ... here.
17296 * predict.h (propagate_unlikely_bbs_forward): Declare.
17297 * cfgexpand.c (pass_expand::execute): Use it.
17298 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17300 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17301 propagate_unlikely_bbs_forward.
17303 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17305 PR middle-end/81331
17306 * except.c (maybe_add_nop_after_section_switch): New function.
17309 2017-07-19 Tom de Vries <tom@codesourcery.com>
17311 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17313 2017-07-19 Tom de Vries <tom@codesourcery.com>
17315 * gimple.h (gimple_phi_arg): Make assert more strict.
17317 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17319 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17320 * config/rs6000/mmintrin.h: New file.
17321 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17323 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17325 PR tree-optimization/81346
17326 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17328 2017-07-19 Tom de Vries <tom@codesourcery.com>
17330 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17332 2017-07-19 Tom de Vries <tom@codesourcery.com>
17334 * config/nvptx/nvptx-modes.def: Add V2DImode.
17335 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17336 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17337 (nvptx_output_mov_insn): Handle lack of mov.b128.
17338 (nvptx_print_operand): Handle 'H' and 'L' codes.
17339 (nvptx_vector_mode_supported): Allow V2DImode.
17340 (nvptx_preferred_simd_mode): New function.
17341 (nvptx_data_alignment): New function.
17342 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17343 nvptx_preferred_simd_mode.
17344 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17346 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17348 2017-07-19 Tom de Vries <tom@codesourcery.com>
17350 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17351 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17352 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17353 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17354 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17355 (mov<VECIM>_insn): New define_insn.
17356 (define_expand "mov<VECIM>): New define_expand.
17358 2017-07-19 Tom de Vries <tom@codesourcery.com>
17360 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17362 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17364 PR tree-optimization/81346
17365 * fold-const.h (fold_div_compare, range_check_type): Declare.
17366 * fold-const.c (range_check_type): New function.
17367 (build_range_check): Use range_check_type.
17368 (fold_div_compare): No longer static, rewritten into
17369 a match.pd helper function.
17370 (fold_comparison): Don't call fold_div_compare here.
17371 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17372 as helper function.
17374 2017-07-19 Nathan Sidwell <nathan@acm.org>
17376 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17377 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17378 * tree.c (find_decls_types_r, verify_type): Use
17379 TYPE_{MIN,MAX}_VALUE_RAW.
17380 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17381 (hash_tree): Likewise.
17382 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17384 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17387 2017-07-18 Tom de Vries <tom@codesourcery.com>
17389 PR middle-end/81464
17390 * omp-expand.c (expand_omp_for_static_chunk): Handle
17391 equal-argument loop exit phi.
17393 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17396 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17397 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17398 operand 2 predicate.
17399 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17400 operand 2 predicate.
17401 (ror,rol -> rorx splitters): Use const_int_operand as
17402 operand 2 predicate.
17404 2017-06-18 Richard Biener <rguenther@suse.de>
17406 PR tree-optimization/81410
17407 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17408 the gap in the ! slp_perm SLP case after each group.
17410 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17412 PR middle-end/81463
17413 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17416 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17418 PR middle-end/81462
17419 * predict.c (set_even_probabilities): Cleanup; do not affect
17420 probabilities that are already known.
17421 (combine_predictions_for_bb): Call even when count is set.
17423 2017-07-18 Nathan Sidwell <nathan@acm.org>
17425 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17428 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17431 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17432 optimization for loop niter analysis.
17434 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17437 * config/avr/avr.c (avr_optimize_casesi): Don't use
17438 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17440 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17442 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17443 body_cost_vec from _vect_peel_extended_info.
17444 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17445 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17448 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17450 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17452 2017-07-18 Richard Biener <rguenther@suse.de>
17454 PR tree-optimization/80620
17455 PR tree-optimization/81403
17456 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17457 info when re-using a VN table entry.
17459 2017-07-18 Richard Biener <rguenther@suse.de>
17461 PR tree-optimization/81418
17462 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17463 vectype_in. Verify that with lane-reducing reduction operations
17464 we have a single def-use cycle.
17466 2017-07-17 Carl Love <cel@us.ibm.com>
17468 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17470 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17471 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17472 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17473 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17474 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17475 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17476 VMULOSW): New enum "unspec" values.
17477 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17478 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17479 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17480 altivec_vmulosw): New patterns.
17481 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17482 VMULOSW): Add definitions.
17483 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17485 * config/alpha/alpha.c: Include predict.h.
17487 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17489 * tree-vrp.c (compare_assert_loc): Fix comparison function
17490 to return predictable results.
17492 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17494 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17496 (subdi3): Likewise.
17497 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17498 * doc/invoke.texi (mexpand-adddi): Update text.
17500 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17502 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17503 that also clobbers the CC register. The old expand code is moved
17505 (*arc_clzsi2): ... here.
17506 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17507 the CC register. The old expand code is moved to ...
17508 (arc_ctzsi2): ... here.
17510 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17512 * config/arc/arc.opt (mindexed-loads): Use initial value
17513 TARGET_INDEXED_LOADS_DEFAULT.
17514 (mauto-modify-reg): Use initial value
17515 TARGET_AUTO_MODIFY_REG_DEFAULT.
17516 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17517 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17518 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17519 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17521 2017-07-17 Martin Liska <mliska@suse.cz>
17524 * opts.c (finish_options): Do not allow -fgnu-tm
17525 w/ -fsanitize={kernel-,}address. Say sorry.
17527 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17530 * tree-loop-distribution.c (classify_partition): Only assert on
17531 numer of iterations.
17532 (merge_dep_scc_partitions): Delete prameter. Update function call.
17533 (distribute_loop): Remove code handling loop with unknown niters.
17534 (pass_loop_distribution::execute): Skip loop with unknown niters.
17536 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17539 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17540 function sort_partitions_by_post_order.
17542 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17544 PR tree-optimization/81374
17545 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17546 the max index of basic blocks, rather than number of basic blocks.
17548 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17550 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17552 (arc_legitimate_pic_operand_p): Likewise.
17553 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17555 (arc_needs_pcl_p): Likewise.
17556 (arc_legitimate_pc_offset_p): Likewise.
17557 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17558 function is also used in constrains.md.
17559 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17560 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17561 PLUS. Only return true/false in known cases, otherwise assert.
17562 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17563 is already called in arc_legitimate_constant_p.
17564 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17566 (LEGITIMATE_PIC_OPERAND_P): Use
17567 arc_raw_symbolic_reference_mentioned_p function.
17568 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17573 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17574 Andrew Burgess <andrew.burgess@embecosm.com>
17576 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17577 (arc_return_address_register): New function.
17578 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17579 (arc_handle_fndecl_attribute): Add naked attribute.
17580 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17581 (TARGET_WARN_FUNC_RETURN): Likewise.
17582 (arc_allocate_stack_slots_for_args): New function.
17583 (arc_warn_func_return): Likewise.
17584 (machine_function): Change type fn_type.
17585 (arc_compute_function_type): Consider new naked function type,
17586 change function return type.
17587 (arc_must_save_register): Adapt to handle new
17588 arc_compute_function_type's return type.
17589 (arc_expand_prologue): Likewise.
17590 (arc_expand_epilogue): Likewise.
17591 (arc_return_address_regs): Delete.
17592 (arc_return_address_register): New function.
17593 (arc_epilogue_uses): Use above function.
17594 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17595 (arc_function_type): Change encoding, add naked type.
17596 (ARC_INTERRUPT_P): Change to handle the new encoding.
17597 (ARC_FAST_INTERRUPT_P): Likewise.
17598 (ARC_NORMAL_P): Define.
17599 (ARC_NAKED_P): Likewise.
17600 (arc_compute_function_type): Delete prototype.
17601 * config/arc/arc.md (in_ret_delay_slot): Use
17602 arc_return_address_register function.
17603 (simple_return): Likewise.
17604 (p_return_i): Likewise.
17606 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17608 PR tree-optimization/81428
17609 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17610 can't be built for those types.
17612 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17614 Remove stuff dead since r239246.
17616 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17617 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17618 (avr_inform_devices): Remove dead stuff.
17620 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17622 * config/arm/arm_neon.h: Fix softp typo.
17624 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17626 PR tree-optimization/81365
17627 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17628 aggregate moves onto bb predecessor edges, make sure there are no
17629 loads that could alias the lhs in between the start of bb and the
17632 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17635 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17636 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17637 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17639 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17641 PR tree-optimization/81396
17642 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17643 (init_symbolic_number): Initialize it to 1.
17644 (perform_symbolic_merge): Add n_ops from both operands into the new
17646 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17647 without base_addr as useless if they need more than one operation.
17648 (bswap_replace): Handle !bswap case for NULL base_addr.
17650 2017-07-17 Tom de Vries <tom@codesourcery.com>
17653 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17656 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17658 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17659 conditional builtin define __FIX_LEON3FT_B2BST.
17661 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17663 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17664 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17667 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17669 PR rtl-optimization/81424
17670 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17671 to remove potential trapping from operands if -fnon-call-exceptions.
17673 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17675 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17676 profile_proability for scalling.
17677 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17679 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17681 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17683 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17685 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17686 fixpoint arithmetics.
17688 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17690 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17691 fixpoint arithmetics.
17693 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17695 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17696 fixpoint arithmetics.
17698 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17700 * profile-count.h (profile_probability::from_reg_br_prob_note,
17701 profile_probability::to_reg_br_prob_note): New functions.
17702 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17703 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17704 * predict.c (probability_reliable_p): Update.
17705 (edge_probability_reliable_p): Update.
17706 (br_prob_note_reliable_p): Update.
17707 (invert_br_probabilities): Update.
17708 (add_reg_br_prob_note): New function.
17709 (combine_predictions_for_insn): Update.
17710 * asan.c (asan_clear_shadow): Update.
17711 * cfgbuild.c (compute_outgoing_frequencies): Update.
17712 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17713 (update_br_prob_note): Update.
17714 (rtl_verify_edges): Update.
17715 (purge_dead_edges): Update.
17716 (fixup_reorder_chain): Update.
17717 * emit-rtl.c (try_split): Update.
17718 * ifcvt.c (cond_exec_process_insns): Update.
17719 (cond_exec_process_if_block): Update.
17720 (dead_or_predicable): Update.
17721 * internal-fn.c (expand_addsub_overflow): Update.
17722 (expand_neg_overflow): Update.
17723 (expand_mul_overflow): Update.
17724 * loop-doloop.c (doloop_modify): Update.
17725 * loop-unroll.c (compare_and_jump_seq): Update.
17726 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17727 * predict.h: Update.
17728 * reorg.c (mostly_true_jump): Update.
17730 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17731 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17732 * config/arc/arc.c: (emit_unlikely_jump): Update.
17733 * config/arm/arm.c: (emit_unlikely_jump): Update.
17734 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17735 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17736 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17737 (ix86_print_operand): Update.
17738 (ix86_split_fp_branch): Update.
17739 (predict_jump): Update.
17740 * config/ia64/ia64.c (ia64_print_operand): Update.
17741 * config/mmix/mmix.c (mmix_print_operand): Update.
17742 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17743 (rs6000_expand_split_stack_prologue): Update.
17744 * config/rs6000/rs6000.c: Update.
17745 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17746 (s390_expand_vec_movstr): Update.
17747 (s390_expand_cs_tdsi): Update.
17748 (s390_expand_split_stack_prologue): Update.
17749 * config/sh/sh.c (sh_print_operand): Update.
17750 (expand_cbranchsi4): Update.
17751 (expand_cbranchdi4): Update.
17752 * config/sparc/sparc.c (output_v9branch): Update.
17753 * config/spu/spu.c (get_branch_target): Update.
17754 (ea_load_store_inline): Update.
17755 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17756 * config/tilepro/tilepro.c: Update.
17758 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17760 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17761 (gimplify_save_expr): Likewise.
17763 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17765 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17767 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17769 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17771 * timevar.def (TV_IPA_FNSUMMARY): Define.
17773 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17775 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17776 to back store errata sensitive sequence from being generated.
17777 (sqrtdf2_fix): Likewise.
17779 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17781 * tree-ssa-threadupdate.c (compute_path_counts,
17782 update_joiner_offpath_counts): Use profile_probability.
17784 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17787 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17789 * config/arm/arm-c.c (arm_cpu_builtins): Define
17790 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17792 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17794 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17795 array entries to represent __ieee128 versions of the
17796 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17797 scalar_extract_sig, and scalar_insert_exp built-in functions.
17798 (altivec_resolve_overloaded_builtin): Add special case handling
17799 for the __builtin_scalar_insert_exp function, as represented by
17800 the P9V_BUILTIN_VEC_VSIEDP constant.
17801 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17802 exponent support for __ieee128 argument.
17803 (VSESQP): Add scalar extract signature support for __ieee128
17805 (VSTDCNQP): Add scalar test negative support for __ieee128
17807 (VSIEQP): Add scalar insert exponent support for __int128 argument
17808 with __ieee128 result.
17809 (VSIEQPF): Add scalar insert exponent support for __ieee128
17810 argument with __ieee128 result.
17811 (VSTDCQP): Add scalar test data class support for __ieee128
17813 (VSTDCNQP): Add overload support for scalar test negative with
17814 __ieee128 argument.
17815 (VSTDCQP): Add overload support for scalar test data class
17816 __ieee128 argument.
17817 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17818 UNSPEC_VSX_SXSIGDP.
17819 (UNSPEC_VSX_SIEXPQP): New constant.
17820 (xsxexpqp): New insn for VSX scalar extract exponent quad
17822 (xsxsigqp): New insn for VSX scalar extract significand quad
17824 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17825 precision with floating point argument.
17826 (xststdcqp): New expand for VSX scalar test data class quad
17828 (xststdcnegqp): New expand for VSX scalar test negative quad
17830 (xststdcqp): New insn to match expansions for VSX scalar test data
17831 class quad precision and VSX scalar test negative quad precision.
17832 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17833 special case operand checking to enforce that second operand of
17834 VSX scalar test data class with quad precision argument is a 7-bit
17836 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17837 prototypes and descriptions of __ieee128 versions of
17838 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17839 scalar_test_data_class, and scalar_test_neg built-in functions.
17841 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17843 PR tree-optimization/81162
17844 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17845 replace a negate with an add.
17847 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17849 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17851 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17853 * config/arm/arm-c.c (arm_cpu_builtins): Define
17854 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17856 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17858 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17859 (armv8-r): Set ARM Cortex-R52 as default CPU.
17860 * config/arm/arm-tables.opt: Regenerate.
17861 * config/arm/arm-tune.md: Regenerate.
17862 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17864 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17865 extension for -mcpu=cortex-r52.
17867 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17869 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17870 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17871 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17872 (fp-armv8): Define it as FP_ARMv8 only.
17873 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17874 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17876 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17877 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17878 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17879 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17880 than TARGET_FPU_ARMV8.
17881 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17882 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17883 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17885 * config/arm/neon.md (neon_vrint): Likewise.
17886 (neon_vcvt): Likewise.
17887 (neon_<fmaxmin_op><mode>): Likewise.
17888 (<fmaxmin><mode>3): Likewise.
17889 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17890 * config/arm/predicates.md (arm_cond_move_operator): Check against
17891 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17893 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17895 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17896 to top of function.
17898 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17900 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17901 loop in comment with memset.
17903 2017-07-14 Martin Liska <mliska@suse.cz>
17905 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17906 * dwarf2out.c (is_java): Remove the function.
17907 (output_pubname): Remove usage of the function.
17908 (lower_bound_default): Remove usage of DW_LANG_Java.
17909 (gen_compile_unit_die): Likewise.
17910 * gcc.c: Remove compiler defaults for .java and .zip files.
17911 * gimple-expr.c (remove_suffix): Change as there's no longer
17912 extension than 4-letter one.
17913 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17914 (gimplify_save_expr): Likewise.
17915 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17916 as it's possible even for other languages than Java.
17917 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17918 * lto-opts.c (lto_write_options): Remove reference to Java.
17919 * opts.c (strip_off_ending): Update file extension handling.
17920 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17921 * tree-eh.c (lower_resx): Likewise.
17922 * tree.c (free_lang_data_in_type): Remove dead code.
17923 (find_decls_types_r): Likewise.
17924 (build_common_builtin_nodes): Remove Java from a comment.
17925 (verify_type): Remove dead code.
17926 * varasm.c (assemble_external): Remove Java from a comment.
17928 2017-07-14 Martin Liska <mliska@suse.cz>
17930 * opts.c (finish_options): Add quotes.
17931 (common_handle_option): Likewise.
17933 2017-07-14 Martin Liska <mliska@suse.cz>
17935 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17936 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17937 Remove N_SO_PASCAL.
17938 * dwarf2out.c (lower_bound_default): Do not handle
17940 (gen_compile_unit_die): Likewise.
17941 * gcc.c: Remove default extension binding for GNU Pascal.
17942 * stmt.c: Remove Pascal language from a comment.
17943 * xcoffout.c: Likewise.
17945 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17948 * diagnostic-show-locus.c (fixit_cmp): New function.
17949 (layout::layout): Sort m_fixit_hints.
17950 (column_range::column_range): Assert that the values are valid.
17951 (struct char_span): New struct.
17952 (correction::overwrite): New method.
17953 (struct source_line): New struct.
17954 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17955 calls in terms of classes source_line and char_span, and
17956 correction::overwrite.
17957 (selftest::test_overlapped_fixit_printing_2): New function.
17958 (selftest::diagnostic_show_locus_c_tests): Call it.
17960 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17962 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17963 early if there is no lhs.
17965 2017-07-13 Martin Liska <mliska@suse.cz>
17967 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17968 (gen_reference_type_die): Likewise.
17969 * stor-layout.c: Remove Pascal-related comment.
17971 2017-07-13 Martin Liska <mliska@suse.cz>
17973 * opts.c (finish_options): Add quotes to error messages.
17974 (parse_sanitizer_options): Likewise.
17976 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17978 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17980 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17982 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17984 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17986 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17988 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17990 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17993 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17994 provides the hardware capability bits, define the macro
17995 __BUILTIN_CPU_SUPPORTS__.
17996 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17997 if GLIBC does not provide the hardware capability bits. Add a
17998 gcc_unreachable call if the built-in cpu function is neither
17999 __builtin_cpu_is nor __builtin_cpu_supports.
18000 (rs6000_get_function_versions_dispatcher): Change the warning
18001 that an old GLIBC is used which does not export the capability
18002 bits to be an error.
18003 * doc/extend.texi (target_clones attribute): Document the
18004 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18005 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18006 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18007 the macros defined by GCC if the newer GLIBC is available.
18009 2017-07-12 Jeff Law <law@redhat.com>
18011 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18012 remaining includes slightly.
18013 * config/riscv/riscv-builtins.c: Include profile-count.h.
18015 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18018 * config/avr/avr.c (avr_set_current_function): In diagnostic
18019 messages: Quote keywords and (parts of) identifiers.
18020 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18023 2017-07-12 Carl Love <cel@us.ibm.com>
18025 * config/rs6000/rs6000-c.c: Add support for built-in functions
18026 vector bool char vec_revb (vector bool char);
18027 vector bool short vec_revb (vector short char);
18028 vector bool int vec_revb (vector bool int);
18029 vector bool long long vec_revb (vector bool long long);
18030 * doc/extend.texi: Update the built-in documentation file for the
18031 new built-in functions.
18033 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18035 * config/s390/s390.md: Remove movcc splitter.
18037 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18039 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18040 load/store on condition.
18042 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18045 * config/avr/avr.c (avr_encode_section_info)
18046 [progmem && !TREE_READONLY]: Error if progmem object needs
18049 2017-07-11 Michael Collison <michael.collison@arm.com>
18051 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18054 2017-07-11 Carl Love <cel@us.ibm.com>
18056 * config/rs6000/rs6000-c.c: Add support for builtins
18057 vector unsigned int vec_parity_lsbb (vector signed int);
18058 vector unsigned int vec_parity_lsbb (vector unsigned int);
18059 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18060 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18061 vector unsigned long long vec_parity_lsbb (vector signed long long);
18062 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18063 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18064 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18065 * doc/extend.texi: Update the built-in documentation file for the
18066 new built-in functions.
18068 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18070 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18071 (layout::m_primary_loc): New field.
18072 (layout::layout): Initialize new field. Move location filtering
18073 logic from here to...
18074 (layout::maybe_add_location_range): ...this new method. Add
18075 support for filtering to just the lines already specified by other
18077 (layout::will_show_line_p): New method.
18078 (gcc_rich_location::add_location_if_nearby): New method.
18079 (selftest::test_add_location_if_nearby): New test function.
18080 (selftest::diagnostic_show_locus_c_tests): Call it.
18081 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18084 2017-07-11 Tom de Vries <tom@codesourcery.com>
18086 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18087 (bb_first_real_insn): New function.
18088 (nvptx_single): Add extra initialization of broadcasted condition
18091 2017-07-11 Nathan Sidwell <nathan@acm.org>
18093 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18095 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18097 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18098 Binutils doc as TEXI will mess them up.
18099 * doc/invoke.texi (AVR Options): Same here.
18101 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18103 * config/sparc/sparc.opt (mfix-ut700): New option.
18104 (mfix-gr712rc): Likewise.
18105 (sparc_fix_b2bst): New variable.
18106 * doc/invoke.texi (SPARC options): Document them.
18107 (ARM options): Fix warnings.
18108 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18109 instructions to prevent sequences that can trigger the store-store
18110 errata for certain LEON3FT processors.
18111 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18112 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18113 * config/sparc/sparc.md (fix_b2bst): New attribute.
18114 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18116 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18119 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18121 (*rsqrtsf2_sse): Ditto.
18123 (div<mode>3): Macroize insn from divdf3 and divsf3
18124 using MODEF mode iterator.
18126 2017-07-10 Martin Sebor <msebor@redhat.com>
18128 PR tree-optimization/80397
18129 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18130 instead of testing for equality to INTEGER_TYPE.
18132 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18134 * config.gcc: Remove uclibc from arc target spec.
18136 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18138 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18140 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18143 * lto-wrapper.c (remove_option): New function.
18144 (merge_and_complain): Merge PIC/PIE options more realistically.
18146 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18148 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18152 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18153 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18154 * config.in: Regenerate.
18155 * configure: Regenerate.
18156 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18157 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18158 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18159 (TARGET_GASISR_PROLOGUES): ...target mask.
18160 * common/config/avr/avr-common.c
18161 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18162 Set -mgas-isr-prologues.
18163 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18164 INSERT_PASS_BEFORE for it.
18165 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18166 * config/avr/avr.c (avr_option_override)
18167 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18168 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18169 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18170 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18171 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18173 (make_avr_pass_pre_proep): New function.
18174 (emit_push_sfr) <treg>: Add argument to function and use it
18175 instead of TMP_REG.
18176 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18177 and set machine->gasisr.yes.
18178 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18179 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18180 __gcc_isr.n_pushed to .L__stack_usage.
18181 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18182 (avr_asm_final_postscan_insn): ...this new static function.
18183 * config/avr/avr.h (machine_function)
18184 <is_no_gccisr, use_L__stack_usage>: New fields.
18185 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18186 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18187 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18188 (gasisr, *gasisr): New expander and insn.
18189 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18190 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18191 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18193 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18195 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18198 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18200 Move jump-tables out of .text again.
18203 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18204 (ASM_OUTPUT_ADDR_VEC): New function.
18205 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18206 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18207 INSN_ADDRESSes as asm comment.
18208 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18209 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18210 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18211 * config/avr/avr.md (*tablejump): Adjust comment.
18212 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18213 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18215 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18216 (avr_output_addr_vec): New proto.
18217 (avr_log_t) <insn_addresses>: New field.
18219 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18222 * config/i386/i386.c (ix86_function_arg_advance): Set
18223 outgoing_args_on_stack to true if there are outgoing arguments
18225 (ix86_function_arg): Likewise.
18226 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18227 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18228 * config/i386/i386.h (machine_function): Add
18229 outgoing_args_on_stack.
18231 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18233 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18234 supporting pthreds.
18235 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18237 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18239 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18240 (REAL_H): Remove $(MACHMODE_H).
18241 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18243 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18244 $(MACHMODE_H) and double-int.h.
18245 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18247 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18250 2017-07-07 Andrew Pinski <apinski@cavium.com>
18252 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18253 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18255 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18257 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18258 Add warning if GCC was not configured to link against a GLIBC that
18259 exports the hardware capability bits.
18260 (make_resolver_func): Make resolver function private and not a
18261 COMDAT function. Create the name with clone_function_name instead
18262 of make_unique_name.
18265 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18266 correct operand in doing the split.
18268 2017-07-07 Carl Love <cel@us.ibm.com>
18270 * config/rs6000/rs6000-c: Add support for built-in function
18271 vector unsigned short vec_pack_to_short_fp32 (vector float,
18273 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18274 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18275 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18276 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18277 (convert_4f32_8i16): Add define_expand.
18278 * doc/extend.texi: Update the built-in documentation file for the
18279 new built-in function.
18281 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18283 * config/sparc/m8.md: New file.
18284 * config/sparc/sparc.md: Include m8.md.
18286 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18288 * config/sparc/sparc.opt: New option -mvis4b.
18289 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18290 (sparc_option_override): Handle VIS4B.
18291 (enum sparc_builtins): Define
18292 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18293 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18294 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18295 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18296 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18297 (check_constant_argument): New function.
18298 (sparc_vis_init_builtins): Define builtins
18299 __builtin_vis_dictunpack{8,16,32},
18300 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18301 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18302 __builtin_vis_fpcmpde{8,16,32}shl and
18303 __builtin_vis_fpcmpur{8,16,32}shl.
18304 (sparc_expand_builtin): Check that the constant operands to
18305 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18306 constant and in range.
18307 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18309 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18310 (SPARC_IMM5_P): Likewise.
18311 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18312 (enabled): Handle vis4b.
18313 (UNSPEC_DICTUNPACK): New unspec.
18314 (UNSPEC_FPCMPSHL): Likewise.
18315 (UNSPEC_FPUCMPSHL): Likewise.
18316 (UNSPEC_FPCMPDESHL): Likewise.
18317 (UNSPEC_FPCMPURSHL): Likewise.
18318 (cpu_feature): New CPU feature `vis4b'.
18319 (dictunpack{8,16,32}): New insns.
18320 (FPCSMODE): New mode iterator.
18321 (fpcscond): New code iterator.
18322 (fpcsucond): Likewise.
18323 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18324 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18325 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18326 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18327 * config/sparc/constraints.md: Define constraints `q' for unsigned
18328 2-bit integer constants and `t' for unsigned 5-bit integer
18330 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18332 (imm5_operand_dictunpack16): Likewise.
18333 (imm5_operand_dictunpack32): Likewise.
18334 (imm2_operand): Likewise.
18335 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18336 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18337 ditunpack* and fpcmp*shl builtins.
18339 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18341 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18342 * config.in: Add HAVE_AS_SPARC6 define.
18343 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18345 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18347 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18348 (CPP_CPU_SPEC): Handle m8.
18349 (ASM_CPU_SPEC): Likewise.
18350 * config/sparc/sparc-opts.h (enum processor_type): Add
18352 * config/sparc/sparc.c (m8_costs): New struct.
18353 (sparc_option_override): Handle TARGET_CPU_m8.
18354 (sparc32_initialize_trampoline): Likewise.
18355 (sparc64_initialize_trampoline): Likewise.
18356 (sparc_issue_rate): Likewise.
18357 (sparc_register_move_cost): Likewise.
18358 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18359 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18360 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18361 (CPP_CPU_SPEC): Handle M8.
18362 (ASM_CPU_SPEC): Likewise.
18363 (AS_M8_FLAG): Define.
18364 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18365 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18366 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18368 * configure: Regenerate.
18369 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18372 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18374 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18376 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18377 ("*movdi_insn_sp32"): Do not set v3pipe.
18378 ("*movsi_insn"): Likewise.
18379 ("*movdi_insn_sp64"): Likewise.
18380 ("*movsf_insn"): Likewise.
18381 ("*movdf_insn_sp32"): Likewise.
18382 ("*movdf_insn_sp64"): Likewise.
18383 ("*zero_extendsidi2_insn_sp64"): Likewise.
18384 ("*sign_extendsidi2_insn"): Likewise.
18385 ("*mov<VM32:mode>_insn"): Likewise.
18386 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18387 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18388 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18389 ("<vlop:code><VL:mode>3"): Likewise.
18390 ("*not_<vlop:code><VL:mode>3"): Likewise.
18391 ("*nand<VL:mode>_vis"): Likewise.
18392 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18393 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18394 ("one_cmpl<VL:mode>2"): Likewise.
18395 ("faligndata<VM64:mode>_vis"): Likewise.
18396 ("alignaddrsi_vis"): Likewise.
18397 ("alignaddrdi_vis"): Likweise.
18398 ("alignaddrlsi_vis"): Likewise.
18399 ("alignaddrldi_vis"): Likewise.
18400 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18401 ("bmaskdi_vis"): Likewise.
18402 ("bmasksi_vis"): Likewise.
18403 ("bshuffle<VM64:mode>_vis"): Likewise.
18404 ("cmask8<P:mode>_vis"): Likewise.
18405 ("cmask16<P:mode>_vis"): Likewise.
18406 ("cmask32<P:mode>_vis"): Likewise.
18407 ("pdistn<P:mode>_vis"): Likewise.
18408 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18410 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18412 * config/sparc/sparc.md ("subtype"): New insn attribute.
18413 ("*wrgsr_sp64"): Set insn subtype.
18414 ("*rdgsr_sp64"): Likewise.
18415 ("alignaddrsi_vis"): Likewise.
18416 ("alignaddrdi_vis"): Likewise.
18417 ("alignaddrlsi_vis"): Likewise.
18418 ("alignaddrldi_vis"): Likewise.
18419 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18420 ("fexpand_vis"): Likewise.
18421 ("fpmerge_vis"): Likewise.
18422 ("faligndata<VM64:mode>_vis"): Likewise.
18423 ("bshuffle<VM64:mode>_vis"): Likewise.
18424 ("cmask8<P:mode>_vis"): Likewise.
18425 ("cmask16<P:mode>_vis"): Likewise.
18426 ("cmask32<P:mode>_vis"): Likewise.
18427 ("fchksm16_vis"): Likewise.
18428 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18429 ("fmean16_vis"): Likewise.
18430 ("fp<plusminus_insn>64_vis"): Likewise.
18431 ("<plusminus_insn>v8qi3"): Likewise.
18432 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18433 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18434 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18435 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18436 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18437 ("*movqi_insn"): Likewise.
18438 ("*movhi_insn"): Likewise.
18439 ("*movsi_insn"): Likewise.
18440 ("movsi_pic_gotdata_op"): Likewise.
18441 ("*movdi_insn_sp32"): Likewise.
18442 ("*movdi_insn_sp64"): Likewise.
18443 ("movdi_pic_gotdata_op"): Likewise.
18444 ("*movsf_insn"): Likewise.
18445 ("*movdf_insn_sp32"): Likewise.
18446 ("*movdf_insn_sp64"): Likewise.
18447 ("*zero_extendhisi2_insn"): Likewise.
18448 ("*zero_extendqihi2_insn"): Likewise.
18449 ("*zero_extendqisi2_insn"): Likewise.
18450 ("*zero_extendqidi2_insn"): Likewise.
18451 ("*zero_extendhidi2_insn"): Likewise.
18452 ("*zero_extendsidi2_insn_sp64"): Likewise.
18453 ("ldfsr"): Likewise.
18454 ("prefetch_64"): Likewise.
18455 ("prefetch_32"): Likewise.
18456 ("tie_ld32"): Likewise.
18457 ("tie_ld64"): Likewise.
18458 ("*tldo_ldub_sp32"): Likewise.
18459 ("*tldo_ldub1_sp32"): Likewise.
18460 ("*tldo_ldub2_sp32"): Likewise.
18461 ("*tldo_ldub_sp64"): Likewise.
18462 ("*tldo_ldub1_sp64"): Likewise.
18463 ("*tldo_ldub2_sp64"): Likewise.
18464 ("*tldo_ldub3_sp64"): Likewise.
18465 ("*tldo_lduh_sp32"): Likewise.
18466 ("*tldo_lduh1_sp32"): Likewise.
18467 ("*tldo_lduh_sp64"): Likewise.
18468 ("*tldo_lduh1_sp64"): Likewise.
18469 ("*tldo_lduh2_sp64"): Likewise.
18470 ("*tldo_lduw_sp32"): Likewise.
18471 ("*tldo_lduw_sp64"): Likewise.
18472 ("*tldo_lduw1_sp64"): Likewise.
18473 ("*tldo_ldx_sp64"): Likewise.
18474 ("*mov<VM32:mode>_insn"): Likewise.
18475 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18476 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18478 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18480 * config/sparc/sparc.md ("type"): New insn type viscmp.
18481 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18483 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18484 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18485 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18486 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18488 ("n7_vis_logical_11cycle"): Likewise.
18489 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18490 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18491 * config/sparc/niagara.md ("niag_vis"): Likewise.
18492 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18493 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18495 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18497 * config/sparc/sparc.md: New instruction type `bmask'.
18498 (bmaskdi_vis): Use the `bmask' type.
18499 (bmasksi_vis): Likewise.
18500 * config/sparc/ultra3.md (us3_array): Likewise.
18501 * config/sparc/niagara7.md (n7_array): Likewise.
18502 * config/sparc/niagara4.md (n4_array): Likewise.
18503 * config/sparc/niagara2.md (niag2_vis): Likewise.
18504 (niag3_vis): Likewise.
18505 * config/sparc/niagara.md (niag_vis): Likewise.
18507 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18509 * ipa-comdats.c: Remove optimize check from gate.
18510 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18511 for functions not optimized.
18512 (ipa_fn_summary_read): Skip optimize check.
18513 (ipa_fn_summary_write): Likewise.
18514 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18516 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18518 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18519 (check_callers): Give up on caller which is not optimized.
18520 (inline_small_functions): Likewise.
18521 (ipa_inline): Do not give up when not optimizing.
18522 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18523 away unoptimizes cdtors.
18524 (whole_program_function_and_variable_visibility): Do
18525 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18526 * ipa.c (process_references): Do not check optimize.
18527 (symbol_table::remove_unreachable_nodes): Update optimize check.
18528 (set_writeonly_bit): Update optimize check.
18529 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18530 (pass_ipa_single_use::gate): Remove.
18532 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18534 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18535 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18536 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18537 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18538 permute_load, permute_store, adjust_extract, adjust_splat,
18539 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18540 replace_swap_with_copy, dump_swap_insn_table,
18541 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18542 recombine_lvx_pattern, recombine_stvx_pattern,
18543 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18544 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18545 to file rs6000-p8swap.c.
18546 * config/rs6000/rs6000-p8swap.c: New file.
18547 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18548 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18549 and rs6000*-*-* targets.
18551 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18553 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18555 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18557 * lto-wrapper.c (merge_and_complain): Do not merge
18558 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18559 fsigned_zeros, ftrapping_math, fwrapv.
18560 (append_compiler_options): Do not track these options.
18561 (append_linker_options): Likewie
18563 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18565 * cgraphunit.c (cgraph_node::finalize_function): When
18566 !flag_toplevel_reorde set no_reorder flag.
18567 (varpool_node::finalize_decl): Likewise.
18568 (symbol_table::compile): Drop no toplevel reorder path.
18570 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18572 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18573 edges; zero probability is not better than uninitialized.
18575 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18577 * asan.h (asan_sanitize_allocas_p): Declare.
18578 * asan.c (asan_sanitize_allocas_p): New function.
18579 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18580 (handle_builtin_alloca): Likewise.
18581 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18582 if !asan_sanitize_allocas_p.
18583 * params.def (asan-instrument-allocas): Add new option.
18584 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18585 * opts.c (common_handle_option): Disable allocas sanitization for
18588 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18590 * asan.c: Include gimple-fold.h.
18591 (get_last_alloca_addr): New function.
18592 (handle_builtin_stackrestore): Likewise.
18593 (handle_builtin_alloca): Likewise.
18594 (asan_emit_allocas_unpoison): Likewise.
18595 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18596 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18597 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18598 (instrument_builtin_call): Pass gimple iterator to
18599 get_mem_refs_of_builtin_call.
18600 (last_alloca_addr): New global.
18601 * asan.h (asan_emit_allocas_unpoison): Declare.
18602 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18603 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18604 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18605 if function calls alloca.
18606 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18607 * gimple-fold.h (replace_call_with_value): Declare.
18608 * internal-fn.c: Include asan.h.
18609 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18610 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18612 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18614 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18615 (C_SELFTEST_FLAGS): New.
18616 (CPP_SELFTEST_FLAGS): New.
18617 (SELFTEST_DEPS): New, from deps of s-selftest.
18618 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18619 (CPP_SELFTEST_DEPS): New.
18620 (selftest): Add dependency on s-selftest-c++.
18621 (s-selftest): Rename to...
18622 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18623 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18624 than SELFTEST_FLAGS.
18625 (selftest-gdb): Rename to...
18626 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18628 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18629 (selftest-valgrind): Rename to...
18630 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18632 (selftest-valgrind): Reintroduce as an alias for
18633 selftest-c-valgrind.
18634 (s-selftest-c++): New.
18635 (selftest-c++-gdb): New.
18636 (selftest-c++-valgrind): New.
18638 2017-07-06 Olivier Hainque <hainque@adacore.com>
18640 * gcc.c (process_command): When deciding if undefined variables
18641 should be ignored when processing specs, accept "gcc -v" as well.
18643 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18645 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18646 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18648 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18650 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18651 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18652 * config/arm/arm-tables.opt: Regenerate.
18653 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18655 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18657 2017-07-06 Carl Love <cel@us.ibm.com>
18659 * ChangeLog: Clean up from mid air collision
18661 2017-07-06 Carl Love <cel@us.ibm.com>
18663 * config/rs6000/rs6000-c.c: Add support for built-in functions
18664 vector signed int vec_subc (vector signed int, vector signed int);
18665 vector signed __int128 vec_subc (vector signed __int128,
18666 vector signed __int128);
18667 vector unsigned __int128 vec_subc (vector unsigned __int128,
18668 vector unsigned __int128);
18669 vector signed int vec_sube (vector signed int, vector signed int,
18670 vector signed int);
18671 vector unsigned int vec_sube (vector unsigned int,
18672 vector unsigned int,
18673 vector unsigned int);
18674 vector signed __int128 vec_sube (vector signed __int128,
18675 vector signed __int128,
18676 vector signed__int128);
18677 vector unsigned __int128 vec_sube (vector unsigned __int128,
18678 vector unsigned __int128,
18679 vector unsigned __int128);
18680 vector signed int vec_subec (vector signed int, vector signed int,
18681 vector signed int);
18682 vector unsigned int vec_subec (vector unsigned int,
18683 vector unsigned int,
18684 vector unsigned int);
18685 vector signed __int128 vec_subec (vector signed __int128,
18686 vector signed __int128,
18687 vector signed__int128);
18688 vector unsigned __int128 vec_subec (vector unsigned __int128,
18689 vector unsigned __int128,
18690 vector unsigned __int128);
18691 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18692 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18693 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18694 BU_ALTIVEC_OVERLOAD_X definitions.
18695 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18696 * doc/extend.texi: Update the built-in documentation file for the new
18697 built-in functions.
18699 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18702 * diagnostic-show-locus.c (layout::layout): Use start and finish
18703 spelling location for the start and finish of each range.
18704 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18705 Add unused aspect param.
18706 * input.c (expand_location_1): Add "aspect" param, and use it
18707 to access the correct part of the location.
18708 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18710 (expand_location_to_spelling_point): Likewise.
18711 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18712 param, and pass it to expand_location_1.
18714 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18716 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18717 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18718 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18719 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18720 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18721 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18722 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18723 _mm_maskz_getmant_ss): New intrinsics.
18724 (__builtin_ia32_getexpss128_mask): Changed to ...
18725 __builtin_ia32_getexpss128_round ... this.
18726 (__builtin_ia32_getexpsd128_mask): Changed to ...
18727 __builtin_ia32_getexpsd128_round ... this.
18728 * config/i386/i386-builtin-types.def
18729 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18730 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18731 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18732 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18733 __builtin_ia32_getmantss_mask_round): New builtins.
18734 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18735 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18736 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18737 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18738 * config/i386/sse.md
18739 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18740 avx512f_sgetexp<mode><mask_scalar_name>
18741 <round_saeonly_scalar_name> ... this.
18742 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18743 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18744 vgetexp<ssescalarmodesuffix>
18745 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18746 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18747 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18748 avx512f_vgetmant<mode><mask_scalar_name>
18749 <round_saeonly_scalar_name> ... this.
18750 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18751 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18752 vgetmant<ssescalarmodesuffix>
18753 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18754 %0<mask_scalar_operand4>, %1, %2
18755 <round_saeonly_scalar_mask_op4>, %3} ... this.
18756 * config/i386/subst.md (mask_scalar_operand4,
18757 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18758 round_saeonly_scalar_nimm_predicate): New subst attributes.
18760 2017-07-06 Julia Koval <julia.koval@intel.com>
18762 * config/i386/i386.c (ix86_erase_embedded_rounding):
18763 Remove code for old rounding pattern.
18765 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18767 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18769 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18771 * doc/sourcebuild.texi (Test Directives, Variants of
18772 dg-require-support): Add documentation for dg-require-stack-check.
18774 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18776 * config/i386/subst.md (mask_scalar, round_scalar,
18777 round_saeonly_scalar): New meta-templates.
18778 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18779 round_scalar_mask_operand3, round_scalar_mask_op3,
18780 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18781 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18782 round_saeonly_scalar_constraint,
18783 round_saeonly_scalar_prefix): New subst attribute.
18784 * config/i386/sse.md
18785 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18786 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18787 <round_scalar_name> ... this.
18788 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18789 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18790 <round_scalar_name> ... this.
18791 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18792 <sse>_vm<code><mode>3<mask_scalar_name>
18793 <round_saeonly_scalar_name> ... this.
18794 (v<plusminus_mnemonic><ssescalarmodesuffix>
18795 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18796 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18797 v<plusminus_mnemonic><ssescalarmodesuffix>
18798 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18799 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18800 (v<multdiv_mnemonic><ssescalarmodesuffix>
18801 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18802 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18803 v<multdiv_mnemonic><ssescalarmodesuffix>
18804 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18805 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18806 (v<maxmin_float><ssescalarmodesuffix>
18807 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18808 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18809 v<maxmin_float><ssescalarmodesuffix>
18810 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18811 %0<mask_scalar_operand3>, %1, %<iptr>2
18812 <round_saeonly_scalar_mask_op3>} ... this.
18814 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18816 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18817 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18819 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18820 Alan Hayward <alan.hayward@arm.com>
18821 David Sherwood <david.sherwood@arm.com>
18823 * combine.c (simplify_if_then_else): Remove "enum" before
18825 * compare-elim.c (can_eliminate_compare): Likewise.
18826 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18828 (aarch64_lookup_simd_builtin_type): Likewise.
18829 (aarch64_simd_builtin_type): Likewise.
18830 (aarch64_init_simd_builtin_types): Likewise.
18831 (aarch64_simd_expand_args): Likewise.
18832 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18834 (aarch64_reverse_mask): Likewise.
18835 (aarch64_simd_emit_reg_reg_move): Likewise.
18836 (aarch64_gen_adjusted_ldpstp): Likewise.
18837 (aarch64_ccmp_mode_to_code): Likewise.
18838 (aarch64_operands_ok_for_ldpstp): Likewise.
18839 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18840 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18842 (aarch64_min_divisions_for_recip_mul): Likewise.
18843 (aarch64_reassociation_width): Likewise.
18844 (aarch64_get_condition_code_1): Likewise.
18845 (aarch64_simd_emit_reg_reg_move): Likewise.
18846 (aarch64_simd_attr_length_rglist): Likewise.
18847 (aarch64_reverse_mask): Likewise.
18848 (aarch64_operands_ok_for_ldpstp): Likewise.
18849 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18850 (aarch64_gen_adjusted_ldpstp): Likewise.
18851 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18853 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18854 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18855 (arm_lookup_simd_builtin_type): Likewise.
18856 (arm_simd_builtin_type): Likewise.
18857 (arm_init_simd_builtin_types): Likewise.
18858 (arm_expand_builtin_args): Likewise.
18859 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18860 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18861 (ft32_setup_incoming_varargs): Likewise.
18862 (ft32_function_arg): Likewise.
18863 (ft32_function_arg_advance): Likewise.
18864 (ft32_pass_by_reference): Likewise.
18865 (ft32_arg_partial_bytes): Likewise.
18866 (ft32_valid_pointer_mode): Likewise.
18867 (ft32_addr_space_pointer_mode): Likewise.
18868 (ft32_addr_space_legitimate_address_p): Likewise.
18869 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18871 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18872 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18873 (iamcu_alignment): Likewise.
18874 (canonicalize_vector_int_perm): Likewise.
18875 (ix86_noce_conversion_profitable_p): Likewise.
18876 (ix86_mpx_bound_mode): Likewise.
18877 (ix86_operands_ok_for_move_multiple): Likewise.
18878 * config/microblaze/microblaze-protos.h
18879 (microblaze_expand_conditional_branch_reg): Likewise.
18880 * config/microblaze/microblaze.c
18881 (microblaze_expand_conditional_branch_reg): Likewise.
18882 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18884 (rs6000_reassociation_width): Likewise.
18885 (rs6000_invalid_binary_op): Likewise.
18886 (fusion_p9_p): Likewise.
18887 (emit_fusion_p9_load): Likewise.
18888 (emit_fusion_p9_store): Likewise.
18889 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18891 (riscv_hard_regno_mode_ok_p): Likewise.
18892 (riscv_address_insns): Likewise.
18893 (riscv_split_symbol): Likewise.
18894 (riscv_legitimize_move): Likewise.
18895 (riscv_function_value): Likewise.
18896 (riscv_hard_regno_nregs): Likewise.
18897 (riscv_expand_builtin): Likewise.
18898 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18899 (riscv_build_integer): Likewise.
18900 (riscv_split_integer): Likewise.
18901 (riscv_legitimate_constant_p): Likewise.
18902 (riscv_cannot_force_const_mem): Likewise.
18903 (riscv_regno_mode_ok_for_base_p): Likewise.
18904 (riscv_valid_base_register_p): Likewise.
18905 (riscv_valid_offset_p): Likewise.
18906 (riscv_valid_lo_sum_p): Likewise.
18907 (riscv_classify_address): Likewise.
18908 (riscv_legitimate_address_p): Likewise.
18909 (riscv_address_insns): Likewise.
18910 (riscv_load_store_insns): Likewise.
18911 (riscv_force_binary): Likewise.
18912 (riscv_split_symbol): Likewise.
18913 (riscv_force_address): Likewise.
18914 (riscv_legitimize_address): Likewise.
18915 (riscv_move_integer): Likewise.
18916 (riscv_legitimize_const_move): Likewise.
18917 (riscv_legitimize_move): Likewise.
18918 (riscv_address_cost): Likewise.
18919 (riscv_subword): Likewise.
18920 (riscv_output_move): Likewise.
18921 (riscv_canonicalize_int_order_test): Likewise.
18922 (riscv_emit_int_order_test): Likewise.
18923 (riscv_function_arg_boundary): Likewise.
18924 (riscv_pass_mode_in_fpr_p): Likewise.
18925 (riscv_pass_fpr_single): Likewise.
18926 (riscv_pass_fpr_pair): Likewise.
18927 (riscv_get_arg_info): Likewise.
18928 (riscv_function_arg): Likewise.
18929 (riscv_function_arg_advance): Likewise.
18930 (riscv_arg_partial_bytes): Likewise.
18931 (riscv_function_value): Likewise.
18932 (riscv_pass_by_reference): Likewise.
18933 (riscv_setup_incoming_varargs): Likewise.
18934 (riscv_print_operand): Likewise.
18935 (riscv_elf_select_rtx_section): Likewise.
18936 (riscv_save_restore_reg): Likewise.
18937 (riscv_for_each_saved_reg): Likewise.
18938 (riscv_register_move_cost): Likewise.
18939 (riscv_hard_regno_mode_ok_p): Likewise.
18940 (riscv_hard_regno_nregs): Likewise.
18941 (riscv_class_max_nregs): Likewise.
18942 (riscv_memory_move_cost): Likewise.
18943 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18944 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18945 (rl78_addr_space_address_mode): Likewise.
18946 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18948 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18949 (rs6000_reassociation_width): Likewise.
18950 (rs6000_invalid_binary_op): Likewise.
18951 (fusion_p9_p): Likewise.
18952 (emit_fusion_p9_load): Likewise.
18953 (emit_fusion_p9_store): Likewise.
18954 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18955 (ok_for_simple_move_operands): Likewise.
18956 (ok_for_simple_move_strict_operands): Likewise.
18957 (ok_for_simple_arith_logic_operands): Likewise.
18958 (visium_legitimize_reload_address): Likewise.
18959 (visium_select_cc_mode): Likewise.
18960 (output_cbranch): Likewise.
18961 (visium_split_double_move): Likewise.
18962 (visium_expand_copysign): Likewise.
18963 (visium_expand_int_cstore): Likewise.
18964 (visium_expand_fp_cstore): Likewise.
18965 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18966 (visium_function_arg): Likewise.
18967 (visium_function_arg_advance): Likewise.
18968 (visium_libcall_value): Likewise.
18969 (visium_setup_incoming_varargs): Likewise.
18970 (visium_legitimate_constant_p): Likewise.
18971 (visium_legitimate_address_p): Likewise.
18972 (visium_legitimize_address): Likewise.
18973 (visium_secondary_reload): Likewise.
18974 (visium_register_move_cost): Likewise.
18975 (visium_memory_move_cost): Likewise.
18976 (prepare_move_operands): Likewise.
18977 (ok_for_simple_move_operands): Likewise.
18978 (ok_for_simple_move_strict_operands): Likewise.
18979 (ok_for_simple_arith_logic_operands): Likewise.
18980 (visium_function_value_1): Likewise.
18981 (rtx_ok_for_offset_p): Likewise.
18982 (visium_legitimize_reload_address): Likewise.
18983 (visium_split_double_move): Likewise.
18984 (visium_expand_copysign): Likewise.
18985 (visium_expand_int_cstore): Likewise.
18986 (visium_expand_fp_cstore): Likewise.
18987 (visium_split_cstore): Likewise.
18988 (visium_select_cc_mode): Likewise.
18989 (visium_split_cbranch): Likewise.
18990 (output_cbranch): Likewise.
18991 (visium_print_operand_address): Likewise.
18992 * expmed.c (flip_storage_order): Likewise.
18993 * expmed.h (emit_cstore): Likewise.
18994 (flip_storage_order): Likewise.
18995 * genrecog.c (validate_pattern): Likewise.
18996 * hsa-gen.c (gen_hsa_addr): Likewise.
18997 * internal-fn.c (expand_arith_overflow): Likewise.
18998 * ira-color.c (allocno_copy_cost_saving): Likewise.
18999 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19000 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19001 (process_invariant_for_inheritance): Likewise.
19002 * lra-eliminations.c (move_plus_up): Likewise.
19003 * omp-low.c (lower_oacc_reductions): Likewise.
19004 * simplify-rtx.c (simplify_subreg): Likewise.
19005 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19006 (TARGET_CHKP_BOUND_MODE): Likewise..
19007 * targhooks.c (default_chkp_bound_mode): Likewise.
19008 (default_setup_incoming_vararg_bounds): Likewise.
19009 * targhooks.h (default_chkp_bound_mode): Likewise.
19010 (default_setup_incoming_vararg_bounds): Likewise.
19011 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19012 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19013 (have_whole_vector_shift): Likewise.
19014 * tree-vect-stmts.c (vectorizable_load): Likewise.
19015 * doc/tm.texi: Regenerate.
19017 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19019 Graceful degrade if Binutils PR21472 is not available.
19022 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19023 .rodata in flash test fails.
19024 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19025 * confgure: Regenerate.
19026 * config.in: Regenerate.
19027 * config/avr/avr.c (avr_asm_named_section)
19028 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19029 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19030 (avr_asm_init_sections): Same.
19032 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19034 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19035 (fma<VH:mode>4_intrinsic): Likewise.
19036 (*fmsub<VCVTF:mode>4): Likewise.
19037 (*fmsub<VH:mode>4_intrinsic): Likewise.
19039 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19042 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19043 Don't depend on "optimize > 0".
19044 (out_movhi_r_mr, out_movqi_mr_r): Same.
19045 (out_movhi_mr_r, out_movqi_r_mr): Same.
19046 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19047 io_address_operand on "optimize > 0".
19049 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19051 * tree-loop-distribution.c: Add general explanantion on the pass.
19052 (generate_loops_for_partition): Mark distributed loop.
19053 (pg_add_dependence_edges): New parameter. Handle alias data
19054 dependence specially and record it in the parameter if asked.
19055 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19056 (init_partition_graph_vertices, add_partition_graph_edge): New.
19057 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19058 (free_partition_graph_vdata, build_partition_graph): New.
19059 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19060 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19061 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19062 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19063 (version_for_distribution_p, finalize_partitions): New.
19064 (distribute_loop): Handle alias data dependence specially. Factor
19065 out loop fusion code as functions and call these functions.
19067 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19069 * tree-loop-distribution.c (classify_partition): New parameter and
19070 better handle reduction statement.
19071 (rdg_build_partitions): Revise comment.
19072 (distribute_loop): Compute statements in all partitions and pass it
19073 to classify_partition.
19075 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19077 * tree-loop-distribution.c (enum partition_type): New.
19078 (struct partition): New field type.
19079 (partition_merge_into): Add parameter. Update partition type.
19080 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19081 (build_rdg_partition_for_vertex): Compute partition type.
19082 (rdg_build_partitions): Dump partition type.
19083 (distribute_loop): Update calls to partition_merge_into.
19085 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19087 * tree-loop-distribution.c (struct ddr_hasher): New.
19088 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19090 (classify_partition): Call get_data_dependence.
19091 (pg_add_dependence_edges): Ditto.
19092 (distribute_loop): Release data dependence hash table.
19094 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19096 * tree-loop-distribution.c (ref_base_address): Delete.
19097 (similar_memory_accesses): Rename ...
19098 (share_memory_accesses): ... to this. Check if partitions access
19099 the same memory reference.
19100 (distribute_loop): Call share_memory_accesses.
19102 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19104 * tree-loop-distribution.c (struct partition): New field recording
19105 its data reference.
19106 (partition_alloc, partition_free): Init and release data refs.
19107 (partition_merge_into): Merge data refs.
19108 (build_rdg_partition_for_vertex): Collect data refs for partition.
19109 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19111 (distribute_loop): Remve data refs from vertice data of partition
19114 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19116 * tree-loop-distribution.c (params.h): Include header file.
19117 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19118 (datarefs_vec): New global var.
19119 (create_rdg_vertices): Use datarefs_vec directly.
19120 (free_rdg): Don't free data references.
19121 (build_rdg): Update use. Don't free data references.
19122 (distribute_loop): Compute global variable for data references.
19123 Bail out if there are too many data references.
19125 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19127 * tree-loop-distribution.c (loop_nest): New global var.
19128 (build_rdg): Use loop directly, rather than loop nest.
19129 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19131 (distribute_loop): Compute global variable loop nest. Update use.
19133 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19135 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19136 (partition_merge_into): New parameter. Dump reason for fusion.
19137 (distribute_loop): Update use of partition_merge_into.
19139 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19141 * tree-loop-distribution.c (bb_top_order_index): New.
19142 (bb_top_order_index_size, bb_top_order_cmp): New.
19143 (stmts_from_loop): Use topological order.
19144 (pass_loop_distribution::execute): Compute and release topological
19145 order for basic blocks.
19147 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19149 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19152 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19154 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19155 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19156 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19157 * internal-fn.def (LOOP_DIST_ALIAS): New.
19158 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19159 (fold_loop_internal_call): ... this.
19160 (vect_loop_dist_alias_call): New function.
19161 (set_uid_loop_bbs): Call fold_loop_internal_call.
19162 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19165 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19168 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19169 Require dead FLAGS_REG at the beginning of a peephole.
19171 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19174 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19175 arguments in the call to __builtin_ia32_sbb_u32.
19176 (_subborrow_u64): Swap _X and _Y arguments in the call to
19177 __builtin_ia32_sbb_u64.
19179 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19182 * tree-vrp.c (compare_assert_loc): Turn into a function template
19183 with stable template parameter. Only test if a->e is NULL,
19184 !a->e == !b->e has been verified already. Use e == NULL or
19185 e != NULL instead of e or ! e tests. If stable is true, don't use
19186 iterative_hash_expr, on the other side allow a or b or both NULL
19187 and sort the NULLs last.
19188 (process_assert_insertions): Sort using compare_assert_loc<false>
19189 instead of compare_assert_loc, later sort using
19190 compare_assert_loc<true> before calling process_assert_insertions_for
19191 in a loop. Use break instead of continue once seen NULL pointer.
19193 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19195 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19196 Cortex-R7 and Cortex-R8 processors.
19198 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19200 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19201 uninitialized while src is not.
19203 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19205 * common/config/arm/arm-common.c: Adjust include path for
19207 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19208 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19209 (arm-cpu-data.h): Likewise.
19210 (arm-cpu-cdata.h): Likewise.
19211 * config/arm/arm-cpu.h: Delete.
19212 * config/arm/arm-cpu-cdata.h: Delete.
19213 * config/arm/arm-cpu-data.h: Delete.
19215 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19217 * config/arm/arm-cpus.in (cortex-a55): New.
19218 (cortex-a75): Likewise.
19219 (cortex-a75.cortex-a55): Likewise.
19220 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19222 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19223 * config/arm/arm-cpu-cdata.h: Regenerate.
19224 * config/arm/arm-cpu-data.h: Regenerate.
19225 * config/arm/arm-cpu.h: Regenerate.
19226 * config/arm/arm-tables.opt: Regenerate.
19227 * config/arm/arm-tune.md: Regenerate.
19229 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19231 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19233 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19235 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19238 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19240 PR tree-optimization/81292
19241 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19242 full_string_p, also call adjust_related_strinfos if the adjustment
19243 is simple, otherwise invalidate related strinfos.
19245 2017-07-04 Martin Liska <mliska@suse.cz>
19248 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19249 newly created variable as DECL_IGNORED_P.
19251 2017-07-04 Martin Liska <mliska@suse.cz>
19254 * ipa-inline.c (inline_small_functions):
19255 Use xstrdup_for_dump.
19257 2017-07-04 Tom de Vries <tom@codesourcery.com>
19259 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19261 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19264 * config/darwin.c (darwin_function_switched_text_sections):
19267 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19269 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19271 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19273 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19275 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19277 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19278 min_profitable_iters, and th as inclusive lower bounds.
19279 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19280 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19281 for min_profitable_iters and min_profitable_estimate.
19282 (vect_transform_loop): Treat th as an inclusive lower bound.
19283 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19285 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19288 * config/darwin.c (darwin_function_switched_text_sections):
19289 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19290 in two pieces, and suppress the use of buf.
19292 2017-07-03 Nathan Sidwell <nathan@acm.org>
19294 * hash-table.h (hash_table_mod1): Fix indentation.
19296 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19298 PR middle-end/81290
19299 * predict.c (force_edge_cold): Be more careful about propagation
19301 * profile-count.h (profile_probability::guessed,
19302 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19304 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19306 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19308 * doc/invoke.texi (rcpc architecture extension): Document it.
19310 2017-07-03 Richard Biener <rguenther@suse.de>
19312 PR tree-optimization/60510
19313 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19314 the scalar reduction PHI and use it.
19315 (vectorizable_reduction): Properly guard the single_defuse_cycle
19316 path for non-SLP reduction chains where we cannot use it.
19317 Rework reduc_def/index and vector type deduction. Rework
19318 vector operand gathering during reduction op code-gen.
19319 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19320 chains dissolve the chain and leave it to non-SLP reduction
19323 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19325 * tree-data-ref.h (dr_alignment): Declare.
19326 * tree-data-ref.c (dr_alignment): New function.
19327 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19328 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19330 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19332 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19334 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19335 and base_misalignment fields.
19336 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19337 * tree-data-ref.c: Include builtins.h.
19338 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19339 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19340 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19341 * tree-vect-data-refs.c: Include tree-cfg.h.
19342 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19343 fields instead of calculating an alignment here.
19344 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19345 innermost_loop_behavior fields.
19347 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19349 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19351 (DR_STEP_ALIGNMENT): New macro.
19352 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19353 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19354 (create_data_ref): Print it.
19355 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19356 to tell whether the step preserves vector (mis)alignment.
19357 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19358 Move the check for an integer step and generalise to all INTEGER_CST.
19359 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19360 Print the outer step alignment.
19362 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19364 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19365 with offset_alignment.
19366 (DR_ALIGNED_TO): Delete.
19367 (DR_OFFSET_ALIGNMENT): New macro.
19368 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19369 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19370 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19371 (create_data_ref): Likewise.
19372 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19373 (vect_analyze_data_refs): Likewise.
19374 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19375 creating dummy innermost behavior.
19377 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19379 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19380 with a "innermost_loop_behavior *" and refeence tree.
19381 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19382 (create_data_ref): Update call accordingly.
19383 * tree-predcom.c (find_looparound_phi): Likewise.
19385 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19387 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19388 fields with dr_wrt_vec_loop.
19389 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19390 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19391 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19392 (vect_dr_behavior): New function.
19393 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19394 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19395 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19396 track whether the step preserves the misalignment.
19397 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19398 Use vect_dr_behavior.
19399 (vect_setup_realignment): Update call accordingly.
19400 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19401 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19402 call to vect_create_addr_base_for_vector_ref.
19403 (vect_create_cond_for_align_checks): Likewise.
19404 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19405 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19406 (vect_recog_mask_conversion_pattern): Likewise.
19407 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19408 (new_stmt_vec_info): Remove redundant zeroing.
19410 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19412 * common/config/arm/arm-common.c (arm_be8_option): New function.
19413 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19414 (ISA_ARMv6): Add isa_bit_be8.
19415 * config/arm/arm.h (arm_be8_option): Add prototype.
19416 (BE8_SPEC_FUNCTION): New define.
19417 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19418 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19419 (mlittle-endian): Similarly.
19420 (mbe8, mbe32): New options.
19421 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19422 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19424 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19426 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19428 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19430 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19431 (cleanup_tree_cfg_bb): Use it.
19432 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19434 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19436 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19439 * loop-doloop.c (add_test): Update profile.
19441 2017-07-03 Martin Liska <mliska@suse.cz>
19444 * sanopt.c (rewrite_usage_of_param): New function.
19445 (sanitize_rewrite_addressable_params): Likewise.
19446 (pass_sanopt::execute): Call rewrite_usage_of_param.
19448 2017-07-03 Richard Biener <rguenther@suse.de>
19450 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19451 back to using VIEW_CONVERT_EXPR.
19453 2017-07-03 Martin Liska <mliska@suse.cz>
19456 * doc/extend.texi: Document when a resolver function is
19457 generated for target_clones.
19459 2017-07-03 Martin Liska <mliska@suse.cz>
19461 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19462 and shadow memory of auto variables which are subject of
19463 use-after-scope sanitization.
19464 (asan_expand_mark_ifn): Add do set only when is_poison.
19466 2016-07-03 Richard Biener <rguenther@suse.de>
19468 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19470 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19471 (vectorizable_reduction): Perform reduction PHI creation when
19472 visiting a reduction PHI and adjust and simplify code generation
19473 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19474 (vect_transform_loop): Visit reduction PHIs.
19475 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19476 defs into the SLP tree.
19477 (vect_build_slp_tree): Reduction defs terminate the recursion.
19478 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19480 (vect_get_vec_defs_for_stmt_copy): Export.
19481 (vect_get_vec_defs): Likewise.
19482 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19484 (vect_get_vec_defs_for_stmt_copy): Declare.
19485 (vect_get_vec_defs): Likewise.
19487 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19489 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19490 parameter with a "loop" parameter and use it instead of the
19491 loop containing DR_STMT. Don't check simple_iv when doing
19492 BB analysis. Describe the two analysis modes in the comment.
19494 2017-07-03 Tom de Vries <tom@codesourcery.com>
19496 PR tree-optimization/69468
19497 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19498 (find_same_succ_bb): Handle ignore_edge_flags.
19500 2017-07-03 Tom de Vries <tom@codesourcery.com>
19502 PR tree-optimization/81192
19503 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19505 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19507 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19509 2017-07-03 Tom de Vries <tom@codesourcery.com>
19511 PR tree-optimization/81192
19512 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19513 BB_SAME_SUCC (bb) == NULL.
19515 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19517 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19520 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19522 * dumpfile.c: Include profile-count.h
19523 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19525 (insert_cond_bb): Update profile.
19526 * tree-cfg.h (insert_cond_bb): Update prototype.
19527 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19528 * tree-dump.c: Do not include tree-cfg.
19530 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19532 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19534 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19536 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19539 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19541 * tree-complex.c (expand_complex_div_wide): update profile.
19543 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19544 Alan Hayward <alan.hayward@arm.com>
19545 David Sherwood <david.sherwood@arm.com>
19547 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19548 (CORETYPES_H): New define.
19549 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19550 (insn-modes-inline.h, s-modes-inline-h): New rules.
19551 (generated_files): Add insn-modes-inline.h.
19552 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19553 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19554 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19555 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19556 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19557 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19558 (build/gencodes.o, build/genconditions.o): Likewise.
19559 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19560 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19561 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19562 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19563 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19564 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19565 * coretypes.h: Include everything up to real.h for generators.
19566 Include insn-modes.h first. Include wide-int-print.h after
19567 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19568 * machmode.h: Don't include insn-modes.h here.
19569 * function-tests.c: Remove includes of signop.h, machmode.h,
19570 double-int.h and wide-int.h.
19572 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19574 * optc-save-gen.awk: Likewise.
19575 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19576 * godump.c: Remove include of wide-int-print.h.
19577 * pretty-print.h: Likewise.
19578 * wide-int-print.cc: Likewise.
19579 * wide-int.cc: Likewise.
19580 * hash-map-tests.c: Remove include of signop.h.
19581 * hash-set-tests.c: Likewise.
19582 * rtl-tests.c: Likewise.
19583 * mkconfig.sh: Remove include of machmode.h.
19584 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19586 (emit_insn_modes_inline_h): ...this new function. Emit the code
19587 into an insn-modes-inline.h header file, adding appropriate
19588 include guards and end comments.
19589 (emit_insn_modes_c_header): Remove include of machmode.h.
19590 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19592 (main): Handle -i flag and call emit_insn_modes_inline_h when
19595 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19597 * tree-ssa-strlen.c (strinfo): Rename the length field to
19598 nonzero_chars. Add a full_string_p field.
19599 (compare_nonzero_chars, zero_length_string_p): New functions.
19600 (get_addr_stridx): Add an offset_out parameter.
19601 Use compare_nonzero_chars.
19602 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19603 (new_strinfo): Update after above changes to strinfo.
19604 (set_endptr_and_length): Set full_string_p.
19605 (get_string_length): Update after above changes to strinfo.
19606 (unshare_strinfo): Update call to new_strinfo.
19607 (maybe_invalidate): Likewise.
19608 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19609 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19610 as a uhwi instead of an shwi. Update after above changes to
19611 strinfo and new_strinfo.
19612 (zero_length_string): Assert that chainsi contains full strings.
19613 Use zero_length_string_p. Update call to new_strinfo.
19614 (adjust_related_strinfos): Update after above changes to strinfo.
19615 Copy full_string_p from origsi.
19616 (adjust_last_stmt): Use zero_length_string_p.
19617 (handle_builtin_strlen): Update after above changes to strinfo and
19618 new_strinfo. Install the lhs as the string length if the previous
19619 entry didn't describe a full string.
19620 (handle_builtin_strchr): Update after above changes to strinfo
19622 (handle_builtin_strcpy): Likewise.
19623 (handle_builtin_strcat): Likewise.
19624 (handle_builtin_malloc): Likewise.
19625 (handle_pointer_plus): Likewise.
19626 (handle_builtin_memcpy): Likewise. Track nonzero characters
19627 that aren't necessarily followed by a nul terminator.
19628 (handle_char_store): Likewise.
19630 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19632 PR tree-optimization/80769
19633 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19634 for malloc and calloc. Document the new invariant that all related
19635 strinfos have delayed lengths or none do.
19636 (verify_related_strinfos): Move earlier in file.
19637 (set_endptr_and_length): New function, split out from...
19638 (get_string_length): ...here. Also set the lengths of related
19640 (zero_length_string): Assert that chainsi has known (rather than
19642 (adjust_related_strinfos): Likewise.
19644 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19646 PR tree-optimization/81136
19647 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19648 assert that two references with the same misalignment have the same
19649 compile-time misalignment if those compile-time misalignments
19652 2017-07-01 Andi Kleen <ak@linux.intel.com>
19654 * print-tree.c (print_node): Print all attributes.
19656 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19658 * cfg.c (scale_bbs_frequencies): New function.
19659 * cfg.h (scale_bbs_frequencies): Declare it.
19660 * cfgloopanal.c (single_likely_exit): Cleanup.
19661 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19663 (scale_loop_profile): Likewise.
19664 (loop_version): Likewise.
19665 (create_empty_loop_on_edge): Update.
19666 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19667 scale_loop_frequencies, scale_loop_profile, loopify,
19668 loop_version): Update prototypes.
19669 * modulo-sched.c (sms_schedule): Update.
19670 * predict.c (unlikely_executed_edge_p): Also check probability.
19671 (probably_never_executed_edge_p): Fix typo.
19672 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19673 * tree-parloops.c (gen_parallel_loop): Update.
19674 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19675 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19676 * tree-ssa-loop-split.c (split_loop): Update.
19677 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19678 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19679 (vect_loop_versioning): Update.
19680 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19682 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19684 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19686 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19688 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19689 to keep profile consistent.
19691 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19693 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19694 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19695 * profile-count.h (max_safe_multiplier): Make unsigned.
19696 (profile_count::guessed_zero): New.
19698 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19700 * bb-reorder.c (fix_up_crossing_landing_pad,
19701 fix_crossing_conditional_branches): Use make_single_succ_edge
19702 to keep profile consistent.
19704 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19706 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19709 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19712 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19713 the right scopes, make sure cond_jump isn't preserved between multiple
19714 iterations. Search for fallthru edge whenever there are 3+ edges and
19715 use find_fallthru_edge for it.
19717 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19719 Patch by Alexander Monakov <amonakov@ispras.ru>
19720 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19721 probabilities consistently.
19723 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19725 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19726 emit_cmp_and_jump_insns.
19728 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19731 * tree-inline.c (expand_call_inline): Combine profile statuses.
19733 2017-06-30 Andrew Pinski <apinski@cavium.com>
19735 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19736 fold_stmt returned true.
19738 2017-06-30 Nathan Sidwell <nathan@acm.org>
19740 * ggc.h (empty_string): Delete.
19741 * cfgexpand.c (expand_asm_stmt): Use plain "".
19742 * optabs.c (expand_asm_memory_barrier): Likewise.
19743 * stringpool.c (empty_string): Delete.
19744 (digit_vector, digit_string): Delete.
19745 (ggc_alloc_string): Use plain "", don't optimize single digit
19746 strings. Use ggc_alloc_atomic.
19748 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19750 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19751 comparison set and one other set, use the cost of the non-comparison
19754 2017-06-30 Nathan Sidwell <nathan@acm.org>
19756 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19759 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19761 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19762 loops. Remove now unneeded calls to gimple_switch_set_label() that
19763 just set removed labels to NULL_TREE.
19765 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19767 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19768 (set_range_info): ...here. Only call set_range_info_raw if domain
19770 (set_nonzero_bits): Call set_range_info_raw.
19771 * tree-ssanames.h (set_range_info_raw): New.
19773 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19776 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19777 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19778 of nonimmediate_operand and <store_mask_constraint> instead of m
19779 for the input operand. For V8FI iterator, always split if input
19780 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19781 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19782 <store_mask_predicate> instead of register_operand and
19783 <store_mask_constraint> instead of v for the input operand. Make
19784 sure both operands aren't MEMs for if not <mask_applied>.
19786 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19788 * lto-wrapper.c (copy_file) Close both file descriptors before
19791 2017-06-30 Martin Liska <mliska@suse.cz>
19794 * multiple_target.c (create_dispatcher_calls): Make ifunc
19795 also for function that don't have calls or are not referenced.
19797 2017-06-30 Richard Biener <rguenther@suse.de>
19799 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19800 analyze the first scalar stmt. Move vector type computation
19801 for the BB case here from ...
19802 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19803 live operation processing in the SLP case properly.
19805 2017-06-30 Richard Biener <rguenther@suse.de>
19807 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19809 2017-06-30 Martin Liska <mliska@suse.cz>
19812 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19813 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19815 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19817 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19820 2017-06-30 Richard Biener <rguenther@suse.de>
19822 PR tree-optimization/81249
19823 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19824 condition reduction result to original scalar type.
19826 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19828 * profile-count.h (enum profile_quality): Fix typos and whitespace
19831 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19833 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19834 type for branch probabilities.
19836 2017-06-29 Julian Brown <julian@codesourcery.com>
19837 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19839 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19840 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19841 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19842 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19844 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19846 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19847 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19848 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19849 CC usage from generic code to here.
19850 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19851 CC usage into the target macros.
19853 2017-06-29 Maya Rashish <coypu@sdf.org>
19855 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19858 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19860 * arm/arm-builtins.c: Include profile-count.h
19861 * except.c (sjlj_emit_function_enter): Use
19862 profile_probability::unlikely.
19864 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19866 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19867 and tocrel_offset be pointer args rather than implicitly using
19869 (legitimate_constant_pool_address_p, rs6000_emit_move,
19870 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19871 tocrel_offset and use in toc_relative_expr_p call.
19872 (print_operand, print_operand_address): Use static tocrel_base_oac
19873 and tocrel_offset_oac.
19874 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19877 2017-06-29 Maya Rashish <coypu@sdf.org>
19879 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19881 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19883 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19884 objects, take into account only the alignment of 'op0' and 'mode1' if
19887 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19889 * ccmp.c (ccmp_tree_comparison_p): New function.
19890 (ccmp_candidate_p): Update to use above function.
19891 (get_compare_parts): New function.
19892 (expand_ccmp_next): Update to use new functions.
19893 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19895 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19896 take mode as argument.
19897 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19898 * expr.c (expand_expr_real_1): Pass mode as argument.
19900 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19902 * combine.c (combine_instructions): Print insns to dump_file, together
19905 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19907 * asan.c (asan_emit_stack_protection): Update.
19908 (create_cond_insert_point): Update.
19909 * auto-profile.c (afdo_propagate_circuit): Update.
19910 * basic-block.h (struct edge_def): Turn probability to
19911 profile_probability.
19912 (EDGE_FREQUENCY): Update.
19913 * bb-reorder.c (find_traces_1_round): Update.
19914 (better_edge_p): Update.
19915 (sanitize_hot_paths): Update.
19916 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19917 (make_single_succ_edge): Update.
19918 (check_bb_profile): Update.
19919 (dump_edge_info): Update.
19920 (update_bb_profile_for_threading): Update.
19921 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19922 probabilitycount to 0.
19923 * cfgbuild.c (compute_outgoing_frequencies): Update.
19924 * cfgcleanup.c (try_forward_edges): Update.
19925 (outgoing_edges_match): Update.
19926 (try_crossjump_to_edge): Update.
19927 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19928 (expand_gimple_tailcall): Update.
19929 (construct_init_block): Use make_single_succ_edge.
19930 (construct_exit_block): Use make_single_succ_edge.
19931 * cfghooks.c (verify_flow_info): Update.
19932 (redirect_edge_succ_nodup): Update.
19933 (split_edge): Update.
19934 (account_profile_record): Update.
19935 * cfgloopanal.c (single_likely_exit): Update.
19936 * cfgloopmanip.c (scale_loop_profile): Update.
19937 (set_zero_probability): Remove.
19938 (duplicate_loop_to_header_edge): Update.
19939 * cfgloopmanip.h (loop_version): Update prototype.
19940 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19941 (force_nonfallthru_and_redirect): Update.
19942 (update_br_prob_note): Update.
19943 (rtl_verify_edges): Update.
19944 (purge_dead_edges): Update.
19945 (rtl_lv_add_condition_to_bb): Update.
19946 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19947 * cgraphunit.c (init_lowered_empty_function): Update.
19948 (cgraph_node::expand_thunk): Update.
19949 * cilk-common.c: Include profile-count.h
19950 * dojump.c (inv): Remove.
19951 (jumpifnot): Update.
19952 (jumpifnot_1): Update.
19953 (do_jump_1): Update.
19955 (do_jump_by_parts_greater_rtx): Update.
19956 (do_compare_rtx_and_jump): Update.
19957 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19958 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19959 * dwarf2cfi.c: Include profile-count.h
19960 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19961 (sjlj_emit_dispatch_table): Likewise.
19962 * explow.c: Include profile-count.h
19963 * expmed.c (emit_store_flag_force): Update.
19964 (do_cmp_and_jump): Update.
19965 * expr.c (compare_by_pieces_d::generate): Update.
19966 (compare_by_pieces_d::finish_mode): Update.
19967 (emit_block_move_via_loop): Update.
19968 (store_expr_with_bounds): Update.
19969 (store_constructor): Update.
19970 (expand_expr_real_2): Update.
19971 (expand_expr_real_1): Update.
19972 * expr.h (try_casesi, try_tablejump): Update prototypes.
19973 * gimple-pretty-print.c (dump_probability): Update.
19974 (dump_profile): New.
19975 (dump_gimple_label): Update.
19976 (dump_gimple_bb_header): Update.
19977 * graph.c (draw_cfg_node_succ_edges): Update.
19978 * hsa-gen.c (convert_switch_statements): Update.
19979 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19980 (find_if_case_1): Update.
19981 (find_if_case_2): Update.
19982 * internal-fn.c (expand_arith_overflow_result_store): Update.
19983 (expand_addsub_overflow): Update.
19984 (expand_neg_overflow): Update.
19985 (expand_mul_overflow): Update.
19986 (expand_vector_ubsan_overflow): Update.
19987 * ipa-cp.c (good_cloning_opportunity_p): Update.
19988 * ipa-split.c (split_function): Use make_single_succ_edge.
19989 * ipa-utils.c (ipa_merge_profiles): Update.
19990 * loop-doloop.c (add_test): Update.
19991 (doloop_modify): Update.
19992 * loop-unroll.c (compare_and_jump_seq): Update.
19993 (unroll_loop_runtime_iterations): Update.
19994 * lra-constraints.c (lra_inheritance): Update.
19995 * lto-streamer-in.c (input_cfg): Update.
19996 * lto-streamer-out.c (output_cfg): Update.
19997 * mcf.c (adjust_cfg_counts): Update.
19998 * modulo-sched.c (sms_schedule): Update.
19999 * omp-expand.c (expand_omp_for_init_counts): Update.
20000 (extract_omp_for_update_vars): Update.
20001 (expand_omp_ordered_sink): Update.
20002 (expand_omp_for_ordered_loops): Update.
20003 (expand_omp_for_generic): Update.
20004 (expand_omp_for_static_nochunk): Update.
20005 (expand_omp_for_static_chunk): Update.
20006 (expand_cilk_for): Update.
20007 (expand_omp_simd): Update.
20008 (expand_omp_taskloop_for_outer): Update.
20009 (expand_omp_taskloop_for_inner): Update.
20010 * omp-simd-clone.c (simd_clone_adjust): Update.
20011 * optabs.c (expand_doubleword_shift): Update.
20012 (expand_abs): Update.
20013 (emit_cmp_and_jump_insn_1): Update.
20014 (expand_compare_and_swap_loop): Update.
20015 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20016 * predict.c (predictable_edge_p): Update.
20017 (edge_probability_reliable_p): Update.
20018 (set_even_probabilities): Update.
20019 (combine_predictions_for_insn): Update.
20020 (combine_predictions_for_bb): Update.
20021 (propagate_freq): Update.
20022 (estimate_bb_frequencies): Update.
20023 (force_edge_cold): Update.
20024 * profile-count.c (profile_count::dump): Add missing space into dump.
20025 (profile_count::debug): Add newline.
20026 (profile_count::differs_from_p): Explicitly convert to unsigned.
20027 (profile_count::stream_in): Update.
20028 (profile_probability::dump): New member function.
20029 (profile_probability::debug): New member function.
20030 (profile_probability::differs_from_p): New member function.
20031 (profile_probability::differs_lot_from_p): New member function.
20032 (profile_probability::stream_in): New member function.
20033 (profile_probability::stream_out): New member function.
20034 * profile-count.h (profile_count_quality): Rename to ...
20035 (profile_quality): ... this one.
20036 (profile_probability): New.
20037 (profile_count): Update.
20038 * profile.c (compute_branch_probabilities): Update.
20039 * recog.c (peep2_attempt): Update.
20040 * sched-ebb.c (schedule_ebbs): Update.
20041 * sched-rgn.c (find_single_block_region): Update.
20042 (compute_dom_prob_ps): Update.
20043 (schedule_region): Update.
20044 * sel-sched-ir.c (compute_succs_info): Update.
20045 * stmt.c (struct case_node): Update.
20046 (do_jump_if_equal): Update.
20047 (get_outgoing_edge_probs): Update.
20048 (conditional_probability): Update.
20049 (emit_case_dispatch_table): Update.
20050 (expand_case): Update.
20051 (expand_sjlj_dispatch_table): Update.
20052 (emit_case_nodes): Update.
20053 * targhooks.c: Update.
20054 * tracer.c (better_p): Update.
20055 (find_best_successor): Update.
20056 * trans-mem.c (expand_transaction): Update.
20057 * tree-call-cdce.c: Update.
20058 * tree-cfg.c (gimple_split_edge): Upate.
20059 (move_sese_region_to_fn): Upate.
20060 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20061 * tree-eh.c (lower_resx): Upate.
20062 (cleanup_empty_eh_move_lp): Upate.
20063 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20064 * tree-inline.c (copy_edges_for_bb): Update.
20065 (copy_cfg_body): Update.
20066 * tree-parloops.c (gen_parallel_loop): Update.
20067 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20068 (gimple_gen_time_profiler): Update.
20069 * tree-ssa-dce.c (remove_dead_stmt): Update.
20070 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20071 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20072 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20073 (unloop_loops): Update.
20074 (try_peel_loop): Update.
20075 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20076 * tree-ssa-loop-split.c (connect_loops): Update.
20077 (split_loop): Update.
20078 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20079 (hoist_guard): Update.
20080 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20081 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20082 (value_replacement): Update.
20083 * tree-ssa-reassoc.c (branch_fixup): Update.
20084 * tree-ssa-tail-merge.c (replace_block_by): Update.
20085 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20086 (create_edge_and_update_destination_phis): Update.
20087 (compute_path_counts): Update.
20088 (recompute_probabilities): Update.
20089 (update_joiner_offpath_counts): Update.
20090 (freqs_to_counts_path): Update.
20091 (duplicate_thread_path): Update.
20092 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20093 (struct switch_conv_info): Update.
20094 (gen_inbound_check): Update.
20095 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20096 (vect_do_peeling): Update.
20097 (vect_loop_versioning): Update.
20098 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20099 (optimize_mask_stores): Update.
20100 * ubsan.c (ubsan_expand_null_ifn): Update.
20101 * value-prof.c (gimple_divmod_fixed_value): Update.
20102 (gimple_divmod_fixed_value_transform): Update.
20103 (gimple_mod_pow2): Update.
20104 (gimple_mod_pow2_value_transform): Update.
20105 (gimple_mod_subtract): Update.
20106 (gimple_mod_subtract_transform): Update.
20107 (gimple_ic): Update.
20108 (gimple_stringop_fixed_value): Update.
20109 (gimple_stringops_transform): Update.
20110 * value-prof.h: Update.
20112 2017-06-29 Carl Love <cel@us.ibm.com>
20114 * config/rs6000/rs6000-c.c: Add support for built-in functions
20115 vector signed int vec_signed (vector float);
20116 vector signed long long vec_signed (vector double);
20117 vector signed int vec_signed2 (vector double, vector double);
20118 vector signed int vec_signede (vector double);
20119 vector signed int vec_signedo (vector double);
20120 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20121 instruction generator.
20122 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20123 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20124 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20126 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20127 vunsignede_v2df): Add define_expands.
20128 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20129 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20130 VEC_UNSIGNEDO): Add definitions.
20131 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20132 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20133 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20134 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20135 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20136 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20137 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20138 vec_unsignede, vec_unsignedo): Add builtin defines.
20139 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20141 * doc/extend.texi: Update the built-in documentation file for the
20142 new built-in functions.
20144 2017-06-29 Richard Biener <rguenther@suse.de>
20146 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20147 reduction chains to LOOP_VINFO_REDUCTIONS.
20148 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20149 SLP reductions after processing reduction chains.
20151 2017-06-29 Nathan Sidwell <nathan@acm.org>
20153 * builtins.c (fold_builtin_FUNCTION): Use
20154 lang_hooks.decl_printable_name.
20156 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20158 PR middle-end/81194
20159 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20160 with only one label.
20161 * stmt.c (expand_case): Assert NCASES is greater than one.
20163 2017-06-29 Richard Biener <rguenther@suse.de>
20165 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20167 (group_case_labels): Likewise.
20168 (find_taken_edge): Push sanity checking on val to workers...
20169 (find_taken_edge_cond_expr): ... here
20170 (find_taken_edge_switch_expr): ... and here, handle cases
20171 with just a default label.
20172 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20173 (group_case_labels): Likewise.
20174 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20175 group_case_labels does anything cleanup the CFG again.
20177 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20179 PR tree-optimization/81196
20180 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20181 exit condition comparing two IVs.
20183 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20185 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20186 profile to the dummy entry at the end of the list of architectures.
20187 * config/arm/arm-cpu-cdata.h: Regenerated.
20189 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20190 Michael Collison <michael.collison@arm.com>
20193 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20195 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20196 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20197 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20198 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20199 of shift when the shift amount is masked with constant equal to
20200 the size of the mode.
20201 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20204 2017-06-29 Martin Liska <mliska@suse.cz>
20206 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20208 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20211 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20214 (edge_growth_cache_entry::edge_growth_cache_entry): New
20216 (reset_edge_growth_cache): Update to use constructor.
20218 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20220 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20221 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20222 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20224 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20226 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20227 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20229 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20231 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20232 (*-linux-uclibc*): Add t-uclibc tmake_file.
20233 * config/t-musl: New.
20234 * config/t-uclibc: New.
20236 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20238 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20240 (gen_comm_data): Emit architectural setting of arch_prof.
20241 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20243 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20244 (armv8-m.base, armv8-m.main): Likewise.
20245 * arm-protos.h (arm_build_target): Add profile field.
20246 (arch_option): Likewise.
20247 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20249 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20250 arm_active_target.profile.
20252 2017-06-28 Richard Biener <rguenther@suse.de>
20254 PR middle-end/81227
20255 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20256 TYPE_OVERFLOW_WRAPS.
20257 * match.pd (negate_expr_p): Likewise.
20258 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20259 fold_build2, not fold_binary.
20261 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20263 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20264 Convert memory address to Pmode.
20265 (aarch64_print_operand): Assert MEM operands are always Pmode.
20267 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20270 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20271 Remove redundant if.
20272 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20273 * config/arm/aarch-common-protos.h
20274 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20275 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20277 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20280 * multiple_target.c (create_dispatcher_calls): Set the default
20281 clone to be static, not public.
20283 2017-06-28 Richard Biener <rguenther@suse.de>
20285 * tree-vect-loop.c (vectorizable_reduction): Move special
20286 cond reduction IV var creation ...
20287 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20288 parameter. Use STMT_VINFO_VECTYPE.
20289 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20292 2017-06-28 Martin Liska <mliska@suse.cz>
20295 * ipa-visibility.c (non_local_p): Handle visibility.
20297 2017-06-28 Martin Liska <mliska@suse.cz>
20300 * common.opt: Add IntegerRange to various options.
20301 * opt-functions.awk (integer_range_info): New function.
20302 * optc-gen.awk: Add integer_range_info to cl_options struct.
20303 * opts-common.c (decode_cmdline_option): Handle
20304 CL_ERR_INT_RANGE_ARG.
20305 (cmdline_handle_error): Likewise.
20306 * opts.c (print_filtered_help): Show valid interval in
20307 when --help is provided.
20308 * opts.h (struct cl_option): Add range_min and range_max fields.
20309 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20311 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20313 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20314 (x * C EQ/NE y * C): New transformation.
20316 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20318 * genmultilib (combination_space): Accept '+' in option names.
20320 2017-06-28 Martin Liska <mliska@suse.cz>
20323 * asan.c (instrument_derefs): Bail out inner references
20324 that are hard register variables.
20326 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20329 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20330 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20332 2017-06-28 Richard Biener <rguenther@suse.de>
20334 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20335 (vect_get_slp_defs): Adjust.
20336 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20338 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20340 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20341 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20342 (vectorizable_reduction): Adjust.
20343 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20345 (vect_get_slp_defs): Likewise.
20346 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20347 (vectorizable_bswap): Adjust.
20348 (vectorizable_call): Likewise.
20349 (vectorizable_conversion): Likewise.
20350 (vectorizable_assignment): Likewise.
20351 (vectorizable_shift): Likewise.
20352 (vectorizable_operation): Likewise.
20353 (vectorizable_store): Likewise.
20354 (vectorizable_condition): Likewise.
20355 (vectorizable_comparison): Likewise.
20357 2017-06-28 Michael Collison <michael.collison@arm.com>
20360 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20362 (arm_gen_movmemqi): Removed unused variable 'i'.
20363 Convert 'for' loop into 'while' loop.
20364 (arm_expand_prologue): Remove last unnecessary set of insn.
20365 (thumb_pop): Remove unused variable 'pushed_words'.
20366 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20368 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20370 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20371 * config/s390/s390-protos.h: Add prototype of
20372 s390_rel_address_ok_p.
20373 * config/s390/s390.c (s390_got_symbol): New function.
20374 (s390_rel_address_ok_p): New function.
20375 (legitimize_pic_address): Use s390_rel_address_ok_p.
20376 (s390_load_got): Use s390_got_symbol.
20377 (s390_option_override): Issue error if
20378 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20379 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20381 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20383 2017-06-27 Andrew Pinski <apinski@cavium.com>
20385 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20386 (X * copysign (1.0, X)): New pattern.
20387 (X * copysign (1.0, -X)): New pattern.
20388 (copysign (-1.0, CST)): New pattern.
20390 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20392 * genmultilib (combination_space): Remove variable.
20393 Validate reuse rules against regular expression for any sequence
20394 of multilib options in any order.
20396 2017-06-27 Michael Collison <michael.collison@arm.com>
20398 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20399 call aarch64_split_simd_combine.
20400 * (aarch64_combine_internal<mode>): Delete pattern.
20401 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20402 Allow register and subreg operands.
20404 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20406 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20407 specific need, just fallback on defaults.
20408 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20410 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20411 Olivier Hainque <hainque@adacore.com>
20413 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20415 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20416 targets. Pick a default if no particular attempt applied.
20417 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20420 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20422 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20423 (x86_64-wrs-vxworks7): Likewise.
20425 2017-06-27 Marek Polacek <polacek@redhat.com>
20428 * ubsan.c (instrument_null): Check get_base_address's result for null.
20430 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20432 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20434 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20436 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20437 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20438 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20439 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20440 New function types.
20441 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20442 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20443 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20444 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20445 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20446 BUILT_IN_FEUPDATEENV): New builtins.
20447 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20448 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20449 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20450 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20452 (builtin_structptr_types): Adjust size.
20453 * tree.c (builtin_structptr_types): Add four entries.
20455 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20456 Olivier Hainque <hainque@adacore.com>
20458 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20459 (TLS_SYM): New local macro, forcing reference to __tls__ on
20460 link command lines for VxWorks 7 RTPs, triggering initialization
20462 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20463 OS features TLS support, true for RTPs on VxWorks 7.
20464 * config/vxworks.c (vxworks_override_options): Setup emutls
20467 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20469 * predict.c (test_prediction_value_range): Use -1U instead of -1
20470 to avoid narrowing conversion warning.
20471 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20472 to avoid narrowing conversion warning.
20473 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20475 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20477 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20479 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20480 64bit configurations.
20481 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20482 (SIZE_TYPE): Likewise.
20483 * config/vxworks.c (vxworks_emutls_var_fields): Use
20484 long_unsigned_type_node instead of unsigned_type_node as the offset
20485 field type, which is "pointer" mode in emutls.c.
20487 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20490 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20492 PR middle-end/81207
20493 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20494 gimple_vuse copying separately from gimple_vdef copying.
20496 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20498 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20501 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20502 Olivier Hainque <hainque@adacore.com>
20504 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20505 for all vxworks7 targets.
20506 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20507 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20508 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20509 variations for VX6/VX7 and 32/64bits later on in ...
20510 (VXWORKS_LIB_SPEC): Leverage new macros.
20511 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20512 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20514 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20516 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20517 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20519 2017-06-26 Carl Love <cel@us.ibm.com>
20521 * config/rs6000/rs6000-c.c: Add support for built-in functions
20522 vector bool char vec_reve (vector bool char);
20523 vector signed char vec_reve (vector signed char);
20524 vector unsigned char vec_reve (vector unsigned char);
20525 vector bool int vec_reve (vector bool int);
20526 vector signed int vec_reve (vector signed int);
20527 vector unsigned int vec_reve (vector unsigned int);
20528 vector bool long long vec_reve (vector bool long long);
20529 vector signed long long vec_reve (vector signed long long);
20530 vector unsigned long long vec_reve (vector unsigned long long);
20531 vector bool short vec_reve (vector bool short);
20532 vector signed short vec_reve (vector signed short);
20533 vector double vec_reve (vector double);
20534 vector float vec_reve (vector float);
20535 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20536 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20537 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20538 (altivec_vreve): New pattern.
20539 * config/rs6000/altivec.h (vec_reve): New define.
20540 * doc/extend.texi (vec_rev): Update the built-in documentation file
20541 for the new built-in functions.
20543 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20545 PR tree-optimization/71815
20546 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20548 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20550 (slsr_process_phi): Likewise.
20551 (replace_uncond_cands_and_profitable_phis): Don't replace a
20552 multiply candidate with a stride of 1 (copy or cast).
20553 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20555 (lowest_cost_path): Likewise.
20556 (total_savings): Likewise.
20558 2017-06-26 Richard Biener <rguenther@suse.de>
20561 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20562 Use def_builtin_pure for all gather builtins.
20564 2017-06-26 Richard Biener <rguenther@suse.de>
20566 PR tree-optimization/81203
20567 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20568 non-dominating BBs.
20570 2017-06-26 Marek Polacek <polacek@redhat.com>
20573 * doc/invoke.texi: Document -Wmultistatement-macros.
20575 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20577 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20578 arm_neon_ok_no_float_abi effective target.
20580 2017-06-26 Richard Biener <rguenther@suse.de>
20582 PR tree-optimization/80928
20583 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20584 (copy_bbs): Set BB_DUPLICATED flag early.
20585 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20587 (execute_on_shrinking_pred): Likewise.
20588 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20589 BB_DUPLICATED blocks.
20590 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20591 iterate over all PHIs considering removal of *gsi.
20593 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20595 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20598 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20600 * config/rs6000/rs6000-string.c: (expand_block_clear,
20601 do_load_for_compare, select_block_compare_mode,
20602 compute_current_alignment, expand_block_compare,
20603 expand_strncmp_align_check, expand_strn_compare,
20604 expand_block_move, rs6000_output_load_multiple)
20605 Move functions related to string/block move/compare
20606 to a separate file.
20607 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20608 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20609 for this function which is now used in two files.
20610 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20611 * config.gcc: Add rs6000-string.o to extra_objs for
20612 targets powerpc*-*-* and rs6000*-*-*.
20614 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20617 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20618 32-bit, since indexed is not valid for DImode.
20619 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20620 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20621 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20622 (define_peephole2 for Altivec d-form store): Likewise.
20625 * multiple_target.c (create_dispatcher_calls): Only create the
20626 dispatcher call if the function is the default clone of a
20627 versioned function.
20629 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20631 PR middle-end/80902
20632 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20633 a call, force the call to not be a tail call.
20635 2017-06-23 Jeff Law <law@redhat.com>
20637 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20640 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20642 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20643 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20644 handling for early expansion of vector shifts (sl,sr,sra,rl).
20645 (builtin_function_type): Add vector shift right instructions
20646 to the unsigned argument list.
20648 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20650 rtl-optimizatoin/79286
20651 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20652 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20653 trap. PIC register plus a const unspec without offset can never trap.
20655 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20657 * tree.h (builtin_structptr_type): New type.
20658 (builtin_structptr_types): Declare new array.
20659 * tree.c (builtin_structptr_types): New array.
20660 (free_lang_data, build_common_tree_nodes): Use it.
20662 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20665 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20668 2017-06-22 Matt Turner <mattst88@gmail.com>
20670 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20671 Lake models to skylake case. Assume skylake for unknown
20672 models with clflushopt.
20674 2017-06-22 Jeff Law <law@redhat.com>
20676 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20677 frame sizes that do not satisfy aarch64_uimm12_shift.
20679 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20681 * profile-count.h (apply_probability,
20682 apply_scale, probability_in): Fix checks for zero.
20684 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20686 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20687 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20689 2016-06-22 Richard Biener <rguenther@suse.de>
20691 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20692 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20693 REDUC_MAX_EXPR support.
20694 (vectorizable_reduction): Likewise.
20695 (vect_create_epilog_for_reduction): Likewise.
20697 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20699 * match.pd (A / (1 << B) -> A >> B): New.
20700 * generic-match-head.c: Include optabs-tree.h.
20701 * gimple-match-head.c: Likewise.
20702 * optabs-tree.h (target_supports_op_p): New.
20703 * optabs-tree.c (target_supports_op_p): New.
20705 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20707 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20708 $gcc_cv_ld --help output.
20709 (gcc_cv_ld_demangle): Likewise.
20710 (gcc_cv_ld_eh_frame_hdr): Likewise.
20711 (gcc_cv_ld_pie): Likewise.
20712 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20713 (gcc_cv_ld_buildid): Likewise.
20714 (gcc_cv_ld_sysroot): Likewise.
20715 (ld_bndplt_support): Likewise.
20716 (ld_pushpopstate_support): Likewise.
20717 * configure: Regenerate.
20718 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20720 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20723 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20724 operands indexes to avoid gap between operands and match_dups.
20726 2017-06-21 Andrew Pinski <apinski@cavium.com>
20728 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20729 Increment Arith_shift and Arith_shift_reg by 1.
20730 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20732 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20733 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20734 (aarch64_strip_extend): Add new argument and test for it.
20735 (aarch64_cheap_mult_shift_p): New function.
20736 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20737 add a cost if it is true.
20738 Update calls to aarch64_strip_extend.
20739 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20741 2017-06-21 Andrew Pinski <apinski@cavium.com>
20743 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20745 (thunderxt88): Likewise.
20746 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20747 (thunderx_prefetch_tune): New variable.
20748 (thunderx2t99_prefetch_tune): Update for the correct values.
20749 (thunderxt88_tunings): New variable.
20750 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20751 generic_prefetch_tune.
20752 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20754 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20756 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20757 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20758 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20759 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20760 (aarch64_atomic_cas<mode>, GPI): Likewise.
20762 2017-06-21 Martin Liska <mliska@suse.cz>
20764 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20765 statements on cold and hot labels.
20766 * predict.c (tree_estimate_probability_bb): Remove the
20767 prediction from this place.
20769 2017-06-21 Martin Liska <mliska@suse.cz>
20771 PR tree-optimization/79489
20772 * gimplify.c (maybe_add_early_return_predict_stmt): New
20774 (gimplify_return_expr): Call the function.
20775 * predict.c (tree_estimate_probability_bb): Remove handling
20777 * predict.def: Update comment about early return predictor.
20778 * gimple-predict.h (is_gimple_predict): New function.
20779 * predict.def: Change default value of early return to 66.
20780 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20782 * passes.def: Put pass_strip_predict_hints to the beginning of
20785 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20787 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20788 FUNCTION_DECL declarations.
20789 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20791 (dwaf2out_decl): Likewise.
20792 * godump.c (go_early_global_decl): Skip call to the real debug hook
20793 for FUNCTION_DECL declarations.
20794 * passes.c (rest_of_decl_compilation): Skip call to the
20795 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20796 -fdump-go-spec is passed.
20798 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20800 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20801 Reorder for compactness.
20802 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20803 (def_builtin_pure, def_builtin_pure2): New functions.
20804 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20806 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20808 * match.pd (nop_convert): New predicate.
20809 ((A +- CST1) +- CST2): Allow some NOP conversions.
20811 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20814 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20815 with ctors/dtors if GOVD_SHARED is set.
20817 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20819 * config/aarch64/aarch64.md (movti_aarch64):
20820 Emit mov rather than orr.
20821 (movtf_aarch64): Likewise.
20822 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20823 Emit mov rather than orr.
20825 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20827 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20828 Swap alternatives, make integer dup more expensive.
20830 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20832 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20833 Return true for non-tls symbols.
20835 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20837 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20838 (cortex-a75): Likewise.
20839 (cortex-a75.cortex-a55): Likewise.
20840 * config/aarch64/aarch64-tune.md: Regenerate.
20841 * doc/invoke.texi (-mtune): Document new values for -mtune.
20843 2017-06-21 Tom de Vries <tom@codesourcery.com>
20845 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20846 stack_size feature.
20847 (Effective-Target Keywords, Other attributes): Suggest using
20848 dg-add-options stack_size feature to get stack limit in stack_size
20849 effective target documentation.
20851 2017-06-21 Julian Brown <julian@codesourcery.com>
20852 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20854 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20855 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20856 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20858 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20859 attribute type list for neon_multiply.
20860 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20861 attribute type list for neon_multiply.
20862 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20863 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20864 attribute type list for neon_multiply.
20865 * config/arm/types.md (crypto_pmull): Add.
20866 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20867 attribute type list.
20869 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20871 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20874 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20876 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20877 to make sure not to dereference a NULL cost_classes_ptr pointer.
20879 2017-06-20 Carl Love <cel@us.ibm.com>
20881 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20882 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20883 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20884 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20885 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20886 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20887 VMULOSW): New enum "unspec" values.
20888 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20889 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20890 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20891 altivec_vmulosw): New patterns.
20892 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20893 VMULOSW): Add definitions.
20895 2017-06-20 Julia Koval <julia.koval@intel.com>
20897 * config/i386/i386.c: Fix rounding expand for new pattern.
20898 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20900 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20902 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20903 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20905 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20907 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20910 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20912 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20913 architecture, then by implementer ID.
20914 * config/aarch64/aarch64-tune.md: Regenerate.
20916 2017-06-20 Richard Biener <rguenther@suse.de>
20918 PR middle-end/81097
20919 * fold-const.c (split_tree): Fold to type before negating.
20921 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20923 * diagnostic-show-locus.c
20924 (selftest::test_fixit_deletion_affecting_newline): New function.
20925 (selftest::diagnostic_show_locus_c_tests): Call it.
20927 2017-06-20 Andreas Schwab <schwab@suse.de>
20930 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20933 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20935 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20936 __ARM_FEATURE_COPROC according to support.
20938 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20940 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20941 Rewritten to avoid overflow for > 32-bit pointers.
20944 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20945 by removing enum keyword.
20946 (ubsan_type_descriptor): Likewise. Formatting fix.
20949 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20950 splitter): Require TARGET_SSE2 in the condition.
20952 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20955 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20956 for doing vector set of SFmode on ISA 3.0.
20957 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20958 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20960 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20961 SFmode value into a V4SF variable that was extracted from another
20962 V4SF variable without converting the element to double precision
20963 and back to single precision vector format.
20964 (vsx_insert_extract_v4sf_p9_2): Likewise.
20966 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20968 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20969 in UWHI to avoid undefined overflow.
20972 * ubsan.h (enum ubsan_encode_value_phase): New.
20973 (ubsan_encode_value): Change second argument to
20974 enum ubsan_encode_value_phase with default value of
20975 UBSAN_ENCODE_VALUE_GENERIC.
20976 * ubsan.c (ubsan_encode_value): Change second argument to
20977 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20978 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20979 create_tmp_var_raw instead of create_tmp_var and use a
20981 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20982 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20983 ubsan_encode_value callers.
20986 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20987 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20988 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20990 2017-06-19 Richard Biener <rguenther@suse.de>
20992 PR middle-end/81118
20993 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20994 estimates if we changed anything.
20996 2017-06-19 Richard Biener <rguenther@suse.de>
20998 PR tree-optimization/80887
20999 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21000 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21001 simplified lookups, then reset mprts_hook.
21002 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21004 (try_to_simplify): Likewise.
21006 2017-06-19 Martin Liska <mliska@suse.cz>
21009 * gimplify.c (gimplify_switch_expr):
21010 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21012 2017-06-19 Martin Liska <mliska@suse.cz>
21014 * doc/install.texi: Document that PGO runs in 4 stages.
21016 2017-06-19 Martin Liska <mliska@suse.cz>
21019 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21020 to dispatcher function name.
21021 * multiple_target.c (replace_function_decl): New function.
21022 (create_dispatcher_calls): Redirect both edges and references.
21024 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21026 * profile-count.c (profile_count::dump): Dump quality.
21027 (profile_count::differs_from_p): Update for unsigned val.
21028 * profile-count.h (profile_count_quality): New enum.
21029 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21031 2017-06-19 Richard Biener <rguenther@suse.de>
21033 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21034 struct function as arg.
21035 (estimate_numbers_of_iterations): Export overload with loop arg.
21036 (free_numbers_of_iterations_estimates_loop): Use an overload of
21037 free_numbers_of_iterations_estimates instead.
21038 * tree-cfg.c (remove_bb): Adjust.
21039 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21040 * tree-parloops.c (gen_parallel_loop): Likewise.
21041 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21043 (tree_unroll_loops_completely): Likewise.
21044 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21045 Use an overload instead and export.
21046 (estimated_loop_iterations): Adjust.
21047 (max_loop_iterations): Likewise.
21048 (likely_max_loop_iterations): Likewise.
21049 (estimate_numbers_of_iterations): Take struct function as arg
21051 (loop_exits_before_overflow): Adjust.
21052 (free_numbers_of_iterations_estimates_loop): Use an overload.
21053 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21054 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21056 2017-06-19 Richard Biener <rguenther@suse.de>
21059 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21060 RANGE_EXPR conservatively.
21062 2017-06-16 Carl Love <cel@us.ibm.com>
21064 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21065 definitions for vec_float, vec_float2, vec_floato,
21066 vec_floate built-ins.
21067 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21068 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21070 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21071 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21072 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21073 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21074 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21075 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21076 vec_floato): Add builtin defines.
21077 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21078 Update the built-in documentation file for the new built-in
21081 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21083 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21084 (mthumb): Mark as the negative of -marm.
21086 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21088 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21090 (ARM Options, -mtune): Document that this accepts the same
21091 extension options as -mcpu.
21092 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21094 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21096 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21097 permitted extensions.
21099 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21101 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21102 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21103 (armv8-m.main): Add option +nodsp.
21104 * config/arm/arm-cpu-cdata.h: Regenerated.
21106 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21108 * config/arm/t-fuchsia: New file.
21109 * config.gcc (arm*-*-fuchsia*): Use it.
21111 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21113 * config/arm/t-symbian: Rewrite for new option infrastructure.
21115 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21117 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21118 (MULTILIB_REQUIRED): Likewise.
21120 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21122 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21123 (MULTILIB_RESUE): Likewise.
21124 (MULTILIB_MATCHES): Likewise.
21125 (MULTLIB_REQUIRED): Likewise.
21127 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21129 * config/arm/t-rtems: Rewrite for new option framework.
21131 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21133 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21134 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21135 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21136 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21137 * config/arm/t-multilib: ... here.
21138 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21139 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21140 armv7-a and armv8*-a when A-profile libraries have not been built.
21141 * config/arm/t-rmprofile: Rewrite.
21143 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21145 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21146 with a backslash. Remove the backslash after substituting unescaped
21148 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21150 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21152 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21153 the driver to pass through the default setting of -mfloat-abi.
21154 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21156 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21157 (all_feat_combs): New rule.
21158 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21160 * config/arm/t-aprofile: Rewrite.
21162 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21164 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21165 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21166 fpu is not specified by the user/command-line.
21167 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21168 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21169 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21170 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21171 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21172 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21174 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21176 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21177 * config/arm/t-arm-elf: Rewritten.
21179 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21181 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21182 have some floating-point instructions.
21183 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21184 (TARGET_MAYBE_HARD_FLOAT): New macro.
21185 * config/arm/arm-builtins.c (arm_init_builtins): Use
21186 TARGET_MAYBE_HARD_FLOAT.
21187 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21191 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21192 (configargs.h): Include it.
21193 (arm_print_hint_for_fpu_option): New function.
21194 (arm_parse_fpu_option): New function.
21195 (candidate_extension): New class.
21196 (arm_canon_for_multilib): New function.
21197 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21198 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21199 (ARCH_CANONICAL_SPECS): New macro.
21200 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21202 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21204 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21205 are set after handling multilib fragments. Set target_cpu_default2
21208 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21210 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21212 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21213 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21215 (check_arch): Likewise.
21216 * config/arm/arm.c (arm_configure_build_target): Handle
21217 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21218 options in the default.
21220 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21222 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21223 when an option is an alias of another.
21224 * config/arm/parsecpu.awk (optalias): New parser token.
21225 (gen_comm_data): Mark non-alias options as such. Emit entries
21226 for extension aliases.
21227 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21228 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21229 (armv6kz, armv6zk, armv6t2): Likewise.
21230 (armv7): Make vfpv3-d16 an alias.
21231 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21233 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21234 Sort in canonical order.
21235 (armv8-a): Sort in canonical order.
21236 (armv8.1-a, armv8.2-a): Likewise.
21237 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21239 (cortex-a9): Sort in canonical order.
21240 * config/arm/arm.c (selftests.h): Include it.
21241 (arm_test_cpu_arch_data): New function.
21242 (arm_run_self_tests): New function.
21243 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21244 (targetm): Move declaration to the end of the file.
21245 * arm-cpu-cdata.h: Regenerated.
21247 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21249 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21250 call to target_mode_check describing the type of option passed.
21251 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21252 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21253 arm_parse_cpu_option_name to match parameters against list of
21255 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21256 arm_arch_core_flags data structure.
21257 * config/arm/arm-cpu_cdata.h: Regenerated.
21259 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21261 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21263 (arm_print_hint_for_cpu_option): Likewise.
21264 (arm_print_hint_for_arch_option): Likewise.
21265 (arm_parse_cpu_option_name): Likewise.
21266 (arm_parse_arch_option_name): Likewise.
21267 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21268 of entries in the all_fpus list.
21269 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21270 (arm_parse_cpu_option_name): Declare.
21271 (arm_parse_arch_option_name): Declare.
21272 (arm_parse_option_features): Declare.
21273 (arm_intialize_isa): Declare.
21274 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21276 (gen_comm_data): ... here. Make definitions non-static.
21277 * config/arm/arm-cpu-data.h: Regenerated.
21278 * config/arm/arm-cpu-cdata.h: Regenerated.
21280 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21282 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21283 (cpu_arch_extension): New structure.
21284 (cpu_arch_option, arch_option, cpu_option): New structures.
21285 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21286 architecture types.
21287 (gen_data): Generate new format data tables.
21288 * config/arm/arm.c (cpu_tune): New structure.
21289 (cpu_option, processors): Delete.
21290 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21291 (arm_print_hint_for_cpu_option): ... this and ...
21292 (arm_print_hint_for_arch_option): ... this.
21293 (arm_parse_arch_cpu_name): Delete. Replace with ...
21294 (arm_parse_cpu_option_name): ... this and ...
21295 (arm_parse_arch_option_name): ... this.
21296 (arm_unrecognized_feature): Change type of target parameter to
21298 (arm_parse_arch_cpu_features): Delete. Replace with ...
21299 (arm_parse_option_features): ... this.
21300 (arm_configure_build_target): Rework to use new configuration data
21302 (arm_print_tune_info): Rework for new configuration data tables.
21303 * config/arm/arm-cpu-data.h: Regenerated.
21304 * config/arm/arm-cpu.h: Regenerated.
21306 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21308 * Makefile.in (OBJS): Move sbitmap.o from here ...
21309 (OBJS-libcommon): ... to here.
21311 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21313 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21314 (ISA_ALL_CRYPTO): New macro.
21315 (ISA_ALL_SIMD): New macro
21316 (ISA_ALL_FP): New macro.
21317 * config/arm/arm.c (fpu_bitlist): Update initializer.
21318 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21320 (arm9e): Add fpu. Add option for nofp
21321 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21322 (arm926ej-s, arm1026ej-s): Likewise.
21323 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21324 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21325 neon-fp16, neon-vfpv4, nofp and nosimd.
21326 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21327 (cortex-a8): Add fpu. Add option for nofp.
21328 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21329 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21330 (cortex-r4f): Add fpu.
21331 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21332 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21334 (cortex-r8): Likewise.
21335 (cortex-m4): Add fpu. Add option for nofp.
21336 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21337 (cortex-a17.cortex-a7): Likewise.
21338 (cortex-a32): Add fpu. Add options for crypto and nofp.
21339 (cortex-a35, cortex-a53): Likewise.
21340 (cortex-a57): Add fpu. Add option for crypto.
21341 (cortex-a72, cortex-a73): Likewise.
21342 (exynos-m1): Likewise.
21343 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21344 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21345 (cortex-m33): Add fpu. Add option for nofp.
21346 * config/arm/arm-cpu-cdata.h: Regenerated
21347 * config/arm/arm-cpu-data.h: Regenerated.
21349 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21351 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21352 (armv5te, armv5tej): Likewise.
21353 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21354 (armv7): Add options fp and vfpv3-d16.
21355 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21356 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21358 (armv7ve): Likewise.
21359 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21360 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21361 (armv8-a): Add nocrypto option.
21362 (armv8.1-a, armv8.2-a): Likewise.
21363 (armv8-m.main): add options fp, fp.dp and nofp.
21365 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21367 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21369 (armv8-a+crc): Delete.
21370 (armv8.1-a): Add options simd, crypto and nofp.
21371 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21372 (armv8.2-a+fp16): Delete.
21373 (armv8-m.main): Add option dsp.
21374 (armv8-m.main+dsp): Delete.
21375 (cortex-a8): Add fpu. Add nofp option.
21376 (cortex-a9): Add fpu. Add nofp and nosimd options.
21377 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21378 link to main cpu and architecture data structures.
21379 (gen_comm_data): Only put isa attributes from the main architecture
21381 (option): New statement for architecture and CPU entries.
21382 * arm.c (struct cpu_option): New structure.
21383 (struct processors): Add entry for options.
21384 (arm_unrecognized_feature): New function.
21385 (arm_parse_arch_cpu_name): Ignore any characters after the first
21387 (arm_parse_arch_cpu_feature): New function.
21388 (arm_configure_build_target): Separate out any CPU and architecture
21389 features and parse separately. Don't error out if -mfpu=auto is
21390 used with only an architecture string.
21391 (arm_print_asm_arch_directives): New function.
21392 (arm_file_start): Call it.
21393 * config/arm/arm-cpu-cdata.h: Regenerated.
21394 * config/arm/arm-cpu-data.h: Likewise.
21395 * config/arm/arm-tables.opt: Likewise.
21397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21399 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21400 assembler when it is not -mfpu=auto.
21402 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21404 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21405 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21406 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21407 (ASM_CPU_SPEC): Rewrite.
21408 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21409 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21410 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21411 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21412 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21413 copied string is NUL-terminated. Also strip any characters prefixed
21415 (arm_rewrite_selected_arch): New function.
21416 (arm_rewrite_march): New function.
21418 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21420 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21421 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21422 (march, mcpu, mtune): Convert to string-based options.
21423 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21424 (arm_parse_arch_cpu_name): New function.
21425 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21426 identify selected architecture or CPU.
21427 (arm_option_save): New function.
21428 (TARGET_OPTION_SAVE): Redefine.
21429 (arm_option_restore): Restore string options.
21430 (arm_option_print): Print string options.
21432 2017-06-16 Martin Sebor <msebor@redhat.com>
21434 PR tree-optimization/80933
21435 PR tree-optimization/80934
21436 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21437 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21438 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21439 (gimple_fold_builtin): Call them.
21441 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21443 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21444 as unlikely; update profile.
21446 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21448 * predict.c (force_edge_cold): Handle declaring edges impossible
21451 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21453 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21455 (try_unroll_loop_completely): Fix reporting.
21457 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21459 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21461 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21464 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21465 if given a non-constant argument for an intrinsic which requires a
21468 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21470 * profile.c (compare_freqs): New function.
21471 (branch_prob): Sort edge list.
21472 (find_spanning_tree): Assume that the list is priority sorted.
21474 2017-06-16 Richard Biener <rguenther@suse.de>
21476 PR tree-optimization/81090
21477 * passes.def (pass_record_bounds): Remove.
21478 * tree-pass.h (make_pass_record_bounds): Likewise.
21479 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21480 make_pass_record_bounds): Likewise.
21481 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21482 not free niter estimates at the beginning but at the end.
21483 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21485 2017-06-16 Richard Biener <rguenther@suse.de>
21487 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21488 initializer to workaround ICE in host GCC 4.8.
21490 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21492 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21494 (clone_inlined_nodes): Update.
21496 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21498 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21499 prefetch settings, and enable prefetching by default at -O3.
21501 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21503 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21504 Set flag_prefetch_loop_arrays according to tuning data.
21506 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21508 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21509 New tune structure.
21510 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21511 [Unrelated to main purpose of the patch] Place the pointer field last
21512 to enable type checking errors when tune structure are wrongly merged.
21513 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21514 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21515 (thunderx2t99_prefetch_tune): New tune constants.
21516 (tune_params *_tunings): Update all tunings (no functional change).
21517 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21518 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21519 from tunings structures.
21521 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21524 * ubsan.c (instrument_null): Add T argument, use it instead
21525 of computing it based on IS_LHS.
21526 (instrument_object_size): Likewise.
21527 (pass_ubsan::execute): Adjust instrument_null and
21528 instrument_object_size callers to pass gimple_get_lhs or
21529 gimple_assign_rhs1 result to it. Use instrument_null instead of
21530 calling get_base_address and instrument_mem_ref. Handle
21531 aggregate call arguments for object-size sanitization.
21533 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21535 PR tree-optimization/81089
21536 * tree-vrp.c (is_masked_range_test): Validate operands of
21539 2017-06-15 Martin Sebor <msebor@redhat.com>
21542 * dumpfile.c (dump_register): Avoid calling memset to initialize
21543 a class with a default ctor.
21544 * gcc.c (struct compiler): Remove const qualification.
21545 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21546 * hash-table.h: Ditto.
21547 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21549 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21550 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21552 * params.h (struct param_info): Make struct members non-const.
21553 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21554 with default initialization.
21555 * vec.h (vec_copy_construct, vec_default_construct): New helper
21557 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21558 with vec_copy_construct.
21559 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21560 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21561 * doc/invoke.texi (-Wclass-memaccess): Document.
21563 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21565 * emit-rtl.h (is_leaf): Update comment about local
21566 register allocator.
21568 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21571 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21572 for a variable to have a section before checking if the section has a
21574 Set section to.persistent if persistent attribute is set.
21575 Warn if .persistent attribute is used on an automatic variable.
21577 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21579 PR rtl-optimization/80474
21580 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21582 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21584 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21587 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21589 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21590 (casesi): Emit bounds checking as RTL.
21591 (casesi_internal_mips16_<mode>): Remove bounds checking.
21593 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21595 * config/xtensa/xtensa.c (xtensa_option_override): Append
21596 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21597 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21598 xtensa_doloop_hooks): Define unconditionally.
21599 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21601 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21602 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21603 for it in xtensa_option_override.
21604 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21605 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21607 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21609 * doc/cppopts.texi: Document '-' special value to -MF.
21611 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21613 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21614 (cortex_a53_fconst): Likewise.
21615 (cortex_a53_fpmul): Likewise.
21616 (cortex_a53_f_load_64): Likewise.
21617 (cortex_a53_f_load_many): Likewise.
21618 (cortex_a53_advsimd_alu): Likewise.
21619 (cortex_a53_advsimd_alu_q): Likewise.
21620 (cortex_a53_advsimd_mul): Likewise.
21621 (cortex_a53_advsimd_mul_q): Likewise.
21622 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21623 Add missing fmul, r2f_cvt and fconst cases.
21625 2017-06-14 Richard Biener <rguenther@suse.de>
21627 PR middle-end/81088
21628 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21630 (fold_binary_loc): When associating do not treat pre-existing
21631 TREE_OVERFLOW on literal constants as a reason to allow
21632 TREE_OVERFLOW on associated literal constants.
21634 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21636 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21637 (MASK_FEATURES): New macro.
21638 * config/sparc/sparc.c (sparc_option_override): Remove the special
21639 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21641 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21643 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21644 a division of 0 if non-call exceptions are enabled.
21646 2017-06-14 Andrew Pinski <apinski@cavium.com>
21647 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21650 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21651 Improve vector initialization code gen for only variable case.
21653 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21655 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21657 2017-06-14 Richard Biener <rguenther@suse.de>
21659 PR tree-optimization/81083
21660 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21663 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21665 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21666 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21667 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21668 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21669 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21670 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21672 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21674 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21675 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21677 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21679 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21681 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21683 * config/rs6000/t-rtems: Don't handle SPE.
21685 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21687 * config/rs6000/t-linux: Don't handle SPE.
21689 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21691 * config/rs6000/eabispe.h: Delete file.
21693 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21695 * config/rs6000/t-spe: Delete file.
21697 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21699 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21700 (rs6000_legitimate_offset_address_p): Return false for anything in
21701 V2SImode or V2SFmode.
21703 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21705 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21706 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21707 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21709 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21710 (rs6000_legitimate_offset_address_p): Ditto.
21711 (rs6000_emit_move): Ditto.
21712 (rs6000_init_builtins): Remove V4HI_type_node.
21714 2017-06-13 Martin Liska <mliska@suse.cz>
21717 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21718 (gate_asan): Likewise.
21719 * asan.h (asan_no_sanitize_address_p): Remove the function.
21720 (sanitize_flags_p): New function.
21721 * builtins.def: Fix coding style.
21722 * common.opt: Use renamed enum value.
21723 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21724 * doc/extend.texi: Document no_sanitize attribute.
21725 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21726 to SANITIZE_UNDEFINED_NONDEFAULT.
21727 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21728 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21729 Use sanitize_flags_p.
21730 * gimplify.c (gimplify_function_tree): Likewise.
21731 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21732 * opts.c (parse_no_sanitize_attribute): New function.
21733 (common_handle_option): Use renamed enum value.
21734 * opts.h (parse_no_sanitize_attribute): Declare.
21735 * tree.c (sanitize_flags_p): New function.
21736 * tree.h: Declared here.
21737 * tsan.c: Use sanitize_flags_p.
21738 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21739 (instrument_mem_ref): Likewise.
21740 (instrument_bool_enum_load): Likewise.
21741 (do_ubsan_in_current_function): Remove the function.
21742 (pass_ubsan::execute): Use sanitize_flags_p.
21743 * ubsan.h: Remove do_ubsan_in_current_function
21744 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21745 (dump_function_to_file): Use it here.
21747 2017-06-13 Martin Jambor <mjambor@suse.cz>
21749 PR tree-optimization/80803
21750 PR tree-optimization/81063
21751 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21752 (propagate_subaccesses_across_link): Enqueue subtree whenever
21753 necessary instead of relying on the caller.
21755 2017-06-13 Martin Jambor <mjambor@suse.cz>
21757 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21758 that have a first_link.
21759 (sort_and_splice_var_accesses): Do not check first_link before
21761 (subtree_mark_written_and_enqueue): Likewise.
21762 (propagate_all_subaccesses): Likewise and do not stop at first
21763 parent with a first_link.
21765 2017-06-13 Martin Jambor <mjambor@suse.cz>
21767 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21770 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21772 * match.pd: New pattern.
21774 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21776 * tree-vrp.c (is_masked_range_test): New function.
21777 (register_edge_assert_for): Determine ranges for
21780 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21782 PR tree-optimization/67328
21783 * fold-const.c (maskable_range_p): New function.
21784 (build_range_check): Generate bittests if possible.
21786 2017-06-13 Martin Liska <mliska@suse.cz>
21788 * gimple-pretty-print.c (dump_probability): Add new argument.
21789 (dump_edge_probability): Dump both probability and count.
21790 (dump_gimple_label): Likewise.
21791 (dump_gimple_bb_header): Likewise.
21793 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21796 * config/avr/avr-devices.c: Fix indentation.
21797 * config/avr/gen-avr-mmcu-specs.c: Dito.
21799 2017-06-13 Richard Biener <rguenther@suse.de>
21801 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21802 instead get vector type from stmt_info.
21803 (vectorizable_reduction): Adjust. Remove dead code.
21805 2017-06-13 Richard Biener <rguenther@suse.de>
21807 PR middle-end/81065
21808 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21809 case of C * (x * C2 + C3).
21810 (fold_addr_of_array_ref_difference): Properly fold index difference.
21812 2017-06-12 David S. Miller <davem@davemloft.net>
21815 * config/sparc/sparc.md (return expander): Emit frame blockage if
21816 function uses alloca.
21818 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21820 * combine.c (make_field_assignment): Check len rather than the mode
21821 precision when calling force_to_mode.
21823 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21825 Support multilibs and devices that see flash in RAM address range.
21828 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21829 (avr_mcu_t) <flash_pm_offset>: New field.
21830 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21831 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21832 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21833 (AVR_TINY_PM_OFFSET): Remove macro.
21834 * config/avr/avr.opt (-mshort-calls): New option.
21835 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21836 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21837 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21838 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21839 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21840 instead of avr_arch->have_jmp_call.
21841 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21842 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21843 avr_arch->flash_pm_offset to define.
21844 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21845 new field flash_pm_offset. Add entry for avrxmega3.
21846 (avr_texinfo): Add entry for avrxmega3.
21847 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21848 attiny212, attiny214,
21849 attiny412, attiny414, attiny416, attiny417,
21850 attiny814, attiny816, attiny817,
21851 attiny1614, attiny1616, attiny1617,
21852 attiny3214, attiny3216, attiny3217.
21853 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21854 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21855 (avr_print_operand_address) [AVR_TINY]: Same.
21856 (avr_asm_init_sections) <readonly_data_section>: Only patch
21857 callback if avr_arch->flash_pm_offset = 0.
21858 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21859 for rodata if avr_arch->flash_pm_offset != 0.
21860 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21861 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21862 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21863 (m_options): Append opt_rcall.
21864 (m_dirnames): Append dir_rcall.
21865 * config/avr/t-multilib: Regenerate.
21867 * configure.ac [target=avr]: Check whether avrxmega3 default
21868 linker description file works as needed.
21869 * configure: Regenerate.
21870 * doc/avr-mmcu.texi: Regenerate.
21871 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21872 <__AVR_ARCH__>: Document avrxmega3 and 103.
21873 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21874 <__AVR_SHORT_CALLS__>: Document it.
21875 <__AVR_PM_BASE_ADDRESS__>: Document it.
21876 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21877 (AVR Variable Attributes) <progmem>: Document this is
21878 not needed for avrxmega3.
21879 (AVR Named Address Spaces) <__flash>: Dito.
21881 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21883 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21885 2017-06-12 Doug Rupp <rupp@adacore.com>
21887 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21888 Append vxworks-stdint.h to the tm_file list.
21889 * config/vxworks-stdint.h: New file.
21891 2017-06-12 Martin Liska <mliska@suse.cz>
21893 PR tree-optimization/81041
21894 * tree-profile.c (gimple_gen_ic_func_profiler):
21895 Create an extra BB in profile-generate
21896 (gimple_gen_time_profiler): Likewise.
21898 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21900 PR tree-optimization/81003
21901 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21902 (update_range_test): Use it instead of force_gimple_operand_gsi.
21904 2017-06-12 Richard Biener <rguenther@suse.de>
21906 PR tree-optimization/81053
21907 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21908 with backedge value not defined in loop. Simplify def stmt
21911 2017-06-11 Tom de Vries <tom@codesourcery.com>
21914 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21916 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21917 nvptx_cannot_force_const_mem.
21919 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21921 * opts.c (finish_options): Move test for flag_split_stack after
21922 it has been initialized.
21924 2017-06-11 Jason Merrill <jason@redhat.com>
21926 * tree.h (id_equal): New.
21927 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21928 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21929 instead of strcmp of IDENTIFIER_POINTER.
21931 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21933 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21934 (mark_all_inlined_calls_cdtor): Fix formating.
21935 (inline_transform): Rescale profile before inlining.
21937 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21939 * cgraph.h (cgraph_edge::clone): Update prototype.
21940 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21941 (cgraph_node::create_clone): Update.
21942 (cgraph_node::create_version_clone): Update.
21943 * tree-inline.c (copy_bb): Update.
21944 (expand_call_inline): Update.
21946 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21948 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21949 factored out from ...
21950 (rs6000_emit_prologue): ... here.
21952 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21954 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21955 factored out from ...
21956 (rs6000_emit_prologue): ... here.
21958 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21960 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21962 (handle_missing_profiles): Fix computation of tp_first_run.
21963 (counts_to_freqs): Do not touch freqs when count is 0.
21965 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21967 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21970 2017-06-10 Tom de Vries <tom@codesourcery.com>
21972 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21973 attributes): Document signal effective target.
21975 2017-06-10 Tom de Vries <tom@codesourcery.com>
21977 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21978 Document effective target stack_size.
21980 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21982 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21983 to the edit_context if they can be auto-applied.
21985 2017-06-9 Ian Lance Taylor <iant@golang.org>
21987 * opts.c (finish_options): If -fsplit-stack, disable implicit
21988 -forder-blocks-and-partition.
21989 * doc/invoke.texi (Optimize Options): Document that when using
21990 -fsplit-stack -forder-blocks-and-partition is not implicitly
21993 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21995 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21996 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21997 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21998 * builtins.def (abort, trap, unreachable): Declare cold.
21999 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22000 * tree-core.h (ECF_COLD): New.
22001 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22002 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22004 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22006 * predict.c (unlikely_executed_stmt_p): Cleanup.
22008 2017-06-09 Richard Biener <rguenther@suse.de>
22010 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22011 model if the ref is always written to.
22013 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22015 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22017 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22019 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22022 2017-06-09 Tom de Vries <tom@codesourcery.com>
22025 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22026 "target cannot support label values" when encountering LABEL_REF.
22028 2017-06-09 Martin Liska <mliska@suse.cz>
22030 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22031 (gimple_gen_ic_func_profiler): Emit direct comparison
22032 of __gcov_indirect_call_callee with NULL.
22033 (gimple_gen_time_profiler): Change probability from
22034 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22036 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22038 * profile.c (edge_gcov_counts): Turn to pointer.
22039 (compute_branch_probabilities, compute_branch_probabilities): Update.
22040 (branch_prob): Do not clear edge_gcov_count.
22041 * profile.h (edge_gcov_counts): Turn to pointer.
22042 (edge_gcov_count): Update.
22044 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22046 * gimple.h (gimple_check_failed): Mark cold.
22048 2017-06-09 Richard Biener <rguenther@suse.de>
22050 PR tree-optimization/66623
22051 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22052 refactor check_reduction into two parts, properly computing
22053 whether we have to check reduction validity for outer loop
22056 2017-06-09 Richard Biener <rguenther@suse.de>
22058 PR tree-optimization/79483
22059 * graphite-scop-detection.c (order): New global.
22060 (get_order): Compute bb to order mapping that satisfies code
22061 generation constraints.
22062 (cmp_pbbs): New helper.
22063 (build_scops): Start domwalk at entry block, sort generated
22066 2017-06-09 Richard Biener <rguenther@suse.de>
22068 PR middle-end/81007
22069 * ipa-polymorphic-call.c
22070 (ipa_polymorphic_call_context::restrict_to_inner_class):
22071 Skip FIELD_DECLs with error_mark_node type.
22072 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22075 2017-06-09 Martin Liska <mliska@suse.cz>
22077 * predict.c (struct branch_predictor): New struct.
22078 (test_prediction_value_range): New test.
22079 (predict_c_tests): New function.
22080 * selftest-run-tests.c (selftest::run_tests): Run the function.
22081 * selftest.h: Declare new tests.
22083 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22086 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22087 gen_add3_insn did not fail.
22088 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22089 r0, construct that number in a temporary reg and add that reg to r0.
22090 If asked to put the result in r0 as well, fail.
22092 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22094 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22095 for early expansion of vec_eqv.
22097 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22099 PR middle-end/81005
22100 * ubsan.c (instrument_null): Avoid pointless code temporary.
22101 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22103 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22107 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22110 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22113 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22115 * predict.c (unlikely_executed_edge_p): Move ahead.
22116 (probably_never_executed_edge_p): Use it.
22118 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22120 PR middle-end/79988
22121 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22122 gimple_call_builtin_p call.
22124 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22126 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22127 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22128 rtl_check_failed_type2, rtl_check_failed_code1,
22129 rtl_check_failed_code2, rtl_check_failed_code_mode,
22130 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22131 rtvec_check_failed_bounds, rtl_check_failed_flag,
22132 _fatal_insn_not_found, _fatal_insn): Likewise.
22133 * tree.h (tree_contains_struct_check_failed,
22134 tree_check_failed, tree_not_check_failed,
22135 tree_class_check_failed, tree_range_check_failed,
22136 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22137 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22138 tree_operand_check_failed, omp_clause_check_failed,
22139 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22142 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22144 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22145 flag_branch_probabilities.
22146 * ipa-inline.c (edge_badness): Likewise.
22147 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22148 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22149 * predict.c (maybe_hot_frequency_p): Likewise.
22150 (probably_never_executed): Likewise.
22151 * sched-ebb.c (schedule_ebbs): Likewise.
22152 * sched-rgn.c (find_single_block_region): Likewise.
22153 * tracer.c (tail_duplicate): Likewise.
22155 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22157 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22158 longer requires x_flag_profile_use.
22160 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22162 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22163 instead of flag_reorder_blocks_and_partition.
22164 * dbxout.c (dbxout_function_end): Likewise.
22165 * dwarf2out.c (gen_subprogram_die): Likewise.
22166 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22167 * hw-doloop.c (reorg_loops): Likewise.
22168 * varasm.c (assemble_start_function,
22169 assemble_end_function): Likewise.
22170 (decide_function_section): Do not check for
22171 flag_reorder_blocks_and_partition.
22173 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22175 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22177 (chkp_get_hard_register_fake_addr_expr): Ditto.
22178 (chkp_build_addr_expr): Add check for hard reg case.
22179 (chkp_parse_array_and_component_ref): Ditto.
22180 (chkp_find_bounds_1): Ditto.
22181 (chkp_process_stmt): Don't generate bounds store for
22184 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22186 * predict.c (maybe_hot_bb_p): Do not check profile status.
22187 (maybe_hot_edge_p): Likewise.
22188 (probably_never_executed): Check for zero counts even if profile
22190 (unlikely_executed_edge_p): New function.
22191 (unlikely_executed_stmt_p): New function.
22192 (unlikely_executed_bb_p): New function.
22193 (set_even_probabilities): Use unlikely predicates.
22194 (combine_predictions_for_bb): Likewise.
22195 (predict_paths_for_bb): Likewise.
22196 (predict_paths_leading_to_edge): Likewise.
22197 (determine_unlikely_bbs): New function.
22198 (estimate_bb_frequencies): Use it.
22199 (compute_function_frequency): Use zero counts even if profile is
22201 * profile-count.h: Fix typo.
22203 2017-08-08 Julia Koval <julia.koval@intel.com>
22205 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22206 _mm512_mask_cvtsepi16_storeu_epi8,
22207 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22208 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22209 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22210 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22211 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22212 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22213 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22214 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22215 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22216 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22217 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22218 __builtin_ia32_pmovuswb256mem_mask,
22219 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22220 __builtin_ia32_pmovwb512mem_mask): New builtins.
22222 2017-08-08 Julia Koval <julia.koval@intel.com>
22224 PR target/73350,80862
22225 * config/i386/subst.md (round): Fix round pattern.
22226 * config/i386/i386.c (ix86_erase_embedded_rounding):
22227 Fix erasing rounding for the fixed pattern.
22229 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22231 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22233 2017-06-08 Martin Liska <mliska@suse.cz>
22235 PR gcov-profile/80911
22236 * gcov.c (block_info::block_info): New constructor.
22238 2017-06-07 Carl Love <cel@us.ibm.com>
22240 * config/rs6000/rs6000-c: The return type of the following
22241 built-in functions was implemented as int not long long. Fix sign
22242 of return value for the unsigned version of vec_mulo and vec_mule.
22243 vector unsigned long long vec_bperm (vector unsigned long long,
22244 vector unsigned char)
22245 vector signed long long vec_mule (vector signed int,
22247 vector unsigned long long vec_mule (vector unsigned int,
22248 vector unsigned int)
22249 vector signed long long vec_mulo (vector signed int,
22251 vector unsigned long long vec_mulo (vector unsigned int,
22252 vector unsigned int)
22253 * doc/extend.texi: Fix the documentation for the built-in
22256 2017-06-07 Carl Love <cel@us.ibm.com>
22259 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22262 2017-06-07 Carl Love <cel@us.ibm.com>
22264 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22265 support, Generate doublehv for signed int/float for BE case only.
22267 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22269 * doc/invoke.texi (mcx16): Rewrite.
22271 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22273 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22274 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22275 *mov<mode>_softfloat, and an anonymous splitter): Use
22276 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22278 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22280 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22282 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22283 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22284 (rs6000_debug_reg_global): Adjust.
22285 (rs6000_init_hard_regno_mode_ok): Adjust.
22286 (rs6000_dbx_register_number): Adjust.
22287 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22288 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22289 Remove SPE_ACC and SPEFSCR.
22290 (REG_ALLOC_ORDER): Ditto.
22291 (FRAME_POINTER_REGNUM): Change to 111.
22292 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22293 (REG_CLASS_NAMES): Ditto.
22294 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22295 (REGISTER_NAMES): Ditto.
22296 (ADDITIONAL_REG_NAMES): Ditto.
22297 (rs6000_reg_names): Ditto.
22298 * config/rs6000/rs6000.md: Renumber some register number
22301 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22303 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22305 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22306 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22308 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22309 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22310 Delete the SPE high registers.
22311 (REG_ALLOC_ORDER): Ditto.
22312 (enum reg_class): Remove SPE_HIGH_REGS.
22313 (REG_CLASS_NAMES): Ditto.
22314 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22315 (REGISTER_NAMES): Ditto.
22316 (rs6000_reg_names): Ditto.
22317 * doc/tm.texi.in: Remove SPE as example.
22318 * doc/tm.texi: Regenerate.
22320 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22322 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22323 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22324 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22325 * config/rs6000/rs6000.md (type): Remove "brinc".
22327 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22329 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22330 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22331 * config/rs6000/linuxspe.h: Delete file.
22332 * config/rs6000/rs6000.md: Don't include spe.md.
22333 * config/rs6000/spe.h: Delete file.
22334 * config/rs6000/spe.md: Delete file.
22335 * config/rs6000/t-rs6000: Remove spe.md.
22337 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22339 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22340 (reg_or_none500mem_operand): Delete.
22341 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22342 instead of reg_or_none500mem_operand.
22344 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22346 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22347 handling of SPE flags.
22348 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22350 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22352 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22354 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22355 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22356 paired_divv2sf3): Similar.
22357 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22358 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22359 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22361 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22362 Rename the paired_* instruction patterns.
22363 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22365 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22366 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22367 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22368 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22369 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22370 PAIRED_VECTOR_MODE.
22371 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22372 (spe_func_has_64bit_regs_p): Delete.
22373 (spe_expand_predicate_builtin): Delete.
22374 (spe_expand_evsel_builtin): Delete.
22375 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22376 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22377 (invalid_e500_subreg): Delete.
22378 (rs6000_legitimize_address): Always force_reg op2 as well, for
22379 paired single memory accesses.
22380 (rs6000_member_type_forces_blk): Delete.
22381 (rs6000_spe_function_arg): Delete.
22382 (rs6000_expand_unop_builtin): Delete SPE handling.
22383 (rs6000_expand_binop_builtin): Ditto.
22384 (spe_expand_stv_builtin): Delete.
22385 (bdesc_2arg_spe): Delete.
22386 (spe_expand_builtin): Delete.
22387 (spe_expand_predicate_builtin): Delete.
22388 (spe_expand_evsel_builtin): Delete.
22389 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22390 (spe_init_builtins): Delete.
22391 (spe_func_has_64bit_regs_p): Delete.
22392 (savres_routine_name): Delete "info" parameter. Adjust callers.
22393 (rs6000_emit_stack_reset): Ditto.
22394 (rs6000_dwarf_register_span): Delete.
22395 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22396 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22397 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22399 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22401 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22402 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22403 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22404 mulv2sf3, divv2sf3): Delete expanders.
22406 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22408 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22410 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22412 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22413 * config/rs6000/rs6000.c: Ditto.
22415 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22417 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22418 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22419 comparison_operator.
22421 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22423 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22424 * config/rs6000/rs6000.opt: Ditto.
22425 * config/rs6000/t-rtems: Ditto.
22427 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22429 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22430 TARGET_E500_SINGLE by 0, simplify.
22431 * config/rs6000/rs6000.c: Ditto.
22432 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22433 (spe_build_register_parallel): Delete.
22434 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22435 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22436 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22437 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22438 (E500_CONVERT): Delete.
22439 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22441 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22443 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22444 * config/rs6000/dfp.md: Ditto.
22445 (negdd2, *negdd2_fpr): Merge.
22446 (absdd2, *absdd2_fpr): Merge.
22447 (negtd2, *negtd2_fpr): Merge.
22448 (abstd2, *abstd2_fpr): Merge.
22449 * config/rs6000/e500.h: Delete file.
22450 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22451 TARGET_FPRS by 1 and simplify.
22452 * config/rs6000/rs6000-c.c: Ditto.
22453 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22454 TARGET_DF_SPE by 0.
22455 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22457 * config/rs6000/rs6000.md: Ditto.
22458 (floatdidf2, *floatdidf2_fpr): Merge.
22459 (move_from_CR_gt_bit): Delete.
22460 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22461 (E500_CR_IOR_COMPARE): Delete.
22462 (All patterns that require !TARGET_FPRS): Delete.
22463 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22465 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22467 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22469 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22471 * graphds.c (add_edge): Intitialize edge's attached data.
22472 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22473 pointer parameter. Call pointed function on each edge during
22474 graph traversing. Skip traversing the edge when the function
22476 (graphds_dfs, graphds_scc): Ditto.
22477 (for_each_edge): New parameter. Pass the new parameter to callback
22479 * graphds.h (skip_edge_callback): New function pointer type.
22480 (graphds_dfs, graphds_scc): New function pointer parameter.
22481 (graphds_edge_callback, for_each_edge): New parameter.
22483 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22485 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22486 out code checking if runtime alias check is possible to below ...
22487 Call the new function.
22488 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22489 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22491 2017-06-07 Marek Polacek <polacek@redhat.com>
22494 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22495 TYPE_OVERFLOW_WRAPS check.
22497 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22499 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22500 if versioning is required.
22501 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22502 peeling with the check for versioning.
22504 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22506 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22507 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22508 Set true to new parameter if new ssa variable is defined.
22509 (vect_gen_vector_loop_niters): Refactor. Set range information
22510 for the new vector loop bound variable.
22511 (vect_do_peeling): Ditto.
22513 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22515 * tree-affine.c (ssa.h): Include header file.
22516 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22517 has wrapping overflow behavior.
22519 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22521 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22523 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22525 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22526 (tree_to_aff_combination): ... here.
22528 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22530 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22531 reg_pressure model function.
22532 (ivopts_global_cost_for_size): Delete.
22533 (determine_set_costs, iv_ca_recount_cost): Call new model function
22534 ivopts_estimate_reg_pressure.
22536 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22538 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22539 expensive than udiv. Remove floating point cases from mod.
22541 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22543 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22544 Increase idiv cost.
22546 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22548 * config/aarch64/aarch64.md
22549 (copysignsf3): Fix mask generation.
22551 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22553 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22555 (class dump_manager): Add register_dumps method.
22556 * dumpfile.c: Include langhooks.h.
22557 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22558 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22559 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22560 (dump_manager::dump_register): Start with 512 entries instead of 32.
22561 (dump_manager::register_dumps): New method.
22562 * toplev.c (general_init): Instead of invoking register_dumps
22563 langhook, invoke register_dumps method on the dump manager.
22564 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22567 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22569 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22570 Say that # requires an associated define_split to exist, and that
22571 the define_split must be suitable for use after register allocation.
22573 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22575 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22576 (compute_outgoing_frequencies): Also initialize zero counts.
22577 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22578 around loops; preserve more of profile when nothing changes.
22580 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22582 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22584 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22585 * config/arm/arm-cpu-cdata.h: Regenerate.
22586 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22587 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22588 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22589 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22590 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22592 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22594 * config/arm/t-rmprofile: Likewise.
22595 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22597 2017-06-06 David S. Miller <davem@davemloft.net>
22600 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22601 blockage if function uses alloca.
22603 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22605 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22606 New "uid" fields to hold pretty-print IDs of group and ref.
22607 Memory references are now identified as <group_id>:<ref_id>
22608 instead of using [random] addresses.
22609 (dump_mem_details): Simplify, no functional change.
22610 (dump_mem_ref): Simplify and make output more concise.
22611 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22612 (find_or_create_group): Initialize group uid.
22613 (record_ref): Initialize ref uid. Improve debug output.
22614 (prune_group_by_reuse, should_issue_prefetch_p,)
22615 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22616 (mark_nontemporal_store, determine_loop_nest_reuse):
22617 Improve debug output.
22619 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22621 * dbgcnt.def (prefetch): New debug counter.
22622 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22623 (schedule_prefetches): Stop issueing prefetches if debug counter
22626 2017-06-06 Tom de Vries <tom@codesourcery.com>
22628 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22629 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22631 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22633 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22634 Use aarch64_reg_or_zero predicate for operand 4.
22635 (aarch64_compare_and_swap<mode> define_insn_and_split):
22636 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22637 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22639 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22641 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22642 (arm_compute_save_core_reg_mask): This.
22643 (thumb1_compute_save_reg_mask): Rename into ...
22644 (thumb1_compute_save_core_reg_mask): This.
22645 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22646 (arm_compute_frame_layout): Likewise.
22648 2017-06-06 Richard Biener <rguenther@suse.de>
22650 PR tree-optimization/80974
22651 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22652 keep or clear leaders SSA info.
22654 2017-06-06 Tom de Vries <tom@codesourcery.com>
22656 * config/nvptx/nvptx.c (split_mode_p): New function.
22657 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22659 2017-06-06 Tom de Vries <tom@codesourcery.com>
22661 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22663 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22666 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22669 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22671 * shrink-wrap.c (handle_simple_exit): Update profile.
22672 (try_shrink_wrapping): Upate profile.
22674 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22676 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22677 (tree_guess_outgoing_edge_probabilities): New.
22678 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22679 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22681 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22683 * ipa-split.c (split_function): Initialize return bb profile.
22685 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22687 * profile.c (compute_branch_probabilities): Also initialize
22688 EXIT_BLOCK profile.
22690 2017-06-06 Richard Biener <rguenther@suse.de>
22692 PR tree-optimization/80928
22693 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22694 (vect_analyze_loop_operations): Properly guard analysis for
22696 (vect_transform_loop): Likewise.
22697 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22698 (vect_model_induction_cost): Do not cost for pure SLP.
22699 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22700 of induction in inner loop vectorization.
22701 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22702 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22703 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22705 (vect_analyze_slp_cost_1): Cost induction.
22706 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22707 (vect_get_slp_vect_defs): Likewise.
22708 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22709 (vect_transform_stmt): Handle SLP reductions.
22710 * tree-vectorizer.h (vectorizable_induction): Adjust.
22712 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22714 * config/rs6000/rs6000.c (make_resolver_func): Update
22715 init_lowered_empty_function call.
22717 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22719 * doc/invoke.texi: Document the -fprofile-abs-path option.
22720 * common.opt (fprofile-abs-path): New option.
22721 * gcov-io.h (gcov_write_filename): Declare.
22722 * gcov-io.c (gcov_write_filename): New function.
22723 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22724 * profile.c (output_location): Likewise.
22726 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22728 * shring-wrap.c: Revert accidental commit.
22730 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22732 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22734 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22736 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22738 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22739 profile in callgraph edge.
22740 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22741 (apply_scale): Likewise.
22742 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22743 Also scale profile when inlining function with zero profile.
22744 (initialize_cfun): Update exit block profile even when it is zero.
22745 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22746 when profile is read.
22748 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22750 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22751 (CLONE_*): New constants to define the processors we can generate
22752 code for with the target_clone attribute.
22753 (rs6000_clone_map): New array to identify which clone processors
22754 the current program is running on.
22755 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22756 target_clone attribute.
22757 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22758 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22759 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22760 (cpu_expand_builtin): Add support for target_clone attribute.
22761 (rs6000_valid_attribute_p): Allow "default" attribute.
22762 (get_decl_name): New debug function to simplify printing the
22763 current function name in debugging statements.
22764 (rs6000_clone_priority): New functions to support the target_clone
22765 attribute, and be able to generate code to switch between ISA 2.05
22766 through ISA 3.0 (power6 through power9).
22767 (rs6000_compare_version_priority): Likewise.
22768 (rs6000_get_function_versions_dispatcher): Likewise.
22769 (make_resolver_func): Likewise.
22770 (add_condition_to_bb): Likewise.
22771 (dispatch_function_versions): Likewise.
22772 (rs6000_generate_version_dispatcher_body): Likewise.
22773 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22774 (fusion_gpr_load_p): Fix a spacing issue.
22775 * doc/extend.texi (Common Function Attributes): Document that the
22776 PowerPC supports the target_clone attribute.
22778 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22780 * config/arm/arm.h: explain F symbol found in description of ARM
22781 register allocation in its legend.
22783 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22785 * config/mips/frame-header-opt.c: Include profile-count.h.
22786 * config/riscv/riscv.c: Include profile-count.h
22788 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22790 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22792 (sm_set_flag_if_changed): Add bbs field.
22793 (execute_sm_if_changed_flag_set): Pass BBS.
22794 (execute_sm): Update.
22796 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22798 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22801 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22803 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22804 (peephole2): New peephole2 to emit the above.
22805 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22807 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22809 * config/aarch64/aarch64.c (define_peephole2 above
22810 *sub_<shift>_<mode>): New peephole.
22812 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22814 * config/i386/i386.c (make_resolver_func): Update.
22815 * Makefile.in: Add profile-count.h and profile-count.o
22816 * auto-profile.c (afdo_indirect_call): Update to new API.
22817 (afdo_set_bb_count): Update.
22818 (afdo_propagate_edge): Update.
22819 (afdo_propagate_circuit): Update.
22820 (afdo_calculate_branch_prob): Update.
22821 (afdo_annotate_cfg): Update.
22822 * basic-block.h: Include profile-count.h
22823 (struct edge_def): Turn count to profile_count.
22824 (struct basic_block_def): Likewie.
22825 (REG_BR_PROB_BASE): Move to profile-count.h
22826 (RDIV): Move to profile-count.h
22827 * bb-reorder.c (max_entry_count): Turn to profile_count.
22828 (find_traces): Update.
22829 (rotate_loop):Update.
22830 (connect_traces):Update.
22831 (sanitize_hot_paths):Update.
22832 * bt-load.c (migrate_btr_defs): Update.
22833 * cfg.c (RDIV): Remove.
22834 (init_flow): Use alloc_block.
22835 (alloc_block): Uninitialize count.
22836 (unchecked_make_edge): Uninitialize count.
22837 (check_bb_profile): Update.
22838 (dump_edge_info): Update.
22839 (dump_bb_info): Update.
22840 (update_bb_profile_for_threading): Update.
22841 (scale_bbs_frequencies_int): Update.
22842 (scale_bbs_frequencies_gcov_type): Update.
22843 (scale_bbs_frequencies_profile_count): New.
22844 * cfg.h (update_bb_profile_for_threading): Update.
22845 (scale_bbs_frequencies_profile_count): Declare.
22846 * cfgbuild.c (compute_outgoing_frequencies): Update.
22847 (find_many_sub_basic_blocks): Update.
22848 * cfgcleanup.c (try_forward_edges): Update.
22849 (try_crossjump_to_edge): Update.
22850 * cfgexpand.c (expand_gimple_tailcall): Update.
22851 (construct_exit_block): Update.
22852 * cfghooks.c (verify_flow_info): Update.
22853 (dump_bb_for_graph): Update.
22854 (split_edge): Update.
22855 (make_forwarder_block): Update.
22856 (duplicate_block): Update.
22857 (account_profile_record): Update.
22858 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22859 (get_estimated_loop_iterations): Update.
22860 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22861 (single_likely_exit): Update.
22862 * cfgloopmanip.c (scale_loop_profile): Update.
22864 (set_zero_probability): Update.
22865 (lv_adjust_loop_entry_edge): Update.
22866 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22867 (purge_dead_edges): Update.
22868 (rtl_account_profile_record): Update.
22869 * cgraph.c (cgraph_node::create): Uninitialize count.
22870 (symbol_table::create_edge): Uninitialize count.
22871 (cgraph_update_edges_for_call_stmt_node): Update.
22872 (cgraph_edge::dump_edge_flags): Update.
22873 (cgraph_node::dump): Update.
22874 (cgraph_edge::maybe_hot_p): Update.
22875 * cgraph.h: Include profile-count.h
22876 (create_clone), create_edge, create_indirect_edge): Update.
22877 (cgraph_node): Turn count to profile_count.
22878 (cgraph_edge0: Likewise.
22879 (make_speculative, clone): Update.
22880 (create_edge): Update.
22881 (init_lowered_empty_function): Update.
22882 * cgraphclones.c (cgraph_edge::clone): Update.
22883 (duplicate_thunk_for_node): Update.
22884 (cgraph_node::create_clone): Update.
22885 * cgraphunit.c (cgraph_node::analyze): Update.
22886 (cgraph_node::expand_thunk): Update.
22887 * final.c (dump_basic_block_info): Update.
22888 * gimple-streamer-in.c (input_bb): Update.
22889 * gimple-streamer-out.c (output_bb): Update.
22890 * graphite.c (print_global_statistics): Update.
22891 (print_graphite_scop_statistics): Update.
22892 * hsa-brig.c: Include basic-block.h.
22893 * hsa-dump.c: Include basic-block.h.
22894 * hsa-gen.c (T sum_slice): Update.
22895 (convert_switch_statements):Update.
22896 * hsa-regalloc.c: Include basic-block.h.
22897 * ipa-chkp.c (chkp_produce_thunks): Update.
22898 * ipa-cp.c (struct caller_statistics): Update.
22899 (init_caller_stats): Update.
22900 (gather_caller_stats): Update.
22901 (ipcp_cloning_candidate_p): Update.
22902 (good_cloning_opportunity_p): Update.
22903 (get_info_about_necessary_edges): Update.
22904 (dump_profile_updates): Update.
22905 (update_profiling_info): Update.
22906 (update_specialized_profile): Update.
22907 (perhaps_add_new_callers): Update.
22908 (decide_about_value): Update.
22909 (ipa_cp_c_finalize): Update.
22910 * ipa-devirt.c (struct odr_type_warn_count): Update.
22911 (struct decl_warn_count): Update.
22912 (struct final_warning_record): Update.
22913 (possible_polymorphic_call_targets): Update.
22914 (ipa_devirt): Update.
22915 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22916 * ipa-icf.c (sem_function::merge): Update.
22917 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22918 * ipa-inline.c (compute_uninlined_call_time): Update.
22919 (compute_inlined_call_time): Update.
22920 (want_inline_small_function_p): Update.
22921 (want_inline_self_recursive_call_p): Update.
22922 (edge_badness): Update.
22923 (lookup_recursive_calls): Update.
22924 (recursive_inlining): Update.
22925 (inline_small_functions): Update.
22926 (dump_overall_stats): Update.
22927 (dump_inline_stats): Update.
22928 * ipa-profile.c (ipa_profile_generate_summary): Update.
22929 (ipa_propagate_frequency): Update.
22930 (ipa_profile): Update.
22931 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22932 * ipa-utils.c (ipa_merge_profiles): Update.
22933 * loop-doloop.c (doloop_modify): Update.
22934 * loop-unroll.c (report_unroll): Update.
22935 (unroll_loop_runtime_iterations): Update.
22936 * lto-cgraph.c (lto_output_edge): Update.
22937 (lto_output_node): Update.
22938 (input_node): Update.
22939 (input_edge): Update.
22940 (merge_profile_summaries): Update.
22941 * lto-streamer-in.c (input_cfg): Update.
22942 * lto-streamer-out.c (output_cfg): Update.
22943 * mcf.c (create_fixup_graph): Update.
22944 (adjust_cfg_counts): Update.
22945 (sum_edge_counts): Update.
22946 * modulo-sched.c (sms_schedule): Update.
22947 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22948 * predict.c (maybe_hot_count_p): Update.
22949 (probably_never_executed): Update.
22950 (dump_prediction): Update.
22951 (combine_predictions_for_bb): Update.
22952 (propagate_freq): Update.
22953 (handle_missing_profiles): Update.
22954 (counts_to_freqs): Update.
22955 (rebuild_frequencies): Update.
22956 (force_edge_cold): Update.
22957 * predict.h: Include profile-count.h
22958 (maybe_hot_count_p, counts_to_freqs): UPdate.
22959 * print-rtl-function.c: Do not include cfg.h
22960 * print-rtl.c: Include basic-block.h
22961 * profile-count.c: New file.
22962 * profile-count.h: New file.
22963 * profile.c (is_edge_inconsistent): Update.
22964 (correct_negative_edge_counts): Update.
22965 (is_inconsistent): Update.
22966 (set_bb_counts): Update.
22967 (read_profile_edge_counts): Update.
22968 (compute_frequency_overlap): Update.
22969 (compute_branch_probabilities): Update; Initialize and deinitialize
22971 (branch_prob): Update.
22972 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22973 (edge_gcov_count): New.
22974 (bb_gcov_count): New.
22975 * shrink-wrap.c (try_shrink_wrapping): Update.
22976 * tracer.c (better_p): Update.
22977 * trans-mem.c (expand_transaction): Update.
22978 (ipa_tm_insert_irr_call): Update.
22979 (ipa_tm_insert_gettmclone_call): Update.
22980 * tree-call-cdce.c: Update.
22981 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22982 (gimple_duplicate_sese_tail): Update.
22983 (gimple_account_profile_record): Update.
22984 (execute_fixup_cfg): Update.
22985 * tree-inline.c (copy_bb): Update.
22986 (copy_edges_for_bb): Update.
22987 (initialize_cfun): Update.
22988 (freqs_to_counts): Update.
22989 (copy_cfg_body): Update.
22990 (expand_call_inline): Update.
22991 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22992 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22993 (try_unroll_loop_completely): Update.
22994 (try_peel_loop): Update.
22995 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22996 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22997 * tree-ssa-loop-split.c (connect_loops): Update.
22998 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22999 * tree-ssa-reassoc.c (branch_fixup): Update.
23000 * tree-ssa-tail-merge.c (replace_block_by): Update.
23001 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23002 (compute_path_counts): Update.
23003 (update_profile): Update.
23004 (recompute_probabilities): Update.
23005 (update_joiner_offpath_counts): Update.
23006 (estimated_freqs_path): Update.
23007 (freqs_to_counts_path): Update.
23008 (clear_counts_path): Update.
23009 (ssa_fix_duplicate_block_edges): Update.
23010 (duplicate_thread_path): Update.
23011 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23012 (struct switch_conv_info): Update.
23013 * tree-tailcall.c (decrease_profile): Update.
23014 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23015 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23016 * value-prof.c (check_counter): Update.
23017 (gimple_divmod_fixed_value): Update.
23018 (gimple_mod_pow2): Update.
23019 (gimple_mod_subtract): Update.
23020 (gimple_ic_transform): Update.
23021 (gimple_stringop_fixed_value): Update.
23022 * value-prof.h (gimple_ic): Update.
23024 2017-06-02 Carl Love <cel@us.ibm.com>
23026 * config/rs6000/rs6000-c: Add support for built-in functions
23027 vector double vec_doublee (vector signed int);
23028 vector double vec_doublee (vector unsigned int);
23029 vector double vec_doublee (vector float);
23030 vector double vec_doubleh (vector signed int);
23031 vector double vec_doubleh (vector unsigned int);
23032 vector double vec_doubleh (vector float);
23033 vector double vec_doublel (vector signed int);
23034 vector double vec_doublel (vector unsigned int);
23035 vector double vec_doublel (vector float);
23036 vector double vec_doubleo (vector signed int);
23037 vector double vec_doubleo (vector unsigned int);
23038 vector double vec_doubleo (vector float);.
23039 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23040 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23042 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23043 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23044 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23046 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23047 vec_doublel, vec_doubleh.
23048 * doc/extend.texi: Update the built-in documentation file for the
23049 new built-in functions.
23051 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23054 * ipa-inline-analysis.c (free_growth_caches): Set
23055 edge_removal_hook_holder to NULL after removing it.
23057 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23059 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23060 comparision with zero.
23062 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23063 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23064 for early expansion of vec_min and vec_max builtins.
23065 (builtin_function_type): Add min/max unsigned variants to those
23066 identified as having unsigned arguments.
23068 2017-06-02 Olivier Hainque <hainque@adacore.com>
23070 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23072 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23074 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23075 Use VALL_F16 iterator rather than VALL.
23077 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23079 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23080 Emit CBNZ inside loop when doing a strong exchange and comparing
23081 against zero. Generate the CC flags after the loop.
23083 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23085 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23086 (dl_section_ref): New.
23087 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23088 On AIX, append an expression to subtract the size of the
23089 section length to dl_section_ref.
23091 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23093 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23094 for early expansion of vector absolute builtins.
23096 2017-06-02 Richard Biener <rguenther@suse.de>
23098 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23099 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23101 2017-06-02 Richard Biener <rguenther@suse.de>
23103 PR tree-optimization/80948
23104 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23105 stmt order as well.
23107 2017-06-02 Richard Biener <rguenther@suse.de>
23109 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23111 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23112 for inductions as relevant.
23114 2017-06-02 Richard Biener <rguenther@suse.de>
23116 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23117 (vectorizable_induction): ... this. Remove dead code.
23119 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23121 * builtins. (expand_builtin_alloca): Remove second parameter and
23122 infer its value from the first parameter instead.
23123 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23125 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23127 PR rtl-optimization/80903
23128 * loop-doloop.c (add_test): Unshare sequence.
23130 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23132 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23134 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23136 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23138 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23139 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23140 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23141 (xlogue_layout::get_stub_rtx): Make static.
23142 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23143 (xlogue_layout::compute_stub_managed_regs): Rename to...
23144 (xlogue_layout::count_stub_managed_regs): ...this.
23145 (xlogue_layout::is_stub_managed_reg): New function.
23146 (xlogue_layout::m_stub_names): Rename to...
23147 (xlogue_layout::s_stub_names): ...this, make static.
23148 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23149 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23150 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23151 xlogue_layout::s_stub_names): Instantiate statics.
23152 (stub_managed_regs): Remove.
23153 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23154 (disable_call_ms2sysv_xlogues): Rename to...
23155 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23156 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23158 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23159 change after reload_completed.
23160 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23162 (ix86_expand_prologue): Likewise.
23163 (ix86_expand_epilogue): Likewise.
23164 (ix86_expand_split_stack_prologue): Likewise.
23165 (ix86_compute_frame_layout): Remove frame parameter ...
23166 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23167 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23169 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23170 (ix86_frame): Move from here ...
23171 * config/i386/i386.h (ix86_frame): ... to here.
23172 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23173 complete ix86_frame data structure instead. Remove some_ld_name.
23175 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23177 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23178 symbols that hold a DECL_VALUE_EXPR.
23180 2017-06-01 Martin Jambor <mjambor@suse.cz>
23182 PR tree-optimization/80898
23183 * tree-sra.c (process_subtree_disqualification): Removed.
23184 (disqualify_candidate): Do not acll
23185 process_subtree_disqualification.
23186 (subtree_mark_written_and_enqueue): New function.
23187 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23188 RHS has been disqualified and re-queue LHS if necessary. Apart
23189 from that, ignore disqualified RHS.
23191 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23193 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23194 address fetch for z10 or later.
23196 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23198 * config/arc/arc.md (tst_movb): Add guard when splitting.
23200 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23202 * config/arc/arc.c (arc_can_eliminate): Test against
23203 arc_frame_pointer_needed.
23205 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23207 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23208 to prevent store reordering.
23209 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23210 (type): Add block type.
23211 (stack_tie): Define special instruction to be used in
23214 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23216 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23217 constraint. It is not valid for the pattern.
23218 (noncommutative_binary_comparison): Likewise.
23220 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23222 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23225 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23227 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23228 be used by the reg-alloc.
23230 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23232 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23233 reg-alloc when having mul64 or mul32x16 instructions.
23234 (mulsidi3): Likewise.
23235 (umulsidi3): Likewise.
23236 (mulsi32x16): New pattern.
23237 (mulsi64): Likewise.
23238 (mulsidi64): Likewise.
23239 (umulsidi64): Likewise.
23240 (MUL32x16_REG): Define.
23241 (mul64_600): Use MUL32x16_REG.
23242 (mac64_600): Likewise.
23243 (umul64_600): Likewise.
23244 (umac64_600): Likewise.
23246 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23248 * config/arc/arc.md (mulsi3_700): Make it commutative.
23250 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23252 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23254 (*sign_extendsidi2_insn): Likewise for movstosw.
23256 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23258 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23259 the type of the input discriminant value. Convert the
23260 discriminant value of signedness vary.
23262 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23264 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23265 Add to -Wall section.
23267 2017-06-01 Richard Biener <rguenther@suse.de>
23269 PR middle-end/66313
23270 * fold-const.c (fold_plusminus_mult_expr): If the factored
23271 factor may be zero use a wrapping type for the inner operation.
23272 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23273 and handle moved defs.
23274 (process_assignment): Properly guard the unary op case. Return a
23275 tri-state indicating that moving the stmt before the call may allow
23276 to continue. Pass through to_move.
23277 (find_tail_calls): Handle moving unrelated defs before
23280 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23283 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23284 splitter result in the canonical way.
23286 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23288 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23289 also for 32bit target. Update insn attributes.
23290 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23292 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23294 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23295 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23296 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23297 (_mm_maskz_min_ss): New intrinsics.
23299 2017-05-31 Martin Liska <mliska@suse.cz>
23301 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23302 Change comment style to one we normally use.
23303 (vectorizable_reduction): Likewise.
23304 (vectorizable_induction): Likewise.
23305 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23306 (vectorizable_call): Likewise.
23307 (vectorizable_simd_clone_call): Likewise.
23308 (vectorizable_conversion): Likewise.
23309 (vectorizable_assignment): Likewise.
23310 (vectorizable_shift): Likewise.
23311 (vectorizable_operation): Likewise.
23312 (vectorizable_store): Likewise.
23313 (vectorizable_load): Likewise.
23314 * tree-vectorizer.h: Likewise.
23316 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23318 * passes.c (emergency_dump_function): New.
23319 * tree-pass.h (emergency_dump_function): Declare.
23320 * plugin.c (plugins_internal_error_function): Remove.
23321 * plugin.h (plugins_internal_error_function): Remove declaration.
23322 * toplev.c (internal_error_function): New static function. Use it...
23323 (general_init): ...here.
23325 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23327 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23328 (arc_rtx_costs): Add costs for new patterns.
23329 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23330 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23332 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23334 * tree-ssa-strlen.c (get_next_strinfo): New function.
23335 (get_stridx_plus_constant): Use it.
23336 (zero_length_string): Likewise.
23337 (adjust_related_strinfos): Likewise.
23338 (adjust_last_stmt): Likewise.
23340 2017-05-31 Richard Biener <rguenther@suse.de>
23343 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23344 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23346 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23348 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23349 loop_vinfo argument and use of dependence distance vectors.
23350 Check instead whether the two references differ only in their
23351 initial value and assume that they have the same alignment if the
23352 difference is a multiple of the vector alignment.
23353 (vect_analyze_data_refs_alignment): Update call accordingly.
23355 2017-05-31 Martin Liska <mliska@suse.cz>
23358 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23360 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23362 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23363 (create_intersect_range_checks): Move from ...
23364 * tree-data-ref.c (create_intersect_range_checks_index)
23365 (create_intersect_range_checks): ... to here.
23366 (create_runtime_alias_checks): New function factored from ...
23367 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23368 here. Call above function.
23369 * tree-data-ref.h (create_runtime_alias_checks): New function.
23371 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23373 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23374 segment length for dr_b and compute it in wide_int.
23376 2017-05-31 Richard Biener <rguenther@suse.de>
23378 PR tree-optimization/80906
23379 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23380 and pass through iv_map.
23381 (copy_bb_and_scalar_dependences): Adjust.
23382 (translate_pending_phi_nodes): Likewise.
23383 (copy_loop_close_phi_args): Handle code-generating IVs instead
23386 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23388 * diagnostic-color.c (color_dict): Add "type-diff".
23389 (parse_gcc_colors): Update comment.
23390 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23391 -fdiagnostics-show-template-tree and -fno-elide-type.
23392 (GCC_COLORS): Add type-diff to example.
23394 (-fdiagnostics-show-template-tree): New.
23395 (-fno-elide-type): New.
23396 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23397 the pp_format_decoder callback. Call any m_format_postprocessor's
23399 (pretty_printer::pretty_printer): Initialize
23400 m_format_postprocessor.
23401 (pretty_printer::~pretty_printer): Delete any
23402 m_format_postprocessor.
23403 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23404 (class format_postprocessor): New class.
23405 (struct pretty_printer::format_decoder): Document the new parameters.
23406 (struct pretty_printer::m_format_postprocessor): New field.
23407 * tree-diagnostic.c (default_tree_printer): Update for new
23408 bool and const char ** params.
23409 * tree-diagnostic.h (default_tree_printer): Likewise.
23411 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23413 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23414 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23415 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23416 rs6000_gen_cell_microcode code.
23417 (rs6000_final_prescan_insn): Delete.
23418 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23419 "warn-cell-microcode" entries.
23420 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23421 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23422 throughout. Change cc_reg_not_micro_cr0_operand to
23423 cc_reg_not_cr0_operand throughout.
23424 (*extendhi<mode>2_noload): Delete.
23425 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23426 (mwarn-cell-microcode): Delete.
23427 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23428 -mgen-cell-microcode and -mwarn-cell-microcode.
23430 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23433 * config/i386/constraints.md (Yd): New constraint.
23435 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23436 and (?Yd, r) alternatives. Update insn attributes.
23437 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23438 and (?*Yd, r) alternatives. Update insn attributes.
23439 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23441 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23443 * gimplify.c (gimplify_modify_expr): Don't create a
23444 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23447 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23449 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23451 2017-05-30 Richard Biener <rguenther@suse.de>
23453 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23454 and reduc_def fields.
23455 (STMT_VINFO_REDUC_TYPE): New define.
23456 (STMT_VINFO_REDUC_DEF): Likewise.
23457 (vect_force_simple_reduction): Adjust prototype.
23458 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23459 (vect_is_simple_reduction): Remove check_reduction argument.
23460 (vect_force_simple_reduction): Adjust and set
23461 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23462 (vectorizable_reduction): Do not re-do reduction analysis
23463 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23464 * tree-parloops.c (gather_scalar_reductions): Adjust.
23466 2017-05-30 Richard Biener <rguenther@suse.de>
23468 PR middle-end/80901
23469 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23472 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23474 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23475 Introduce unknown_misalignment parameter and remove vf.
23476 (vect_peeling_hash_get_lowest_cost):
23477 Pass unknown_misalignment parameter.
23478 (vect_enhance_data_refs_alignment):
23479 Fix unsupportable data ref treatment.
23481 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23483 * tree-vect-data-refs.c (vect_get_data_access_cost):
23484 Workaround for SLP handling.
23485 (vect_enhance_data_refs_alignment):
23486 Compute costs for doing no peeling at all, compare to the best
23487 peeling costs so far and avoid peeling if cheaper.
23489 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23491 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23492 Return peeling info and set costs to zero for unlimited cost
23494 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23495 with unknown misalignment. Compute and costs for unknown
23496 misalignment, compare them to the costs for known misalignment
23497 and choose the cheapest for peeling.
23499 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23501 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23502 (vect_get_peeling_costs_all_drs): Create function.
23503 (vect_peeling_hash_get_lowest_cost):
23504 Use vect_get_peeling_costs_all_drs.
23505 (vect_peeling_supportable): Create function.
23506 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23508 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23510 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23511 DR_HAS_NEGATIVE_STEP.
23512 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23513 (vect_enhance_data_refs_alignment): Use.
23514 (vect_duplicate_ssa_name_ptr_info): Use.
23515 * tree-vectorizer.h (dr_misalignment): Use.
23516 (known_alignment_for_access_p): Use.
23518 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23521 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23523 (has_section_name): New function.
23525 2017-05-30 Martin Liska <mliska@suse.cz>
23528 * auto-profile.c (get_function_decl_from_block): Fix
23531 2017-05-30 Richard Biener <rguenther@suse.de>
23533 PR middle-end/80876
23534 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23536 2017-05-30 Martin Liska <mliska@suse.cz>
23538 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23539 * dumpfile.h (struct dump_file_info): Remove ctors.
23541 2017-05-30 Martin Liska <mliska@suse.cz>
23543 * predict.def: Fix GNU coding style.
23545 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23547 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23548 Mark 'to' argument with ATTRIBUTE_UNUSED.
23550 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23552 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23553 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23554 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23557 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23559 * doc/install.texi (Options specification): Restore entry of
23560 --enable-sjlj-exceptions.
23562 2017-05-27 Michael Eager <eager@eagercon.com>
23565 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23567 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23569 * config/microblaze/microblaze.h
23570 (FIXED_REGISTERS): Update in macro.
23571 (CALL_USED_REGISTERS): Update in macro.
23573 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23575 * doc/install.texi: Add links to macOS binary distributions.
23577 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23581 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23583 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23585 2017-05-26 Martin Liska <mliska@suse.cz>
23587 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23589 2017-05-26 Martin Liska <mliska@suse.cz>
23591 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23592 always leading ';; '.
23593 (dump_bb_info): Likewise.
23594 (brief_dump_cfg): Likewise.
23595 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23596 * dumpfile.c: Remove usage of TDF_VERBOSE.
23597 * dumpfile.h (enum dump_kind): Likewise.
23598 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23599 * print-tree.c (debug_verbose): Remove.
23600 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23601 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23602 * tree-diagnostic.c (default_tree_printer): Replace
23603 TDF_DIAGNOSTIC with TDF_SLIM.
23605 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23607 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23608 in parameter loop, rather than loop_vinfo.
23609 (create_intersect_range_checks): Ditto.
23610 (vect_create_cond_for_alias_checks): Update call to above functions.
23612 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23614 PR tree-optimization/80815
23615 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23616 for merging runtime alias checks. Handle negative DR_STEPs.
23618 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23620 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23622 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23623 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23624 out code pruning runtime alias checks.
23625 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23626 factored out from above.
23627 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23629 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23631 (prune_runtime_alias_test_list): New decalaration.
23633 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23635 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23636 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23637 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23638 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23639 (operator==, comp_dr_with_seg_len_pair): Ditto.
23640 (vect_prune_runtime_alias_test_list): Ditto.
23642 2017-05-26 Martin Liska <mliska@suse.cz>
23645 * params.def: Bound partial-inlining-entry-probability param.
23647 2017-05-26 Marek Polacek <polacek@redhat.com>
23650 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23653 2017-05-26 Richard Biener <rguenther@suse.de>
23655 PR tree-optimization/80842
23656 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23659 2017-05-26 Richard Biener <rguenther@suse.de>
23661 PR tree-optimization/80844
23662 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23664 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23666 * doc/md.texi (Machine Constraints): Update x86 family
23667 machine constraints section to match 'config/i386/constraints.md'.
23669 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23671 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23673 2017-05-25 Nathan Sidwell <nathan@acm.org>
23675 * doc/invoke.texi (--enable-languages): Update documentation.
23677 2017-05-25 Martin Liska <mliska@suse.cz>
23679 * dumpfile.c: Add TDF_FOLDING.
23680 * dumpfile.h (enum dump_kind): Likewise.
23681 * genmatch.c (dt_simplify::gen_1): Use it.
23683 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23685 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23687 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23689 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23690 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23692 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23694 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23695 * match.pd (X == C): Rewrite it here.
23696 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23697 with_certain_nonzero_bits2): New predicates.
23698 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23700 2017-05-24 Nathan Sidwell <nathan@acm.org>
23702 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23705 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23708 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23710 * config/powerpcspe: New port. Files are copied from the rs6000
23711 port, with "rs6000" in filenames replaced by "powerpcspe".
23713 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23715 PR rtl-optimization/80754
23716 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23718 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23720 * config/sparc/sparc.md (length): Return the correct value for -mflat
23721 sibcalls to match output_sibcall.
23723 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23727 * config/rs6000/rs6000.c (struct machine_function): Add new field
23729 (rs6000_get_separate_components): Init that field, use it.
23730 (rs6000_components_for_bb): Use the field.
23732 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23734 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23736 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23738 PR middle-end/80823
23739 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23741 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23744 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23745 address against address_operand predicate.
23746 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23748 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23750 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23751 parameters passed indirectly.
23753 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23755 * config/i386/i386.md (*movdi_internal): Remove SSE4
23756 alternative 18 (?r, *v). Update insn attributes.
23757 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23758 Update insn attributes.
23759 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23760 Update insn attributes.
23761 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23762 alternative 1 (r, v). Remove isa attribute.
23763 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23764 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23765 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23767 2017-05-23 Tom de Vries <tom@codesourcery.com>
23769 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23772 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23774 * cgraphunit.c (symbol_table::process_new_functions): Update.
23775 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23776 (inline_generate_summary): Rename to ...
23777 (ipa_fn_summary_generate): ... this one.
23778 (inline_read_summary): Rename to ...
23779 (ipa_fn_summary_read): ... this one.
23780 (inline_write_summary): Rename to ...
23781 (ipa_fn_summary_write): ... this one.
23782 (inline_free_summary): Rename to ...
23783 (ipa_free_fn_summary): ... this one.
23784 (pass_data_local_fn_summary, pass_local_fn_summary,
23785 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23786 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23787 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23788 make_pass_ipa_fn_summary): New.
23789 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23790 inline_write_summary, inline_free_summary): Remove.
23791 (ipa_free_fn_summary) : New.
23792 * ipa-inline.c (ipa_inline): Update.
23793 (pass_ipa_inline): Do not generate summaries.
23794 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23796 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23797 and add pass_ipa_fn_summary.
23798 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23800 (make_pass_inline_parameters): Remove.
23802 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23804 * omp-low.c (struct omp_context): Remove "default_kind" member.
23807 * omp-offload.c (execute_oacc_device_lower): Remove the
23808 parallelism dimensions function attributes for unparallelized
23809 OpenACC kernels constructs.
23811 2017-05-23 Martin Liska <mliska@suse.cz>
23813 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23815 (cgraph_edge::make_speculative): Likewise.
23816 (cgraph_edge::resolve_speculation): Likewise.
23817 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23818 (cgraph_node::dump): Likewise.
23819 * cgraph.h: Likewise.
23820 * cgraphunit.c (analyze_functions): Likewise.
23821 (symbol_table::compile): Likewise.
23822 * ipa-cp.c (print_all_lattices): Likewise.
23823 (determine_versionability): Likewise.
23824 (initialize_node_lattices): Likewise.
23825 (ipcp_verify_propagated_values): Likewise.
23826 (estimate_local_effects): Likewise.
23827 (update_profiling_info): Likewise.
23828 (create_specialized_node): Likewise.
23829 (perhaps_add_new_callers): Likewise.
23830 (decide_about_value): Likewise.
23831 (decide_whether_version_node): Likewise.
23832 (identify_dead_nodes): Likewise.
23833 (ipcp_store_bits_results): Likewise.
23834 * ipa-devirt.c (dump_targets): Likewise.
23835 (ipa_devirt): Likewise.
23836 * ipa-icf.c (sem_item::dump): Likewise.
23837 (sem_function::equals): Likewise.
23838 (sem_variable::equals): Likewise.
23839 (sem_item_optimizer::read_section): Likewise.
23840 (sem_item_optimizer::execute): Likewise.
23841 (congruence_class::dump): Likewise.
23842 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23843 (dump_inline_summary): Likewise.
23844 (estimate_node_size_and_time): Likewise.
23845 (inline_analyze_function): Likewise.
23846 * ipa-inline-transform.c (inline_call): Likewise.
23847 * ipa-inline.c (report_inline_failed_reason): Likewise.
23848 (want_early_inline_function_p): Likewise.
23849 (edge_badness): Likewise.
23850 (update_edge_key): Likewise.
23851 (inline_small_functions): Likewise.
23852 * ipa-profile.c (ipa_profile): Likewise.
23853 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23854 (ipa_make_edge_direct_to_target): Likewise.
23855 (remove_described_reference): Likewise.
23856 (ipa_impossible_devirt_target): Likewise.
23857 (propagate_controlled_uses): Likewise.
23858 (ipa_print_node_params): Likewise.
23859 (ipcp_transform_function): Likewise.
23860 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23861 (propagate_pure_const): Likewise.
23862 * ipa-reference.c (generate_summary): Likewise.
23863 (read_write_all_from_decl): Likewise.
23864 (propagate): Likewise.
23865 (ipa_reference_read_optimization_summary): Likewise.
23866 * ipa-utils.c (ipa_merge_profiles): Likewise.
23867 * ipa.c (walk_polymorphic_call_targets): Likewise.
23868 (symbol_table::remove_unreachable_nodes): Likewise.
23869 (ipa_single_use): Likewise.
23870 * passes.c (execute_todo): Likewise.
23871 * predict.c (drop_profile): Likewise.
23872 * symtab.c (symtab_node::get_dump_name): New function.
23873 (symtab_node::dump_name): Likewise.
23874 (symtab_node::dump_asm_name): Likewise.
23875 (symtab_node::dump_references): Likewise.
23876 (symtab_node::dump_referring): Likewise.
23877 (symtab_node::dump_base): Likewise.
23878 (symtab_node::debug_symtab): Likewise.
23879 * tree-sra.c (convert_callers_for_node): Likewise.
23880 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23881 * value-prof.c (init_node_map): Likewise.
23883 2017-05-23 Martin Liska <mliska@suse.cz>
23885 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23886 and symtab_node::debug_symtab to symbol_table::debug.
23887 * cgraphunit.c (analyze_functions): Use the renamed function.
23888 (symbol_table::compile): Likewise.
23889 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23890 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23891 * passes.c (execute_todo): Likewise.
23892 * symtab.c (symbol_table::dump): New function.
23893 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23895 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23897 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23898 that nonconst implies exec.
23900 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23902 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23903 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23904 (inline_edge_summary_vec): Turn into ...
23905 (ipa_call_summaries): ... this one.
23906 (redirect_to_unreachable, edge_set_predicate,
23907 evaluate_properties_for_edge, inline_summary_alloc,
23908 reset_ipa_call_summary, reset_inline_summary,
23909 inline_summary_t::duplicate): Update.
23910 (inline_edge_duplication_hook): Turn to ...
23911 (ipa_call_summary_t::duplicate): ... this one.
23912 (inline_edge_removal_hook): Turn to ...
23913 (ipa_call_summary_t::remove): ... this one.
23914 (dump_inline_edge_summary): Turn to ...
23915 (dump_ipa_call_summary): ... this one.
23916 (estimate_function_body_sizes): Update.
23917 (inline_update_callee_summaries): Update.
23918 (remap_edge_change_prob): Update.
23919 (remap_edge_summaries): Update.
23920 (inline_merge_summary): Update.
23921 (do_estimate_edge_time): Update.
23922 (inline_generate_summary): Update.
23923 (inline_read_section): Update.
23924 (inline_read_summary): Update.
23925 (inline_free_summary): Update.
23926 * ipa-inline.c (can_inline_edge_p): Update.
23927 (compute_inlined_call_time): Update.
23928 (want_inline_small_function_p): Update.
23929 (edge_badness): Update.
23930 (early_inliner): Update.
23931 * ipa-inline.h (inline_edge_summary): Turn to ...
23932 (ipa_call_summary): ... this one.
23933 (ipa_call_summary_t): New class.
23934 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23935 (ipa_call_summaries): New.
23936 (inline_edge_summary): Remove.
23937 (estimate_edge_growth): Update.
23938 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23939 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23940 * ipa-split.c (execute_split_functions): Update.
23941 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23943 2017-05-23 Tom de Vries <tom@codesourcery.com>
23945 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23946 attributes): Document rdrand effective target.
23948 2017-05-23 Tom de Vries <tom@codesourcery.com>
23950 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23951 attributes): Sort alphabetically.
23953 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23955 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23957 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23960 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23961 V2DF/V2DI splat into two separate patterns, one that handles
23962 registers, and the other that only handles memory. Drop support
23963 for splatting from a GPR on ISA 2.07 and then splitting the
23964 splat into direct move and splat.
23965 (vsx_splat_<mode>_reg): Likewise.
23966 (vsx_splat_<mode>_mem): Likewise.
23968 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23970 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23972 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23974 PR middle-end/80809
23975 * omp-low.c (finish_taskreg_remap): New function.
23976 (finish_taskreg_scan): If unit size of ctx->record_type
23977 is non-constant, unshare the size expression and replace
23978 decls in it with possible outer var refs.
23980 PR middle-end/80809
23981 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23982 GOVD_SHARED rather than GOVD_PRIVATE with it.
23983 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23984 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23986 PR middle-end/80853
23987 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23988 as last argument to build_outer_var_ref for pointer bases of array
23989 section reductions.
23991 2017-05-19 Martin Sebor <msebor@redhat.com>
23993 * print-tree.c (print_node): Print DECL_READ_P flag.
23995 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23997 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23998 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23999 * cgraph.c: Likewise.
24000 * cgraphunit.c: Likewise.
24001 * gengtype.c: Likewise.
24002 * ipa-cp.c: Likewise.
24003 * ipa-devirt.c: Likewise.
24004 * ipa-icf.c: Likewise.
24005 * ipa-predicate.c: Likewise.
24006 * ipa-profile.c: Likewise.
24007 * ipa-prop.c: Likewise.
24008 * ipa-split.c: Likewise.
24010 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24011 edge_predicate_pool, dump_inline_hints,
24012 inline_summary::account_size_time, redirect_to_unreachable,
24013 edge_set_predicate, set_hint_predicate,
24014 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24015 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24016 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24017 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24018 ipa_call_summary_t::remove, initialize_growth_caches,
24019 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24020 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24021 mark_modified, unmodified_parm_1, unmodified_parm,
24022 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24023 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24024 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24025 will_be_nonconstant_predicate, record_modified_bb_info,
24026 get_minimal_bb, record_modified, param_change_prob,
24027 phi_result_unknown_predicate, predicate_for_phi_result,
24028 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24029 estimate_function_body_sizes, compute_inline_parameters,
24030 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24031 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24032 inline_update_callee_summaries, remap_edge_change_prob,
24033 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24034 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24035 inline_analyze_function, inline_summary_t::insert,
24036 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24037 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24038 inline_free_summary): Move to ipa-fnsummary.h
24039 (predicate_t): Remove.
24040 * ipa-fnsummary.c: New file.
24041 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24042 (enum inline_hints_vals, inline_hints, agg_position_info,
24043 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24044 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24045 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24046 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24047 inline_read_summary, inline_write_summary, inline_free_summary,
24048 inline_analyze_function, initialize_inline_failed,
24049 inline_merge_summary, inline_update_overall_summary,
24050 compute_inline_parameters): Move to ipa-fnsummary.h
24051 * ipa-fnsummary.h: New file.
24052 * ipa-inline-transform.h: Include ipa-inline.h.
24053 * ipa-inline.c: LIkewise.
24055 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24057 * ipa-inline.c (edge_badness): Use inlined_time instead of
24058 inline_summaries->get.
24060 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24062 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24064 2017-05-22 Nathan Sidwell <nathan@acm.org>
24066 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24067 (fdump-lang): Document 'raw' option.
24068 * dumpfile.h (TDI_tu): Delete.
24069 * dumpfile.c (dump_files): Remove translation-unit.
24070 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24072 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24074 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24075 command option from $(AWK) call.
24076 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24077 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24078 [FORMAT]: Remove handling of variable.
24079 * config/avr/t-multilib: Regenerate.
24081 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24083 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24085 (dump_inline_summary): Do not print self_time.
24086 (estimate_function_body_sizes): Do not set self_time.
24087 (compute_inline_parameters): Likewise.
24088 (inline_read_section, inline_write_summary): Do not stream self_time.
24089 * ipa-inline.h (inline_summary): Drop self_time.
24091 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24093 * ipa-inline-analysis.c (account_size_time): Rename to ...
24094 (inline_summary::account_size_time): ... this one.
24095 (reset_ipa_call_summary): Turn to ...
24096 (ipa_call_summary::reset): ... this one.
24097 (reset_inline_summary): Turn to ...
24098 (inline_summary::reset): ... this one.
24099 (inline_summary_t::remove): Update.
24100 (inline_summary_t::duplicate): Update.
24101 (ipa_call_summary_t::remove): Update.
24102 (dump_inline_summary): Update.
24103 (estimate_function_body_sizes): Update.
24104 (compute_inline_parameters): Update.
24105 (estimate_node_size_and_time): Update.
24106 (inline_merge_summary): Update.
24107 (inline_update_overall_summary): Update.
24108 (inline_read_section): Update.
24109 (inline_write_summary): Update.
24110 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24111 add account_size_time and reset member functions.
24112 (ipa_call_summary): Add reset function.
24113 * ipa-predicate.h (predicate::operator &): Constify.
24115 2017-05-22 Richard Biener <rguenther@suse.de>
24117 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24119 2017-05-19 Jason Merrill <jason@redhat.com>
24121 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24123 2017-05-19 Marek Polacek <polacek@redhat.com>
24126 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24127 TYPE_OVERFLOW_WRAPS checks.
24129 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24131 * tree-core.h (enum omp_clause_default_kind): Add
24132 "OMP_CLAUSE_DEFAULT_PRESENT".
24133 * tree-pretty-print.c (dump_omp_clause): Handle it.
24134 * gimplify.c (enum gimplify_omp_var_data): Add
24135 "GOVD_MAP_FORCE_PRESENT".
24136 (gimplify_adjust_omp_clauses_1): Map it to
24137 "GOMP_MAP_FORCE_PRESENT".
24138 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24140 * gimplify.c (oacc_default_clause): Clarify.
24142 2017-05-19 Nathan Sidwell <nathan@acm.org>
24144 LANG_HOOK_REGISTER_DUMPS
24145 * toplev.c (general_init): Call register dump lang hook.
24146 * doc/invoke.texi: Document -fdump-lang option family.
24147 * dumpfile.c (dump_files): Remove class dump here.
24148 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24149 * dumpfile.h (tree_dump_index): Remove TDI_class.
24150 * langhooks-def.h (lhd_register_dumps): Declare.
24151 (LANG_HOOKS_REGISTER_DUMPS): Define.
24152 (LANG_HOOKS_INITIALIZER): Add it.
24153 * langhooks.c (lhd_register_dumps): Define.
24154 * langhooks.h (struct lang_hooks): Add register_dumps.
24156 2017-05-19 Nathan Sidwell <nathan@acm.org>
24158 * context.h (context::set_passes): New.
24159 * context.c (context::context): Do not create pass manager.
24160 * toplev.c (general_init): Create pass manager here.
24162 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24164 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24165 use this splitter if two add or or instructions would also work for
24166 the constant we want to generate.
24168 2017-05-19 Richard Biener <rguenther@suse.de>
24171 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24172 predicate evaluation.
24174 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24176 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24178 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24181 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24183 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24184 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24185 (inline_edge_summary_vec): Turn into ...
24186 (ipa_call_summaries): ... this one.
24187 (redirect_to_unreachable, edge_set_predicate,
24188 evaluate_properties_for_edge, inline_summary_alloc,
24189 reset_ipa_call_summary, reset_inline_summary,
24190 inline_summary_t::duplicate): Update.
24191 (inline_edge_duplication_hook): Turn to ...
24192 (ipa_call_summary_t::duplicate): ... this one.
24193 (inline_edge_removal_hook): Turn to ...
24194 (ipa_call_summary_t::remove): ... this one.
24195 (dump_inline_edge_summary): Turn to ...
24196 (dump_ipa_call_summary): ... this one.
24197 (estimate_function_body_sizes): Update.
24198 (inline_update_callee_summaries): Update.
24199 (remap_edge_change_prob): Update.
24200 (remap_edge_summaries): Update.
24201 (inline_merge_summary): Update.
24202 (do_estimate_edge_time): Update.
24203 (inline_generate_summary): Update.
24204 (inline_read_section): Update.
24205 (inline_read_summary): Update.
24206 (inline_free_summary): Update.
24207 * ipa-inline.c (can_inline_edge_p): Update.
24208 (compute_inlined_call_time): Update.
24209 (want_inline_small_function_p): Update.
24210 (edge_badness): Update.
24211 (early_inliner): Update.
24212 * ipa-inline.h (inline_edge_summary): Turn to ...
24213 (ipa_call_summary): ... this one.
24214 (ipa_call_summary_t): New class.
24215 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24216 (ipa_call_summaries): New.
24217 (inline_edge_summary): Remove.
24218 (estimate_edge_growth): Update.
24219 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24220 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24221 * ipa-split.c (execute_split_functions): Update.
24222 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24224 2017-05-19 Richard Biener <rguenther@suse.de>
24226 PR middle-end/80764
24227 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24229 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24231 * config/rs6000/rs6000.c (struct machine_function): Add field
24232 fpr_is_wrapped_separately.
24233 (rs6000_get_separate_components): Use 64 components. Handle the
24234 new FPR components.
24235 (rs6000_components_for_bb): Handle the FPR components.
24236 (rs6000_emit_prologue_components): Handle the FPR components.
24237 (rs6000_emit_epilogue_components): Handle the FPR components.
24238 (rs6000_set_handled_components): Handle the FPR components.
24239 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24240 that are already separately shrink-wrapped.
24241 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24242 that are already separately shrink-wrapped.
24244 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24247 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24250 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24251 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24252 cases where the register allocator uses a move and an offsettable
24253 memory operation to/from a FPR register on ISA 2.06/2.07.
24254 (define_peephole2 for Altivec d-form store): Likewise.
24256 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24259 * config/i386/mmx.md (*mov<mode>_internal): Enable
24260 alternatives 11, 12, 13 and 14 also for 32bit targets.
24261 Remove alternatives 15, 16, 17 and 18.
24262 * config/i386/sse.md (vec_concatv2di): Change
24263 alternative (!x, *y) to (x, ?!*Yn).
24265 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24267 * dumpfile.h (enum dump_kind): Remove stray comma.
24269 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24271 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24272 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24273 predicate::num_conditions
24274 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24275 (CHANGED): turn into predicate::changed.
24276 (agg_position_info): Move to ipa-predicate.h
24277 (add_condition, predicate::add_clause, predicate::operator &=,
24278 predicate::or_with, predicate::evaluate, predicate::probability,
24279 dump_condition, dump_clause, predicate::dump,
24280 predicate::remap_after_duplication, predicate::remap_after_inlining,
24281 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24282 (evaluate_conditions_for_known_args): Update.
24283 (set_cond_stmt_execution_predicate): Update.
24284 * ipa-inline.h: Include ipa-predicate.h
24285 (condition, inline_param_summary, conditions, agg_position_info,
24286 predicate): Move to ipa-predicate.h
24287 * ipa-predicate.c: New file.
24288 * ipa-predicate.h: New file.
24290 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24292 * final.c (leaf_function_p): Check we are not in a sequence.
24294 2017-05-18 Martin Liska <mliska@suse.cz>
24296 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24297 * dumpfile.c (dump_register): Use new enum dump_kind.
24298 (get_dump_file_name): Likewise.
24299 (dump_enable_all): Likewise.
24300 (dump_switch_p_1): Likewise.
24301 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24302 * dumpfile.h (enum dump_kind): New enum type.
24303 (struct dump_file_info): Create constructor and
24304 format fields and comments.
24305 * passes.c (pass_manager::register_one_dump_file):
24307 * statistics.c (statistics_early_init): Likewise.
24308 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24309 TDF_TREE with TDF_SLIM.
24310 (gather_memory_references_ref): Likewise.
24312 2017-05-18 Martin Liska <mliska@suse.cz>
24314 * vec.h (struct vnull): Use it.
24316 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24318 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24319 (true_predicate, false_predicate, true_predicate_p,
24320 false_predicate_p): Remove.
24321 (single_cond_predicate, not_inlined_predicate): Turn to member function
24323 (add_condition): Update.
24324 (add_clause): Turn to...
24325 (predicate::add_clause): ... this one; update; allow passing NULL
24327 (and_predicates): Turn to ...
24328 (predicate::operator &=): ... this one.
24329 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24330 (or_predicates): Turn to ...
24331 (predicate::or_with): ... this one.
24332 (evaluate_predicate): Turn to ...
24333 (predicate::evaluate): ... this one.
24334 (predicate_probability): Turn to ...
24335 (predicate::probability): ... this one.
24336 (dump_condition): Update.
24337 (dump_predicate): Turn to ...
24338 (predicate::dump): ... this one.
24339 (account_size_time): Update.
24340 (edge_set_predicate): Update.
24341 (set_hint_predicate): UPdate.
24342 (evaluate_conditions_for_known_args): Update.
24343 (evaluate_properties_for_edge): Update.
24344 (remap_predicate_after_duplication): Turn to...
24345 (predicate::remap_after_duplication): ... this one.
24346 (remap_hint_predicate_after_duplication): Update.
24347 (inline_summary_t::duplicate): UPdate.
24348 (dump_inline_edge_summary): Update.
24349 (dump_inline_summary): Update.
24350 (set_cond_stmt_execution_predicate): Update.
24351 (set_switch_stmt_execution_predicate): Update.
24352 (compute_bb_predicates): Update.
24353 (will_be_nonconstant_expr_predicate): Update.
24354 (will_be_nonconstant_predicate): Update.
24355 (phi_result_unknown_predicate): Update.
24356 (predicate_for_phi_result): Update.
24357 (array_index_predicate): Update.
24358 (estimate_function_body_sizes): Update.
24359 (estimate_node_size_and_time): Update.
24360 (estimate_ipcp_clone_size_and_time): Update.
24361 (remap_predicate): Rename to ...
24362 (predicate::remap_after_inlining): ... this one.
24363 (remap_hint_predicate): Update.
24364 (inline_merge_summary): Update.
24365 (inline_update_overall_summary): Update.
24366 (estimate_size_after_inlining): Update.
24367 (read_predicate): Rename to ...
24368 (predicate::stream_in): ... this one.
24369 (read_inline_edge_summary): Update.
24370 (write_predicate): Rename to ...
24371 (predicate::stream_out): ... this one.
24372 (write_inline_edge_summary): Update.
24373 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24374 (clause_t): Turn to uint32_t
24375 (predicate): Turn to class; implement constructor and operators
24377 (size_time_entry): Update.
24378 (inline_summary): Update.
24379 (inline_edge_summary): Update.
24381 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24383 * fold-const.c (fold_binary_loc): Move transformation...
24384 * match.pd (C - X CMP X): ... here.
24386 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24388 * config/sparc/sparc.c (sparc_option_override): Set function
24389 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24390 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24392 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24394 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24396 2017-05-18 Marek Polacek <polacek@redhat.com>
24399 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24400 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24401 gimple_assign_load_p.
24403 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24405 PR middle-end/80692
24406 * real.c (do_compare): Give decimal_do_compare preference over
24407 comparing just the signs.
24409 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24411 * doc/md.texi (Canonicalization of Instructions): Describe the
24412 canonical form of instructions that inherently set a condition
24415 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24417 PR middle-end/80775
24418 * tree-cfg.c: Move deletion of unreachable case statements to after
24419 the merging of consecutive case labels.
24421 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24423 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24424 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24425 restoring of callee-saved registers.
24427 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24429 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24430 * config/visium/visium.c (single_set_and_flags): Likewise.
24431 * config/visium/visium.md (Substitutions): Likewise.
24433 2017-05-17 Martin Liska <mliska@suse.cz>
24435 * cfg.c: Introduce dump_flags_t type and
24436 use it instead of int type.
24438 * cfghooks.c: Likewise.
24439 * cfghooks.h (struct cfg_hooks): Likewise.
24440 * cfgrtl.c: Likewise.
24441 * cfgrtl.h: Likewise.
24442 * cgraph.c (cgraph_node::get_body): Likewise.
24443 * coretypes.h: Likewise.
24444 * domwalk.c: Likewise.
24445 * domwalk.h: Likewise.
24446 * dumpfile.c (struct dump_option_value_info): Likewise.
24447 (dump_enable_all): Likewise.
24448 (dump_switch_p_1): Likewise.
24449 (opt_info_switch_p): Likewise.
24450 * dumpfile.h (enum tree_dump_index): Likewise.
24451 (struct dump_file_info): Likewise.
24452 * genemit.c: Likewise.
24453 * generic-match-head.c: Likewise.
24454 * gengtype.c (open_base_files): Likewise.
24455 * gimple-pretty-print.c: Likewise.
24456 * gimple-pretty-print.h: Likewise.
24457 * graph.c (print_graph_cfg): Likewise.
24458 * graphite-scop-detection.c (dot_all_sese): Likewise.
24459 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24460 * loop-unroll.c (report_unroll): Likewise.
24461 * passes.c (pass_manager::register_one_dump_file): Likewise.
24462 * print-tree.c: Likewise.
24463 * statistics.c: Likewise.
24464 * tree-cfg.c: Likewise.
24465 * tree-cfg.h: Likewise.
24466 * tree-dfa.c: Likewise.
24467 * tree-dfa.h: Likewise.
24468 * tree-dump.c (dump_function): Likewise.
24469 * tree-dump.h (struct dump_info): Likewise.
24470 * tree-pretty-print.c: Likewise.
24471 * tree-pretty-print.h: Likewise.
24472 * tree-ssa-live.c: Likewise.
24473 * tree-ssa-live.h: Likewise.
24474 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24475 * tree-vect-loop.c: Likewise.
24476 * tree-vect-slp.c: Likewise.
24478 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24479 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24481 PR tree-optimization/80457
24482 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24483 of all arguments to a statement as scalar_to_vec operations.
24484 (vectorizable_call): Adjust call to vect_model_simple_cost for
24486 (vectorizable_conversion): Likewise.
24487 (vectorizable_assignment): Likewise.
24488 (vectorizable_shift): Likewise.
24489 (vectorizable_operation): Likewise.
24490 (vectorizable_comparison): Likewise.
24491 (vect_is_simple_cond): Record the def types for operands.
24492 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24493 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24494 for statement argument count.
24496 2017-05-16 Carl Love <cel@us.ibm.com>
24498 * config/rs6000/rs6000-c: Add support for built-in functions
24499 vector unsigned long long vec_bperm (vector unsigned long long,
24500 vector unsigned char)
24501 vector signed long long vec_mule (vector signed int,
24503 vector unsigned long long vec_mule (vector unsigned int,
24504 vector unsigned int)
24505 vector signed long long vec_mulo (vector signed int,
24507 vector unsigned long long vec_mulo (vector unsigned int,
24508 vector unsigned int)
24509 vector signed char vec_sldw (vector signed char,
24510 vector signed char,
24512 vector unsigned char vec_sldw (vector unsigned char,
24513 vector unsigned char,
24515 vector signed short vec_sldw (vector signed short,
24516 vector signed short,
24518 vector unsigned short vec_sldw (vector unsigned short,
24519 vector unsigned short,
24521 vector signed int vec_sldw (vector signed int,
24524 vector unsigned int vec_sldw (vector unsigned int,
24525 vector unsigned int,
24527 vector signed long long vec_sldw (vector signed long long,
24528 vector signed long long,
24530 vector unsigned long long vec_sldw (vector unsigned long long,
24531 vector unsigned long long,
24533 * config/rs6000/rs6000-c: Add support for built-in functions
24534 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24535 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24536 * doc/extend.texi: Update the built-in documentation for the
24537 new built-in functions.
24539 2017-05-16 Marek Polacek <polacek@redhat.com>
24543 * tree.c (save_expr): Don't fold the expression.
24545 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24547 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24548 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24549 and (?*y,m). Update insn attributes.
24551 2017-05-16 Martin Liska <mliska@suse.cz>
24553 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24554 flags argument of print_gimple_stmt, print_gimple_expr,
24555 print_generic_stmt and print_generic_expr.
24556 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24557 * coretypes.h: Likewise.
24558 * except.c (dump_eh_tree): Likewise.
24559 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24560 * gimple-pretty-print.h: Likewise.
24561 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24562 (backprop::push_to_worklist): Likewise.
24563 (backprop::pop_from_worklist): Likewise.
24564 (backprop::process_use): Likewise.
24565 (backprop::intersect_uses): Likewise.
24566 (note_replacement): Likewise.
24567 * gimple-ssa-store-merging.c
24568 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24569 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24570 (pass_store_merging::execute): Likewise.
24571 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24572 (ssa_base_cand_dump_callback): Likewise.
24573 (dump_incr_vec): Likewise.
24574 (replace_refs): Likewise.
24575 (replace_mult_candidate): Likewise.
24576 (create_add_on_incoming_edge): Likewise.
24577 (create_phi_basis): Likewise.
24578 (insert_initializers): Likewise.
24579 (all_phi_incrs_profitable): Likewise.
24580 (introduce_cast_before_cand): Likewise.
24581 (replace_one_candidate): Likewise.
24582 * gimplify.c (gimplify_expr): Likewise.
24583 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24584 (set_rename): Likewise.
24585 (rename_uses): Likewise.
24586 (copy_loop_phi_nodes): Likewise.
24587 (add_close_phis_to_merge_points): Likewise.
24588 (copy_loop_close_phi_args): Likewise.
24589 (copy_cond_phi_args): Likewise.
24590 (graphite_copy_stmts_from_block): Likewise.
24591 (translate_pending_phi_nodes): Likewise.
24592 * graphite-poly.c (print_pdr): Likewise.
24593 (dump_gbb_cases): Likewise.
24594 (dump_gbb_conditions): Likewise.
24595 (print_scop_params): Likewise.
24596 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24597 (build_cross_bb_scalars_use): Likewise.
24598 (gather_bbs::before_dom_children): Likewise.
24599 * hsa-dump.c (dump_hsa_immed): Likewise.
24600 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24601 (get_replacement_map): Likewise.
24602 * ipa-inline-analysis.c (dump_condition): Likewise.
24603 (estimate_function_body_sizes): Likewise.
24604 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24605 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24606 * ipa-prop.c (ipa_dump_param): Likewise.
24607 (ipa_print_node_jump_functions_for_edge): Likewise.
24608 (ipa_modify_call_arguments): Likewise.
24609 (ipa_modify_expr): Likewise.
24610 (ipa_dump_param_adjustments): Likewise.
24611 (ipa_dump_agg_replacement_values): Likewise.
24612 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24613 * ipa-pure-const.c (check_stmt): Likewise.
24614 (pass_nothrow::execute): Likewise.
24615 * ipa-split.c (execute_split_functions): Likewise.
24616 * omp-offload.c (dump_oacc_loop_part): Likewise.
24617 (dump_oacc_loop): Likewise.
24618 * trans-mem.c (tm_log_emit): Likewise.
24619 (tm_memopt_accumulate_memops): Likewise.
24620 (dump_tm_memopt_set): Likewise.
24621 (dump_tm_memopt_transform): Likewise.
24622 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24623 (print_loop): Likewise.
24624 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24625 (chkp_gather_checks_info): Likewise.
24626 (chkp_get_check_result): Likewise.
24627 (chkp_remove_check_if_pass): Likewise.
24628 (chkp_use_outer_bounds_if_possible): Likewise.
24629 (chkp_reduce_bounds_lifetime): Likewise.
24630 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24631 (chkp_mark_completed_bounds): Likewise.
24632 (chkp_register_incomplete_bounds): Likewise.
24633 (chkp_mark_invalid_bounds): Likewise.
24634 (chkp_maybe_copy_and_register_bounds): Likewise.
24635 (chkp_build_returned_bound): Likewise.
24636 (chkp_get_bound_for_parm): Likewise.
24637 (chkp_build_bndldx): Likewise.
24638 (chkp_get_bounds_by_definition): Likewise.
24639 (chkp_generate_extern_var_bounds): Likewise.
24640 (chkp_get_bounds_for_decl_addr): Likewise.
24641 * tree-chrec.c (chrec_apply): Likewise.
24642 * tree-data-ref.c (dump_data_reference): Likewise.
24643 (dump_subscript): Likewise.
24644 (dump_data_dependence_relation): Likewise.
24645 (analyze_overlapping_iterations): Likewise.
24646 * tree-inline.c (expand_call_inline): Likewise.
24647 (tree_function_versioning): Likewise.
24648 * tree-into-ssa.c (dump_defs_stack): Likewise.
24649 (dump_currdefs): Likewise.
24650 (dump_names_replaced_by): Likewise.
24651 (dump_update_ssa): Likewise.
24652 (update_ssa): Likewise.
24653 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24654 * tree-parloops.c (build_new_reduction): Likewise.
24655 (try_create_reduction_list): Likewise.
24656 (ref_conflicts_with_region): Likewise.
24657 (oacc_entry_exit_ok_1): Likewise.
24658 (oacc_entry_exit_single_gang): Likewise.
24659 * tree-pretty-print.h: Likewise.
24660 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24661 (get_scalar_evolution): Likewise.
24662 (add_to_evolution): Likewise.
24663 (get_loop_exit_condition): Likewise.
24664 (analyze_evolution_in_loop): Likewise.
24665 (analyze_initial_condition): Likewise.
24666 (analyze_scalar_evolution): Likewise.
24667 (instantiate_scev): Likewise.
24668 (number_of_latch_executions): Likewise.
24669 (gather_chrec_stats): Likewise.
24670 (final_value_replacement_loop): Likewise.
24671 (scev_const_prop): Likewise.
24672 * tree-sra.c (dump_access): Likewise.
24673 (disqualify_candidate): Likewise.
24674 (create_access): Likewise.
24675 (reject): Likewise.
24676 (maybe_add_sra_candidate): Likewise.
24677 (create_access_replacement): Likewise.
24678 (analyze_access_subtree): Likewise.
24679 (analyze_all_variable_accesses): Likewise.
24680 (sra_modify_assign): Likewise.
24681 (initialize_constant_pool_replacements): Likewise.
24682 (find_param_candidates): Likewise.
24683 (decide_one_param_reduction): Likewise.
24684 (replace_removed_params_ssa_names): Likewise.
24685 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24686 * tree-ssa-copy.c (dump_copy_of): Likewise.
24687 (copy_prop_visit_cond_stmt): Likewise.
24688 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24689 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24690 (record_equivalences_from_stmt): Likewise.
24691 * tree-ssa-dse.c (compute_trims): Likewise.
24692 (delete_dead_call): Likewise.
24693 (delete_dead_assignment): Likewise.
24694 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24695 (forward_propagate_into_cond): Likewise.
24696 (pass_forwprop::execute): Likewise.
24697 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24698 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24700 (move_computations_worker): Likewise.
24701 (execute_sm): Likewise.
24702 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24703 (remove_exits_and_undefined_stmts): Likewise.
24704 (remove_redundant_iv_tests): Likewise.
24705 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24706 (adjust_iv_update_pos): Likewise.
24707 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24708 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24709 (value_replacement): Likewise.
24710 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24711 * tree-ssa-pre.c (print_pre_expr): Likewise.
24712 (get_representative_for): Likewise.
24713 (create_expression_by_pieces): Likewise.
24714 (insert_into_preds_of_block): Likewise.
24715 (eliminate_insert): Likewise.
24716 (eliminate_dom_walker::before_dom_children): Likewise.
24717 (eliminate): Likewise.
24718 (remove_dead_inserted_code): Likewise.
24719 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24720 * tree-ssa-reassoc.c (get_rank): Likewise.
24721 (eliminate_duplicate_pair): Likewise.
24722 (eliminate_plus_minus_pair): Likewise.
24723 (eliminate_not_pairs): Likewise.
24724 (undistribute_ops_list): Likewise.
24725 (eliminate_redundant_comparison): Likewise.
24726 (update_range_test): Likewise.
24727 (optimize_range_tests_var_bound): Likewise.
24728 (optimize_vec_cond_expr): Likewise.
24729 (rewrite_expr_tree): Likewise.
24730 (rewrite_expr_tree_parallel): Likewise.
24731 (linearize_expr): Likewise.
24732 (break_up_subtract): Likewise.
24733 (linearize_expr_tree): Likewise.
24734 (attempt_builtin_powi): Likewise.
24735 (attempt_builtin_copysign): Likewise.
24736 (transform_stmt_to_copy): Likewise.
24737 (transform_stmt_to_multiply): Likewise.
24738 (dump_ops_vector): Likewise.
24739 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24740 (print_scc): Likewise.
24741 (set_ssa_val_to): Likewise.
24742 (visit_reference_op_store): Likewise.
24743 (visit_use): Likewise.
24744 (sccvn_dom_walker::before_dom_children): Likewise.
24745 (run_scc_vn): Likewise.
24746 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24748 (expr_hash_elt::print): Likewise.
24749 (const_and_copies::pop_to_marker): Likewise.
24750 (const_and_copies::record_const_or_copy_raw): Likewise.
24751 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24752 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24753 (dump_predicates): Likewise.
24754 (find_uninit_use): Likewise.
24755 (warn_uninitialized_phi): Likewise.
24756 (pass_late_warn_uninitialized::execute): Likewise.
24757 * tree-ssa.c (verify_vssa): Likewise.
24758 (verify_ssa): Likewise.
24759 (maybe_optimize_var): Likewise.
24760 * tree-vrp.c (dump_value_range): Likewise.
24761 (dump_all_value_ranges): Likewise.
24762 (dump_asserts_for): Likewise.
24763 (register_edge_assert_for_2): Likewise.
24764 (vrp_visit_cond_stmt): Likewise.
24765 (vrp_visit_switch_stmt): Likewise.
24766 (vrp_visit_stmt): Likewise.
24767 (vrp_visit_phi_node): Likewise.
24768 (simplify_cond_using_ranges_1): Likewise.
24769 (fold_predicate_in): Likewise.
24770 (evrp_dom_walker::before_dom_children): Likewise.
24771 (evrp_dom_walker::push_value_range): Likewise.
24772 (evrp_dom_walker::pop_value_range): Likewise.
24773 (execute_early_vrp): Likewise.
24775 2017-05-16 Richard Biener <rguenther@suse.de>
24777 * dwarf2out.c (loc_list_from_tree_1): Do not create
24778 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24780 2017-05-16 Richard Biener <rguenther@suse.de>
24782 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24784 (note_variable_value_in_expr): If we resolved the decl ref
24785 do not push to the stack.
24787 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24789 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24790 operations in fast-math mode.
24791 (vaddq_f16): Likewise.
24792 (vmul_f16): Likewise.
24793 (vmulq_f16): Likewise.
24794 (vsub_f16): Likewise.
24795 (vsubq_f16): Likewise.
24796 * config/arm/neon.md (add<mode>3): New.
24798 (fma:<VH:mode>3): New. Also remove outdated comment.
24801 2017-05-16 Martin Liska <mliska@suse.cz>
24805 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24806 (odr_types_equivalent_p): Likewise.
24808 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24810 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24812 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24815 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24816 non-interunit SSE move alternatives with '?'.
24817 (zero-extendsidi peephole2): New peephole to skip intermediate
24818 general register in SSE zero-extend sequence.
24820 2017-05-15 Jeff Law <law@redhat.com>
24822 * reorg.c (relax_delay_slots): Create a new variable to hold
24823 the temporary target rather than clobbering TARGET_LABEL.
24825 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24826 missing argument to extract_bit_field call.
24827 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24829 2017-05-15 Martin Liska <mliska@suse.cz>
24832 * gcc.c (process_command): Do not allow empty argument of -o option.
24834 2017-05-15 Renlin Li <renlin.li@arm.com>
24836 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24837 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24838 * config/aarch64/constraints.md (Usf): Add long call check.
24839 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24840 (call_value): Likewise.
24841 (sibcall): Likewise.
24842 (sibcall_value): Likewise.
24844 (call_value_insn): New.
24845 (sibcall_insn): Update rtx pattern.
24846 (sibcall_value_insn): Likewise.
24847 (call_internal): Remove.
24848 (call_value_internal): Likewise.
24849 (sibcall_internal): Likewise.
24850 (sibcall_value_internal): Likewise.
24851 (call_reg): Likewise.
24852 (call_symbol): Likewise.
24853 (call_value_reg): Likewise.
24854 (call_value_symbol): Likewise.
24856 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24859 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24861 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24863 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24864 compatible with CCGOCmode and with CCZmode.
24866 2017-05-14 Martin Sebor <msebor@redhat.com>
24868 PR middle-end/77671
24869 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24870 (gimple_fold_builtin_snprintf): Same.
24871 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24872 (gimple_fold_builtin_snprintf): Same.
24873 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24874 of character types.
24875 (is_call_safe): New function.
24876 (try_substitute_return_value): Call it.
24877 (try_simplify_call): New function.
24878 (pass_sprintf_length::handle_gimple_call): Call it.
24880 2017-05-14 Martin Sebor <msebor@redhat.com>
24882 PR middle-end/80669
24883 * builtins.c (expand_builtin_stpncpy): Simplify.
24885 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24887 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24888 * config/i386/i386.h
24889 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24890 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24891 (struct machine_function): Add new members call_ms2sysv,
24892 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24893 (struct machine_frame_state): New fields sp_realigned and
24894 sp_realigned_offset.
24895 * config/i386/i386.c
24896 (enum xlogue_stub): New enum.
24897 (enum xlogue_stub_sets): New enum.
24898 (class xlogue_layout): New class.
24899 (struct ix86_frame): New fields stack_realign_allocate_offset,
24900 stack_realign_offset and outlined_save_offset. Modify comments to
24901 detail stack layout when using out-of-line stubs.
24902 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24903 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24904 -mcall-ms2sysv-xlogues.
24905 (stub_managed_regs): New static variable.
24906 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24907 registers managed by out-of-line stub.
24908 (disable_call_ms2sysv_xlogues): New function.
24909 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24910 m->call_ms2sysv when appropriate and compute frame layout for
24912 (sp_valid_at, fp_valid_at): New inline functions.
24913 (choose_basereg): New function.
24914 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24916 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24917 Use align parameter of choose_baseaddr to generated aligned SSE movs
24919 (pro_epilogue_adjust_stack): Modify to track
24920 machine_frame_state::sp_realigned.
24921 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24922 (ix86_nsaved_sseregs): Likewise.
24923 (ix86_emit_save_regs): Likewise.
24924 (ix86_emit_save_regs_using_mov): Likewise.
24925 (ix86_emit_save_sse_regs_using_mov): Likewise.
24926 (get_scratch_register_on_entry): Likewise.
24927 (gen_frame_set): New function.
24928 (gen_frame_load): Likewise.
24929 (gen_frame_store): Likewise.
24930 (emit_outlined_ms2sysv_save): Likewise.
24931 (emit_outlined_ms2sysv_restore): Likewise.
24932 (ix86_expand_prologue): Modify stack re-alignment code and call
24933 emit_outlined_ms2sysv_save when appropriate.
24934 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24935 parameter rtx_insn *insn, which allows the function to be used to only
24936 generate the notes.
24937 (ix86_expand_epilogue): Modify validity checks of frame and stack
24938 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24939 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24940 * config/i386/predicates.md
24941 (save_multiple): New predicate.
24942 (restore_multiple): Likewise.
24943 * config/i386/sse.md
24944 (save_multiple<mode>): New pattern.
24945 (save_multiple_realign<mode>): Likewise.
24946 (restore_multiple<mode>): Likewise.
24947 (restore_multiple_and_return<mode>): Likewise.
24948 (restore_multiple_leave_return<mode>): Likewise.
24949 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24951 2017-05-14 Julia Koval <julia.koval@intel.com>
24953 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24954 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24955 (__builtin_ia32_xsetbv): New builtins.
24956 * config/i386/i386.c (ix86_expand_special_args_builtin):
24958 (ix86_expand_builtin): Special expand for new intrinsics.
24959 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24960 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24961 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24963 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24965 * cfganal.c (inverted_post_order_compute): Change argument type
24967 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24968 * df-core.c (rest_of_handle_df_initialize): Adjust.
24969 (rest_of_handle_df_finish): Likewise.
24970 (df_analyze_1): Likewise.
24971 (df_analyze): Likewise.
24972 (loop_inverted_post_order_compute): Change argument to be a vec *.
24973 (df_analyze_loop): Adjust.
24974 (df_get_n_blocks): Likewise.
24975 (df_get_postorder): Likewise.
24976 * df.h (struct df_d): Change field to be a vec.
24977 * lcm.c (compute_laterin): Adjust.
24978 (compute_available): Likewise.
24979 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24980 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24981 * tree-ssa-pre.c (compute_antic): Likewise.
24983 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24985 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24986 (depth_first_search::depth_first_search): Change structure init
24987 function to this constructor.
24988 (depth_first_search::add_bb): Rename function to this member.
24989 (depth_first_search::execute): Likewise.
24990 (flow_dfs_compute_reverse_finish): Adjust.
24992 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24994 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24995 (longest_simple_path): Likewise.
24996 * shrink-wrap.c (spread_components): Likewise.
24997 (disqualify_problematic_components): Likewise.
24998 (emit_common_heads_for_components): Likewise.
24999 (emit_common_tails_for_components): Likewise.
25000 (insert_prologue_epilogue_for_components): Likewise.
25002 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25004 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25007 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25009 * df-core.c (df_set_blocks): Start using auto_bitmap.
25010 (df_compact_blocks): Likewise.
25011 * df-problems.c (df_rd_confluence_n): Likewise.
25012 * df-scan.c (df_insn_rescan_all): Likewise.
25013 (df_process_deferred_rescans): Likewise.
25014 (df_update_entry_block_defs): Likewise.
25015 (df_update_exit_block_uses): Likewise.
25016 (df_entry_block_bitmap_verify): Likewise.
25017 (df_exit_block_bitmap_verify): Likewise.
25018 (df_scan_verify): Likewise.
25019 * lra-constraints.c (lra_constraints): Likewise.
25020 (undo_optional_reloads): Likewise.
25021 (lra_undo_inheritance): Likewise.
25022 * lra-remat.c (calculate_gen_cands): Likewise.
25023 (do_remat): Likewise.
25024 * lra-spills.c (assign_spill_hard_regs): Likewise.
25025 (spill_pseudos): Likewise.
25026 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25027 (bitmap_set_subtract_values): Likewise.
25029 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25031 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25032 management with auto_bitmap.
25033 (fix_inter_tick): Likewise.
25034 (fix_recovery_deps): Likewise.
25035 * ira.c (add_store_equivs): Likewise.
25036 (find_moveable_pseudos): Likewise.
25037 (split_live_ranges_for_shrink_wrap): Likewise.
25038 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25039 (rtx_reuse_manager::seen_def_p): Likewise.
25040 (rtx_reuse_manager::set_seen_def): Likewise.
25041 * print-rtl.h (class rtx_reuse_manager): Likewise.
25043 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25045 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25047 (migrate_btr_def): Likewise.
25048 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25049 * df-core.c (loop_post_order_compute): Likewise.
25050 (loop_inverted_post_order_compute): Likewise.
25051 * hsa-common.h: Likewise.
25052 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25053 * init-regs.c (initialize_uninitialized_regs): Likewise.
25054 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25055 (inline_small_functions): Likewise.
25056 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25057 * ira.c (combine_and_move_insns): Likewise.
25058 (build_insn_chain): Likewise.
25059 * loop-invariant.c (find_invariants): Likewise.
25060 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25061 * predict.c (tree_predict_by_opcode): Likewise.
25062 (predict_paths_leading_to): Likewise.
25063 (predict_paths_leading_to_edge): Likewise.
25064 (estimate_loops_at_level): Likewise.
25065 (estimate_loops): Likewise.
25066 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25067 (spread_components): Likewise.
25068 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25069 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25070 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25071 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25072 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25073 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25074 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25075 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25076 (mark_threaded_blocks): Likewise.
25077 (thread_through_all_blocks): Likewise.
25078 * tree-ssa.c (verify_ssa): Likewise.
25079 (execute_update_addresses_taken): Likewise.
25080 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25082 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25084 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25086 (post_order_compute): Likewise.
25087 (inverted_post_order_compute): Likewise.
25088 (pre_and_rev_post_order_compute_fn): Likewise.
25090 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25092 * genrecog.c (int_set::int_set): Explicitly construct our
25093 auto_vec base class.
25094 * vec.h (auto_vec::auto_vec): New constructor.
25096 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25098 * bitmap.h (class auto_bitmap): New constructor taking
25099 bitmap_obstack * argument.
25101 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25103 * bitmap.h (class auto_bitmap): Change type of m_bits to
25104 bitmap_head, and adjust ctor / dtor and member operators.
25106 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25108 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25109 when returned register mode doesn't match original mode.
25111 2017-05-12 Jeff Law <law@redhat.com>
25112 Jakub Jelinek <jakub@redhat.com>
25114 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25115 we look for cc setter after the compare-elim changes.
25116 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25117 within the vector to match what compare-elim now expects.
25118 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25119 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25121 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25122 after the compare-elim changes.
25123 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25124 the vector to match what compare-elim now expects.
25125 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25126 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25127 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25128 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25129 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25131 * config/visium/visium.c (single_set_and_flags): Fix where
25132 we look for cc setter after the compare-elim changes.
25133 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25134 with the vector to match what compare-elim now expects.
25135 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25136 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25137 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25138 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25139 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25140 (neg<mode>2_insn_set_overflow): Likewise.
25142 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25144 PR middle-end/79794
25145 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25146 maybe_expand_insn call, set ops[0].target. If still set after call,
25147 set alt_rtl. Add extra arg to recursive calls.
25148 (extract_bit_field): Add alt_rtl argument. Pass to
25150 * expmed.h (extract_bit_field): Fix prototype.
25151 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25152 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25153 to extract_bit_field_calls.
25154 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25155 Pass alt_rtl to extract_bit_field calls.
25156 * calls.c (store_unaligned_arguments_into_psuedos)
25157 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25158 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25160 * optabs.h (struct expand_operand): Add target bitfield.
25162 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25164 * compare-elim.c (try_eliminate_compare): Canonicalize
25165 operation with embedded compare to
25166 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25167 (set (reg) (operation)].
25169 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25171 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25174 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25175 cost of adding a carry flag for ADC instruction.
25176 [case MINUS]: Ignore the cost of subtracting a carry flag
25177 for SBB instruction.
25179 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25181 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25183 * config/rs6000/bmiintrin.h: New file.
25184 * config/rs6000/bmi2intrin.h: New file.
25185 * config/rs6000/x86intrin.h: New file.
25187 2017-05-12 Jeff Law <law@redhat.com>
25189 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25192 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25194 PR middle-end/80707
25195 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25197 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25199 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25200 early expansion of vector divide builtins.
25201 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25202 builtins identified as having unsigned arguments.
25204 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25206 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25207 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25208 expansion of vector logical operations (and, andc, or, xor,
25211 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25213 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25214 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25216 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25218 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25219 early GIMPLE expansion of vector multiplies.
25221 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25223 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25224 TARGET_HAVE_MOVT conditional.
25225 (movt splitter): Likewise.
25227 2017-05-12 Richard Biener <rguenther@suse.de>
25229 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25230 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25231 Fold all stmts not inplace.
25233 2017-05-12 Richard Biener <rguenther@suse.de>
25235 PR tree-optimization/80713
25236 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25237 inserted_exprs bit for not removed stmts.
25239 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25241 PR middle-end/69921
25242 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25243 parallelized" attribute for parallelized OpenACC kernels.
25244 * omp-offload.c (execute_oacc_device_lower): Use it.
25246 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25247 Set "oacc kernels" attribute.
25248 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25249 parameter. Adjust all users.
25250 (oacc_fn_attrib_kernels_p): Remove function.
25251 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25252 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25253 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25254 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25255 assert "oacc kernels" attribute is set.
25257 2017-05-11 Carl Love <cel@us.ibm.com>
25259 * config/rs6000/rs6000-c: Add support for built-in functions
25260 vector unsigned char vec_popcnt (vector signed char)
25261 vector unsigned char vec_popcnt (vector unsigned char)
25262 vector unsigned short vec_popcnt (vector signed short)
25263 vector unsigned short vec_popcnt (vector unsigned short)
25264 vector unsigned int vec_popcnt (vector signed int)
25265 vector unsigned int vec_popcnt (vector unsigned int)
25266 vector unsigned long long vec_popcnt (vector signed long long)
25267 vector unsigned long long vec_popcnt (vector unsigned long long)
25268 vector signed long long vec_slo (vector signed long long,
25269 vector signed char)
25270 vector signed long long vec_slo (vector signed long long,
25271 vector unsigned char)
25272 vector unsigned long long vec_slo (vector unsigned long long,
25273 vector signed char)
25274 vector unsigned long long vec_slo (vector unsigned long long,
25275 vector unsigned char)
25276 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25277 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25278 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25279 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25280 * doc/extend.texi: Update the built-in documentation file for the
25281 new built-in functions.
25283 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25285 * attribs.h (sorted_attr_string): Move machine independent
25286 functions for target clone support from the i386 port to common
25287 code. Rename ix86_function_versions to common_function_versions.
25288 Rename make_name to make_unique_name.
25289 (common_function_versions): Likewise.
25290 (make_unique_name): Likewise.
25291 (make_dispatcher_decl): Likewise.
25292 (is_function_default_version): Likewise.
25293 * attribs.c (attr_strcmp): Likewise.
25294 (sorted_attr_string): Likewise.
25295 (common_function_versions): Likewise.
25296 (make_unique_name): Likewise.
25297 (make_dispatcher_decl): Likewise.
25298 (is_function_default_version): Likewise.
25299 * config/i386/i386.c (attr_strcmp): Likewise.
25300 (sorted_attr_string): Likewise.
25301 (ix86_function_versions): Likewise.
25302 (make_name): Likewise.
25303 (make_dispatcher_decl): Likewise.
25304 (is_function_default_version): Likewise.
25305 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25307 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25310 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25311 Account for direct move costs for vec_construct of integer
25314 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25317 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25318 (UNSPEC_STX_ATOMIC): Ditto.
25319 (loaddi_via_sse): New insn.
25320 (storedi_via_sse): Ditto.
25321 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25322 Update corresponding peephole2 patterns.
25323 (atomic_storedi_fpu): Ditto.
25325 2017-05-11 Julia Koval <julia.koval@intel.com>
25327 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25328 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25330 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25331 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25332 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25334 2017-05-11 Nathan Sidwell <nathan@acm.org>
25336 * graphite-poly.c: Include dumpfile.h.
25338 * dumpfle.h (dump_function): Declare here ...
25339 * tree-dump.h (dump_function): ... not here.
25340 * dumpfile.c: #include tree-cfg.h.
25341 (dump_function): Move here from ...
25342 * tree-dump.c (dump_function): ... here.
25343 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25344 * graphite-poly.c: Don't include tree-dump.h.
25345 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25346 * print-tree.c: Likewise.
25347 * stor-layout.c: Likewise.
25348 * tree-nested.c: Likewise.
25350 * dumpfile.c (dump_start): Use TDF_FLAGS.
25351 (dump_enable_all): Fix TDF_KIND check thinko.
25353 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25355 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25356 array entries to represent two legal parameterizations of the
25357 overloaded __builtin_cmpb function, as represented by the
25358 P6_OV_BUILTIN_CMPB constant.
25359 (altivec_resolve_overloaded_builtin): Add special case handling
25360 for the __builtin_cmpb function, as represented by the
25361 P6_OV_BUILTIN_CMPB constant.
25362 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25363 (BU_P6_64BIT_2): New macro.
25364 (BU_P6_OVERLOAD_2): New macro
25365 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25366 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25367 (CMPB): Add overload support to represent both 32-bit and 64-bit
25368 compare-bytes function.
25369 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25370 support for TARGET_CMPB.
25371 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25372 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25373 documentation of the __builtin_cmpb overloaded built-in function.
25375 2017-05-11 Richard Biener <rguenther@suse.de>
25377 PR tree-optimization/80705
25378 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25379 bases are not vectorizable.
25381 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25383 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25384 when counting register pressure.
25386 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25388 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25389 (iv_ca_more_deps): Renamed to ...
25390 (iv_ca_compare_deps): ... this.
25391 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25393 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25395 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25397 (determine_group_iv_costs): ... here.
25398 (find_inv_vars_cb): Record inv var if it's not recorded before.
25400 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25402 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25403 (get_shiftadd_cost): Ditto.
25405 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25407 * tree-ssa-address.c: Include header file.
25408 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25410 (add_to_parts): Refactor.
25411 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25412 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25415 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25417 PR tree-optimization/53090
25418 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25420 (extract_cond_operands): Detect condition with IV on both sides
25421 and return COMP_IV_EXPR_2.
25422 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25423 (rewrite_use_compare): Simplify by removing call to function
25424 extract_cond_operands.
25426 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25428 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25429 (extract_cond_operands): Detect condition comparing against non-
25430 invariant bound and return appropriate enum value.
25431 (find_interesting_uses_cond): Update use of extract_cond_operands.
25432 Handle its return value accordingly.
25433 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25435 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25437 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25438 nonlinear iv_use computation in loop invariant sensitive way.
25440 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25442 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25443 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25445 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25447 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25448 (dump_cand): Support iv_cand.inv_exprs.
25449 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25451 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25454 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25456 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25458 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25459 as local function. Include necessary header files.
25460 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25462 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25464 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25466 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25468 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25469 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25470 RSHIFT_EXPR and BIT_NOT_EXPR.
25472 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25474 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25475 (adjust_setup_cost): New parameter supporting round up adjustment.
25476 (struct address_cost_data): Delete.
25477 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25478 (split_address_cost, ptr_difference_cost): Delete.
25479 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25480 (struct ainc_cost_data): New struct.
25481 (get_address_cost_ainc): New function.
25482 (get_address_cost, get_computation_cost): Reimplement.
25483 (determine_group_iv_cost_address): Record inv_expr for all uses of
25485 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25486 (iv_ca_has_deps): Reimplemented to ...
25487 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25489 (iv_ca_extend): Call iv_ca_more_deps.
25491 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25493 * tree-ssa-address.c (struct mem_address): Move to header file.
25494 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25495 * tree-ssa-address.h (struct mem_address): Move from C file.
25496 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25498 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25500 * tree-affine.h (aff_combination_type): New interface.
25501 (aff_combination_zero_p): Remove static.
25502 (aff_combination_const_p): New interface.
25503 (aff_combination_singleton_var_p): New interfaces.
25505 2017-05-11 Richard Biener <rguenther@suse.de>
25507 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25508 Skip unreachable blocks and destinations.
25509 (eliminate): Move stmt removal and fixup ...
25510 (fini_eliminate): ... here. Skip inserted exprs.
25511 (pass_pre::execute): Move fini_pre after fini_eliminate.
25512 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25513 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25514 PRE to get rid of dead code that has invalid SSA form and
25515 split critical edges again.
25517 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25519 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25521 2017-05-11 Richard Biener <rguenther@suse.de>
25523 * passes.c (execute_function_todo): Verify loops if they are
25524 said to be up-to-date.
25525 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25526 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25528 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25531 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25532 handle calling assemble_external ourself.
25535 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25536 modes with zero size. Enhance comment.
25538 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25540 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25541 built-ins for vec_xl and vec_xst with short and char pointer
25544 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25546 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25547 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25548 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25549 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25550 (_mm_maskz_min_round_ss): New intrinsics.
25551 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25552 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25553 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25554 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25555 (__builtin_ia32_minss_mask_round): New builtins.
25556 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25557 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25558 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25560 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25561 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25563 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25566 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25568 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25569 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25570 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25571 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25572 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25573 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25574 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25575 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25576 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25577 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25578 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25579 (__builtin_ia32_mulss_mask_round): New builtins.
25580 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25581 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25582 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25584 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25585 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25587 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25590 2017-05-10 Julia Koval <julia.koval@intel.com>
25592 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25593 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25594 (_mm256_setr_m128i): New intrinsics.
25596 2017-05-10 Julia Koval <julia.koval@intel.com>
25598 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25599 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25600 (_mm_maskz_rcp14_ss): New intrinsics.
25601 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25602 (__builtin_ia32_rcp14ss_mask): New builtins.
25603 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25605 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25607 PR tree-optimization/51513
25608 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25609 (assert_unreachable_fallthru_edge_p): Use it.
25610 (group_case_labels_stmt): Likewise.
25611 * tree-cfg.h: Prototype it.
25612 * stmt.c: Include cfghooks.h and tree-cfg.h.
25613 (emit_case_dispatch_table) <gap_label>: New local variable.
25614 Use it to fill dispatch table gaps.
25615 Test for default_label before updating probabilities.
25616 (expand_case) <default_label>: Remove unneeded initialization.
25617 Test for unreachable default case statement and remove its edge.
25618 Set default_label accordingly.
25619 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25621 2017-05-10 Carl Love <cel@us.ibm.com>
25623 * config/rs6000/rs6000-c: Add support for built-in functions
25624 vector signed char vec_neg (vector signed char)
25625 vector signed short int vec_neg (vector short int)
25626 vector signed int vec_neg (vector signed int)
25627 vector signed long long vec_neg (vector signed long long)
25628 vector float vec_neg (vector float)
25629 vector double vec_neg (vector double)
25630 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25632 * config/rs6000/altivec.h: Add define for vec_neg
25633 * doc/extend.texi: Update the built-in documentation for the
25634 new built-in functions.
25636 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25638 PR tree-optimization/77644
25639 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25641 2017-05-10 Nathan Sidwell <nathan@acm.org>
25643 * dumpfile.h (TDI_lang_all): New.
25644 (TDF_KIND): New. Renumber others
25645 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25647 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25649 (get_dump_file_name): Adjust suffix generation.
25650 (dump_enable_all): Use TDF_KIND.
25651 * doc/invoke.texi (-fdump-lang-all): Document.
25653 * dumpfile.h: Tabify.
25655 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25658 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25659 Move member access before delete.
25661 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25663 * tree-inline.c (expand_call_inline): Split block at stmt
25666 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25669 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25670 are now unused after splitting mov{sf,sd}_hardfloat.
25671 (f32_lr2): Likewise.
25672 (f32_lm): Likewise.
25673 (f32_lm2): Likewise.
25674 (f32_li): Likewise.
25675 (f32_li2): Likewise.
25676 (f32_lv): Likewise.
25677 (f32_sr): Likewise.
25678 (f32_sr2): Likewise.
25679 (f32_sm): Likewise.
25680 (f32_sm2): Likewise.
25681 (f32_si): Likewise.
25682 (f32_si2): Likewise.
25683 (f32_sv): Likewise.
25684 (f32_dm): Likewise.
25685 (f32_vsx): Likewise.
25686 (f32_av): Likewise.
25687 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25688 For movsf, order stores so the VSX stores occur before the GPR
25689 store which encourages the register allocator to use a traditional
25690 FPR instead of a GPR. For movsd, order the stores so that the GPR
25691 store comes before the VSX stores to allow the power6 to work.
25692 This is due to the power6 not having a 32-bit integer store
25693 instruction from a FPR.
25694 (movsf_hardfloat): Likewise.
25695 (movsd_hardfloat): Likewise.
25697 2017-05-09 Martin Sebor <msebor@redhat.com>
25699 PR translation/80280
25700 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25703 PR translation/80280
25704 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25705 data member added in r247778.
25706 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25708 2017-05-09 Nathan Sidwell <nathan@acm.org>
25710 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25712 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25715 2017-05-09 Marek Polacek <polacek@redhat.com>
25717 * doc/invoke.texi: Fix typo.
25719 2017-05-09 Richard Biener <rguenther@suse.de>
25721 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25722 (vrp_val_is_min): Likewise.
25723 (set_value_range_to_value): Likewise.
25724 (set_value_range_to_nonnegative): Likewise.
25725 (gimple_assign_nonzero_p): Likewise.
25726 (gimple_stmt_nonzero_p): Likewise.
25727 (vrp_int_const_binop): Likewise. Remove unreachable case.
25728 (adjust_range_with_scev): Adjust comments.
25729 (compare_range_with_value): Likewise.
25730 (extract_range_from_phi_node): Likewise.
25731 (test_for_singularity): Likewise.
25733 2017-05-09 Richard Biener <rguenther@suse.de>
25735 * tree-vrp.c (get_single_symbol): Add assert that we don't
25736 get overflowed constants as invariant part.
25737 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25738 checks. Use wi::cmp instead of recursing for integer constants.
25739 (compare_values): Just ignore whether we assumed undefined
25740 overflow instead of failing the compare.
25741 (extract_range_for_var_from_comparison_expr): Add comment before the
25742 TREE_NO_WARNING sets.
25743 (test_for_singularity): Likewise.
25744 (extract_range_from_comparison): Do not disable optimization
25745 when we assumed undefined overflow.
25746 (extract_range_basic): Remove init of unused var.
25748 2017-05-09 Richard Biener <rguenther@suse.de>
25750 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25751 (extract_range_from_multiplicative_op_1): Adjust.
25752 (extract_range_from_binary_expr_1): Use int_const_binop.
25754 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25757 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25758 rs6000_store_data_bypass_p in seven define_bypass directives and
25759 in several comments.
25760 * config/rs6000/rs6000-protos.h: Add prototype for
25761 rs6000_store_data_bypass_p function.
25762 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25763 function implements slightly different (rs6000-specific) semantics
25764 than store_data_bypass_p, returning false rather than aborting
25765 with assertion error when arguments do not satisfy the
25766 requirements of store data bypass.
25767 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25768 rs6000_store_data_bypass_p.
25770 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25772 * config/xtensa/xtensa-protos.h
25773 (xtensa_initial_elimination_offset): New declaration.
25774 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25775 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25776 macro definition, add case for FRAME_POINTER_REGNUM when
25777 FRAME_GROWS_DOWNWARD.
25778 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25779 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25780 xtensa_initial_elimination_offset.
25782 2017-05-08 Nathan Sidwell <nathan@acm.org>
25784 * doc/invoke.texi: Alphabetize -fdump options.
25786 2017-05-08 Martin Sebor <msebor@redhat.com>
25788 PR translation/80280
25789 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25791 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25793 * target.def (compute_frame_layout): New optional target hook.
25794 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25795 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25796 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25798 * reload1.c (verify_initial_elim_offsets): Likewise.
25799 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25800 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25801 (arm_get_frame_offsets): Split up into this ...
25802 (arm_compute_frame_layout): ... and this function.
25804 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25806 * config/aarch64/constraints.md (Usa): New constraint.
25807 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25809 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25811 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25812 with_multilib_list after it has been checked.
25814 2017-05-08 Richard Biener <rguenther@suse.de>
25816 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25817 (bitmap_set_subtract_values): Likewise.
25819 2017-05-08 Richard Biener <rguenther@suse.de>
25821 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25822 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25824 (gimple_stmt_nonzero_warnv_p): Rename to ...
25825 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25827 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25828 (extract_range_basic): Adjust, do not disable propagation on
25829 strict overflow sensitive simplification.
25830 (vrp_visit_cond_stmt): Likewise.
25832 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25834 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25835 body size unconditionally.
25837 2017-05-07 Jeff Law <law@redhat.com>
25840 2017-05-06 Jeff Law <law@redhat.com>
25841 PR tree-optimization/78496
25842 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25845 PR tree-optimization/78496
25846 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25847 (simplify_stmt_using_ranges): Call it.
25848 (vrp_dom_walker::before_dom_children): Extract equivalences
25849 from an ASSERT_EXPR with an equality comparison against a
25852 2017-05-06 Jeff Law <law@redhat.com>
25854 PR tree-optimization/78496
25855 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25858 PR tree-optimization/78496
25859 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25860 (simplify_stmt_using_ranges): Call it.
25861 (vrp_dom_walker::before_dom_children): Extract equivalences
25862 from an ASSERT_EXPR with an equality comparison against a
25865 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25867 * lra-constraints.c (lra_copy_reg_equiv): New function.
25868 (split_reg): Use it to copy equivalence information from the
25869 original register to the spill register.
25871 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25873 PR rtl-optimization/75964
25874 * simplify-rtx.c (simplify_const_relational_operation): Remove
25875 invalid handling of comparisons of integer ABS.
25877 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25879 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25880 initialize to zero.
25881 (init_regs): Remove declaration.
25882 (function_arg_advance_32): Initialize error_p as boolean variable.
25884 2017-05-05 Nathan Sidwell <nathan@acm.org>
25886 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25887 lines. Use for (;;).
25889 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25891 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25892 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25893 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25894 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25895 VF=2 that require versioning.
25897 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25899 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25902 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25904 * diagnostic.h (diagnostic_override_option_index): Convert from
25905 macro to inline function.
25907 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25909 * diagnostic.c (last_module_changed_p): New function.
25910 (set_last_module): New function.
25911 (diagnostic_report_current_module): Convert macro usage to
25912 the above functions.
25913 * diagnostic.h (diagnostic_context::last_module): Strengthen
25914 from const line_map * to const line_map_ordinary *.
25915 (diagnostic_last_module_changed): Delete macro.
25916 (diagnostic_set_last_module): Delete macro.
25918 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25920 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25921 with diagnostic_report_diagnostic.
25922 (diagnostic_n_impl_richloc): Likewise.
25923 * diagnostic.h (report_diagnostic): Delete macro.
25924 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25925 with diagnostic_report_diagnostic.
25926 * substring-locations.c (format_warning_va): Likewise.
25928 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25930 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25931 save/restor of format_spec. Move option-printing code to...
25932 (print_option_information): ...this new function, and
25933 reimplement by simply printing to the pretty_printer,
25934 rather than appending to the format string.
25936 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25938 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25939 handling logic into...
25940 (update_effective_level_from_pragmas): ...this new function.
25942 2017-05-04 Andrew Waterman <andrew@sifive.com>
25944 * config/riscv/riscv.opt (mstrict-align): New option.
25945 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25946 (SLOW_UNALIGNED_ACCESS): Define.
25947 (riscv_slow_unaligned_access): Declare.
25948 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25950 (riscv_slow_unaligned_access): New variable.
25951 (rocket_tune_info): Set slow_unaligned_access to true.
25952 (optimize_size_tune_info): Set slow_unaligned_access to false.
25953 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25954 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25955 (riscv_option_override): Set riscv_slow_unaligned_access.
25956 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25958 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25960 * config/riscv/riscv.md: Unify indentation.
25962 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25967 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25969 (extendsi<mode>2): Add support for doing sign extension via
25970 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25971 don't have ISA 3.0 instructions.
25972 (extendsi<mode>2 splitter): Likewise.
25973 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25974 generate the normal insns since SImode can now go in vector
25975 registers. Disallow the special UNSPECs needed for previous
25976 machines to hide SImode being used. Add new insns
25977 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25978 (fix_trunc<mode>si2_stfiwx): Likewise.
25979 (fix_trunc<mode>si2_internal): Likewise.
25980 (fixuns_trunc<mode>si2): Likewise.
25981 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25982 (fctiw<u>z_<mode>_smallint): Likewise.
25983 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25984 of floating point to 32-bit integer from doing a direct move to
25985 the GPR registers to do a store.
25986 (fctiwz_<mode>): Break long line.
25988 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25990 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25991 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25992 (addr_list, addr_offset_valid_p): New.
25993 (split_address_groups): Check offset validity with above function.
25994 (gt-tree-ssa-loop-ivopts.h): Include header file.
25996 2017-05-05 Nathan Sidwell <nathan@acm.org>
25998 * config.gcc (arm*-*-*): Add missing 'fi'.
26000 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26002 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26003 included in -fopt-info does not matter.
26004 * doc/optinfo.texi (-fopt-info): Fix description of default
26005 behavour. Explicitly say order of options included in -fopt-info
26008 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26010 * config.gcc: Allow combinations of aprofile and rmprofile values for
26011 --with-multilib-list.
26012 * config/arm/t-multilib: New file.
26013 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26014 variables. Remove setting of ISA and floating-point ABI in
26015 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26016 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26017 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26018 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26020 * config/arm/t-rmprofile: Likewise except for the matches changes.
26021 * doc/install.texi (--with-multilib-list): Document the combination of
26022 aprofile and rmprofile values and warn about pitfalls in doing that.
26024 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26026 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26027 (movdi_aarch64): Likewise.
26029 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26031 PR tree-optimization/80632
26032 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26034 (build_arrays): Initialize it for virtual phis.
26035 (fix_phi_nodes): Use it for virtual phis.
26037 PR tree-optimization/80558
26038 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26039 [x, y] op z into [x op, y op z] for op & or | if conditions
26042 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26043 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26046 * config/arm/arm.md (use_literal_pool): Remove.
26047 (64-bit immediate split): No longer takes cost into consideration
26048 if arm_disable_literal_pool is enabled.
26049 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26050 used when arm_disable_literal_pool is enabled.
26051 (arm_max_const_double_inline_cost): Remove use of
26052 arm_disable_literal_pool.
26053 (push_minipool_fix): Add assert.
26054 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26055 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26056 (no_literal_pool_sf_immediate): New.
26058 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26060 PR tree-optimization/80613
26061 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26062 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26064 2017-05-05 Richard Biener <rguenther@suse.de>
26066 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26068 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26070 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26071 of this flag from insn conditions due to removal from r247495.
26073 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26075 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26077 (arm_early_store_addr_dep_ptr): Likewise.
26078 * config/arm/aarch-common-protos.h
26079 (arm_early_load_addr_dep_ptr): Add prototype.
26080 (arm_early_store_addr_dep_ptr): Likewise.
26081 * config/arm/cortex-a53.md: Add new bypasses.
26083 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26085 * tree.c (next_type_uid): Change type to unsigned.
26086 (type_hash_canon): Decrement back next_type_uid if
26087 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26088 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26091 2017-05-04 Martin Sebor <msebor@redhat.com>
26093 * builtins.c: Fix a trivial typo in a comment.
26095 PR middle-end/79234
26096 * builtins.c (check_sizes): Adjust to handle reading past the end.
26097 Avoid printing excessive upper bound of ranges. Use %E to print
26098 tree nodes instead of converting them to %wu.
26099 (expand_builtin_memchr): New function.
26100 (compute_dest_size): Rename...
26101 (compute_objsize): ...to this.
26102 (expand_builtin_memcpy): Adjust.
26103 (expand_builtin_mempcpy): Adjust.
26104 (expand_builtin_strcat): Adjust.
26105 (expand_builtin_strcpy): Adjust.
26106 (check_strncat_sizes): Adjust.
26107 (expand_builtin_strncat): Adjust.
26108 (expand_builtin_strncpy): Adjust and simplify.
26109 (expand_builtin_memset): Adjust.
26110 (expand_builtin_bzero): Adjust.
26111 (expand_builtin_memcmp): Adjust.
26112 (expand_builtin): Handle memcmp.
26113 (maybe_emit_chk_warning): Check strncat just once.
26115 2017-05-04 Martin Sebor <msebor@redhat.com>
26117 PR preprocessor/79214
26118 PR middle-end/79222
26119 PR middle-end/79223
26120 * builtins.c (check_sizes): Add inlining context and issue
26121 warnings even when -Wno-system-headers is set.
26122 (check_strncat_sizes): Same.
26123 (expand_builtin_strncat): Same.
26124 (expand_builtin_memmove): New function.
26125 (expand_builtin_stpncpy): Same.
26126 (expand_builtin): Handle memmove and stpncpy.
26128 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26130 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26131 which is not used any more.
26133 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26135 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26137 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26139 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26140 (cortexa53_tunings): Likewise.
26141 (cortexa57_tunings): Likewise.
26142 (cortexa72_tunings): Likewise.
26143 (cortexa73_tunings): Likewise.
26145 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26147 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26148 Set loop alignment to 8.
26150 2017-05-04 Martin Sebor <msebor@redhat.com>
26152 PR translation/80280
26153 * builtins.c (expand_builtin_object_size): Add missing quoting to
26154 %D and like directives.
26155 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26156 (hsa_type_for_tree_type): Same.
26157 (verify_function_arguments): Same.
26158 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26159 * varasm.c (get_section): Same.
26162 2017-05-04 Martin Sebor <msebor@redhat.com>
26164 PR translation/80280
26165 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26167 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26169 * config/aarch64/aarch64.c (generic_addrcost_table):
26170 Change HI/TI mode setting.
26172 2017-05-04 Martin Jambor <mjambor@suse.cz>
26174 PR tree-optimization/80622
26175 * tree-sra.c (comes_initialized_p): New function.
26176 (build_accesses_from_assign): Only set write lazily when
26177 comes_initialized_p is false.
26178 (analyze_access_subtree): Use comes_initialized_p.
26179 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26180 instead of testing for PARM_DECL.
26182 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26184 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26185 constraint on operand 0 to allow more general addressing modes.
26186 Adjust output template.
26187 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26189 * config/aarch64/aarch64-protos.h
26190 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26191 * config/aarch64/constraints.md (Dp): New address constraint.
26192 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26195 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26197 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26198 update use of estimate_ipcp_clone_size_and_time.
26199 (estimate_local_effects): Update use of
26200 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26201 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26202 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26203 Return nonspecialized time.
26205 2017-05-04 Richard Biener <rguenther@suse.de>
26207 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26208 for the last VUSE which def dominates the PHI. Directly call
26210 (get_continuation_for_phi_1): Remove.
26212 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26214 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26215 to explain the use of truncating division. Cap the number of
26216 iterations to the maximum given by nb_iterations_upper_bound,
26219 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26221 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26222 * configure: Regenerate.
26223 * config.in: Regenerate.
26224 * config/i386/driver-mingw32.c: new file.
26225 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26226 * config.host: Link driver-mingw32.o on MinGW host.
26227 * doc/install.texi: Document new --enable-mingw-wildcard configure
26230 2017-05-04 Marek Polacek <polacek@redhat.com>
26232 PR tree-optimization/80612
26233 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26235 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26236 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26238 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26239 (movt splitter): Likewise.
26240 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26241 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26242 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26243 block for Thumb-1 with MOVT.
26244 (thumb2_legitimate_address_p): Move code block ...
26245 (can_avoid_literal_pool_for_label_p): ... into this new function.
26246 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26248 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26249 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26250 "M-profile targets with the MOVT instruction".
26252 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26254 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26255 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26256 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26258 2017-05-04 Martin Liska <mliska@suse.cz>
26260 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26261 variable cond_code.
26263 2017-05-04 Richard Biener <rguenther@suse.de>
26265 * tree.c (array_at_struct_end_p): Handle arrays at struct
26266 end with flexarrays more conservatively. Refactor and treat
26267 arrays of arrays or aggregates more strict. Fix
26268 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26269 * tree.c (array_at_struct_end_p): Adjust prototype.
26270 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26271 * gimple-fold.c (get_range_strlen): Likewise.
26272 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26274 2017-05-04 Richard Biener <rguenther@suse.de>
26276 PR tree-optimization/31130
26277 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26279 (supports_overflow_infinity): Likewise.
26280 (is_negative_overflow_infinity): Likewise.
26281 (is_positive_overflow_infinity): Likewise.
26282 (is_overflow_infinity): Likewise.
26283 (stmt_overflow_infinity): Likewise.
26284 (overflow_infinity_range_p): Likewise.
26285 (usable_range_p): Remove as always returning true.
26286 (make_overflow_infinity): Remove.
26287 (negative_overflow_infinity): Likewise.
26288 (positive_overflow_infinity): Likewise.
26289 (avoid_overflow_infinity): Likewise.
26290 (set_value_range): Adjust accordingly.
26291 (set_value_range_to_nonnegative): Likewise, remove now unused
26292 overflow_infinity arg.
26293 (vrp_operand_equal_p): Adjust.
26294 (update_value_range): Likewise.
26295 (range_int_cst_singleton_p): Likewise.
26296 (operand_less_p): Likewise.
26297 (compare_values_warnv): Likewise.
26298 (extract_range_for_var_from_comparison_expr): Likewise.
26299 (vrp_int_const_binop): Likewise.
26300 (zero_nonzero_bits_from_vr): Likewise.
26301 (extract_range_from_multiplicative_op_1): Likewise.
26302 (extract_range_from_binary_expr_1): Likewise.
26303 (extract_range_from_unary_expr): Likewise.
26304 (extract_range_from_comparison): Likewise.
26305 (extract_range_basic): Likewise.
26306 (adjust_range_with_scev): Likewise.
26307 (compare_ranges): Likewise.
26308 (compare_range_with_value): Likewise.
26309 (dump_value_range): Likewise.
26310 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26312 (simplify_cond_using_ranges): Adjust.
26314 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26316 * brig-builtins.def: Added a builtin for class_f64.
26317 * builtin-types.def: Added a builtin type needed by class_f64.
26319 2017-05-03 Jason Merrill <jason@redhat.com>
26321 * timevar.def: Add TV_CONSTEXPR.
26323 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26325 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26327 2017-05-03 Martin Jambor <mjambor@suse.cz>
26329 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26330 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26331 * ipa-cp.c (ipcp_propagate_stage): Do not call
26332 ipa_update_after_lto_read.
26333 * ipa-inline.c (ipa_inline): Likewise.
26335 2017-05-03 Martin Jambor <mjambor@suse.cz>
26337 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26338 tag. Added a default constructor and a destructor.
26339 (ipa_edge_args_sum_t): New class;
26340 (ipa_edge_args_sum): Declare.
26341 (ipa_edge_args_vector): Remove declaration.
26342 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26343 (ipa_free_edge_args_substructures): Remove declaration.
26344 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26345 (ipa_edge_args_info_available_for_edge_p): Likewise.
26346 * ipa-prop.c (ipa_edge_args_vector): Removed.
26347 (edge_removal_hook_holder): Likewise.
26348 (edge_duplication_hook_holder): Likewise.
26349 (ipa_edge_args_sum): New variable.
26350 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26351 ipa_edge_args_vector.
26352 (ipa_free_edge_args_substructures): Likewise.
26353 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26354 ipa_edge_args_vector.
26355 (ipa_edge_removal_hook): Turned into method
26356 ipa_edge_args_sum_t::remove.
26357 (ipa_edge_duplication_hook): Turned into method
26358 ipa_edge_args_sum_t::duplicate.
26359 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26360 registering edge hooks.
26361 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26362 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26363 ipa_edge_args_sum instead of ipa_edge_args_vector.
26364 * ipa-profile.c (ipa_profile): Likewise.
26366 2017-05-03 Martin Jambor <mjambor@suse.cz>
26368 * symbol-summary.h (function_summary): New method exists.
26369 (function_summary::symtab_removal): Deallocate through release.
26370 (call_summary): New class.
26371 (gt_ggc_mx): New overload.
26372 (gt_pch_nx): Likewise.
26373 (gt_pch_nx): Likewise.
26375 2017-05-03 Jeff Law <law@redhat.com>
26377 PR tree-optimization/78496
26378 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26379 from simplify_cond_using_ranges. Split off code to walk
26380 backwards through casts into ...
26381 (simplify_cond_using_ranges_2): New function.
26382 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26383 (execute_vrp): After identifying jump threads, call
26384 simplify_cond_using_ranges_2.
26386 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26389 * ipa-inline.h (inline_summary): Add ctor.
26390 (create_ggc): Do not use ggc_cleared_alloc.
26392 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26393 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26395 * gcc.c (handle_braces): Support escaping in switch matching
26397 * doc/invoke.texi (Spec Files): Document it.
26398 Remove superfluous @code markup in items.
26400 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26402 * diagnostic-show-locus.c (struct column_range): New struct.
26403 (get_affected_columns): New function.
26404 (get_printed_columns): New function.
26405 (struct correction): New struct.
26406 (correction::ensure_capacity): New function.
26407 (correction::ensure_terminated): New function.
26408 (struct line_corrections): New struct.
26409 (line_corrections::~line_corrections): New dtor.
26410 (line_corrections::add_hint): New function.
26411 (layout::print_trailing_fixits): Reimplement in terms of the new
26413 (selftest::test_overlapped_fixit_printing): New function.
26414 (selftest::diagnostic_show_locus_c_tests): Call it.
26416 2017-05-03 Nathan Sidwell <nathan@acm.org>
26418 Canonicalize canonical type hashing
26419 * tree.h (type_hash_canon_hash): Declare.
26420 * tree.c (type_hash_list, attribute_hash_list): Move into
26421 type_hash_canon_hash.
26422 (build_type_attribute_qual_variant): Break out hash code calc into
26423 type_hash_canon_hash.
26424 (type_hash_canon_hash): New. Generic type hash computation.
26425 (build_range_type_1, build_array_type_1, build_function_type,
26426 build_method_type_directly, build_offset_type, build_complex_type,
26427 make_vector_type): Call it.
26429 2017-05-03 Richard Biener <rguenther@suse.de>
26431 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26432 When all DRs have unknown misaligned do not always peel
26433 when there is a store but apply the same costing model as if
26434 there were only loads.
26436 2017-05-03 Richard Biener <rguenther@suse.de>
26439 PR tree-optimization/80492
26440 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26441 compare_base_decls returning dont-know properly.
26443 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26445 * config/arm/iterators.md (CCSI): New mode iterator.
26446 (arch): New mode attribute.
26447 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26448 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26449 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26450 code iterator for success result mode.
26451 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26452 the corresponding new insn generators.
26454 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26457 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26458 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26460 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26462 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26463 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26464 (DDR_A): Wrap DDR argument in brackets.
26465 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26466 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26467 (DDR_REVERSED_P): Likewise.
26469 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26471 PR tree-optimization/79472
26472 * tree-switch-conversion.c (struct switch_conv_info): Add
26473 contiguous_range and default_case_nonstandard fields.
26474 (collect_switch_conv_info): Compute contiguous_range and
26475 default_case_nonstandard fields, don't clear final_bb if
26476 contiguous_range and only the default case doesn't have the required
26478 (check_all_empty_except_final): Set default_case_nonstandard instead
26479 of failing if contiguous_range and the default case doesn't have empty
26481 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26482 and only the default case doesn't have the required constants. Skip
26484 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26485 if default_case_nonstandard.
26486 (build_constructors): Build constant 1 just once. Assert that default
26487 values aren't inserted in between cases if contiguous_range. Skip
26489 (build_arrays): Skip virtual phis.
26490 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26491 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26492 Handle virtual phis.
26493 (gen_inbound_check): Handle default_case_nonstandard case.
26494 (process_switch): Adjust check_final_bb caller. Call
26495 gather_default_values with the first non-default case instead of
26496 default case if default_case_nonstandard.
26498 2017-05-02 Nathan Sidwell <nathan@acm.org>
26500 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26501 check. Fix formatting.
26503 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26505 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26506 errors when comparing specialized and unspecialized times.
26508 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26510 * diagnostic-show-locus.c
26511 (layout::should_print_annotation_line_p): Make private.
26512 (layout::print_annotation_line): Make private.
26513 (layout::annotation_line_showed_range_p): Make private.
26514 (layout::show_ruler): Make private.
26515 (layout::print_source_line): Make private. Pass in line and
26516 line_width, rather than calling location_get_source_line. Drop
26518 (layout::print_leading_fixits): New method.
26519 (layout::print_any_fixits): Rename to...
26520 (layout::print_trailing_fixits): ...this, and make private.
26521 Don't print newline fixits.
26522 (diagnostic_show_locus): Move logic for printing one row into...
26523 (layout::print_line): ...this new function. Move the
26524 location_get_source_line call and error-handling from
26525 print_source_line to here. Call print_leading_fixits, and rename
26526 print_any_fixits to print_trailing_fixits.
26527 (selftest::test_fixit_insert_containing_newline): Update now that
26528 newlines are partially supported.
26529 (selftest::test_fixit_insert_containing_newline_2): New test.
26530 (selftest::test_fixit_replace_containing_newline): Update comments.
26531 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26532 * edit-context.c (class added_line): New class.
26533 (class edited_line): Describe newline handling in comment.
26534 (edited_line::actually_edited_p): New method.
26535 (edited_line::print_content): Delete redundant decl.
26536 (edited_line::m_predecessors): New field.
26537 (edited_file::print_content): Call edited_line::print_content.
26538 (edited_file::print_diff): Update to support newlines.
26539 (edited_file::print_diff_hunk): Likewise.
26540 (edited_file::print_run_of_changed_lines): New function.
26541 (edited_file::print_diff_line): Convert to...
26542 (print_diff_line): ...this.
26543 (edited_file::get_effective_line_count): New function.
26544 (edited_line::edited_line): Initialize new field m_predecessors.
26545 (edited_line::~edited_line): Clean up m_predecessors.
26546 (edited_line::apply_fixit): Handle newlines.
26547 (edited_line::get_effective_line_count): New function.
26548 (edited_line::print_content): New function.
26549 (edited_line::print_diff_lines): New function.
26550 (selftest::test_applying_fixits_insert_containing_newline): New
26552 (selftest::test_applying_fixits_replace_containing_newline): New
26554 (selftest::insert_line): New function.
26555 (selftest::test_applying_fixits_multiple_lines): Add example of
26557 (selftest::edit_context_c_tests): Call the new tests.
26559 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26561 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26562 parameter cand. Update dump information.
26563 (get_computation_cost): Update uses.
26565 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26567 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26568 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26569 (get_computation_at, rewrite_use_address): Update use of
26570 get_computation_aff.
26572 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26574 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26575 (get_computation): Delete.
26576 (get_computation_cost): Implement like get_computation_cost_at.
26577 Use get_computation_at.
26578 (get_computation_cost_at): Delete.
26579 (rewrite_use_nonlinear_expr): Use get_computation_at.
26580 (rewrite_use_compare, remove_unused_ivs): Ditto.
26582 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26584 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26586 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26588 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26589 (ivopts_global_cost_for_size): Rename parameter and update uses.
26590 (iv_ca_recount_cost): Update uses.
26591 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26592 candidates seperately in n_invs and n_cands.
26593 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26595 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26597 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26598 (find_inv_vars_cb): New.
26599 (find_depends): Renamed to ...
26600 (find_inv_vars): ... this.
26601 (add_candidate_1, force_var_cost): Call find_inv_vars.
26602 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26604 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26606 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26607 inv_vars. Add inv_exprs.
26608 (struct iv_cand): Rename depends_on to inv_vars.
26609 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26610 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26611 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26612 (dump_cand): Dump inv_vars.
26613 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26614 (record_invariant, find_depends, add_candidate_1): Ditto.
26615 (set_group_iv_cost, force_var_cost): Ditto.
26616 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26617 (get_computation_cost_at, get_computation_cost): Ditto.
26618 (determine_group_iv_cost_generic): Ditto.
26619 (determine_group_iv_cost_address): Ditto.
26620 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26621 (determine_group_iv_costs): Ditto.
26622 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26623 (iv_ca_set_remove_invariants): Renamed to ...
26624 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26625 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26626 (iv_ca_set_add_invariants): Renamed to ...
26627 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26628 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26629 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26630 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26631 (create_new_ivs): Remove useless dump.
26633 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26635 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26637 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26638 (iv_ca_set_no_cp, create_new_iv): Ditto.
26640 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26642 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26644 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26646 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26647 function tree_check2.
26649 2017-05-02 Martin Liska <mliska@suse.cz>
26651 * doc/gcov.texi: Add missing preposition.
26652 * gcov.c (function_info::function_info): Properly fill up
26653 all member variables.
26655 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26657 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26659 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26661 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26663 2017-05-02 Martin Liska <mliska@suse.cz>
26666 * lto-streamer-in.c (lto_read_tree_1): Remove
26667 LTO_STREAMER_DEBUG.
26668 * lto-streamer.c (struct tree_hash_entry): Likewise.
26669 (struct tree_entry_hasher): Likewise.
26670 (tree_entry_hasher::hash): Likewise.
26671 (tree_entry_hasher::equal): Likewise.
26672 (lto_streamer_init): Likewise.
26673 (lto_orig_address_map): Likewise.
26674 (lto_orig_address_get): Likewise.
26675 (lto_orig_address_remove): Likewise.
26676 * lto-streamer.h: Likewise.
26677 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26678 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26680 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26682 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26683 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26684 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26685 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26686 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26687 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26688 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26689 (mm_maskz_sub_ss): New intrinsics.
26690 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26691 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26692 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26693 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26694 (__builtin_ia32_subss_mask_round): New builtins.
26695 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26696 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26697 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26699 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26700 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26702 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26705 2017-05-02 Martin Jambor <mjambor@suse.cz>
26707 PR tree-optimization/78687
26708 * tree-sra.c (access): New field parent.
26709 (process_subtree_disqualification): New function.
26710 (disqualify_candidate): Call it.
26711 (build_accesses_from_assign): Reset write flag if creating an
26713 (build_access_subtree): Fill in parent field and also prpagate
26714 down grp_write flag.
26715 (create_artificial_child_access): New parameter set_grp_write, set
26716 grp_write to its value.
26717 (propagate_subaccesses_across_link): Also propagate grp_write flag
26719 (propagate_all_subaccesses): Push the closest parent back to work
26720 queue if add_access_to_work_queue returned true.
26722 2017-05-02 Richard Biener <rguenther@suse.de>
26724 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26725 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26726 -fstrict-overflow documentation.
26727 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26728 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26729 flag_strict_overflow.
26730 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26731 * lto-opts.c (lto_write_options): Do not stream it.
26732 * lto-wrapper.c (merge_and_complain): Do not handle it.
26733 * opts.c (default_options_table): Do not set -fstrict-overflow.
26734 (finish_options): Likewise do not clear it when sanitizing.
26735 * simplify-rtx.c (simplify_const_relational_operation): Do not
26736 test flag_strict_overflow.
26738 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26740 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26741 using enabled attribute.
26742 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26743 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26744 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26745 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26746 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26747 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26748 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26749 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26750 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26751 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26753 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26755 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26757 2017-05-02 Richard Biener <rguenther@suse.de>
26759 PR tree-optimization/80591
26761 2017-04-10 Richard Biener <rguenther@suse.de>
26763 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26766 2017-05-02 Richard Biener <rguenther@suse.de>
26768 PR tree-optimization/80549
26769 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26770 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26771 headers if they do not have a preheader.
26773 2017-05-02 Martin Liska <mliska@suse.cz>
26776 * common.opt: Fix typo.
26777 * doc/invoke.texi: Likewise.
26779 2017-05-01 Jan Beulich <jbeulich@suse.com>
26781 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26782 swapping, add (x,x,m,x,n) alternative.
26784 2017-05-01 Nathan Sidwell <nathan@acm.org>
26786 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26787 unnecessary unadjusted_alignment check.
26789 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26792 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26794 * gimplify.c (gimplify_cilk_detach): New function.
26795 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26796 * tree-core.h: Document EXPR_CILK_SPAWN.
26797 * tree.h (EXPR_CILK_SPAWN): Define.
26799 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26801 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26802 to use new fixit_hint representation, using the "replace" logic.
26803 (get_line_span_for_fixit_hint): Likewise.
26804 (layout::print_any_fixits): Likewise.
26805 (selftest::test_one_liner_many_fixits): Rename to...
26806 (selftest::test_one_liner_many_fixits_1): ...this, and update
26807 comment and expected output to reflect that the multiple fix-it
26808 hints are now consolidated into one insertion.
26809 (selftest::test_one_liner_many_fixits_2): New test.
26810 (selftest::test_diagnostic_show_locus_one_liner): Update for
26812 (selftest::test_fixit_consolidation): Update for fix-it API
26814 * diagnostic.c (print_parseable_fixits): Likewise.
26815 * edit-context.c (edited_line::m_line_events): Convert from
26816 auto_vec <line_event *> to auto_vec <line_event>.
26817 (class line_event): Convert from abstract base class to a concrete
26818 class, taking over the role of replace_event.
26819 (class insert_event): Delete.
26820 (class replace_event): Rename to class line_event. Convert to
26822 (edit_context::add_fixits): Reimplement.
26823 (edit_context::apply_insert): Delete.
26824 (edit_context::apply_replace): Rename to...
26825 (edit_context::apply_fixit): ...this. Convert to half-open range.
26826 (edited_file::apply_insert): Delete.
26827 (edited_file::apply_replace): Rename to...
26828 (edited_file::apply_fixit): ...this.
26829 (edited_line::~edited_line): Drop deletion of events.
26830 (edited_line::apply_insert): Delete.
26831 (edited_line::apply_replace): Rename to...
26832 (edited_line::apply_fixit): ...this. Convert to half-open range.
26833 Update for change to type of m_line_events.
26834 * edit-context.h (edit_context::apply_insert): Delete.
26835 (edit_context::apply_replace): Rename to...
26836 (edit_context::apply_fixit): ...this.
26838 2017-05-01 Martin Sebor <msebor@redhat.com>
26840 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26843 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26846 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26847 __get_cpuid_max returns 0.
26848 (__get_cpuid_count): Ditto.
26850 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26852 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26853 replacement expression is another instance of one of its arguments.
26855 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26858 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26859 check for stack push/pop autoinc.
26860 * config/i386/i386.c (ix86_agi_dependent): Return false
26861 if the only reason why modified_in_p returned true is that
26862 addr is SP based and set_insn is a push or pop.
26864 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26866 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26869 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26872 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26873 (account_size_time): Use two predicates - exec_pred and
26875 (evaluate_conditions_for_known_args): Compute both clause and
26877 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26878 (inline_summary_t::duplicate): Update.
26879 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26881 (compute_inline_parameters): Likewise.
26882 (estimate_edge_size_and_time): Update caluclation of time.
26883 (estimate_node_size_and_time): Compute both time and nonspecialized
26885 (estimate_ipcp_clone_size_and_time): Update.
26886 (inline_merge_summary): Update.
26887 (do_estimate_edge_time): Update.
26888 (do_estimate_edge_size): Update.
26889 (do_estimate_edge_hints): Update.
26890 (inline_read_section, inline_write_summary): Stream both new predicates.
26891 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26893 (compute_inlined_call_time): Cleanup.
26894 (big_speedup_p): Update.
26895 (edge_badness): Update.
26896 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26897 (size_time_entry): Replace predicate by exec_predicate and
26898 nonconst_predicate.
26899 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26900 (estimate_edge_time): Return also nonspec_time.
26901 (reset_edge_growth_cache): Update.
26903 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26905 PR rtl-optimization/80491
26906 * ifcvt.c (noce_process_if_block): When looking for x setter
26907 with missing else_bb, don't check only the insn right before
26908 cond_earliest, but look for the last insn that x is modified in
26909 within the same bb.
26911 PR rtl-optimization/80491
26912 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26914 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26916 PR tree-optimization/80487
26917 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26919 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26921 PR tree-optimization/79697
26922 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26923 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26924 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26926 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26927 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26929 2017-04-28 Martin Sebor <msebor@redhat.com>
26931 PR tree-optimization/80523
26932 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26933 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26935 (maybe_warn, format_directive, parse_directive): Use new functions.
26936 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26938 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26940 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26942 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26944 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26945 target_header_dir): Set correctly.
26946 * configure: Regenerated.
26947 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26948 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26949 instead of SYSTEM_HEADER_DIR.
26951 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26953 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26954 (estimate_local_effects): Likewise.
26955 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26956 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26957 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26958 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26959 do_estimate_edge_time, estimate_edge_time): Likewise.
26960 * ipa-inline-analysis.c (estimate_node_size_and_time,
26961 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26962 (estimate_time_after_inlining): Remove.
26964 2017-04-28 Martin Liska <mliska@suse.cz>
26966 * doc/gcov.texi: Enhance documentation of gcov.
26968 2017-04-28 Martin Liska <mliska@suse.cz>
26970 * doc/gcov.texi: Sort options in alphabetic order.
26971 * doc/gcov-dump.texi: Likewise.
26972 * doc/gcov-tool.texi: Likewise.
26973 * gcov.c (print_usage): Likewise.
26974 * gcov-dump.c (print_usage): Likewise.
26975 * gcov-tool.c (print_merge_usage_message): Likewise.
26976 (print_rewrite_usage_message): Likewise.
26977 (print_overlap_usage_message): Likewise.
26979 2017-04-28 Martin Liska <mliska@suse.cz>
26981 PR gcov-profile/53915
26982 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26984 2017-04-28 Martin Liska <mliska@suse.cz>
26986 PR gcov-profile/79891
26987 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26988 is marked by compiler as living on a line.
26989 (get_cycles_count): Remove usage of the union.
26990 (output_intermediate_file): Likewise.
26991 (find_source): Fix GNU coding style.
26992 (accumulate_line_counts): Remove old non-all block mode.
26993 (output_lines): Remove usage of the union.
26994 * profile.c (output_location): Include all BBs, even if
26995 belonging to a same line (and file) as a previous BB.
26997 2017-04-28 Martin Liska <mliska@suse.cz>
26999 * gcov.c (process_args): Handle new argument 'w'.
27000 (read_graph_file): Assign ID to BBs.
27001 (output_branch_count): Display BB # if verbose flag is set.
27002 (output_lines): Likewise for arcs.
27003 (print_usage): Add '--verbose' option help.
27004 * doc/gcov.texi: Document --verbose (-w) option.
27006 2017-04-28 Martin Liska <mliska@suse.cz>
27008 * gcov.c (struct block_location_info): New struct.
27009 (process_file): Fill up the new structure.
27010 (read_graph_file): Replace usage of encoding by the newly added
27012 (add_line_counts): Likewise.
27013 (accumulate_line_counts): Remove usage of the union.
27014 (function_info::function_info): New function.
27015 (function_info::~function_info): Likewise.
27016 (process_file): Call delete instead of release_function.
27017 (release_function): Release the function.
27018 (release_structures): Call delete instead of release_function.
27019 (solve_flow_graph): Replace usage of num_blocks.
27020 (find_exception_blocks): Likewise.
27021 (output_lines): Fix GNU coding style.
27023 2017-04-28 Martin Liska <mliska@suse.cz>
27026 * coverage.c (coverage_remove_note_file): New function.
27027 * coverage.h: Declare the function.
27028 * toplev.c (finalize): Clean if an error has been seen.
27030 2017-04-28 Martin Liska <mliska@suse.cz>
27032 PR gcov-profile/80031
27033 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27034 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27035 * gcov.c (read_graph_file): Read just number of blocks.
27036 * profile.c (branch_prob): Do not stream 0 flags per a basic
27039 2017-04-28 Martin Liska <mliska@suse.cz>
27041 * gcov-dump.c (tag_*): Add new argument to declarations.
27042 (dump_gcov_file): Likewise.
27043 (tag_blocks): Add and use new argument depth.
27044 (tag_arcs): Likewise.
27045 (tag_lines): Likewise.
27046 (tag_counters): Likewise.
27047 (tag_summary): Likewise.
27048 (dump_working_sets): Use depth to do a proper indentation.
27050 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27053 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27054 * symtab.c (symtab_node::debug_symtab): Move definition here.
27056 2017-04-28 Richard Biener <rguenther@suse.de>
27058 * lto-streamer.h (LTO_major_version): Bump to 7.
27060 2017-04-28 Richard Biener <rguenther@suse.de>
27062 * tree-vrp.c (assert_info): New struct.
27063 (add_assert_info): New helper.
27064 (register_edge_assert_for_2): Refactor to add asserts to a vector
27066 (register_edge_assert_for_1): Likewise.
27067 (register_edge_assert_for): Likewise.
27068 (finish_register_edge_assert_for): New helper actually registering
27069 asserts where live on edge.
27070 (find_conditional_asserts): Adjust.
27071 (find_switch_asserts): Likewise.
27072 (evrp_dom_walker::try_find_new_range): Generalize.
27073 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27075 2017-04-27 Marek Polacek <polacek@redhat.com>
27078 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27079 arg10 and arg11 to itype.
27081 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27083 * doc/extend.texi (Object Size Checking): Improve grammar.
27085 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27088 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27089 that the logic for permitting reciprocal estimates matches that
27092 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27095 * tree.c (type_cache_hasher::equal): Only compare
27096 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27097 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27098 non-aggregate element types.
27099 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27100 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27102 2017-04-27 Richard Biener <rguenther@suse.de>
27104 PR middle-end/80533
27105 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27106 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27107 keeping a reference to a trailing array.
27109 2017-04-27 Richard Biener <rguenther@suse.de>
27111 PR middle-end/80539
27112 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27113 being in loop-closed SSA form conservatively.
27114 (chrec_fold_multiply_poly_poly): Likewise.
27116 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27118 PR middle-end/79665
27119 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27120 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27122 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27125 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27126 (aarch64_function_arg_alignment): Return unsigned int again, but still
27127 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27128 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27129 Don't emit -Wpsabi note.
27130 (aarch64_function_arg_boundary): Likewise.
27131 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27134 2017-04-26 Nathan Sidwell <nathan@acm.org>
27136 * tree.h (crc32_unsigned_n): Declare.
27137 (crc32_unsigned, crc32_unsigned): Make inline.
27138 * tree.c (crc32_unsigned_bits): Replace with ...
27139 (crc32_unsigned_n): ... this.
27140 (crc32_unsigned, crc32_byte): Remove.
27141 (crc32_string): Remove unnecessary braces.
27143 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27145 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27146 * ipa-inline-analysis.c (MAX_TIME): Remove.
27147 (account_size_time): Use sreal for time.
27148 (dump_inline_summary): Update.
27149 (estimate_function_body_sizes): Update.
27150 (estimate_edge_size_and_time): Update.
27151 (estimate_calls_size_and_time): Update.
27152 (estimate_node_size_and_time): Update.
27153 (inline_merge_summary): Update.
27154 (inline_update_overall_summary): Update.
27155 (estimate_time_after_inlining): Update.
27156 (inline_read_section): Update.
27157 (inline_write_summary): Update.
27158 * ipa-inline.c (compute_uninlined_call_time): Update.
27159 (compute_inlined_call_time): Update.
27160 (recursive_inlining): Update.
27161 (inline_small_functions): Update.
27162 (dump_overall_stats): Update.
27163 * ipa-inline.h: Include sreal.h.
27164 (size_time_entry): Turn time to sreal.
27165 (inline_summary): Turn self_time nad time to sreal.
27167 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27169 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27171 (sreal::stream_out, sreal::stream_in): New.
27172 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27174 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27176 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27179 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27182 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27184 (dimode_scalar_chain::compute_convert_gain): Ditto.
27185 (dimode_scalar_chain::make_vector_copies): Ditto.
27186 (dimode_scalar_chain::convert_reg): Ditto.
27187 (dimode_scalar_chain::convert_insn): Ditto.
27188 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27189 (VI248_AVX512BW_1): New mode iterator.
27190 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27191 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27194 2017-04-25 Martin Sebor <msebor@redhat.com>
27196 PR tree-optimization/80497
27197 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27198 constants are representable in HOST_WIDE_INT.
27199 (parse_directive): Ditto.
27201 2017-04-25 Martin Sebor <msebor@redhat.com>
27204 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27205 (new_zero_array): Adjust signature.
27206 (dom_info::dom_init): Used unsigned rather that size_t.
27207 (dom_info::dom_info): Same.
27209 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27210 Jakub Jelinek <jakub@redhat.com>
27213 * config/arm/arm.c: Include gimple.h.
27214 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27215 returns negative, increment ncrn only if it returned positive.
27216 (arm_needs_doubleword_align): Return int instead of bool,
27217 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27218 members, but if there is any such non-FIELD_DECL
27219 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27220 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27221 returns negative, increment nregs only if it returned positive.
27222 (arm_setup_incoming_varargs): Likewise.
27223 (arm_function_arg_boundary): Emit -Wpsabi note if
27224 arm_needs_doubleword_align returns negative, return
27225 DOUBLEWORD_ALIGNMENT only if it returned positive.
27227 2017-04-25 Marek Polacek <polacek@redhat.com>
27230 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27231 first argument to type.
27233 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27236 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27237 type checks to test for compatibility instead of equality.
27239 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27240 Jakub Jelinek <jakub@redhat.com>
27243 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27245 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27246 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27247 the alignment computation, but return their maximum in warn_alignment.
27248 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27249 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27251 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27252 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27255 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27257 * config/arc/simdext.md (dmpyh): Fix typo.
27259 2017-04-25 Richard Biener <rguenther@suse.de>
27261 PR tree-optimization/80492
27262 * alias.c (compare_base_decls): Handle registers with asm
27263 specification conservatively.
27264 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27265 compare_base_decls returning dont-know properly.
27267 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27269 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27270 (legitimate_offset_address_p): New function.
27271 (arc_legitimate_address_p): Use above function.
27273 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27275 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27277 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27279 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27280 ACCH registers whenever they are available.
27282 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27284 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27285 double regs fix when not used.
27287 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27289 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27291 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27292 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27294 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27296 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27297 when emitting short ADD instructions.
27299 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27301 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27303 (cmpsi_cc_c_insn): Likewise.
27304 (cbranchsi4_scratch): Compute proper instruction length using
27305 compact_hreg_operand.
27306 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27308 2017-04-25 Richard Biener <rguenther@suse.de>
27310 PR middle-end/80509
27311 * passes.c (pass_manager::pass_manager): Initialize
27312 m_name_to_pass_map.
27314 2017-04-25 Richard Biener <rguenther@suse.de>
27316 PR tree-optimization/79201
27317 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27319 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27322 * config/s390/vector.md: Split MEM->GPR vector moves for
27323 non-s_operand addresses.
27325 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27328 * config/s390/predicates.md (reload_const_wide_int_operand): New
27330 * config/s390/s390.md ("movti"): Remove d/P alternative.
27331 ("movti_bigconst"): New pattern definition.
27333 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27336 * s390-protos.h (s390_expand_cs_hqi): Removed.
27337 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27338 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27339 modes as well as CCZ1mode and CCZmode.
27340 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27341 signature of s390_emit_compare_and_swap.
27342 (s390_expand_cs_hqi): Likewise, make static.
27343 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27344 compare-and-swap, in some cases.
27345 (s390_expand_cs): Wrapper function.
27346 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27348 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27349 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27350 patterns for small and large integers. Forbid symref memory operands.
27351 Move expander to s390.c. Require cc register.
27352 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27353 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27354 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27355 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27356 symref memory operands. Remove CC mode and call s390_match_ccmode
27358 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27360 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27362 * config/s390/s390.md (define_peephole2): New peephole to help
27363 combining the load-and-test pattern with volatile memory.
27365 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27367 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27368 with CCZmode for TARGET_Z196.
27370 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27372 PR rtl-optimization/80501
27373 * combine.c (make_compound_operation_int): Set subreg_code to SET
27374 even for AND with mask of the sign bit of mode.
27376 PR rtl-optimization/80500
27377 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27378 sum's initial value.
27380 2017-04-25 Julian Brown <julian@codesourcery.com>
27381 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27383 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27385 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27387 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27389 2017-04-25 Julian Brown <julian@codesourcery.com>
27390 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27392 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27393 (thunderx2t99_sha): New Reservation.
27395 2017-04-25 Julian Brown <julian@codesourcery.com>
27396 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27398 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27399 type for 1-element load.
27401 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27403 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27405 2017-04-24 Martin Jambor <mjambor@suse.cz>
27407 PR tree-optimization/80293
27408 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27409 char arrays not totally scalarizable if it is false.
27410 (analyze_all_variable_accesses): Pass correct value in the new
27411 parameter. Add a statistics counter.
27413 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27415 PR middle-end/79931
27416 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27418 2017-04-24 Richard Biener <rguenther@suse.de>
27420 PR tree-optimization/80494
27421 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27422 out for complex types.
27424 2017-04-24 Richard Biener <rguenther@suse.de>
27426 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27427 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27428 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27429 (DFS): Adjust and never fail.
27430 (sccvn_dom_walker::fail): Remove.
27431 (sccvn_dom_walker::before_dom_children): Adjust.
27432 (run_scc_vn): Likewise and never fail.
27433 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27434 (pass_fre::execute): Likewise.
27436 2017-04-24 Richard Biener <rguenther@suse.de>
27438 PR tree-optimization/79725
27439 * tree-ssa-sink.c (statement_sink_location): Return whether
27440 failure reason was zero uses. Move that check later.
27441 (sink_code_in_bb): Deal with zero uses by removing the stmt
27444 2017-04-24 Richard Biener <rguenther@suse.de>
27447 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27448 pointer-based references.
27450 2017-04-24 Richard Biener <rguenther@suse.de>
27453 * pass_manager.h (pass_manager::operator new): Remove.
27454 (pass_manager::operator delete): Likewise.
27455 * passes.c (pass_manager::operator new): Remove.
27456 (pass_manager::operator delete): Likewise.
27457 (pass_manager::pass_manager): Zero individual pass members.
27459 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27462 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27463 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27464 Check "XEXP (src, 1)" operand here.
27465 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27466 Check "XEXP (src, 1)" operand here.
27467 (dimode_scalar_chain::make_vector_copies): Detect count register
27468 of a shift instruction. Zero extend count register from QImode
27469 to DImode to satisfy vector shift pattern count operand predicate.
27470 Substitute vector shift count operand with a DImode copy.
27471 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27474 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27476 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27477 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27478 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27479 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27480 (UNSPEC_NOREX_MEM): Remove definition.
27482 2017-04-21 Richard Biener <rguenther@suse.de>
27484 PR tree-optimization/79547
27485 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27486 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27487 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27488 without any constraints.
27490 2017-04-21 Richard Biener <rguenther@suse.de>
27492 PR tree-optimization/78847
27493 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27495 2017-04-21 Richard Biener <rguenther@suse.de>
27497 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27498 (build_distinct_type_copy): Likewise.
27499 (build_variant_type_copy): Likewise.
27500 * tree.c (build_qualified_type): Pass down mem-stat info.
27501 (build_distinct_type_copy): Likewise.
27502 (build_variant_type_copy): Likewise.
27504 2017-04-21 Richard Biener <rguenther@suse.de>
27506 PR tree-optimization/80237
27507 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27509 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27510 for a simplified result.
27512 2016-04-21 Richard Biener <rguenther@suse.de>
27514 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27515 sth as strict as a simple_iv but a chrec without symbols and an
27516 operand defined in the loop we are peeling (and not some subloop).
27517 (propagate_constants_for_unrolling): Propagate all constants.
27519 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27522 * config/i386/i386.c (print_reg): Remove assert for disalowed
27523 regno values, call output_operand_lossage instead.
27525 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27528 * config/i386/constraints.md (Yc): New register constraint.
27529 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27530 Use Yc constraint for alternative 2 of operand 0. Remove
27531 preferred_for_speed attribute.
27533 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27535 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27536 lastprivate clauses in SIMT case.
27538 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27540 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27542 2017-04-20 Richard Biener <rguenther@suse.de>
27544 PR tree-optimization/57796
27545 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27546 as N scalar stores.
27547 (vect_model_load_cost): Cost gathers as N scalar loads.
27549 2017-04-20 Richard Biener <rguenther@suse.de>
27551 * ggc-page.c (ggc_allocated_p): Rename to ...
27552 (safe_lookup_page_table_entry): ... this and return the lookup
27554 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27556 2017-04-20 Richard Biener <rguenther@suse.de>
27558 PR tree-optimization/80453
27559 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27560 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27561 from the conditions.
27562 (vn_phi_eq): Pass them down.
27563 (vn_phi_lookup): Record them.
27564 (vn_phi_insert): Likewise.
27566 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27568 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27569 uninitialized variable warning to avoid buffer overrun.
27571 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27574 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27575 is suppressed for '{ 0 }' in C.
27577 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27579 * BASE-VER: Set to 8.0.0.
27581 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27583 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27584 priority .init_array and .fini_array section with SECTION_NOTYPE
27587 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27589 PR middle-end/80423
27590 * tree.h (build_array_type): Add typeless_storage default argument.
27591 * tree.c (type_cache_hasher::equal): Also compare
27592 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27593 (build_array_type): Add typeless_storage argument, set
27594 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27596 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27597 (build_array_type): Likewise. Add typeless_storage argument.
27599 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27600 Jakub Jelinek <jakub@redhat.com>
27602 PR tree-optimization/80426
27603 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27604 operation on symbolic operands, also compute the overflow for the
27605 invariant part when the operation degenerates into a negation.
27607 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27610 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27611 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27614 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27616 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27619 * config/avr/avr.c (tree.h): Include it.
27620 (cgraph.h): Include it.
27621 (avr_encode_section_info): Don't warn for uninitialized progmem
27622 variable if it's just an alias.
27624 2017-04-19 Richard Biener <rguenther@suse.de>
27627 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27628 when needed by AutoPGO.
27630 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27633 * doc/lto.texi: Remove an extra 'that'.
27635 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27637 PR rtl-optimization/80429
27638 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27639 are only used in debug insns.
27641 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27642 Vladimir Makarov <vmakarov@redhat.com>
27644 * config/sparc/predicates.md (input_operand): Add comment. Return
27645 true for any memory operand when LRA is in progress.
27646 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27648 2017-04-18 Jeff Law <law@redhat.com>
27651 * mips.md ({return,simple_return}_internal): Do not overwrite
27654 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27656 PR tree-optimization/80443
27657 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27658 instead of adding 1, subtract -1 and similarly instead of subtracting
27661 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27663 PR rtl-optimization/80357
27664 * haifa-sched.c (tmp_bitmap): New variable.
27665 (model_recompute): Handle duplicate use records.
27666 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27667 (free_global_sched_pressure_data): Free it.
27669 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27672 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27673 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27674 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27675 instead of SYSTEM_HEADER_DIR.
27677 2017-04-18 Jeff Law <law@redhat.com>
27679 PR middle-end/80422
27680 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27681 predecessors after walking up the insn chain.
27683 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27686 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27687 sizetype type into debug info.
27689 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27692 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27693 unneeded test for TARGET_UPPER_REGS_SF.
27694 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27696 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27699 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27700 instead of gsi_after_labels.
27702 2017-04-18 Jeff Law <law@redhat.com>
27704 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27708 2017-04-13 Jeff Law <law@redhat.com>
27709 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27710 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27712 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27715 * config/avr/avr.c (intl.h): Include it.
27716 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27718 2017-04-18 Martin Liska <mliska@suse.cz>
27720 PR gcov-profile/78783
27721 * gcov-tool.c (gcov_output_files): Validate that destination
27722 file is either removed by the tool or by a user.
27724 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27725 Guy Benyei <guybe@mellanox.com>
27727 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27728 block, and do not negate it, the stored id is already negative.
27730 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27732 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27734 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27737 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27738 masks of options that should be turned off if the VSX vector
27739 options are turned off.
27740 (OTHER_P8_VECTOR_MASKS): Likewise.
27741 (OTHER_VSX_VECTOR_MASKS): Likewise.
27742 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27743 rs6000_disable_incompatible_switches to validate no type switches
27745 (rs6000_incompatible_switch): New function to disallow turning on
27746 other vector options if -mno-vsx, -mno-power8-vector, or
27747 -mno-power9-vector are specified.
27749 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27751 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27753 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27755 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27756 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27757 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27758 (ARG_POINTER_CFA_OFFSET): Likewise.
27760 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27762 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27763 conditions to take advantage of various optimizations.
27765 2017-04-13 Jeff Law <law@redhat.com>
27767 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27768 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27769 (zero_extendsidi2_dext): Likewise.
27771 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27774 * fold-const.c (fold_ternary_loc): Revert
27775 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27778 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27780 PR rtl-optimization/80343
27781 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27782 new scratch pseudo.
27784 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27787 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27788 to ubsan_encode_value.
27790 2017-04-13 Jeff Law <law@redhat.com>
27792 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27793 appearing in DEBUG_INSNs.
27795 2017-04-13 Martin Liska <mliska@suse.cz>
27797 PR gcov-profile/80413
27798 * gcov-io.c (gcov_write_string): Copy to buffer just when
27799 allocated size is greater than zero.
27801 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27804 * dwarf2out.c (decls_for_scope): Ignore declarations of
27805 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27807 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27810 * ipa-visibility.c (non_local_p): Fix typos.
27811 (localize_node): When localizing symbol in same comdat group,
27812 dissolve the group only when we know external symbols are going
27814 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27816 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27818 PR tree-optimization/79390
27819 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27820 order does not result in usable sequence, retry with reversed operand
27826 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27827 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27828 op0 instead of fold_convert_loc (loc, type, arg0).
27830 2017-04-12 Jeff Law <law@redhat.com>
27832 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27833 has a delay slot in the generated code.
27835 * config/cris/cris.md (cris_preferred_reload_class): Return
27836 GENNONACR_REGS rather than GENERAL_REGS.
27838 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27841 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27842 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27843 signedness of the result type.
27845 2017-04-12 Richard Biener <rguenther@suse.de>
27846 Jeff Law <law@redhat.com>
27848 PR tree-optimization/80359
27849 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27850 trim stores to TARGET_MEM_REFs.
27852 2017-04-12 Richard Biener <rguenther@suse.de>
27854 PR tree-optimization/79390
27855 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27856 threading case even more.
27858 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27861 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27862 for quad_address_p for TImode, instead of just not indexed_address.
27864 2017-04-12 Richard Biener <rguenther@suse.de>
27865 Bernd Edlinger <bernd.edlinger@hotmail.de>
27867 PR middle-end/79671
27868 * alias.c (component_uses_parent_alias_set_from): Handle
27869 TYPE_TYPELESS_STORAGE.
27870 (get_alias_set): Likewise.
27871 * tree-core.h (tree_type_common): Add typeless_storage flag.
27872 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27873 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27874 for types containing members with TYPE_TYPELESS_STORAGE.
27875 (place_field): Likewise.
27876 (layout_type): Likewise for ARRAY_TYPE.
27877 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27878 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27879 TYPE_TYPELESS_STORAGE.
27880 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27882 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27885 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27886 first argument to type.
27888 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27892 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27893 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27894 (rs6000_expand_binop_builtin): Likewise.
27895 (rs6000_expand_ternop_builtin): Likewise; also add missing
27896 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27897 vshasigma built-ins.
27898 * doc/extend.texi: Document that vec_xxpermdi's third argument
27899 must be a constant.
27901 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27903 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27904 Use shift_const cost parameter when calculating gain of STV shifts.
27906 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27908 PR rtl-optimization/70478
27909 * lra-constraints.c (process_alt_operands): Check memory for
27910 disfavoring memory insn operand.
27912 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27914 PR middle-end/80100
27915 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27916 left shift in unsigned HOST_WIDE_INT type.
27918 PR rtl-optimization/80385
27919 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27920 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27923 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27924 if they have any depend clauses.
27926 2017-04-11 Martin Liska <mliska@suse.cz>
27929 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27930 * ipa-split.c (split_function): Create a local comdat symbol
27931 if caller is in a comdat group.
27933 2017-04-11 Martin Liska <mliska@suse.cz>
27936 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27939 2017-04-11 Martin Sebor <msebor@redhat.com>
27941 PR middle-end/80364
27942 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27943 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27945 (directive::set_width, directive::set_precision, format_character):
27947 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27950 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27953 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27954 conflict, set target->arch_name instead of target->cpu_name.
27956 2017-04-11 Richard Biener <rguenther@suse.de>
27958 PR tree-optimization/80374
27959 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27960 build_zero_cst, remove fold_convertible_p check again.
27962 2017-04-11 Martin Liska <mliska@suse.cz>
27965 * ubsan.c (instrument_object_size): Do not instrument register
27968 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27971 * config/i386/i386-builtin-types.def
27972 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27973 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27974 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27975 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27976 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27977 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27978 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27979 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27980 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27981 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27982 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27983 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27984 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27985 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27986 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27987 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27988 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27989 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27990 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27991 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27992 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27993 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27994 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27995 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27996 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27997 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27998 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27999 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28000 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28001 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28002 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28003 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28004 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28005 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28006 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28007 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28008 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28009 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28010 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28011 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28012 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28013 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28014 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28015 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28016 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28017 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28019 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28020 flag to second_arg_count, handle 4 argument function type _COUNT
28021 aliases, handle second_arg_count on second argument rather than last.
28023 2017-04-10 Jeff Law <law@redhat.com>
28025 PR tree-optimization/80374
28026 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28027 record anything if we can not convert integer_zero_node to the
28030 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28033 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28034 Enhance special handling given to the TARGET_P9_MINMAX option in
28035 relation to certain other options.
28037 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28039 PR tree-optimization/80153
28040 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28041 remove POINTER_PLUS_EXPR's base part directly, rather than through
28044 2017-04-10 Richard Biener <rguenther@suse.de>
28045 Bin Cheng <bin.cheng@arm.com>
28047 PR tree-optimization/80153
28048 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28049 the first element of pointer type aff_tree. Build result expr in
28051 (add_elt_to_tree): Convert to type unconditionally. Remove other
28052 fold_convert calls.
28053 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28054 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28056 2017-04-10 Richard Biener <rguenther@suse.de>
28058 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28061 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28063 PR rtl-optimization/70478
28064 * lra-constraints.c (curr_small_class_check): New.
28065 (update_and_check_small_class_inputs): New.
28066 (process_alt_operands): Update curr_small_class_check. Disfavor
28067 alternative insn memory operands. Check available regs for small
28070 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28073 * config/mips/mips.opt (-mvirt): Update description.
28074 * doc/invoke.texi (-mvirt): Likewise.
28076 2017-04-10 Richard Biener <rguenther@suse.de>
28078 PR middle-end/80362
28079 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28080 looking for NEGATE_EXPR in -A / -B to A / B folding.
28082 2017-04-10 Martin Liska <mliska@suse.cz>
28084 PR gcov-profile/80224
28085 * gcov.c (print_usage): Fix usage string.
28086 (get_gcov_intermediate_filename): Remove.
28087 (output_gcov_file): Use both for normal and intermediate format.
28088 (generate_results): Do not initialize special file for
28089 intermediate format.
28091 2017-04-10 Richard Biener <rguenther@suse.de>
28093 PR tree-optimization/80304
28094 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28097 2017-04-10 Nathan Sidwell <nathan@acm.org>
28100 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28101 (rs6000_init_builtins): Use it.
28103 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28105 * config/arm/arm.md (<mrc>): Add mode to SET source.
28106 (<mrrc>): Likewise.
28108 2017-04-10 Richard Biener <rguenther@suse.de>
28110 PR middle-end/80344
28111 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28113 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28116 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28117 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28118 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28119 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28120 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28121 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28122 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28123 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28124 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28125 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28126 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28127 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28128 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28129 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28130 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28131 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28132 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28133 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28134 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28135 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28136 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28137 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28138 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28140 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28142 PR rtl-optimization/70478
28143 * lra-constraints.c: Reverse the last patch.
28145 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28147 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28148 Add comment for WCHAR_T.
28150 2017-04-08 Martin Liska <mliska@suse.cz>
28153 2017-04-07 Martin Liska <mliska@suse.cz>
28156 * ipa-split.c (split_function): Add function part to a same comdat
28159 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28162 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28164 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28166 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28167 (vec_pairing): ...to this.
28168 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28169 (rs6000_sched_init): Adjust for name change.
28170 (struct rs6000_sched_context): Likewise.
28171 (rs6000_init_sched_context): Likewise.
28172 (rs6000_set_sched_context): Likewise.
28174 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28180 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28182 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28183 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28184 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28186 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28188 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28190 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28192 PR rtl-optimization/70703
28193 * ira-color.c (update_conflict_hard_regno_costs): Use
28194 int64_t instead of HOST_WIDE_INT.
28196 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28198 PR rtl-optimization/70478
28199 * lra-constraints.c (process_alt_operands): Disfavor alternative
28200 insn memory operands.
28202 2017-04-07 Jeff Law <law@redhat.com>
28204 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28205 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28207 2017-04-07 Martin Liska <mliska@suse.cz>
28210 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28211 Show error message instead of an ICE.
28213 2017-04-07 Martin Liska <mliska@suse.cz>
28216 * ipa-split.c (split_function): Add function part to a same comdat
28219 2017-04-07 Richard Biener <rguenther@suse.de>
28221 PR middle-end/80341
28222 * tree.c (get_unwidened): Also handle ! for_type case for
28224 * convert.c (do_narrow): Split out from ...
28225 (convert_to_integer_1): ... here. Do not pass final truncation
28226 type to get_unwidened for TRUNC_DIV_EXPR.
28228 2017-04-07 Richard Biener <rguenther@suse.de>
28230 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28232 (aff_combination_const): Adjust.
28233 (aff_combination_scale): Likewise.
28234 (aff_combination_add_elt): Likewise.
28235 (aff_combination_add_cst): Likewise.
28236 (aff_combination_convert): Likewise.
28237 (add_elt_to_tree): Likewise. Remove unused argument.
28238 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28240 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28242 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28244 * config/arm/arm.c (arm_default_short_enums): Use
28245 ARM_DEFAULT_SHORT_ENUMS.
28246 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28248 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28251 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28252 members with redundant out-of-class redeclaration.
28254 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28257 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28258 * config/i386/i386.md (*zero_extendsidi2):
28259 Add (?*x,*x) and (?*v,*v) alternatives.
28261 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28264 * config/i386/i386.c (ix86_expand_builtin)
28265 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28266 mode from insn data. Convert operands to insn operand mode.
28267 Copy operands that don't satisfy insn predicate to a register.
28269 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28271 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28274 2017-04-06 Richard Biener <rguenther@suse.de>
28276 PR tree-optimization/80334
28277 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28278 preserve alignment of accesses.
28280 2017-04-06 Richard Biener <rguenther@suse.de>
28282 PR tree-optimization/80262
28283 * tree-sra.c (build_ref_for_offset): Preserve address-space
28285 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28286 Drop useless address-space information on MEM_REF offsets.
28288 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28290 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28292 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28294 PR rtl-optimization/70703
28295 * ira-color.c (update_conflict_hard_regno_costs): Use
28296 HOST_WIDE_INT instead of long.
28298 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28301 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28302 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28304 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28305 for x86_64 target. Handle -m3dnowa option.
28307 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28309 PR rtl-optimization/70703
28310 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28311 (update_conflict_hard_regno_costs): Use long instead of unsigned
28312 arithmetic for cost calculation.
28314 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28315 Bernd Edlinger <bernd.edlinger@hotmail.de>
28318 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28321 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28324 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28325 ptr_mode with Pmode throughout.
28326 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28327 into probe_stack_range and use DImode.
28329 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28332 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28333 call_eh_return is true.
28335 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28337 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28338 Initialize last_match_fntype_index.
28340 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28343 * tree-nvr.c: Include internal-fn.h.
28344 (pass_return_slot::execute): Ignore internal calls without
28347 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28348 Richard Biener <rguenther@suse.de>
28351 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28352 captures used multiple times, except for the last use.
28353 * generic-match-head.c: Include gimplify.h.
28355 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28357 PR tree-optimization/79390
28358 * target.h (struct noce_if_info): Declare.
28359 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28360 * target.def (noce_conversion_profitable_p): New target hook.
28361 * ifcvt.h (struct noce_if_info): New type, moved from ...
28362 * ifcvt.c (struct noce_if_info): ... here.
28363 (noce_conversion_profitable_p): Renamed to ...
28364 (default_noce_conversion_profitable_p): ... this. No longer
28366 (noce_try_store_flag_constants, noce_try_addcc,
28367 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28368 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28369 instead of noce_conversion_profitable_p.
28370 * config/i386/i386.c: Include ifcvt.h.
28371 (ix86_option_override_internal): Don't override
28372 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28373 (ix86_noce_conversion_profitable_p): New function.
28374 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28375 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28376 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28377 * doc/tm.texi: Regenerated.
28379 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28381 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28384 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28387 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28388 instructions for small multiply cores.
28390 2017-04-04 Jeff Law <law@redhat.com>
28392 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28394 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28395 that are not set by the loop over the elements.
28397 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28400 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28401 int mode, convert_modes it to mode as unsigned, otherwise use
28402 lowpart_subreg to mode rather than SImode.
28403 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28404 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28405 Use DImode instead of SImode for the shift count operand.
28406 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28409 2017-04-04 Richard Biener <rguenther@suse.de>
28411 PR middle-end/80281
28412 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28413 arithmetic done for the negate or the plus. Simplify.
28414 (A - (-B) -> A + B): Likewise.
28415 * fold-const.c (split_tree): Make sure to not negate pointers.
28417 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28419 PR rtl-optimization/60818
28420 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28421 a compare of comparisons with the thing compared if this results
28422 in a different machine mode.
28424 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28426 * alias.c (base_alias_check): Fix typo in comment.
28427 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28428 * cgraphunit.c (symbol_table::compile): Likewise.
28429 * collect2.c (maybe_run_lto_and_relink): Likewise.
28430 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28431 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28432 * config/avr/avr.c (avr_map_op_t): Likewise.
28433 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28434 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28435 * config/epiphany/epiphany.md (movcc): Likewise.
28436 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28437 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28439 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28440 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28441 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28442 * config/sh/sh.c (sh_rtx_costs): Likewise.
28443 * fold-const.c (fold_truth_andor): Likewise.
28444 * genautomata.c (collapse_flag): Likewise.
28445 * gengtype.h (struct type::u::s): Likewise.
28446 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28447 * input.c (FORMAT_AMOUNT): Likewise.
28448 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28449 (known_aggs_to_agg_replacement_list): Likewise.
28450 * ipa-inline-analysis.c: Likewise.
28451 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28452 * ipa-polymorphic-call.c
28453 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28454 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28455 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28457 * modulo-sched.c (apply_reg_moves): Likewise.
28458 * omp-expand.c (build_omp_regions_1): Likewise.
28459 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28460 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28461 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28462 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28463 * value-prof.c: Likewise.
28464 * var-tracking.c (val_reset): Likewise.
28466 2017-04-03 Richard Biener <rguenther@suse.de>
28468 PR tree-optimization/80275
28469 * fold-const.c (split_address_to_core_and_offset): Handle
28472 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28474 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28475 descriptors is at least equal to that of functions.
28477 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28479 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28481 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28484 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28485 (mov<IMOD4:mode>): New expander.
28486 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28488 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28490 PR rtl-optimization/79405
28491 * fwprop.c (propagations_left): New variable.
28492 (forward_propagate_into): Decrement it.
28493 (fwprop_init): Initialize it.
28494 (fw_prop): If the variable has reached zero, stop propagating.
28495 (fwprop_addr): Ditto.
28497 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28500 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28501 a FUNCTION_DECL, pass it as decl instead of origin to
28504 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28506 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28509 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28512 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28513 TARGET_VSX_SMALL_INTEGER.
28515 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28517 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28518 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28520 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28522 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28523 extraction from odd-numbered MSA register.
28525 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28527 PR middle-end/80173
28528 * expmed.c (store_bit_field_1): Don't attempt to create
28529 a word subreg out of hard registers wider than word if they
28530 have HARD_REGNO_NREGS of 1 for their mode.
28532 PR middle-end/80163
28533 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28534 conversions to integer types wider than word and pointer.
28537 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28538 (rtx_equal_for_cselib_p): Pass 0 to it.
28539 * cselib.c (cselib_hasher::equal): Likewise.
28540 (rtx_equal_for_cselib_1): Add depth argument. If depth
28541 is 128, don't look up VALUE locs and punt. Increment
28542 depth in recursive calls when walking VALUE locs.
28544 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28546 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28547 (make_gcov_file_name): Use the canonical path name for generating
28549 (read_line): Fix handling of files with ascii null bytes.
28551 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28553 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28554 to initialise a vector register instead
28555 of using a const_int.
28557 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28559 PR translation/80189
28560 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28561 diagnostic messages.
28563 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28566 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28567 (dfp_diex_<mode>): Update mode of operand 1.
28568 * doc/extend.texi (dxex, dxexq): Document change to return type.
28569 (diex, diexq): Document change to argument type.
28571 2017-03-30 Martin Jambor <mjambor@suse.cz>
28574 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28575 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28576 it reflects the signature changes performed at the callee side.
28577 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28578 to cgraph_build_function_type_skip_args.
28579 (build_function_decl_skip_args): Adjust call to the above function.
28581 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28584 * config/i386/sse.md
28585 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28586 register as dest whenever it is a MEM not rtx_equal_p to the
28587 corresponding dup operand, and when forcing into reg move the
28588 reg into the memory afterwards.
28589 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28590 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28591 for the force_reg mode.
28592 (avx512vl_vextractf128<mode>): Use register as dest either
28593 always when a MEM, or when it is a MEM not rtx_equal_p to the
28594 corresponding dup operand, or even not when it is a CONST_VECTOR
28595 depending on the mode and lo vs. hi.
28596 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28598 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28599 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28600 Likewise. Require that operands[2] is even.
28601 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28602 Remove extraneous parens. Require that operands[2] is a multiple
28604 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28605 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28606 disallow memory then.
28608 2017-03-30 Richard Biener <rguenther@suse.de>
28610 PR tree-optimization/77498
28611 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28612 to non-constants over backedges.
28614 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28616 PR rtl-optimization/80233
28617 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28618 as last_combined_insn. Do not test for BARRIER_P separately.
28620 2017-03-29 Andreas Schwab <schwab@suse.de>
28623 * calls.c (prepare_call_address): Convert funexp to Pmode before
28624 copying to temp reg.
28626 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28628 PR tree-optimization/80158
28629 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28630 Handle possible future case of more than one alternate
28632 (replace_rhs_if_not_dup): Likewise.
28633 (replace_one_candidate): Likewise.
28635 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28637 PR rtl-optimization/80193
28638 * ira.c (ira): Do not check allocation for LRA.
28640 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28642 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28643 (nvptx_output_simt_exit): Declare.
28644 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28645 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28646 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28647 (nvptx_declare_function_name): ...here. Emit declaration of local
28648 memory space buffer for omp_simt_enter insn.
28649 (nvptx_output_unisimt_switch): New.
28650 (nvptx_output_softstack_switch): New.
28651 (nvptx_output_simt_enter): New.
28652 (nvptx_output_simt_exit): New.
28653 * config/nvptx/nvptx.h (struct machine_function): New fields
28654 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28655 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28656 (UNSPECV_SIMT_EXIT): Ditto.
28657 (omp_simt_enter_insn): New insn.
28658 (omp_simt_enter): New expansion.
28659 (omp_simt_exit): New insn.
28660 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28662 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28663 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28664 (expand_GOMP_SIMT_EXIT): New.
28665 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28666 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28667 (GOMP_SIMT_EXIT): Ditto.
28668 * target-insns.def (omp_simt_enter): New insn.
28669 (omp_simt_exit): Ditto.
28670 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28672 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28673 (lower_rec_input_clauses): Likewise.
28674 (lower_lastprivate_clauses): Handle SIMT privatization.
28676 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28677 (ompdevlow_adjust_simt_enter): New.
28678 (find_simtpriv_var_op): New.
28679 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28680 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28682 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28683 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28684 (copy_decl_for_dup_finish): Ditto.
28686 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28688 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28691 * config/i386/i386.c (ix86_option_override_internal): Always
28692 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28694 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28696 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28698 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28700 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28701 mark new edge's irreducible flag accordign to it.
28702 (vect_do_peeling): Check loop preheader edge's irreducible flag
28703 and pass it to function slpeel_add_loop_guard.
28705 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28707 PR tree-optimization/80218
28708 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28709 Update block frequencies and counts.
28711 2017-03-28 Richard Biener <rguenther@suse.de>
28713 PR tree-optimization/78644
28714 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28715 of a simplification result we may not use it at all.
28717 2017-03-28 Richard Biener <rguenther@suse.de>
28720 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28721 without arguments, generate default definition of a SSA name.
28723 2017-03-28 Richard Biener <rguenther@suse.de>
28725 PR middle-end/80222
28726 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28727 TYPE_REF_CAN_ALIAS_ALL references.
28728 * fold-const.c (fold_indirect_ref_1): Likewise.
28730 2017-03-28 Martin Liska <mliska@suse.cz>
28733 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28734 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28736 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28737 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28739 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28740 (EXTRA_SPECS): Define.
28741 (SUBTARGET_EXTRA_SPECS): Likewise.
28742 (SUBTARGET_CPP_SPEC): Likewise.
28743 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28744 SUBTARGET_EXTRA_SPECS.
28745 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28747 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28749 * config/arc/simdext.md (vst64_insn): Update pattern.
28750 (vld32wh_insn): Likewise.
28751 (vld32wl_insn): Likewise.
28752 (vld64_insn): Likewise.
28753 (vld32_insn): Likewise.
28755 2017-03-28 Marek Polacek <polacek@redhat.com>
28758 * fold-const.c (fold_comparison): Use protected_set_expr_location
28759 instead of SET_EXPR_LOCATION.
28761 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28763 * tree.c (add_expr): Avoid name lookup warning.
28765 2017-03-27 Jeff Law <law@redhat.com>
28767 PR tree-optimization/80216
28768 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28769 function name. Limit recursion depth.
28770 (record_temporary_equivalences): Corresponding changes.
28772 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28774 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28777 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28780 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28782 * cfgcleanup.c (reg_note_cfa_p): New array.
28783 (insns_have_identical_cfa_notes): New function.
28784 (old_insns_match_p): Don't cross-jump in between /f
28785 and non-/f instructions. If both i1 and i2 are frame related,
28786 verify all CFA notes, their order and content.
28788 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28791 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28792 HImode and SImode with zero extend to DImode to one insn.
28793 (bswap<mode>2_extenddi): Likewise.
28794 (bswapsi2_extenddi): Likewise.
28795 (bswaphi2_extendsi): Likewise.
28796 (bswaphi2): Combine bswap HImode and SImode into one insn.
28797 Separate memory insns from swapping register.
28798 (bswapsi2): Likewise.
28799 (bswap<mode>2): Likewise.
28800 (bswaphi2_internal): Delete, no longer used.
28801 (bswapsi2_internal): Likewise.
28802 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28803 store, and gpr<-gpr swap insns.
28804 (bswap<mode>2_store): Likewise.
28805 (bswaphi2_reg): Register only splitter, combine with the splitter.
28806 (bswaphi2 splitter): Likewise.
28807 (bswapsi2_reg): Likewise.
28808 (bswapsi2 splitter): Likewise.
28809 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28810 the insns into load, store, and register/register insns.
28811 (bswapdi2_ldbrx): Likewise.
28812 (bswapdi2_load): Likewise.
28813 (bswapdi2_store): Likewise.
28814 (bswapdi2_reg): Likewise.
28816 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28818 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28819 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28821 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28824 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28826 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28827 special handling for target option conflicts between dform
28828 options (-mpower9-dform, -mpower9-dform-vector,
28829 -mpower9-dform-scalar) and -mno-direct-move.
28831 2017-03-27 Richard Biener <rguenther@suse.de>
28833 PR tree-optimization/80181
28834 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28836 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28838 * config/arc/predicates.md (move_double_src_operand): Replace the
28839 call to move_double_src_operand with a call to address_operand.
28841 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28843 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28844 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28845 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28847 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28849 * config/arc/predicates.md (long_immediate_loadstore_operand):
28850 Consider scaled addresses cases.
28852 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28854 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28855 restored when in interrupt.
28856 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28857 doesn't have delay slot.
28859 2017-03-27 Richard Biener <rguenther@suse.de>
28862 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28863 inlined thunk clones.
28865 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28868 * asan.c (instrument_derefs): Copy over last operand from
28869 original COMPONENT_REF to the new COMPONENT_REF with
28870 DECL_BIT_FIELD_REPRESENTATIVE.
28871 * ubsan.c (instrument_object_size): Likewise.
28873 2017-03-27 Richard Biener <rguenther@suse.de>
28875 PR tree-optimization/80170
28876 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28877 sure DR/SCEV didnt fold in constants we do not see when looking
28878 at the reference base alignment.
28880 2017-03-27 Richard Biener <rguenther@suse.de>
28882 PR middle-end/80171
28883 * gimple-fold.c (fold_ctor_reference): Properly guard against
28884 NULL return value from canonicalize_constructor_val.
28886 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28889 * config/i386/i386.c (ix86_expand_builtin)
28890 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28891 flags reg setting and flags reg using instructions.
28892 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28893 clobbering instructions to zero extend op2.
28895 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28897 * doc/install.texi (Configuration) <--with-aix-soname>:
28898 Update link to AIX ld.
28900 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28902 PR rtl-optimization/80160
28903 PR rtl-optimization/80159
28904 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28905 reg_alternate_class into account.
28907 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28910 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28911 to consider in curr_insn_transform.
28913 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28915 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28916 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28917 and emit_mode_inner.
28919 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28921 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28922 argument to the overloaded builtin variants. Use the new flag to
28923 deprecate certain builtin variants.
28924 * config/s390/s390-builtin-types.def: Add new builtin types.
28925 * config/s390/s390-builtins.h: Support new flags field for
28926 overloaded builtins.
28927 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28928 (s390_macro_to_expand): Enable vector float data type.
28929 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28930 builtins by incrementing the __VEC__ version number.
28931 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28933 (s390_resolve_overloaded_builtin): Emit error messages depending
28934 on the builtin flags.
28935 * config/s390/s390.c (s390_expand_builtin): Support additional
28936 flags argument. Change error message to match the messages
28937 emitted in s390-c.c.
28938 * config/s390/s390.md: New UNSPEC_* constants.
28939 (op_type): Add new instruction types.
28940 * config/s390/vecintrin.h: Add new builtins and test data class
28942 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28943 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28944 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28945 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28946 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28947 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28949 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28950 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28951 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28952 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28954 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28955 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28956 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28957 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28959 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28960 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28961 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28962 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28963 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28964 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28965 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28967 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28968 ("vec_scatter_element<V_HW_4:mode>_DI")
28969 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28970 ("vec_fpint<mode>", "vflls")
28971 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28972 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28973 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28974 ("*vec_cmphe<mode>_cc"): ... these.
28976 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28977 mode constant instead of magic value.
28979 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28981 * config/s390/s390.c (s390_expand_vec_compare): Support other
28982 vector floating point modes than just V2DF.
28983 (s390_expand_vcond): Likewise.
28984 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28985 (s390_cannot_change_mode_class): Prevent mode changes between TF
28986 and V1TF in vector registers.
28987 * config/s390/s390.md (DF, SF): New mode attributes.
28988 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28989 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28990 SFmode support for VRs.
28991 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28993 (VFT, VF_HW): New mode iterators.
28994 (vw, sdx): New mode attributes.
28995 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28996 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28997 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28998 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28999 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29000 also the new vector floating point modes. Renaming to ...
29002 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29003 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29004 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29005 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29006 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29007 ("vec_unordered<mode>"): ... these.
29009 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29010 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29011 ("*vec_extendv2df"): New insn definitions.
29013 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29015 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29016 ("mulditi3_2", "*muldi3_sign"): New patterns.
29017 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29018 rename the pattern definition.
29020 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29022 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29024 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29026 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29028 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29029 instruction if possible.
29030 * config/s390/vector.md (vec_halfnumelts): New mode
29032 ("*vec_vllezlf<mode>"): New pattern.
29034 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29036 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29037 ("popcountv4si2", "popcountv2di2"): Rename to ...
29038 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29039 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29041 ("popcount<mode>2_vxe"): New pattern.
29043 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29045 * common/config/s390/s390-common.c (processor_flags_table): Add
29047 * config.gcc: Add arch12.
29048 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29049 Default to arch12 for unknown CPU model numbers.
29050 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29051 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29052 PROCESSOR_max sanity check.
29053 * config/s390/s390-opts.h (enum processor_type): Add
29055 * config/s390/s390.c (processor_table): Add arch12.
29056 (s390_expand_builtin): Add check for B_VXE flag.
29057 (s390_issue_rate): Add PROCESSOR_ARCH12.
29058 (s390_get_sched_attrmask): Likewise.
29059 (s390_get_unit_mask): Likewise.
29060 (s390_sched_score): Enable z13 scheduling for arch12.
29061 (s390_sched_reorder): Likewise.
29062 (s390_sched_variable_issue): Likewise.
29063 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29065 (s390_tune_attr): Use z13 scheduling also for arch12.
29066 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29067 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29068 (TARGET_VXE_P): New macros.
29069 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29070 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29071 * config/s390/s390.opt: Add arch12 as processor_type.
29073 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29075 * config/s390/s390.md
29076 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29077 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29078 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29080 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29081 Rename expanders to ...
29083 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29084 ("fixuns_truncdddi2_emu"): ... these.
29086 ("fixuns_trunc<mode>si2_emu"): New expander.
29088 ("*fixuns_truncdfdi2_z13"): Rename to ...
29089 ("*fixuns_truncdfdi2_vx"): ... this.
29091 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29093 * config/s390/2964.md: Remove the single element vector compare
29094 instructions which are no longer used.
29095 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29097 (s390_canonicalize_comparison): Remove handling of DFmode
29099 (s390_expand_vec_compare_scalar): Remove function.
29100 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29101 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29103 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29105 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29107 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29109 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29110 will anyway by matched by mov<mode>_64dfp.
29112 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29114 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29115 vlef/vstef. Add missing operand to vleif.
29117 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29119 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29120 pair for all vector types with 64 bit elements.
29121 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29122 * config/s390/vector.md (V_HW_64): ... here.
29123 (V_128_NOSINGLE): New mode iterator.
29124 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29125 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29126 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29127 ("*vec_load_pairv2di"): Change to ...
29128 ("*vec_load_pair<mode>"): ... this one.
29130 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29132 * config/s390/constraints.md: Add comments.
29133 (jKK): Reject element sizes > 8 bytes.
29134 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29136 * config/s390/s390.md: Add the s_operand checks formerly in
29137 s390_split_ok_p to various splitters where they are still
29139 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29140 for 128 bit vectors. Plus two splitters.
29142 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29144 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29147 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29150 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29151 error if the boundary argument is not constant.
29153 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29155 PR rtl-optimization/80112
29156 * loop-doloop.c (doloop_condition_get): Don't check condition
29157 if cmp isn't SET with IF_THEN_ELSE src.
29159 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29161 PR tree-optimization/80158
29162 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29163 replacing a candidate statement, also replace it for the
29164 candidate's alternate interpretation.
29165 (replace_rhs_if_not_dup): Likewise.
29166 (replace_one_candidate): Likewise.
29168 2017-03-24 Richard Biener <rguenther@suse.de>
29170 PR tree-optimization/80167
29171 * graphite-isl-ast-to-gimple.c
29172 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29174 (translate_isl_ast_to_gimple::get_rename): Likewise.
29176 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29178 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29179 handling of certain combinations of target options, including the
29180 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29181 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29183 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29186 * config/arm/arm.md (*load_multiple): Add reload_completed to
29187 matching condition.
29189 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29190 Richard Biener <rguenth@suse.de>
29192 PR tree-optimization/79908
29193 PR tree-optimization/80136
29194 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29195 been cast away, gimplify_and_add suffices.
29197 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29199 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29201 2017-03-23 Richard Biener <rguenther@suse.de>
29203 PR tree-optimization/80032
29204 * gimplify.c (gimple_push_cleanup): Forced unconditional
29205 cleanups still have to go to the conditional_cleanups
29208 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29210 PR tree-optimization/80072
29211 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29213 (next_operand_entry_id): Change type to unsigned int.
29214 (sort_by_operand_rank): Make sure to return the right return value
29215 even if unsigned fields are bigger than INT_MAX.
29216 (struct oecount): Change cnt and id type to unsigned int.
29217 (oecount_hasher::equal): Formatting fix.
29218 (oecount_cmp): Make sure to return the right return value
29219 even if unsigned fields are bigger than INT_MAX.
29220 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29223 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29224 TREE_READONLY on result if writing it more than once.
29227 * doc/invoke.texi (-fsanitize=thread): Document that with
29228 -fnon-call-exceptions atomics are not able to throw
29232 * tsan.c: Include tree-eh.h.
29233 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29234 maybe_clean_or_replace_eh_stmt where needed.
29235 (instrument_memory_accesses): Add cfg_changed argument.
29236 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29237 if it returned true.
29238 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29240 PR rtl-optimization/63191
29241 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29242 wrapper function, moved the whole old content into ...
29243 (ix86_delegitimize_address_1): ... this. New inline function.
29244 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29245 true as last argument instead of ix86_delegitimize_address.
29247 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29249 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29250 cortexa57_branch_cost.
29252 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29254 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29256 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29259 * doc/md.texi (Constraints): Document wA constraint.
29260 * config/rs6000/constraints.md (wA): New.
29261 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29262 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29263 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29264 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29266 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29269 * gimplify.c (is_oacc_declared): New function.
29270 (oacc_default_clause): Use it to set default flags for acc declared
29271 variables inside parallel regions.
29272 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29273 declared variables.
29274 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29275 declare attribute to any decl as necessary.
29277 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29280 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29281 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29282 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29283 (arm_arch_lpae): This.
29284 * config/arm/arm.c (arm_arch7ve): Rename into ...
29285 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29286 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29289 2017-03-22 Martin Liska <mliska@suse.cz>
29292 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29293 error message instead of an ICE.
29295 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29297 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29299 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29301 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29303 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29306 2017-03-21 Martin Sebor <msebor@redhat.com>
29308 * doc/extend.texi: Use "cannot" instead of "can't."
29309 * doc/hostconfig.texi: Same.
29310 * doc/install.texi: Same.
29311 * doc/invoke.texi: Same.
29312 * doc/loop.texi: Same.
29313 * doc/md.texi: Same.
29314 * doc/objc.texi: Same.
29315 * doc/rtl.texi: Same.
29316 * doc/tm.texi: Same.
29317 * doc/tm.texi.in: Same.
29318 * doc/trouble.texi: Same.
29320 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29323 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29324 (collect_checksum_attributes): Set it.
29325 (die_checksum_ordered): Use it.
29327 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29329 PR tree-optimization/79908
29330 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29331 change: For a VA_ARG whose LHS has been cast away, use
29332 force_gimple_operand to construct the side effects.
29334 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29336 PR translation/80001
29337 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29338 more amenable to translation.
29339 (oacc_loop_auto_partitions): Likewise.
29341 2017-03-21 Marek Polacek <polacek@redhat.com>
29342 Martin Sebor <msebor@redhat.com>
29344 PR tree-optimization/80109
29345 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29346 on INTEGRAL_TYPE_P.
29348 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29349 Segher Boessenkool <segher@kernel.crashing.org>
29352 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29353 check reg_used_between_p between insn and one of succ or succ2
29354 depending on if succ is artificial insn not inserted into insn
29357 2017-03-21 Martin Liska <mliska@suse.cz>
29359 PR gcov-profile/80081
29360 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29361 * doc/gcc.texi: Include gcov-dump stuff.
29362 * doc/gcov-dump.texi: New file.
29364 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29366 PR rtl-optimization/79150
29367 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29368 conditional jump, if the jump is the last insn of the loop.
29370 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29371 Richard Biener <rguenth@suse.de>
29373 PR tree-optimization/79908
29374 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29375 been cast away, use force_gimple_operand to construct the side
29378 2017-03-21 Martin Liska <mliska@suse.cz>
29380 PR libfortran/79956
29381 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29384 2017-03-21 Brad Spengler <spender@grsecurity.net>
29387 * plugin.c (htab_hash_plugin): New function.
29388 (add_new_plugin): Use it and adjust.
29389 (parse_plugin_arg_opt): Adjust.
29390 (init_one_plugin): Likewise.
29392 2017-03-21 Richard Biener <rguenther@suse.de>
29394 PR tree-optimization/80032
29395 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29396 if set force the cleanup to happen unconditionally.
29397 (gimplify_target_expr): Push inserted clobbers with force_uncond
29398 to avoid them being removed by control-dependent DCE.
29400 2017-03-21 Richard Biener <rguenther@suse.de>
29402 PR tree-optimization/80122
29403 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29404 va_arg_pack_len when the inlined call stmt requires pack
29406 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29408 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29411 * tsan.c (instrument_builtin_call): If the memory model argument
29412 is not a constant, assume it is valid.
29415 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29418 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29420 PR rtl-optimization/79910
29421 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29422 if its dest is used by an insn before I2 (other than the combined
29423 insns themselves, which are properly handled already).
29425 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29428 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29430 * combine.c (record_used_regs): New static function.
29431 (try_combine): Handle situations where there is an additional
29432 instruction between I2 and I3 which needs to have a LOG_LINK
29436 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29438 * combine.c (try_combine): Delete redundant i1 test. Call
29439 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29441 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29444 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29445 alternatives 13/14.
29447 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29449 PR tree-optimization/80054
29450 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29451 the optimization if a PHI or any of its arguments is not dominated
29452 by the candidate's basis. Use gphi* rather than gimple* as
29454 (replace_profitable_candidates): Clean up a gimple* variable that
29455 should be a gphi* variable.
29457 2017-03-20 Martin Sebor <msebor@redhat.com>
29460 * doc/extend.texi (attribute constructor): Document present limitation.
29462 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29465 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29466 __POWER9_VECTOR__ #ifdef control, change template definition to
29467 use Power9-specific built-in function.
29468 (vec_any_eq): Likewise.
29469 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29470 to control outcomes from this test.
29471 (vector_ae_<mode>p): For VEC_F modes, likewise.
29473 2017-03-20 Ian Lance Taylor <iant@google.com>
29475 * config/i386/i386.c (ix86_function_regparm): Save an extra
29476 register for -fsplit-stack with DECL_STATIC_CHAIN.
29478 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29481 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29482 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29484 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29486 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29488 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29491 2017-03-20 Marek Polacek <polacek@redhat.com>
29494 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29496 2017-03-20 Richard Biener <rguenther@suse.de>
29498 PR tree-optimization/80113
29499 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29500 allocate extra SSA name for PHI def.
29501 (add_close_phis_to_outer_loops): Likewise.
29502 (add_close_phis_to_merge_points): Likewise.
29503 (copy_loop_close_phi_args): Likewise.
29504 (copy_cond_phi_nodes): Likewise.
29506 2017-03-20 Martin Liska <mliska@suse.cz>
29508 PR middle-end/79753
29509 * tree-chkp.c (chkp_build_returned_bound): Do not build
29510 returned bounds for a LHS that's not a BOUNDED_P type.
29512 2017-03-20 Martin Liska <mliska@suse.cz>
29516 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29517 COMPLEX_CST and VECTOR_CST.
29519 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29522 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29523 target operand. A new splitter adds the clobber statement in case
29524 the target operand is dead anyway.
29526 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29528 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29529 to age-old versions of binutils and glibc.
29531 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29533 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29535 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29537 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29539 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29541 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29542 requirement for binutils 2.13.
29544 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29546 * combine.c (try_combine): Delete redundant i1 test. Call
29547 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29549 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29551 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29552 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29554 <riscv64-*-elf>: Re-arrange section
29555 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29556 <riscv32-*-linux>: Likewise.
29557 <riscv64-*-elf>: Likewise
29558 <riscv64-*-linux>: Likewise.
29560 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29563 * aarch64.opt(verbose-cost-dump): Fix typo.
29565 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29568 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29569 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29571 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29573 * reload.c (find_reloads): When reloading a nonoffsettable address,
29574 use RELOAD_OTHER for it and its address reloads.
29576 PR rtl-optimization/79910
29577 * combine.c (record_used_regs): New static function.
29578 (try_combine): Handle situations where there is an additional
29579 instruction between I2 and I3 which needs to have a LOG_LINK
29582 2017-03-17 Jeff Law <law@redhat.com>
29584 PR tree-optimization/71437
29585 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29586 conditional in the hash table first.
29587 (vrp_dom_walker::before_dom_children): Extract condition from
29588 ASSERT_EXPR. Record condition, its inverion and any implied
29589 conditions as well.
29591 2017-03-17 Marek Polacek <polacek@redhat.com>
29592 Markus Trippelsdorf <markus@trippelsdorf.de>
29594 PR tree-optimization/80079
29595 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29598 2017-03-17 Richard Biener <rguenther@suse.de>
29600 PR middle-end/80075
29601 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29602 Properly verify the LHS before the RHS possibly claims to be
29604 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29607 2017-03-17 Martin Jambor <mjambor@suse.cz>
29609 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29610 (List of -O2 options): Likewise.
29611 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29614 2017-03-17 Tom de Vries <tom@codesourcery.com>
29616 * gcov-dump.c (print_usage): Print bug_report_url.
29618 2017-03-17 Richard Biener <rguenther@suse.de>
29620 PR middle-end/80050
29621 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29622 (parser::peek): Likewise.
29624 2017-03-17 Richard Biener <rguenther@suse.de>
29626 PR tree-optimization/80048
29627 * sese.c (free_sese_info): Properly release rename_map and
29628 copied_bb_map elements.
29630 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29632 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29633 Add linked-list forward and backlinks. Insert on
29634 construction, remove on destruction.
29635 (class pass_store_merging): Add m_stores_head field.
29636 (pass_store_merging::terminate_and_process_all_chains):
29637 Iterate over m_stores_head list.
29638 (pass_store_merging::terminate_all_aliasing_chains):
29640 (pass_store_merging::execute): Check for debug stmts first.
29641 Push new chains onto the m_stores_head stack.
29643 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29646 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29647 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29648 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29650 2017-03-16 Jeff Law <law@redhat.com>
29652 PR tree-optimization/71437
29653 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29654 member function. Implementation moved into after_dom_children
29655 member function and into the threader's thread_outgoing_edges
29657 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29658 some code into new thread_outgoing_edges.
29659 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29660 definition. Simplify marker handling (do it here). Assume we always
29661 have the available expression and the const/copies tables.
29662 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29664 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29665 * tree-vrp.c (equiv_stack): No longer file scoped.
29666 (vrp_dom_walker): New class.
29667 (vrp_dom_walker::before_dom_children): New member function.
29668 (vrp_dom_walker::after_dom_children): Likewise.
29669 (identify_jump_threads): Setup domwalker. Use it rather than
29670 walking edges in a random order by hand. Simplify setup/finalization.
29671 (finalize_jump_threads): Remove.
29672 (vrp_finalize): Do not call identify_jump_threads here.
29673 (execute_vrp): Do it here instead and call thread_through_all_blocks
29676 PR tree-optimization/71437
29677 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29679 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29681 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29682 (dom_opt_dom_walker::thread_across_edge): Remove
29683 handle_dominating_asserts argument. All callers changed.
29684 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29685 changes. Remove calls to lhs_of_dominating_assert. Other
29686 uses of handle_dominating_asserts turn into unconditional code
29687 (simplify_control_stmt_condition_1): Likewise.
29688 (simplify_control_stmt_condition): Likewise.
29689 (thread_through_normal_block, thread_across_edge): Likewise.
29690 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29691 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29692 object if it is not an SSA_NAME.
29693 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29694 before calling into the VRP specific simplifiers.
29695 (identify_jump_threads): Remove handle_dominating_asserts
29698 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29701 * tree-diagnostic.c (default_tree_printer): No longer static.
29702 * tree-diagnostic.h (default_tree_printer): New prototype.
29704 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29706 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29707 Change ins into fmov.
29709 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29711 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29712 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29713 Use h_con constraint for operand 1.
29714 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29715 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29717 2017-03-15 Jeff Law <law@redhat.com>
29719 PR tree-optimization/71437
29720 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29721 (record_temporary_equivalences): Use it.
29723 PR tree-optimization/71437
29724 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29725 tree-ssa-scopedtables.
29726 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29727 (record_conditions, record_cond, vuse_eq): Likewise.
29728 (record_edge_info): Adjust to API tweak of record_conditions.
29729 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29730 (record_temporary_equivalences, optimize_stmt): Likewise.
29731 (eliminate_redundant_computations): Likewise.
29732 (record_equivalences_from_stmt): Likewise.
29733 * tree-ssa-scopedtables.c: Include options.h and params.h.
29734 (vuse_eq): New function, moved from tree-ssa-dom.c
29735 (build_and_record_new_cond): Likewise.
29736 (record_conditions): Likewise. Accept vector of conditions rather
29737 than edge_equivalence structure for first argument.
29738 for the first argument.
29739 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29740 from tree-ssa-dom.c.
29741 (avail_exprs_stack::record_cond): Likewise.
29742 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29743 from tree-ssa-dom.c.
29744 (avail_exprs_stack): Add new member functions lookup_avail_expr
29746 (record_conditions): Declare.
29748 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29751 * lra-constraints.c (process_alt_operands): Increase reject for
29752 reloading an input/output operand.
29754 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29757 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29758 insns to convert from signed/unsigned char/short to IEEE 128-bit
29760 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29762 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29765 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29766 subreg of inner mode for values already in registers.
29768 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29770 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29771 iteration reg is used after the loop.
29773 2017-03-14 Martin Sebor <msebor@redhat.com>
29775 PR tree-optimization/79800
29776 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29777 precision in negative-positive range.
29778 (format_floating): Call non-const overload with adjusted precision.
29780 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29783 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29786 2017-03-14 Martin Sebor <msebor@redhat.com>
29788 PR middle-end/80020
29789 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29790 * builtins.def (aligned_alloc): Use it.
29793 * Makefile.in (GTFILES): Add calls.c.
29794 * calls.c: Include "gt-calls.h".
29796 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29798 PR rtl-optimization/79728
29799 * regs.h (struct target_regs): New field
29800 x_contains_allocatable_regs_of_mode.
29801 (contains_allocatable_regs_of_mode): New macro.
29802 * reginfo.c (init_reg_sets_1): Initialize it, and change
29803 contains_reg_of_mode so it includes global regs as well.
29804 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29805 rather than contains_regs_of_mode.
29807 2017-03-14 Martin Liska <mliska@suse.cz>
29809 * doc/invoke.texi: Document options that can't be combined with
29810 -fcheck-pointer-bounds.
29812 2017-03-14 Martin Liska <mliska@suse.cz>
29814 PR middle-end/79831
29815 * doc/invoke.texi (-Wchkp): Document the option.
29817 2017-03-14 Martin Liska <mliska@suse.cz>
29819 * Makefile.in: Install gcov-dump.
29821 2017-03-14 Martin Liska <mliska@suse.cz>
29823 * multiple_target.c (expand_target_clones): Bail out for
29824 an invalid attribute.
29826 2017-03-14 Richard Biener <rguenther@suse.de>
29828 * alias.c (struct alias_set_entry): Pack properly.
29829 * cfgloop.h (struct loop): Likewise.
29830 * cse.c (struct set): Likewise.
29831 * ipa-utils.c (struct searchc_env): Likewise.
29832 * loop-invariant.c (struct invariant): Likewise.
29833 * lra-remat.c (struct cand): Likewise.
29834 * recog.c (struct change_t): Likewise.
29835 * rtl.h (struct address_info): Likewise.
29836 * symbol-summary.h (function_summary): Likewise.
29837 * tree-loop-distribution.c (struct partition): Likewise.
29838 * tree-object-size.c (struct object_size_info): Likewise.
29839 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29840 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29841 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29842 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29843 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29844 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29845 (struct _stmt_vec_info): Likewise.
29847 2017-03-14 Martin Liska <mliska@suse.cz>
29850 * multiple_target.c (create_dispatcher_calls): Check that
29851 a target can create a function dispatcher.
29853 2017-03-14 Martin Liska <mliska@suse.cz>
29856 * multiple_target.c (expand_target_clones): Drop local.local
29857 flag for default implementation.
29859 2017-03-14 Richard Biener <rguenther@suse.de>
29861 PR tree-optimization/80030
29862 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29864 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29866 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29867 gcc_fallthrough() instead of __attribute__((fallthrough));
29869 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29871 * doc/gcc.texi: Remove "up" link to (DIR).
29872 * doc/gccint.texi: Ditto.
29874 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29876 * doc/install.texi (Specific) <avr>: Remove reference to
29879 2017-03-13 Jeff Law <law@redhat.com>
29881 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29882 attribute rather than comments.
29884 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29885 match_scratch operand is highest.
29887 2017-03-13 Martin Liska <mliska@suse.cz>
29889 PR middle-end/78339
29890 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29891 is a CHKP clone, use original declaration.
29893 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29895 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29896 (arc_conditional_register_usage): Use a different allocation order
29897 when optimizing for size.
29898 * common/config/arc/arc-common.c (arc_option_optimization_table):
29899 Section anchors default on when optimizing for size.
29901 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29903 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29905 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29907 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29908 * config/arc/arc.md (cpu_facility): Add cd variant.
29909 (*movqi_insn): Add code density variant.
29910 (*movhi_insn): Likewise.
29911 (*movqi_insn): Likewise.
29912 (*addsi3_mixed): Likewise.
29913 (subsi3_insn): Likewise.
29915 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29917 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29919 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29921 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29922 expressions with MINUS and UNARY ops.
29924 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29927 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29929 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29930 between vec_select and vector argument.
29931 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29932 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29933 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29934 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29935 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29936 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29938 2017-03-13 Richard Biener <rguenther@suse.de>
29941 * params.def (vect-max-peeling-for-alignment): Fix typo.
29943 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29945 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29946 issue that only occurred with binutils below 2.18.
29948 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29950 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29951 refer to binutils 2.11/2.12 minimum.
29953 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29955 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29956 ftp.kernel.org and simplify binutils requirement.
29958 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29960 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29962 (Optimize Options): Ditto. Also remove redundancy.
29964 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29966 PR translation/79848
29967 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29969 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29970 to G_ to avoid double translation.
29972 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29974 PR translation/79923
29975 * auto-profile.c (get_combined_location): Convert leading
29976 character of diagnostics to lower case and remove trailing period.
29977 (read_profile): Likewise for various diagnostics.
29978 * config/arm/arm.c (arm_option_override): Remove trailing period
29979 from various diagnostics.
29980 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29981 (msp430_expand_delay_cycles): Likewise.
29983 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29986 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29987 full command-line argument, rather than just "str".
29988 (aarch64_validate_march): Likewise.
29989 (aarch64_validate_mtune): Likewise.
29991 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29993 PR rtl-optimization/78911
29994 * lra-assigns.c (must_not_spill_p): New function.
29995 (spill_for): Use it.
29997 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29999 PR tree-optimization/79981
30000 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30001 ATOMIC_COMPARE_EXCHANGE ifn result.
30002 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30003 IFN_ATOMIC_COMPARE_EXCHANGE.
30005 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30008 * opts.c (parse_sanitizer_options): Add missing question mark to
30009 "did you mean" message.
30011 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30013 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30015 (VMULEUH_UNS): Likewise.
30016 (VMULOUB_UNS): Likewise.
30017 (VMULOUH_UNS): Likewise.
30018 * config/rs6000/rs6000.c (builtin_function_type): Remove
30019 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30021 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30024 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30025 x with result of extra_parsing_for_operand_code_0.
30026 (function_reader::extra_parsing_for_operand_code_0): Convert
30027 return type from void to rtx, returning x. When reading
30028 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30029 larger size containing struct block_symbol.
30031 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30033 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30034 -mfloat128-hardware without -m64.
30036 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30039 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30040 entries to the case statement that marks unsigned arguments to
30041 overloaded functions.
30043 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30045 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30046 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30048 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30051 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30052 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30054 2017-03-10 Martin Liska <mliska@suse.cz>
30058 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30059 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30062 2017-03-10 Olivier Hainque <hainque@adacore.com>
30064 * tree-switch-conversion (array_value_type): Start by resetting
30065 candidate type to it's main variant.
30067 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30069 PR rtl-optimization/79909
30070 * combine.c (try_combine): Use simplify_replace_rtx on individual
30071 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30072 of the whole CALL_INSN_FUNCTION_USAGE.
30074 PR tree-optimization/79972
30075 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30076 get_range_info on SSA_NAMEs. Formatting fixes.
30078 2017-03-10 Richard Biener <rguenther@suse.de>
30079 Jakub Jelinek <jakub@redhat.com>
30081 PR tree-optimization/77975
30082 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30083 edge to be constant.
30084 (get_val_for): For constant x return it. Formatting fix.
30085 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30086 would use the same bases as the current one.
30088 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30090 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30091 instead of vec_select for V1TImode.
30092 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30094 (VSX_LE_128): Add V1TI to this mode iterator.
30095 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30096 (*vsx_le_perm_store_<mode>): Likewise.
30097 (pre-reload splitter for VSX stores): Likewise.
30098 (post-reload splitter for VSX stores): Likewise.
30099 (*vsx_xxpermdi2_le_<mode>): Likewise.
30100 (*vsx_lxvd2x2_le_<mode>): Likewise.
30101 (*vsx_stxvd2x2_le_<mode>): Likewise.
30103 2017-03-09 Michael Eager <eager@eagercon.com>
30105 Correct failures with --enable-checking=yes,rtl.
30107 * config/microblaze/microblaze.c (microblaze_expand_shift):
30108 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30109 test for const0_rtx.
30110 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30111 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30113 2017-03-09 Richard Biener <rguenther@suse.de>
30115 PR tree-optimization/79977
30116 * graphite-scop-detection.c (scop_detection::merge_sese):
30117 Handle the case of extra exits to blocks dominating the entry.
30119 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30121 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30124 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30126 PR rtl-optimization/79949
30127 * lra-constraints.c (process_alt_operands): Check memory when
30128 trying to predict a cycle. Print about the overall increase.
30130 2017-03-09 Richard Biener <rguenther@suse.de>
30132 PR middle-end/79971
30133 * gimple-expr.c (useless_type_conversion_p): Preserve
30134 TYPE_SATURATING for fixed-point types.
30136 2017-03-09 Richard Biener <rguenther@suse.de>
30139 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30140 alignment of BLKmode params.
30142 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30145 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30146 (VALL_NO_V2Q): Likewise.
30148 * config/aarch64/aarch64-simd.md
30149 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30151 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30152 VALL_NO_V2Q mode iterator.
30153 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30155 2017-03-09 Martin Liska <mliska@suse.cz>
30157 PR tree-optimization/79631
30158 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30159 tree_int_cst_sign_bit just for INTEGER constants.
30161 2017-03-09 Martin Liska <mliska@suse.cz>
30165 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30168 2017-03-09 Marek Polacek <polacek@redhat.com>
30171 * tree.c (inchash::add_expr): Handle TREE_VEC.
30173 2017-03-09 Martin Liska <mliska@suse.cz>
30176 (chkp_narrow_size_and_offset): New function.
30177 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30178 (void chkp_parse_bit_field_ref): New function.
30179 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30180 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30182 2017-03-09 Martin Liska <mliska@suse.cz>
30185 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30186 (chkp_find_bounds_1): Remove gcc_unreachable.
30188 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30191 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30192 BUILT_IN_SYNC*, determine the access type from the size suffix and
30193 always build a MEM_REF with that type. Handle forgotten
30194 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30197 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30198 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30199 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30200 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30201 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30202 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30203 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30204 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30205 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30206 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30207 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30208 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30209 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30210 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30211 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30212 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30213 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30214 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30215 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30216 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30217 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30218 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30219 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30220 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30221 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30222 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30223 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30224 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30225 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30226 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30227 definitions outside of __OPTIMIZE__ guarded section.
30230 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30231 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30232 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30233 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30236 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30238 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30239 ("vfenez<mode>"): Add missing constraints.
30241 2017-03-08 Martin Sebor <msebor@redhat.com>
30244 * config/nds32/nds32.c (nds32_option_override):
30245 Fix misspelled diagnostic.
30247 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30250 * gimplify.c (gimplify_omp_for): Replace index var in outer
30251 taskloop statement with an artificial variable and add
30252 OMP_CLAUSE_PRIVATE clause for it.
30254 2017-03-08 Richard Biener <rguenther@suse.de>
30256 PR tree-optimization/79955
30257 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30258 for accesses that are completely outside of the variable.
30260 2017-03-08 Andrew Haley <aph@redhat.com>
30262 PR tree-optimization/79943
30263 * tree-ssa-loop-split.c (compute_new_first_bound): When
30264 calculating the new upper bound, (END-BEG) should be added, not
30267 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30269 * config/avr/avr.md (setmemhi): Make sure match_dup
30270 operand number comes before match_scratch.
30272 2017-03-08 Richard Biener <rguenther@suse.de>
30274 PR tree-optimization/79920
30275 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30276 with ncopies == 1 to ...
30277 (vect_transform_slp_perm_load): ... here. Properly compute
30278 all element loads by iterating VF times over the group. Do
30279 not handle ncopies (computed in a broken way) in
30280 vect_create_mask_and_perm.
30282 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30285 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30286 is a uniform vector, use uniform_vector_p return value instead of
30287 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30289 2017-03-07 Marek Polacek <polacek@redhat.com>
30291 PR middle-end/79809
30292 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30293 (alloca_call_type): Likewise.
30295 2017-03-07 Martin Liska <mliska@suse.cz>
30297 * gcov.c (process_args): Put comment to correct location.
30299 2017-03-07 Martin Liska <mliska@suse.cz>
30301 PR middle-end/68270
30302 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30303 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30304 (chkp_narrow_bounds_for_field): Likewise.
30305 (chkp_parse_array_and_component_ref): Pass one more argument to
30308 2017-03-07 Richard Biener <rguenther@suse.de>
30310 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30313 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30315 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30316 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30318 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30321 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30322 to end of description.
30323 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30325 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30327 PR rtl-optimization/79901
30328 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30330 (*avx512f_<code><mode>3<mask_name>): ... this.
30331 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30332 iterator instead of VI8_AVX2_AVX512BW.
30334 PR rtl-optimization/79901
30335 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30336 min/max expander, expand it using expand_vec_cond_expr.
30339 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30342 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30345 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30346 to void * for PCH reasons.
30347 * dwarf2out.c (output_loc_operands, output_die): Cast
30348 v.val_vec.array to unsigned char *.
30350 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30353 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30356 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30358 PR rtl-optimization/79571
30359 * lra-constraints.c (process_alt_operands): Calculate static
30360 reject and subtract it from overall when only addresses will be
30363 2017-03-06 Julia Koval <julia.koval@intel.com>
30366 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30367 incoming stack boundary to 128 for 64-bit targets.
30369 2017-03-06 Richard Biener <rguenther@suse.de>
30371 PR tree-optimization/79894
30372 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30373 to NULL after folding it.
30375 2017-03-06 Richard Biener <rguenther@suse.de>
30377 PR tree-optimization/79824
30378 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30379 check disabling peeling for gaps.
30381 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30383 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30384 attributes): Document gettimeofday.
30386 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30388 * config/s390/s390.c (s390_option_override_internal): Set
30389 PARAM_MIN_VECT_LOOP_BOUND
30391 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30393 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30394 * config/s390/s390.md: Likewise.
30396 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30399 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30400 (<avx2_avx512>_perm<mode>): Rename to ...
30401 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30403 (<avx512>_perm<mode>_mask): Rename to ...
30404 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30406 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30407 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30408 instead of VI8F_256_512.
30409 (avx512f_perm<mode>): New define_expand.
30410 (avx512f_perm<mode>_mask): Likewise.
30411 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30412 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30414 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30416 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30417 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30419 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30421 2017-03-06 Martin Liska <mliska@suse.cz>
30424 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30425 when having a SSA NAME w/o VAR_DECL assigned to it.
30427 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30429 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30430 msa_dpsub_<su>_d): Fix MODE for vec_select.
30432 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30434 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30436 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30438 2017-03-06 Richard Biener <rguenther@suse.de>
30440 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30441 * plugin.c (register_plugin_info): Likewise.
30442 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30444 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30446 * config/i386/sse.md (sse_storehps, sse_storelps,
30447 avx_<castmode><avxsizesuffix>_<castmode>,
30448 avx512f_<castmode><avxsizesuffix>_<castmode>,
30449 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30450 in condition that at least one operand is not a MEM.
30452 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30454 PR middle-end/79805
30455 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30456 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30458 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30459 gimple_call_nothrow_p flag based on whether original builtin can throw.
30460 If it can, emit following stmts on the fallthrough edge.
30461 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30462 don't create new bb if inserting just debug stmts on the edge, try to
30463 insert them on the fallthru bb or just reset debug stmts.
30465 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30468 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30469 restore recog_data (including the operand rtxes inside it) around
30470 the call to get_insn_template.
30472 2017-03-03 Martin Sebor <msebor@redhat.com>
30474 PR tree-optimization/79699
30475 * context.c (context::~context): Free MPFR caches to avoid
30476 a memory leak on program exit.
30478 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30480 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30481 Use wide_int::ulow () instead of .elt (0).
30483 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30485 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30486 (*pushxf): Limit oF constraint to 32bit targets and add oC
30487 constraint for 64bit targets.
30488 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30489 (*pushdf): Change rmF constraint to rmC.
30491 2017-03-03 Martin Liska <mliska@suse.cz>
30493 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30494 Remove unused variable.
30496 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30499 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30500 is a memory operand, increase num_memory.
30501 (ix86_expand_args_builtin): Likewise.
30503 2017-03-03 Jan Hubicka <jh@suse.cz>
30506 * ipa-devirt.c (maybe_record_node): Properly handle
30507 __cxa_pure_virtual visibility.
30509 2017-03-03 Martin Liska <mliska@suse.cz>
30511 PR tree-optimization/79803
30512 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30514 (pass_loop_prefetch::execute): Disabled optimization if an
30515 assumption about L1 cache size is not met.
30517 2017-03-03 Martin Liska <mliska@suse.cz>
30519 PR rtl-optimization/79574
30520 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30521 (hash_scan_set): Likewise.
30522 (dump_hash_table): Likewise.
30523 (hoist_code): Likewise.
30525 2017-03-03 Richard Biener <rguenther@suse.de>
30527 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30528 in place of uhigh ().
30529 (fixed_convert_from_real): Likewise.
30531 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30534 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30536 2017-03-03 Richard Biener <rguenther@suse.de>
30538 PR middle-end/79818
30539 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30540 TYPE_OVERFLOW_UNDEFINED check.
30542 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30544 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30546 (vector_ae_<mode>_p): Likewise.
30547 (vector_nez_<mode>_p): Likewise.
30548 (vector_ne_v2di_p): Likewise.
30549 (vector_ae_v2di_p): Likewise.
30550 (vector_ne_<mode>_p): Likewise.
30551 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30553 (vsx_tsqrt<mode>2_fe): Likewise.
30555 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30558 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30560 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30562 PR rtl-optimization/79780
30563 * cprop.c (one_cprop_pass): When second and further conditional trap
30564 in a single basic block is turned into an unconditional trap, turn it
30565 into a deleted note to avoid RTL verification failures.
30567 2017-03-02 Richard Biener <rguenther@suse.de>
30569 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30571 2017-03-02 Richard Biener <rguenther@suse.de>
30573 PR tree-optimization/79345
30575 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30576 param and abort the walk, returning -1 if it is hit.
30577 (walk_aliased_vdefs): Take a limit param and pass it on.
30578 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30579 defaulting to 0 and return a signed int.
30580 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30581 (check_defs): New helper.
30582 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30583 about uninitialized memory.
30584 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30585 bogus uninitialized warning.
30586 (fixed_convert_from_real): Likewise.
30588 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30590 PR tree-optimization/66768
30591 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30592 iv_use if base object can't be determined.
30594 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30596 PR tree-optimization/79345
30597 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30598 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30599 (get_pattern_stats): Initialize it.
30600 * genemit.c (gen_expand): Verify match_scratch numbers come after
30601 match_operand/match_dup numbers.
30602 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30603 match_scratch numbers.
30604 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30606 * config/s390/s390.md (trunctdsd2): Likewise.
30608 2017-03-02 Richard Biener <rguenther@suse.de>
30610 * wide-int.h (wide_int_storage::operator=): Implement in terms
30613 2017-03-02 Richard Biener <rguenther@suse.de>
30615 PR tree-optimization/79777
30616 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30617 the to insert expression to sth existing.
30619 2017-03-01 Martin Sebor <msebor@redhat.com>
30621 PR middle-end/79692
30622 * gimple-ssa-sprintf.c
30623 (directive::known_width_and_precision): New function.
30624 (format_integer): Use it.
30625 (get_mpfr_format_length): Consider the full range of precision
30626 when computing %g output with the # flag. Set the likely byte
30627 count to 3 rather than 1 when precision is indeterminate.
30628 (format_floating): Correct the lower bound of precision.
30630 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30632 * doc/invoke.texi: Document default code model for 64-bit Linux.
30634 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30637 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30638 udiv rather than div since input pattern is unsigned.
30640 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30642 * config/i386/i386.c (print_reg): Warn for values of
30643 unsupported size in integer register.
30645 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30648 * config/rs6000/predicates.md (current_file_function_operand): Do
30649 not allow self calls to be local if the function is replaceable.
30651 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30654 * config/rs6000/altivec.h (vec_ctz and others): Change the
30655 preprocessor macro that controls conditional compilation from
30656 _ARCH_PWR9 to __POWER9_VECTOR__.
30657 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30658 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30659 control (instead of _ARCH_PWR9 control) so that template
30660 definition uses power9-specific function.
30661 (vec_any_eq): Likewise.
30662 (vec_all_ne): Change macro definition to use a power9-specific
30663 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30664 _ARCH_PWR9 control).
30665 (vec_any_eq) Likewise.
30666 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30667 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30668 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30669 support for xvcmpnedp instruction.
30670 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30671 macro expansion so that Power9 implementation of vec_all_ne does
30672 not use the AltiVec predicate framework.
30673 (VCMPNEH_P): Likewise.
30674 (VCMPNEW_P): Likewise.
30675 (VCMPNED_P): Likewise.
30676 (VCMPNEFP_P): Likewise.
30677 (VCMPNEDP_P): Likewise.
30678 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30679 implementation of vec_any_eq to not use AltiVec predicate
30681 (VCMPAEH_P): Likewise.
30682 (VCMPAEW_P): Likewise.
30683 (VCMPAED_P): Likewise.
30684 (VCMPAEFP_P): Likewise.
30685 (VCMPAEDP_P): Likewise.
30686 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30687 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30688 not use the AltiVec predicate framework.
30689 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30690 of vec_any_eq to not use AltiVec predicate framework.
30691 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30692 support for predefined __POWER9_VECTOR__ macro to indicate that
30693 Power9 instruction selection is enabled.
30694 (altivec_overloaded_builtins): Remove extraneous
30695 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30696 function argument types RS6000_BTI_bool_V16QI and
30697 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30698 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30699 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30700 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30701 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30702 Power9 for implementations of vec_cmpne. Change the signature for
30703 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30704 (representing vec_all_ne) to remove the previously described first
30705 argument of type RS6000_BTI_INTSI, as this was an artifact of
30706 reliance on the AltiVec predicate framework, which is no longer
30707 used in the implementation of these functions. Add
30708 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30709 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30710 since, unlike the AltiVec predicate framework implementation, we
30711 do not share function descriptors between vec_alle and vec_anyeq.
30712 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30713 set of modes that receive special treatment even when
30714 TARGET_P9_VECTOR is true. The special treatment emits code that
30715 does not depend on Power9 instructions.
30716 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30717 define_expand to not rely on AltiVec predicate framework.
30718 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30720 (vector_ne_v2di_p): Change this define_expand to not rely on
30721 AltiVec predicate framework.
30722 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30724 (vector_ne_<mode>_p): Change this define_expand to not rely on
30725 AltiVec predicate framework.
30726 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30728 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30729 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30730 define_insn pattern.
30731 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30732 define_insn pattern because the xvcmpne<VSs>. instruction is not
30734 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30735 instruction is not supported.
30737 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30739 * config/nvptx/nvptx.c: Include intl.h.
30741 2017-03-01 Martin Jambor <mjambor@suse.cz>
30744 * ipa-prop.h (ipa_bits): Removed field known.
30745 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30746 to pointers. Adjusted their comments to warn about their sharing.
30747 (ipcp_transformation_summary): Change bits to a vector of pointers.
30748 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30749 (ipa_get_ipa_bits_for_value): Declare.
30750 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30751 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30752 (ipa_bits_hash_table): Likewise.
30753 (ipa_vr_ggc_hash_traits): Likewise.
30754 (ipa_vr_hash_table): Likewise.
30755 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30756 being pointers and vr_known being removed.
30757 (ipa_set_jf_unknown): Likewise.
30758 (ipa_get_ipa_bits_for_value): New function.
30759 (ipa_set_jfunc_bits): Likewise.
30760 (ipa_get_value_range): New overloaded functions.
30761 (ipa_set_jfunc_vr): Likewise.
30762 (ipa_compute_jump_functions_for_edge): Use the above functions to
30763 construct bits and vr parts of jump functions.
30764 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30765 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30767 (ipcp_grow_transformations_if_necessary): Also allocate
30768 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30770 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30771 them. Fix too long lines.
30772 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30773 vr_known being removed.
30774 (ipa_read_jump_function): Use new setter functions to construct bits
30775 and vr parts of jump functions or set them to NULL.
30776 (write_ipcp_transformation_info): Adjust for bits being pointers.
30777 (read_ipcp_transformation_info): Likewise.
30778 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30780 Include gt-ipa-prop.h.
30781 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30783 (ipcp_store_bits_results): Likewise.
30784 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30785 Do not write to existing jump functions but use a temporary instead.
30787 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30790 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30791 attempt to use its first operand as BIT_FIELD_REF base.
30793 2017-03-01 Richard Biener <rguenther@suse.de>
30795 PR middle-end/79721
30796 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30797 interpolating formula in wrapping arithmetic.
30798 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30800 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30802 PR tree-optimization/79734
30803 * tree-vect-generic.c (expand_vector_condition): Optimize
30804 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30805 Handle VEC_COND_EXPR where comparison has different inner width from
30806 type's inner width.
30808 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30810 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30811 markup, and similar issues. Remove @opindex entries for things
30812 that aren't options. Add missing -mmpy-option entries.
30814 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30816 PR tree-optimization/79737
30817 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30818 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30819 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30820 instead of byte_size. Formatting fix.
30821 (shift_bytes_in_array_right): Formatting fix.
30823 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30826 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30827 condition on optimize for the leaf function test.
30829 2017-02-28 Martin Liska <mliska@suse.cz>
30832 * read-rtl-function.c (function_reader::handle_unknown_directive):
30833 Bail out when one uses -flto.
30835 2017-02-28 Martin Liska <mliska@suse.cz>
30837 * common.opt: Replace space with tabular for options of <number>
30839 * config/i386/i386.opt: Show <number> value for
30840 -mlarge-data-threshold.
30841 * opts.c (print_filtered_help): Do not display number in hexadecimal
30844 2017-02-28 Martin Liska <mliska@suse.cz>
30846 * common.opt: Fix --help=option -Q for options which are of
30849 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30851 * config/i386/i386.c (print_reg): Error out for values
30852 of 8-bit size in invalid integer register.
30854 2017-02-28 Martin Sebor <msebor@redhat.com>
30856 PR tree-optimization/79691
30857 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30859 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30862 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30863 gcc_unreachable with output_operand_lossage.
30865 2017-02-28 Richard Biener <rguenther@suse.de>
30867 PR tree-optimization/79740
30868 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30870 (visit_nary_op): Insert the nary into the hashtable if we
30871 pattern-matched sth.
30872 * tree-ssa-pre.c (eliminate_insert): Robustify.
30874 2017-02-28 Richard Biener <rguenther@suse.de>
30876 PR middle-end/79731
30877 * fold-const.c (decode_field_reference): Reject out-of-bound
30880 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30882 * config/i386/i386.c: Include intl.h.
30883 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30884 instead of just cond ? "..." : "...".
30885 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30886 * coverage.c (read_counts_file): Likewise.
30887 * omp-offload.c: Include intl.h.
30888 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30889 of just cond ? "..." : "...".
30890 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30891 of just cond ? "..." : "...".
30893 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30896 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30898 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30899 'tune for' CPU name.
30900 * config/arm/arm-cpu-data.h: Regenerated.
30902 2017-02-28 Richard Biener <rguenther@suse.de>
30904 PR tree-optimization/79732
30905 * tree-inline.c (expand_call_inline): Do not shadow var.
30907 2017-02-28 Richard Biener <rguenther@suse.de>
30909 PR tree-optimization/79723
30910 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30911 address-space properly.
30913 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30915 * doc/optinfo.texi (Optimization groups): Fix option used for
30917 * doc/invoke.texi (-fopt-info): Document "omp".
30918 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30919 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30920 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30921 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30922 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30924 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30926 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30927 OPTGROUP_OMP with "omp".
30929 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30932 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30933 for arithmetic shift of unsigned V2DI.
30935 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30937 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30938 arc/linux.h headers.
30939 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30940 (LINK_SPEC): Likewise.
30941 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30942 (EXTRA_SPECS): Likewise.
30943 (STARTFILE_SPEC): Likewise.
30944 (ENDFILE_SPEC): Likewise.
30945 (LIB_SPEC): Likewise.
30946 (TARGET_SDATA_DEFAULT): Likewise.
30947 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30948 (MULTILIB_DEFAULTS): Likewise.
30949 (DWARF2_UNWIND_INFO): Likewise.
30950 * config/arc/big.h: New file.
30951 * config/arc/elf.h: Likewise.
30952 * config/arc/linux.h: Likewise.
30953 * config/arc/t-uClibc: Remove.
30955 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30957 PR tree-optimization/77536
30958 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30959 (tree_transform_and_unroll_loop): Use above function to compute the
30960 estimated niter of unrolled loop and use it when scaling profile.
30961 Also use count info rather than frequency if it's non-zero.
30962 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30963 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30964 (vect_transform_loop): Call above function.
30966 2017-02-27 Richard Biener <rguenther@suse.de>
30968 PR tree-optimization/45397
30969 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30970 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30971 (visit_nary_op): Add pattern matching for CSEing sign-changed
30972 or truncated operations with wider ones.
30974 2017-02-27 Richard Biener <rguenther@suse.de>
30976 PR tree-optimization/79690
30977 * tree-vect-stmts.c (vectorizable_store): Use vector type
30978 built from the DR with address-space.
30980 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30982 * doc/invoke.texi (Optimize Options): Refine the description
30983 of asan-use-after-return.
30985 2017-02-25 Alan Modra <amodra@gmail.com>
30987 PR rtl-optimization/79584
30988 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30989 base, not ad->base_term, the reg within base. Remove assertion
30990 that ad->base == ad->base_term. Replace gen_int_mode using
30991 bogus mode with const0_rtx.
30993 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30995 PR middle-end/79396
30996 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30997 FMA_EXPR like tcc_binary or tcc_unary.
30999 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31002 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31004 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31005 (output_loc_operands): Handle DW_OP_call_ref and
31006 DW_OP_GNU_variable_value.
31007 (struct variable_value_struct): New type.
31008 (struct variable_value_hasher): Likewise.
31009 (variable_value_hash): New variable.
31010 (string_types): Remove.
31011 (copy_loc_descr): New function.
31012 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31013 (prepend_loc_descr_to_each): New function.
31014 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31015 instead of add_loc_descr_to_each if the first argument is single
31016 location list and the second has multiple.
31017 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31018 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31019 when looking for variable value which doesn't have other location info.
31020 (loc_list_from_tree): Formatting fix.
31021 (gen_array_type_die): Simplify DW_AT_string_length handling.
31022 (adjust_string_types): Remove.
31023 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31024 string_types. Call resolve_variable_values.
31025 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31026 (resolve_addr_in_expr): Likewise. Add A argument.
31027 (copy_deref_exprloc): Remove deref argument. Adjust for the
31028 original expression being DW_OP_GNU_variable_value with optionally
31029 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31030 optionally after it.
31031 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31032 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31033 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31034 (variable_value_hasher::hash, variable_value_hasher::equal): New
31036 (resolve_variable_value_in_expr, resolve_variable_value,
31037 resolve_variable_values, note_variable_value_in_expr,
31038 note_variable_value): New functions.
31039 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31042 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31045 * opts.h (handle_generated_option): Add GENERATED_P argument.
31046 * opts-common.c (handle_option): Adjust function comment.
31047 (handle_generated_option): Add GENERATED_P argument, pass it to
31049 (control_warning_option): Pass false to handle_generated_option
31051 * opts.c (maybe_default_option): Pass true to handle_generated_option
31053 * optc-gen.awk: Likewise.
31055 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31057 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31058 a REG, look at the REG it is a SUBREG of.
31059 (splitter for cmpeqsi_t): Ditto.
31061 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31063 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31064 the special USEs with the pattern of the insn, not the insn itself.
31066 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31069 * doc/invoke.texi: Document -mload-store-pairs.
31071 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31072 Sandra Loosemore <sandra@codesourcery.com>
31074 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31075 argument isn't a CONST_INT.
31076 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31077 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31078 (nios2_validate_compare): Bypass alternate compare logic if *op2
31079 is not a CONST_INT.
31080 (ldstwm_operation_p): Return false if first_base is not a REG or
31081 if first_offset is not a CONST_INT.
31083 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31085 * config/cris/cris.md: Use correct operand in a define_peephole2.
31087 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31089 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31091 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31093 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31094 this_insn if it is an INSN or JUMP_INSN.
31095 (force_offsettable): Look at base, not at addr.
31096 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31097 on things that aren't necessarily CONST_INTs.
31099 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31101 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31102 -mfpmath=sse is the default also for x86-32 targets with SSE2
31103 instruction set when @option{-ffast-math} is enabled
31105 2017-02-24 Jeff Law <law@redhat.com>
31107 PR rtl-optimizatoin/79286
31108 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31111 2017-02-24 Richard Biener <rguenther@suse.de>
31113 PR tree-optimization/79389
31114 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31117 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31119 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31120 function comment to reflect reality.
31121 (loop_exits_before_overflow): Fix typo in function description.
31123 2017-02-24 Richard Biener <rguenther@suse.de>
31125 PR tree-optimization/79389
31126 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31127 properly that a threading opportunity exists. Detect conditional
31128 copy/constant propagation opportunities.
31130 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31132 * config/visium/visium.md (type): Add trap.
31133 (b): New mode attribute.
31134 (*btst): Rename into...
31135 (*btst<mode>): ...this and adjust.
31136 (*cbranchsi4_btst_insn): Rename into...
31137 (*cbranch<mode>4_btst_insn): ...this and adjust.
31138 (trap): New define_insn.
31140 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31142 PR tree-optimization/79389
31143 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31144 (noce_reversed_cond_code): New function.
31145 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31146 reversed_comparison_code. Formatting fix.
31147 (noce_try_store_flag): Test rev_cond != NULL in addition to
31148 reversed_comparison_code.
31149 (noce_try_store_flag_constants): Likewise.
31150 (noce_try_store_flag_mask): Likewise.
31151 (noce_try_addcc): Use rev_cond if non-NULL instead of
31152 reversed_comparison_code.
31153 (noce_try_cmove_arith): Likewise. Formatting fixes.
31154 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31155 (noce_find_if_block): Initialize rev_cond.
31156 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31157 instead of false as last argument never attempt to reverse it
31160 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31162 PR tree-optimization/79663
31163 * tree-predcom.c (combine_chains): Process refs in reverse order
31164 only for ZERO length chains, and add explaining comment.
31166 2017-02-23 Jeff Law <law@redhat.com>
31168 PR tree-optimization/79578
31169 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31170 in call to operand_equal_p.
31172 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31175 * config/i386/cpuid.h: Fix another undefined behavior.
31177 2017-02-23 Richard Biener <rguenther@suse.de>
31179 PR tree-optimization/79683
31180 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31181 vector types for data-refs.
31183 2017-02-23 Martin Liska <mliska@suse.cz>
31185 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31187 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31189 PR middle-end/79665
31190 * internal-fn.c (get_range_pos_neg): Moved to ...
31191 * tree.c (get_range_pos_neg): ... here. No longer static.
31192 * tree.h (get_range_pos_neg): New prototype.
31193 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31194 are known to be in between 0 and signed maximum inclusive, try to
31195 expand both unsigned and signed divmod and use the cheaper one from
31198 2017-02-22 Jeff Law <law@redhat.com>
31200 PR tree-optimization/79578
31201 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31202 to compare base operands.
31204 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31207 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31208 gpc_reg_operand instead of fpr_reg_operand.
31210 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31212 * config/mips/mips.c (mips_return_in_memory): Force FP
31213 vector types to be returned in memory for o32 ABI.
31215 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31217 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31218 instead of DW_TAG_member for static data member declarations and don't
31219 set no_linkage_name for static inline data members.
31220 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31223 2017-02-22 Martin Liska <mliska@suse.cz>
31225 * doc/invoke.texi: Replace inequality signs with square brackets
31228 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31231 * lra-constraints.c (simplify_operand_subreg): Handle
31232 WORD_REGISTER_OPERATIONS targets.
31234 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31237 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31238 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31239 elimination by swapping fld*.
31241 2017-02-22 Richard Biener <rguenther@suse.de>
31243 PR tree-optimization/79673
31244 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31245 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31246 irrelevant address-space qualifiers and avoiding a
31247 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31249 2017-02-22 Richard Biener <rguenther@suse.de>
31251 PR tree-optimization/79666
31252 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31253 to not symbolically negate if that may introduce undefined
31256 2017-02-22 Martin Liska <mliska@suse.cz>
31259 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31260 * data-streamer-out.c (streamer_write_gcov_count_stream):
31262 * value-prof.c (stream_out_histogram_value): Make assert more
31263 precise based on type of counter.
31265 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31268 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31269 Use nonimmediate_operand instead of memory_operand for operand 1.
31270 (float-extend standard_x87sse_constant_load splitter): Ditto.
31272 2017-02-21 Jeff Law <law@redhat.com>
31274 PR tree-optimization/79621
31275 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31276 blocks with edges to themselves.
31278 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31281 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31282 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31283 Use gimple_call_builtin_p.
31286 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31287 on temporarily removed DEBUG_INSNs.
31289 PR tree-optimization/79649
31290 * tree-loop-distribution.c (classify_partition): Give up on
31291 non-generic address space loads/stores.
31293 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31295 * doc/loop.texi (Loop manipulation): Remove nonexistent
31296 tree_ssa_loop_version from the documentation.
31297 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31299 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31302 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31303 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31304 * config/rs6000/rs6000.c: Include except.h.
31305 (rs6000_expand_split_stack_prologue): Call
31306 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31308 2017-02-21 Martin Jambor <mjambor@suse.cz>
31311 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31312 have been analyzed.
31314 2017-02-21 Martin Jambor <mjambor@suse.cz>
31316 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31317 for backward compatibility only.
31318 * doc/invoke.texi (Option Summary): Remove all references to
31319 -fipa-cp-alignment.
31321 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31325 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31327 * lra-constraints.c (curr_insn_transform): Handle
31328 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31330 2017-02-21 Martin Liska <mliska@suse.cz>
31332 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31334 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31336 PR translation/79638
31337 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31339 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31342 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31343 (arm_function_ok_for_sibcall): Return false for an indirect call by
31344 descriptor if all the argument registers are used.
31345 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31346 alignment of the function.
31348 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31350 PR tree-optimization/61441
31351 * simplify-rtx.c (simplify_const_unary_operation): For
31352 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31353 the sNaN unmodified.
31355 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31357 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31358 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31359 instead of SYSTEM_HEADER_DIR.
31361 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31362 Martin Liška <mliska@suse.cz>
31364 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31365 Fix typos and grammar, use active voice, and clarify.
31367 2017-02-20 Marek Polacek <polacek@redhat.com>
31369 PR middle-end/79537
31370 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31373 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31375 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31378 * config/i386/i386.c (ix86_expand_builtin): Handle
31379 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31380 ix86_builtins_isa[fcode].isa as a requirement of those
31381 flags and any other flag in the bitmask.
31382 (ix86_init_mmx_sse_builtins): Use 0 instead of
31383 ~OPTION_MASK_ISA_64BIT as mask.
31384 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31385 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31386 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31387 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31389 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31392 * lra-constraints.c (split_reg): Check requested split mode
31393 is supported by the register.
31395 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31397 * lra-constraints.c (simplify_operand_subreg): Remove early
31400 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31403 * lra-constraints.c (curr_insn_transform): Tighten condition
31404 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31406 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31409 * lra-constraints.c (curr_insn_transform): Handle
31410 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31412 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31415 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31417 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31419 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31422 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31425 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31427 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31430 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31432 * final.c (last_columnnum, override_columnnum): New variables.
31433 (final_start_function): Set last_columnnum, pass it to begin_prologue
31434 hook and pass 0 to dwarf2out_begin_prologue.
31435 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31436 to source_line debug hook.
31437 (notice_source_line): Compute last_columnnum and for debug_column_info
31438 return true on column changes.
31439 * debug.h (struct gcc_debug_hooks): Add column argument to
31440 source_line and begin_prologue hooks.
31441 (debug_nothing_int_charstar_int_bool): Remove prototype.
31442 (debug_nothing_int_int_charstar,
31443 debug_nothing_int_int_charstar_int_bool): New prototypes.
31444 (dwarf2out_begin_prologue): Add column argument.
31445 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31446 begin_prologue hooks.
31447 (debug_nothing_int_charstar_int_bool): Remove.
31448 (debug_nothing_int_int_charstar,
31449 debug_nothing_int_int_charstar_int_bool): New functions.
31450 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31451 through to dwarf2out_source_line.
31452 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31453 (dwarf2out_source_line): Add column argument, emit it if requested.
31454 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31456 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31457 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31458 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31459 through to dwarf2out_begin_prologue.
31460 (vmsdbgout_source_line): Add column argument, pass it through to
31461 dwarf2out_source_line.
31462 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31463 dbxout_source_line caller.
31464 (dbxout_source_line): Add column argument.
31466 * common.opt (gno-column-info, gcolumn-info): New options.
31467 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31468 (check_die): Also test for multiple DW_AT_decl_column attributes.
31469 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31470 DW_AT_decl_column if requested.
31471 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31473 (gen_variable_die): Likewise.
31474 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31475 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31478 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31479 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31480 (ix86_handle_option): Handle OPT_m3dnowa.
31481 * doc/invoke.texi (-m3dnowa): Document.
31482 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31483 -m3dnowa instead of -m3dnow -march=athlon.
31486 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31487 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31489 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31492 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31493 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31494 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31495 generator for vsx_xxpermdi_<mode>_be.
31496 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31497 force big-endian semantics.
31498 (vsx_xxpermdi_<mode>_be): New define_expand with same
31499 implementation as previous version of vsx_xxpermdi_<mode>.
31501 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31503 PR tree-optimization/79327
31504 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31505 variable, its initialization and use.
31507 2017-02-17 Julia Koval <julia.koval@intel.com>
31509 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31510 (OPTION_MASK_ISA_PKU_UNSET): New.
31511 (ix86_handle_option): Handle -mrdpid.
31512 * config/i386/cpuid.h (bit_RDPID): New.
31513 * config/i386/driver-i386.c (host_detect_local_cpu):
31514 Detect RDPID feature.
31515 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31516 * config/i386/i386-c.c (ix86_target_macros_internal):
31518 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31519 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31520 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31521 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31522 * config/i386/i386.md (define_insn "rdpid"): New.
31523 * config/i386/i386.opt Add -mrdpid.
31524 * config/i386/immintrin.h (_rdpid_u32): New.
31526 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31528 PR rtl-optimization/79541
31529 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31530 instead of transforming it into USE.
31532 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31534 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31535 If HONOR_SNANS (SFmode) force the input to a register.
31536 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31537 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31538 an frsp or similar insn.
31540 2017-02-17 Martin Liska <mliska@suse.cz>
31542 PR rtl-optimization/79577
31543 * params.def (selsched-max-sched-times): Increase minimum to 1.
31545 2017-02-17 Martin Liska <mliska@suse.cz>
31547 PR rtl-optimization/79574
31548 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31550 2017-02-17 Martin Liska <mliska@suse.cz>
31552 PR tree-optimization/79529
31553 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31554 ssa_defined_default_def_p to handle cases which are implicitly
31556 * tree-ssa.c (ssa_defined_default_def_p): New function.
31557 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31558 which are implicitly defined.
31559 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31561 2017-02-17 Richard Biener <rguenther@suse.de>
31563 PR middle-end/79576
31564 * params.def (max-ssa-name-query-depth): Limit to 10.
31566 2017-02-17 Richard Biener <rguenther@suse.de>
31568 PR tree-optimization/79552
31569 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31572 2017-02-17 Richard Biener <rguenther@suse.de>
31575 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31577 2017-02-17 Marek Polacek <polacek@redhat.com>
31579 PR middle-end/79536
31580 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31581 (fold_negate_expr): New wrapper.
31583 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31585 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31586 Correct terminology and de-emphasize pre-standard behavior.
31588 2017-02-16 Alan Modra <amodra@gmail.com>
31590 PR rtl-optimization/79286
31591 * ira.c (def_dominates_uses): New function.
31592 (update_equiv_regs): Don't create an equivalence for insns that
31593 may trap where the register def does not dominate the use.
31595 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31597 PR rtl-optimization/78127
31598 * lra.c (lra): Call lra_eliminate before finish the loop after
31601 2017-02-16 Richard Biener <rguenther@suse.de>
31603 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31605 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31606 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31607 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31608 (isl_val_int_from_wi): New function.
31609 (extract_affine_gmp): Rename to ...
31610 (extract_affine_wi): ... this, take a widest_int.
31611 (extract_affine_int): Just wrap extract_affine_wi.
31612 (add_param_constraints): Use isl_val_int_from_wi.
31613 (add_loop_constraints): Likewise, and extract_affine_wi.
31615 2017-02-15 Jeff Law <law@redhat.com>
31617 PR middle-end/79521
31618 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31619 ira_init_register_move_cost_if_necessary.
31621 2017-02-15 Martin Sebor <msebor@redhat.com>
31623 PR middle-end/32003
31624 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31625 removed in a prior commit.
31627 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31629 PR tree-optimization/79347
31630 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31631 counters during peeling.
31633 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31635 * Makefile.in (site.exp): Remove "set ISLVER".
31637 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31640 * real.c (real_from_integer): Call real_convert even for decimal.
31642 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31645 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31647 2017-02-14 Andrew Pinski <apinski@cavium.com>
31649 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31650 cores and change the partno/implementer to be correct.
31651 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31652 the 'B" as the implementer.
31653 * config/aarch64/aarch64-tune.md: Regenerate.
31655 2017-02-14 Carl Love <cel@us.ibm.com>
31657 * config/rs6000/rs6000.c: Add case statement entry to make the
31658 xvcvuxdsp built-in argument unsigned.
31659 * config/rs6000/vsx.md: Fix the source and return operand types so they
31660 match the instruction definitions from the ISA document. Fix typo
31661 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31664 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31667 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31668 member early_clobber_alts.
31669 * lra-lives.c (reg_early_clobber_p): New.
31670 (process_bb_lives): Use it.
31671 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31672 (debug_operand_data): Initialize early_clobber_alts.
31673 (setup_operand_alternative): Set up early_clobber_alts.
31674 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31675 alternatives to new_insn_reg.
31676 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31678 (lra_update_insn_regno_info): Pass the new arg.
31680 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31682 PR middle-end/79505
31683 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31684 (new_oacc_loop_raw): Don't clear already cleared fields.
31687 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31688 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31689 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31691 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31694 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31696 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31699 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31700 the extra instruction to the right place to store 128-bit constant
31703 2017-02-14 Martin Sebor <msebor@redhat.com>
31705 PR middle-end/79448
31706 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31707 warning for strings of unknown length.
31709 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31711 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31713 2017-02-14 Jeff Law <law@redhat.com>
31716 * ira-costs.c (scan_one_insn): Initialize register move costs
31717 for pseudos seen in USE/CLOBBER insns.
31719 PR tree-optimization/79095
31720 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31721 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31722 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31723 if the operands are known to be not equal, then the resulting range
31725 (intersect_ranges): If the new range is ~[0,0] and the old range is
31726 wide, then prefer ~[0,0].
31727 * tree-vrp.c (overflow_comparison_p_1): New function.
31728 (overflow_comparison_p): New function.
31729 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31730 if NAME is used in an overflow test.
31731 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31732 overflow check that can be expressed as an equality test, then adjust
31733 ops to be that equality test.
31735 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31737 * config/s390/s390-builtin-types.def: Remove flags argument.
31738 * config/s390/s390.c (s390_init_builtins): Likewise.
31740 2017-02-14 Martin Liska <mliska@suse.cz>
31742 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31743 vector. Fix trailing white spaces.
31745 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31747 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31750 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31752 PR rtl-optimization/68664
31753 * config/arm/arm.c (arm_sched_can_speculate_insn):
31754 New function. Declare prototype.
31755 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31757 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31759 PR rtl-optimization/68664
31760 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31762 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31764 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31766 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31767 max skip bytes for function, loop and jump.
31769 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31771 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31772 ABS_EXPR for gimple dump.
31774 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31777 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31779 PR tree-optimization/79408
31780 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31781 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31782 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31783 also if rhs1 is INTEGER_CST.
31785 2017-02-14 Richard Biener <rguenther@suse.de>
31787 PR middle-end/79432
31788 * tree-into-ssa.c (insert_phi_nodes): When the function can
31789 have abnormal edges rewrite SSA names with broken use-def
31790 dominance out of SSA and register them for PHI insertion.
31792 2017-02-13 Martin Sebor <msebor@redhat.com>
31794 PR middle-end/79496
31795 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31796 clearing info.nowrite flag when snprintf size argument is a range.
31798 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31800 * cprop.c (cprop_jump): Add missing space in string literal.
31801 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31802 (get_constraint_for_component_ref): Likewise.
31803 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31804 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31805 * lra-constraints.c (process_alt_operands): Likewise.
31806 * ipa-inline.c (inline_small_functions): Likewise.
31807 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31808 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31809 * trans-mem.c (diagnose_tm_1_op): Likewise.
31810 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31811 (grid_parallel_clauses_gridifiable): Likewise.
31813 * config/nvptx/mkoffload.c (process): Add space in between
31816 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31817 "MOD4_SSE_REGS" and "ALL_REGS".
31819 * spellcheck.c (test_data): Add , in between "foo" and "food".
31821 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31824 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31825 boundary crossing check and subsequent code generation agree.
31827 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31829 * config/aarch64/aarch64.c (has_memory_op): Delete.
31830 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31833 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31835 PR rtl-optimization/79388
31836 PR rtl-optimization/79450
31837 * combine.c (distribute_notes): When removing TEM_INSN for which
31838 corresponding dest has last value recorded, invalidate that last
31841 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31843 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31844 of explicit '@'. Add missing assembly comment marker on branch costs
31847 2017-02-13 Nathan Sidwell <nathan@acm.org>
31849 * gengtype-lex.l (<in_struct>): Add '/'.
31851 2017-02-13 Martin Liska <mliska@suse.cz>
31854 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31856 2017-02-13 Richard Biener <rguenther@suse.de>
31858 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31860 * configure: Re-generate.
31861 * config.in: Likewise.
31862 * graphite-dependences.c: Simplify as if
31863 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31864 * graphite-isl-ast-to-gimple.c: Likewise.
31865 * graphite-optimize-isl.c: Likewise.
31866 * graphite-poly.c: Likewise.
31867 * graphite-sese-to-poly.c: Likewise.
31868 * graphite.h: Likewise.
31869 * toplev.c: Include isl/version.h and use isl_version () for
31870 printing the ISL version.
31871 * doc/install.texi: Update ISL requirement.
31873 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31875 * doc/standards.texi (Standards): Update reference to
31878 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31880 * doc/extend.texi (Named Address Spaces): sourceware.org now
31882 * doc/install.texi (Binaries): Ditto.
31885 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31887 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31888 standard terminology "stringize"/"stringizing" throughout.
31889 * doc/cppinternals.texi: Likewise.
31891 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31893 * doc/extend.texi: Fix some spelling mistakes and typos.
31894 * doc/invoke.texi: Likewise.
31896 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31899 * params.def (inline-min-speedup) Change from 10 to 8.
31901 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31903 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31906 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31909 * ipa-inline-analysis.c (get_minimal_bb): New function.
31910 (record_modified): Use it.
31911 (remap_edge_change_prob): Handle also ancestor functions.
31913 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31915 * doc/contrib.texi (Contributors): Remove broken link into
31916 the Mauve CVS repository.
31918 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31920 PR middle-end/79454
31921 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31922 result computation whenever lhs doesn't have vector mode, not
31923 just when it has BLKmode.
31925 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31927 * doc/makefile.texi (profiledbootstrap): Refer to the
31928 installation instructions only in textual form.
31930 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31933 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31935 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31937 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31938 (Specific): Update mingw-w64 reference.
31940 (Specific): Remove broken link to Renesas RX processor.
31942 2017-02-10 Richard Biener <rguenther@suse.de>
31944 * toplev.c (process_options): Do not mention obsolete graphite
31945 options when printing sorry message about missing graphite support.
31946 Mention -floop-nest-optimize.
31948 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31950 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31951 (vtst_p16): Likewise.
31952 (vtstq_p8): Likewise.
31953 (vtstq_p16): Likewise.
31955 (vtstq_p64): Likewise.
31956 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31957 (vset_lane_p64): New.
31958 (vsetq_lane_p64): New.
31960 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31962 PR tree-optimization/79411
31963 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31964 stmt operands are SSA_NAMEs used in abnormal phis.
31965 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31968 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31971 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31974 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31976 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31978 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31980 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31984 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31985 not arbitrary TREE_CONSTANT.
31988 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31989 "omp declare target link" attribute unless is_global_var.
31990 * omp-offload.c (find_link_var_op): Likewise.
31992 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31993 Chung-Lin Tang <cltang@codesourcery.com>
31995 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31997 (gimplify_adjust_omp_clauses): Don't delete TILE.
31998 (gimplify_omp_for): Deal with TILE.
31999 * internal-fn.c (expand_GOACC_TILE): New function.
32000 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32002 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32003 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32005 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32006 avoid DIV for outermost collapse var.
32007 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32008 Remove out of date comments, fix whitespace.
32009 * omp-general.c (omp_extract_for_data): Deal with tiling.
32010 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32011 adjust OLF_DIM_BASE value.
32012 (struct omp_for_data): Add tiling field.
32013 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32014 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32015 for auto loops. Remove default auto determining, moved to
32016 oacc_loop_fixed_partitions.
32017 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32018 stmts, add e_mask field.
32019 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32020 (oacc_thread_numbers): Use oacc_dim_call.
32021 (oacc_xform_tile): New.
32022 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32023 (finish_oacc_loop): Adjust for ifns vector.
32024 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32025 add case for GOACC_TILE fns.
32026 (oacc_loop_xform_loop): Delete.
32027 (oacc_loop_process): Iterate over call list directly, and add
32028 handling for GOACC_TILE fns.
32029 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32031 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32032 vector partitioning to outer loops. Assign 2 partitions to loops
32033 when available. Add TILE handling.
32034 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32035 (execite_oacc_device_lower): Process GOACC_TILE fns,
32036 ignore unknown specs.
32037 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32038 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32039 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32041 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32043 * configure.ac (ACX_BUGURL): Update.
32044 * configure: Regenerate.
32046 2017-02-09 Richard Biener <rguenther@suse.de>
32048 PR tree-optimization/69823
32049 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32050 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32052 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32054 * config/arc/arc-c.def: Add __NPS400__ definition.
32055 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32056 (TARGET_NPS400): Define.
32058 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32060 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32062 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32063 pointer, arch_info.
32064 (arc_cpu_types): Fill the arch_info field with a pointer into the
32065 arc_arch_types table.
32066 (arc_selected_cpu): Declare.
32067 * config/arc/arc.c (arc_selected_cpu): Make global.
32068 (arc_selected_arch): Delete.
32069 (arc_base_cpu): Delete.
32070 (arc_override_options): Remove references to deleted variables,
32071 update access to arch information.
32072 (ARC_OPT): Update access to arch information.
32073 (ARC_OPTX): Likewise.
32074 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32075 (TARGET_ARC600): Update access to arch information.
32076 (TARGET_ARC601): Likewise.
32077 (TARGET_ARC700): Likewise.
32078 (TARGET_EM): Likewise.
32079 (TARGET_HS): Likewise.
32080 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32083 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32086 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32087 condition/operands for integer GE/LE/GEU/LEU operations.
32089 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32091 PR translation/79397
32092 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32095 2017-02-08 Martin Jambor <mjambor@suse.cz>
32098 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32099 whether allocation happened.
32100 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32101 nothing was allocated.
32103 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32105 PR tree-optimization/79408
32106 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32107 constant, but SSA_NAME with a known integer range, use the minimum
32108 of that range instead of op1 to determine if modulo can be replaced
32109 with its first operand.
32111 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32113 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32115 2017-02-08 Richard Biener <rguenther@suse.de>
32117 PR tree-optimization/71824
32118 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32119 Check all loops contained in the merged region.
32121 2017-02-07 Andrew Pinski <apinski@cavium.com>
32123 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32125 2017-02-07 Andrew Pinski <apinski@cavium.com>
32127 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32128 (thunderxt88): Likewise.
32129 (thunderxt81): Disable LSE and change v8.1 to v8.
32130 (thunderxt83): Likewise.
32132 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32133 Richard Biener <rguenther@suse.de>
32135 PR middle-end/79399
32136 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32137 type from int to size_t.
32138 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32140 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32142 PR rtl-optimization/79386
32143 * cprop.c (bypass_conditional_jumps): Initialize
32144 bypass_last_basic_block already before splitting bbs after
32145 unconditional traps...
32146 (bypass_conditional_jumps): ... rather than here.
32149 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32150 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32151 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32152 fix -masm=intel patterns.
32154 2017-02-07 Richard Biener <rguenther@suse.de>
32156 PR tree-optimization/79256
32157 PR middle-end/79278
32158 * builtins.c (get_object_alignment_2): Use min_align_of_type
32159 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32160 and ADJUST_FIELD_ALIGN.
32162 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32164 * doc/tm.texi: Regenerate.
32165 * stor-layout.c (layout_decl): Adjust.
32166 (update_alignment_for_field): Likewise.
32167 (place_field): Likewise.
32168 (min_align_of_type): Likewise.
32169 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32170 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32171 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32172 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32173 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32174 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32175 * config/i386/i386.c (x86_field_alignment): Likewise.
32176 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32177 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32178 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32179 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32180 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32181 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32185 2017-01-30 Richard Biener <rguenther@suse.de>
32187 PR tree-optimization/79256
32188 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32189 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32192 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32194 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32195 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32196 builtins to SImode and emit a zero-extend, if necessary.
32198 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32200 * docs/invoke.texi (RISC-V Options): Alphabetize.
32202 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32204 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32207 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32209 * config/riscv/riscv.c: New file.
32210 * common/config/riscv/riscv-common.c: Likewise.
32211 * config.gcc: Likewise.
32212 * config/riscv/constraints.md: Likewise.
32213 * config/riscv/elf.h: Likewise.
32214 * config/riscv/generic.md: Likewise.
32215 * config/riscv/linux.h: Likewise.
32216 * config/riscv/multilib-generator: Likewise.
32217 * config/riscv/peephole.md: Likewise.
32218 * config/riscv/pic.md: Likewise.
32219 * config/riscv/predicates.md: Likewise.
32220 * config/riscv/riscv-builtins.c: Likewise.
32221 * config/riscv/riscv-c.c: Likewise.
32222 * config/riscv/riscv-ftypes.def: Likewise.
32223 * config/riscv/riscv-modes.def: Likewise.
32224 * config/riscv/riscv-opts.h: Likewise.
32225 * config/riscv/riscv-protos.h: Likewise.
32226 * config/riscv/riscv.h: Likewise.
32227 * config/riscv/riscv.md: Likewise.
32228 * config/riscv/riscv.opt: Likewise.
32229 * config/riscv/sync.md: Likewise.
32230 * config/riscv/t-elf-multilib: Likewise.
32231 * config/riscv/t-linux: Likewise.
32232 * config/riscv/t-linux-multilib: Likewise.
32233 * config/riscv/t-riscv: Likewise.
32234 * configure.ac: Likewise.
32235 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32236 Waterman as RISC-V maintainers.
32237 * doc/install.texi: Add RISC-V entries.
32238 * doc/invoke.texi: Add RISC-V options section.
32239 * doc/md.texi: Add RISC-V constraints section.
32240 * configure: Regenerated.
32242 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32245 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32246 false values to be constant vectors with all 0 or all 1 bits set.
32247 (vcondu<mode><mode>): Likewise.
32248 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32250 (fpmask_comparison_operator): Update comment.
32251 (vecint_comparison_operator): New predicate.
32252 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32253 vector conditionals when the true and false values are constant
32254 vectors with all 0 bits or all 1 bits set.
32256 2017-02-06 Martin Sebor <msebor@redhat.com>
32258 PR tree-optimization/79376
32259 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32261 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32263 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32264 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32265 to simplify split condition.
32267 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32269 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32270 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32273 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32275 PR rtl-optimization/68664
32276 * target.def (can_speculate_insn): New hook.
32277 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32278 * doc/tm.texi: Regenerate.
32279 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32280 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32281 (rs6000_sched_can_speculate_insn): New function.
32283 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32285 PR tree-optimization/79284
32286 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32287 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32288 vectorizable_mask_load_store, vectorizable_operation,
32289 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32290 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32291 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32292 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32293 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32294 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32295 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32296 is_gimple_assign (stmt). Replace another such test with
32297 is_gimple_assign (stmt).
32299 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32302 * config/avr/avr.c (rtl-iter.h): Include it.
32303 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32304 (avr_legitimate_combined_insn): ...and implementation.
32306 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32308 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32309 * config/s390/s390.c (s390_const_operand_ok)
32310 (s390_canonicalize_comparison, s390_extract_part)
32311 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32312 (s390_contiguous_bitmask_p, s390_rtx_costs)
32313 (legitimize_pic_address): Likewise.
32314 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32315 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32316 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32317 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32318 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32320 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32322 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32323 REGNO($0) == REGNO($1).
32325 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32327 * config/s390/linux.h(SIZE_TYPE): Add comment.
32329 2017-02-06 Julian Brown <julian@codesourcery.com>
32330 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32331 Virendra Pathak <virendra.pathak@broadcom.com>
32333 * config/aarch64/aarch64-cores.def: Change the scheduler
32335 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32336 * config/aarch64/thunderx2t99.md: New file.
32338 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32340 * doc/standards.texi (Go Language): Update link to language
32343 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32345 * tree-eh.c (lower_resx): Sanitize profile.
32346 (cleanup_empty_eh_move_lp): Likewise.
32348 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32351 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32353 * cfgloopmanip.h (loop_version): Update prototype.
32354 * modulo-sched.c (sms_schedule): Update call of loop_version.
32355 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32356 * tree-parloops.c (gen_parallel_loop): Likewise.
32357 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32358 * tree-ssa-loop-split.c (split_loop): Likewise.
32359 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32360 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32362 2017-02-05 Martin Liska <mliska@suse.cz>
32365 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32367 (print_operand_address): Initialize a struct to zero.
32369 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32371 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32372 garbage collector only in textual form.
32374 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32376 * doc/extend.texi (x86 specific memory model extensions for
32377 transactional memory): Simplify a phrase.
32379 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32382 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32383 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32384 (atomic_storedi_1): Likewise.
32386 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32388 PR tree-optimization/79338
32389 * tree-parloops.c (gather_scalar_reductions): Don't call
32390 vect_analyze_loop_form for loop->inner before destroying loop's
32393 2017-02-03 Martin Sebor <msebor@redhat.com>
32395 PR tree-optimization/79327
32396 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32397 when precision has resulted in leading zeros.
32398 (format_integer): Adjust the likely counter to assume an unknown
32399 argument that may be zero is non-zero.
32401 2017-02-03 Jason Merrill <jason@redhat.com>
32404 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32405 avoid copying non-taken branch.
32407 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32409 PR tree-optimization/79340
32410 * tree-vect-loop.c (vectorizable_reduction): Release
32411 vec_defs elements after safe_splicing them into other vectors.
32414 PR tree-optimization/79327
32415 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32416 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32418 (format_integer): Use wide_int_to_tree instead of build_int_cst
32419 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32420 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32421 of shortest and longest sequence.
32423 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32425 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32426 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32428 2017-02-03 Walter Lee <walt@tilera.com>
32431 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32432 after initial stackframe link reg save.
32433 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32435 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32438 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32439 wu for stxssp alternative.
32441 2017-02-03 Martin Sebor <msebor@redhat.com>
32443 PR tree-optimization/79352
32444 * gimple-fold.c (get_range_strlen): Add argument.
32445 (get_range_strlen): Change return type to bool.
32446 (get_maxval_strlen): Pass in a dummy argument.
32447 * gimple-fold.h (get_range_strlen): Change return type to bool.
32448 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32449 * tree.h (array_at_struct_end_p): Add argument.
32450 * tree.c (array_at_struct_end_p): Handle it.
32452 2017-02-03 Martin Liska <mliska@suse.cz>
32455 * multiple_target.c (create_dispatcher_calls): Redirect edge
32456 from a caller of a dispatcher.
32457 (expand_target_clones): Make the clones local.
32458 (ipa_target_clone): Do both target clones and resolvers.
32459 (ipa_dispatcher_calls): Remove the pass.
32460 (pass_dispatcher_calls::gate): Likewise.
32461 (make_pass_dispatcher_calls): Likewise.
32462 * passes.def (pass_target_clone): Put as very first IPA early
32465 2017-02-03 Martin Liska <mliska@suse.cz>
32467 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32468 in case of a function with ifunc attribute.
32470 2017-02-03 Martin Liska <mliska@suse.cz>
32472 * cgraph.c (cgraph_node::dump): Dump function version info.
32473 * symtab.c (symtab_node::dump_base): Add missing new line.
32475 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32477 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32478 (ifcombine_ifandif): Use it.
32480 2017-02-03 Martin Liska <mliska@suse.cz>
32482 * doc/invoke.texi: Document default value for
32483 use-after-scope-direct-emission-threshold.
32485 2017-02-03 Martin Liska <mliska@suse.cz>
32487 PR tree-optimization/79339
32488 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32489 (format_floating): Likewise.
32491 2017-02-03 Martin Liska <mliska@suse.cz>
32494 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32496 (ipa_node_params_t::remove): Likewise.
32497 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32498 initialization from removed ipa_node_params_t::insert.
32499 (ipa_node_params::~ipa_node_params): Move from removed
32500 ipa_node_params_t::release.
32501 * symbol-summary.h (symbol_summary::m_released): New member.
32502 Do not release a summary twice. Do not allow to call finalizer
32503 for types of a summary that live in GGC memory.
32505 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32507 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32510 2017-02-02 Martin Sebor <msebor@redhat.com>
32512 PR middle-end/79275
32513 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32514 (format_string): Tighten up the range of output for non-constant
32515 strings and correct the expected range for wide non-constant strings.
32517 2017-02-02 Martin Sebor <msebor@redhat.com>
32519 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32521 PR middle-end/32003
32522 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32524 (-fdump-tree-@var): Add to index and document how to come up
32525 with pass-specific option and dump file names.
32526 (-fdump-passes): Clarify where to look for output.
32528 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32530 PR middle-end/77445
32531 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32532 statistics of the analyzed path; allow threading for speed when
32533 any of BBs along the path are optimized for speed.
32535 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32537 PR middle-end/78468
32538 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32539 settings of the virtual registers.
32542 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32544 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32545 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32548 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32550 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32551 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32553 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32555 * config/s390/s390.md: Add missing comments with the expanded
32557 * config/s390/vector.md: Likewise.
32558 * config/s390/vx-builtins.md: Likewise.
32560 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32563 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32564 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32565 conditions on a single line.
32567 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32569 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32570 __S390_VX__ to __VX__.
32572 2017-02-01 Andrew Pinski <apinski@cavium.com>
32574 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32575 stmt_info to record_stmt_cost.
32576 (vect_get_known_peeling_cost): Pass stmt_info if known to
32578 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32579 cpu_vector_cost field into
32580 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32581 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32582 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32583 splitting of scalar_stmt_cost and vec_stmt_cost.
32584 (thunderx_vector_cost): Likewise.
32585 (cortexa57_vector_cost): LIkewise.
32586 (exynosm1_vector_cost): Likewise.
32587 (xgene1_vector_cost): Likewise.
32588 (thunderx2t99_vector_cost): Improve after the splitting of the two
32590 (aarch64_builtin_vectorization_cost): Update for the splitting of
32591 scalar_stmt_cost and vec_stmt_cost.
32593 2017-02-01 Torvald Riegel <triegel@redhat.com>
32594 Richard Henderson <rth@redhat.com>
32596 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32597 conditional on existance of a fast atomic load.
32598 * optabs-query.c (can_atomic_load_p): New function.
32599 * optabs-query.h (can_atomic_load_p): Declare it.
32600 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32601 no fast atomic load is available for the particular size of access.
32602 (expand_atomic_compare_and_swap): Likewise.
32603 (expand_atomic_load): Likewise.
32604 (expand_atomic_store): Likewise.
32605 (expand_atomic_fetch_op): Likewise.
32606 * testsuite/lib/target-supports.exp
32607 (check_effective_target_sync_int_128): Remove x86 because it provides
32608 no fast atomic load.
32609 (check_effective_target_sync_int_128_runtime): Likewise.
32611 2017-02-01 Richard Biener <rguenther@suse.de>
32613 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32614 (graphite_transform_loops): Provide opt-info for optimized nests.
32615 * tree-parloop.c (parallelize_loops): Provide opt-info for
32616 parallelized loops.
32618 2017-02-01 Richard Biener <rguenther@suse.de>
32620 PR middle-end/79315
32621 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32622 was not set before.
32624 2017-02-01 Richard Biener <rguenther@suse.de>
32626 PR tree-optimization/71824
32627 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32628 Verify the loops are valid in the merged SESE region.
32629 (scop_detection::can_represent_loop_1): Check analyzing the
32630 evolution of the number of iterations in the region succeeds.
32632 2017-01-31 Ian Lance Taylor <iant@golang.org>
32634 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32635 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32637 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32639 PR preprocessor/79210
32640 * input.c (get_substring_ranges_for_loc): Replace line_width
32641 assertion with error-handling.
32643 2017-01-31 Richard Biener <rguenther@suse.de>
32645 PR tree-optimization/77318
32646 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32647 (create_pw_aff_from_tree): Take loop parameter.
32648 (add_condition_to_pbb): Pass loop of the condition to
32649 create_pw_aff_from_tree.
32651 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32653 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32654 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32656 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32660 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32662 (convert_int_to_float128): Likewise.
32663 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32664 (convert_int_to_float128): Likewise.
32665 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32666 (UNSPEC_IEEE128_CONVERT): Likewise.
32667 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32668 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32669 Use local variables for IBM extended format.
32670 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32671 (fix_trunc<mode>si2_fprs): Likewise.
32672 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32673 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32674 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32675 to know that we can now have integers of all sizes in vector
32677 (fix<uns>_<mode>di2_hw): Likewise.
32678 (float<uns>_<mode>si2_hw): Likewise.
32679 (fix_<mode>si2_hw): Likewise.
32680 (fixuns_<mode>si2_hw): Likewise.
32681 (float<uns>_<mode>di2_hw): Likewise.
32682 (float_<mode>di2_hw): Likewise.
32683 (float_<mode>si2_hw): Likewise.
32684 (floatuns_<mode>di2_hw): Likewise.
32685 (floatuns_<mode>si2_hw): Likewise.
32686 (xscvqp<su>wz_<mode>): Delete, no longer used.
32687 (xscvqp<su>dz_<mode>): Likewise.
32688 (xscv<su>dqp_<mode>): Likewise.
32689 (ieee128_mfvsrd_64bit): Likewise.
32690 (ieee128_mfvsrd_32bit): Likewise.
32691 (ieee128_mfvsrwz): Likewise.
32692 (ieee128_mtvsrw): Likewise.
32693 (ieee128_mtvsrd_64bit): Likewise.
32694 (ieee128_mtvsrd_32bit): Likewise.
32696 2017-01-31 Martin Liska <mliska@suse.cz>
32699 * ipa-prop.c (ipa_free_all_node_params): Call release method
32700 instead of ~sumbol_summary to not to trigger double times
32703 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32705 PR tree-optimization/71691
32706 * bitmap.h (class auto_bitmap): New.
32707 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32708 is_maybe_undefined instead of ssa_undefined_value_p.
32710 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32712 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32713 __S390_ARCH_LEVEL__ to __ARCH__.
32715 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32717 PR tree-optimization/79267
32718 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32719 if should_remove_lhs_p is true.
32721 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32724 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32725 (add_alignment_attribute): New.
32726 (base_type_die): Add alignment attribute.
32727 (subrange_type_die): Likewise.
32728 (modified_type_die): Likewise.
32729 (gen_array_type_die): Likewise.
32730 (gen_descr_array_type_die: Likewise.
32731 (gen_enumeration_type_die): Likewise.
32732 (gen_subprogram_die): Likewise.
32733 (gen_variable_die): Likewise.
32734 (gen_field_die): Likewise.
32735 (gen_ptr_to_mbr_type_die): Likewise.
32736 (gen_struct_or_union_type_die): Likewise.
32737 (gen_subroutine_type_die): Likewise.
32738 (gen_typedef_die): Likewise.
32739 (base_type_cmp): Compare alignment attribute.
32741 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32744 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32745 (setb_unsigned) New pattern for setb with CCUNS.
32746 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32747 subfc./subfe sequence to avoid overflow problems. Generate a
32748 shorter sequence with cmpld/setb for power9.
32749 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32750 for generating subfc. instruction.
32751 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32752 now uses this instruction.
32754 2017-01-30 Ian Lance Taylor <iant@google.com>
32757 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32758 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32760 2017-01-30 Martin Sebor <msebor@redhat.com>
32762 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32763 Move constant to the right of a relational operator.
32764 (get_mpfr_format_length, format_character, format_string): Ditto.
32765 (should_warn_p, maybe_warn): Same.
32767 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32769 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32772 * asan.c (get_translation_unit_decl): Remove function.
32773 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32775 2017-01-30 Martin Liska <mliska@suse.cz>
32777 PR gcov-profile/79259
32778 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32779 -fprofile-generate.
32781 2017-01-30 Martin Liska <mliska@suse.cz>
32784 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32785 Initialize variables with NULL value.
32787 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32790 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32792 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32794 2017-01-30 Richard Biener <rguenther@suse.de>
32796 PR tree-optimization/79276
32797 * tree-vrp.c (process_assert_insertions): Properly adjust common
32798 when removing a duplicate.
32800 2017-01-30 Richard Biener <rguenther@suse.de>
32802 PR tree-optimization/79256
32803 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32804 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32806 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32808 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32811 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32812 ("*r<noxa>sbg_<mode>_sll_bitmask")
32813 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32814 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32815 Use contiguous_bitmask_nowrap_operand.
32817 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32820 * config/rs6000/altivec.h (vec_xl): Revise #define.
32821 (vec_xst): Likewise.
32823 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32825 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32827 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32829 PR rtl-optimization/79194
32830 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32831 traps before call to bypass_conditional_jumps.
32833 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32835 PR tree-optimization/71374
32836 * lra-constraints.c (check_conflict_input_operands): New.
32837 (match_reload): Use it.
32839 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32842 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32843 account to calculate conflict_set.
32845 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32847 PR rtl-optimization/78559
32848 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32849 other_insn in combine.
32851 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32853 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32854 uint16_type_node for BT_UINT16.
32856 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32858 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32859 "RTL Tests" to menu.
32860 (GIMPLE Tests): New node.
32861 (RTL Tests): New node.
32863 2017-01-27 Richard Biener <rguenther@suse.de>
32865 PR tree-optimization/79245
32866 * tree-loop-distribution.c (distribute_loop): Apply cost
32867 modeling also to detected patterns.
32869 2017-01-27 Richard Biener <rguenther@suse.de>
32871 PR tree-optimization/71433
32872 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32873 (compare_assert_loc): New function.
32874 (process_assert_insertions): Sort and optimize assert locations
32875 to remove duplicates and push down identical assertions on
32876 edges to their destination block.
32878 2017-01-27 Richard Biener <rguenther@suse.de>
32880 PR tree-optimization/79244
32881 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32882 out SSA names even if abnormal.
32884 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32886 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32887 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32888 instead of MPFR_RNDN.
32890 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32893 * arm.c (arm_option_override): Don't call build_target_option_node
32894 until after doing all option overrides.
32895 (arm_valid_target_attribute_tree): Likewise.
32897 2017-01-27 Martin Liska <mliska@suse.cz>
32899 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32900 for {cd}tors and C++ {cd}tors.
32902 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32904 * config/s390/s390.md ("*setmem_long_and")
32905 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32907 2017-01-26 Martin Sebor <msebor@redhat.com>
32909 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32912 2017-01-26 Martin Sebor <msebor@redhat.com>
32914 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32915 HAVE_DFmode before using XFmode or DFmode.
32916 (parse_directive): Avoid using the z length modifier to avoid
32917 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32919 PR middle-end/78703
32920 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32921 to accept adjustment as an array.
32922 (get_int_range): New function.
32923 (struct directive): Make width and prec arrays.
32924 (directive::set_width, directive::set_precision): Call get_int_range.
32925 (format_integer, format_floating): Handle width and precision ranges.
32926 (format_string, parse_directive): Same.
32928 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32931 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32932 comdat_type_p set, just clone them, but keep the children in the
32936 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32937 which have direct callers with -fvar-tracking-assignments enabled
32939 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32940 inside of type units.
32942 2017-01-26 Martin Sebor <msebor@redhat.com>
32944 PR middle-end/78703
32945 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32947 (struct format_result): Replace number_chars, number_chars_min,
32948 and number_chars_max with a single member of struct result_range.
32950 (format_result::operator+=): Adjust.
32951 (struct fmtresult): Remove bounded. Handle likely and unlikely
32953 (fmtresult::adjust_for_width_or_precision): New function.
32954 (fmtresult:type_max_digits): New function.
32955 (bytes_remaining): Handle likely and unlikely counters.
32956 (min_bytes_remaining): Remove.
32957 (format_percent): Simplify.
32958 (format_integer, format_floating): Set likely and unlikely counters.
32959 (get_string_length, format_character, format_string): Same.
32960 (format_plain, should_warn_p): New function.
32961 (maybe_warn): Call should_warn_p. Update diagnostic messages
32962 and handle those for all directives, including plain strings.
32963 (format_directive): Handle likely and unlikely counters.
32964 Remove unnecessary quoting from diagnostics. Add an informational
32966 (add_bytes): Remove.
32967 (pass_sprintf_length::compute_format_length): Simplify.
32968 (try_substitute_return_value): Handle likely and unlikely counters.
32970 2017-01-26 Carl Love <cel@us.ibm.com>
32972 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32973 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32975 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32978 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32979 endianess for subregs into account.
32980 * lra-constraints.c (lra_constraints): Do risky transformations
32981 always on the first iteration.
32982 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32983 last_call_used_reg_set.
32984 (process_bb_lives): Define and use last_call_used_reg_set.
32985 * lra.c (lra): Always continue after lra_constraints on the first
32988 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32990 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32992 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32994 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32996 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32997 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32998 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32999 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33000 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33001 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33002 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33003 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33004 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33006 2017-01-26 Marek Polacek <polacek@redhat.com>
33009 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33010 for the third operand.
33012 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33014 PR middle-end/79236
33015 * omp-low.c (struct omp_context): Add simt_stmt field.
33016 (scan_omp_for): Return omp_context *.
33017 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33018 context to the _simt_ SIMD stmt.
33019 (lower_omp_for): For combined SIMD with sibling _simt_
33020 SIMD, make sure to use the same decls in _looptemp_
33021 clauses as in the sibling.
33023 2017-01-26 David Sherwood <david.sherwood@arm.com>
33025 PR middle-end/79212
33026 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33029 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33032 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33033 emit fld b; fld a; if possible.
33035 * brig-builtins.def: Update copyright years.
33036 * config/arm/arm_acle_builtins.def: Update copyright years.
33038 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33041 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33042 constraint instead of o for the stxsd instruction.
33044 2017-01-25 Carl Love <cel@us.ibm.com>
33046 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33047 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33049 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33051 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33053 2017-01-25 Richard Biener <rguenther@suse.de>
33055 PR tree-optimization/69264
33056 * target.def (vector_alignment_reachable): Improve documentation.
33057 * doc/tm.texi: Regenerate.
33058 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33060 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33061 earlier changes with respect to TYPE_USER_ALIGN.
33062 (vector_alignment_reachable_p): Likewise. Improve dumping.
33064 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33067 * config/arm/arm.md (xordi3): Force constant operand into a register
33070 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33072 * doc/invoke.texi (-fstore-merging): Correct default optimization
33073 levels at which it is enabled.
33074 (-O): Move -fstore-merging from list to...
33077 2017-01-25 Richard Biener <rguenther@suse.de>
33080 * omp-expand.c: Include debug.h.
33081 (expand_omp_taskreg): Make sure to generate early debug before
33082 outlining anything from a function.
33083 (expand_omp_target): Likewise.
33084 (grid_expand_target_grid_body): Likewise.
33086 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33089 * asan.c (get_translation_unit_decl): New function.
33090 (asan_add_global): Extract modules file name from globals
33091 TRANSLATION_UNIT_DECL name.
33093 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33096 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33097 for long calls with APCS frame and VFP.
33099 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33101 * cfg.c (original_copy_tables_initialized_p): New function.
33102 * cfg.h (original_copy_tables_initialized_p): New decl.
33103 * cfgrtl.c (relink_block_chain): Guard the call to
33104 free_original_copy_tables with a call to
33105 original_copy_tables_initialized_p.
33106 * cgraph.h (symtab_node::native_rtl_p): New decl.
33107 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33108 (symtab_node::needed_p): Don't assert for early assembly output
33109 for __RTL functions.
33110 (cgraph_node::finalize_function): Set "force_output" for __RTL
33112 (cgraph_node::analyze): Bail out early for __RTL functions.
33113 (analyze_functions): Update assertion to support __RTL functions.
33114 (cgraph_node::expand): Bail out early for __RTL functions.
33115 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33117 * function.h (struct function): Update comment for field
33119 * gimple-expr.c: Include "tree-pass.h".
33120 (gimple_has_body_p): Return false for __RTL functions.
33121 * Makefile.in (OBJS): Add run-rtl-passes.o.
33122 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33124 (gcc::pass_manager::get_clean_slate): New accessor.
33125 * passes.c: Include "insn-addr.h".
33126 (should_skip_pass_p): Add logging. Update logic for running
33127 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33128 property-provider override so it is only done for gimple passes.
33130 (skip_pass): New function.
33131 (execute_one_pass): Call skip_pass when skipping passes.
33132 * read-md.c (md_reader::read_char): Support filtering
33133 the input to a subset of line numbers.
33134 (md_reader::md_reader): Initialize fields
33135 m_first_line and m_last_line.
33136 (md_reader::read_file_fragment): New function.
33137 * read-md.h (md_reader::read_file_fragment): New decl.
33138 (md_reader::m_first_line): New field.
33139 (md_reader::m_last_line): New field.
33140 * read-rtl-function.c (function_reader::create_function): Only
33141 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33142 curr_properties. Set DECL_INITIAL to a dummy block.
33143 (read_rtl_function_body_from_file_range): New function.
33144 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33146 * run-rtl-passes.c: New file.
33147 * run-rtl-passes.h: New file.
33149 2017-01-24 Jeff Law <law@redhat.com>
33151 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33154 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33156 PR tree-optimization/79159
33157 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33158 (record_nonwrapping_iv): Improve boundary using above function if no
33159 value range information.
33161 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33162 Martin Jambor <mjambor@suse.cz>
33164 * brig-builtins.def: New file.
33165 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33166 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33167 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33168 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33169 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33170 * builtin-types.def (BT_INT8): New.
33171 (BT_INT16): Likewise.
33172 (BT_UINT8): Likewise.
33173 (BT_UINT16): Likewise.
33174 (BT_FN_ULONG): Likewise.
33175 (BT_FN_UINT_INT): Likewise.
33176 (BT_FN_UINT_ULONG): Likewise.
33177 (BT_FN_UINT_LONG): Likewise.
33178 (BT_FN_UINT_PTR): Likewise.
33179 (BT_FN_ULONG_PTR): Likewise.
33180 (BT_FN_INT8_FLOAT): Likewise.
33181 (BT_FN_INT16_FLOAT): Likewise.
33182 (BT_FN_UINT32_FLOAT): Likewise.
33183 (BT_FN_UINT16_FLOAT): Likewise.
33184 (BT_FN_UINT8_FLOAT): Likewise.
33185 (BT_FN_UINT64_FLOAT): Likewise.
33186 (BT_FN_UINT16_UINT32): Likewise.
33187 (BT_FN_UINT32_UINT16): Likewise.
33188 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33189 (BT_FN_INT_PTR_INT): Likewise.
33190 (BT_FN_UINT_PTR_UINT): Likewise.
33191 (BT_FN_LONG_PTR_LONG): Likewise.
33192 (BT_FN_ULONG_PTR_ULONG): Likewise.
33193 (BT_FN_VOID_UINT64_UINT64): Likewise.
33194 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33195 (BT_FN_INT8_INT8_INT8): Likewise.
33196 (BT_FN_INT16_INT16_INT16): Likewise.
33197 (BT_FN_INT_INT_INT): Likewise.
33198 (BT_FN_UINT_FLOAT_UINT): Likewise.
33199 (BT_FN_FLOAT_UINT_UINT): Likewise.
33200 (BT_FN_ULONG_UINT_UINT): Likewise.
33201 (BT_FN_ULONG_UINT_PTR): Likewise.
33202 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33203 (BT_FN_UINT_UINT_UINT): Likewise.
33204 (BT_FN_VOID_UINT_PTR): Likewise.
33205 (BT_FN_UINT_UINT_PTR: Likewise.
33206 (BT_FN_UINT32_UINT64_PTR): Likewise.
33207 (BT_FN_INT_INT_UINT_UINT): Likewise.
33208 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33209 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33210 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33211 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33212 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33213 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33214 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33215 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33216 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33217 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33218 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33219 * doc/frontends.texi: List BRIG FE.
33220 * doc/install.texi (Testing): Add BRIG tesring requirements.
33221 * doc/invoke.texi (Overall Options): Mention BRIG.
33222 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33224 2017-01-24 Richard Biener <rguenther@suse.de>
33226 PR translation/79208
33227 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33229 2017-01-24 Martin Jambor <mjambor@suse.cz>
33232 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33233 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33234 and known_contexts.
33236 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33238 PR middle-end/79123
33239 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33240 casts from signed to unsigned really don't have a range.
33242 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33244 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33245 GMP_RNDx for compatiblity.
33247 2017-01-24 Martin Liska <mliska@suse.cz>
33250 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33251 that would prevent us to call alloca with -1 as argument.
33253 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33255 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33256 Avoid -Wformat-security warning.
33258 2017-01-23 Andrew Pinski <apinski@cavium.com>
33260 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33263 2017-01-23 Martin Sebor <msebor@redhat.com>
33265 PR middle-end/78703
33266 * gimple-ssa-sprintf.c (warn_level): New global.
33267 (format_integer): Use it here and throughout the rest of the file.
33268 Use the same switch to compute sign as base.
33269 (maybe_warn): New function.
33270 (format_directive): Factor out warnings into maybe_warn.
33271 Add debugging output. Use warn_level.
33272 (add_bytes): Use warn_level.
33273 (pass_sprintf_length::compute_format_length): Add debugging output.
33274 (try_substitute_return_value): Same.
33275 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33277 PR middle-end/78703
33278 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33279 (struct fmtresult, format_integer, format_floating): Adjust.
33280 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33281 (get_string_length, format_string,format_directive): Same.
33282 (pass_sprintf_length::compute_format_length): Same.
33283 (try_substitute_return_value): Simplify slightly.
33285 PR middle-end/78703
33286 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33287 (fmtresult::operator+=): Outlined.
33288 (struct fmtresult): Add ctors.
33289 (struct conversion_spec): Rename...
33290 (struct directive): ...to this. Add and remove data members.
33291 (directive::set_width, directive::set_precision): New functions.
33292 (format_percent): Use fmtresult ctor.
33293 (get_width_and_precision): Remove.
33294 (format_integer): Make naming changes. Avoid computing width and
33296 (format_floating): Same. Adjust indentation.
33297 (format_character, format_none): New functions.
33298 (format_string): Moved character handling to format_character.
33299 (format_directive): Remove arguments, change return type.
33300 (parse_directive): New function.
33301 (pass_sprintf_length::compute_format_length): Move directive
33302 parsing to parse_directive.
33304 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33306 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33307 (assign_assembler_name_if_needed): ... this.
33308 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33309 (assign_assembler_name_if_needed): ... this.
33310 (free_lang_data_in_cgraph): Adjust callers.
33311 * cgraphunit.c (cgraph_node::analyze): Likewise.
33312 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33315 2017-01-23 Richard Biener <rguenther@suse.de>
33317 PR tree-optimization/79088
33318 PR tree-optimization/79188
33319 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33320 resetting loop bounds after last path deletion. Reset loop
33321 bounds of the target loop, make code match the comments.
33322 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33323 Make sure loops need no fixups.
33325 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33327 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33328 exponent support with double type for first argument.
33329 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33330 type returned by __builtin_vec_extract_sig,
33331 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33332 functions from "vector int" to "vector unsigned int" or from
33333 "vector long long int" to "vector unsigned long long int".
33334 Changed type returned by __builtin_vec_extract_exp,
33335 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33336 functions from "vector int" to "vector unsigned int" or from
33337 "vector long long int" to "vector unsigned long long int".
33338 Changed return type of __builtin_vec_test_data_class,
33339 __builtin_vec_test_data_class_sp, and
33340 __builtin_vec_test_data_class_dp from "vector int" to
33341 "vector bool int" or from "vector long long int" to "vector bool
33342 long long int" and changed second argument type from "unsigned
33343 int" to "int". Added new overloaded function forms "vector float
33344 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33345 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33346 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33347 double, vector unsigned long long int)" and "vector double
33348 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33349 long int)". Changed return type of
33350 __builtin_scalar_test_data_class and
33351 __builtin_scalar_test_data_class_sp and
33352 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33353 int" and changed second argument from "unsigned int" to "int".
33354 Changed type returned by __builtin_scalar_test_neg,
33355 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33356 from "int" to "bool int". Added new overloaded function form
33357 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33358 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33359 exponent double-precision with floating point first argument.
33360 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33361 documentation of scalar_test_data_class, scalar_test_neg,
33362 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33363 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33364 vec_test_data_class built-in functions to reflect refinements in
33365 their type signatures.
33367 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33369 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33371 (aarch64_elf_asm_destructor): Likewise.
33373 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33375 PR rtl-optimization/78634
33376 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33377 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33378 * ifcvt.c (noce_try_cmove): Add missing cost check.
33380 PR rtl-optimization/71724
33381 * combine.c (if_then_else_cond): Look for situations where it is
33382 beneficial to undo the work of one of the recursive calls.
33384 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33386 PR tree-optimization/70754
33387 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33388 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33389 combined stmt before it if not NULL.
33390 (combine_chains): Process refs reversely and compute dominance point
33393 2017-01-23 Martin Liska <mliska@suse.cz>
33395 PR tree-optimization/79196
33396 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33397 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33399 (strlen_optimize_stmt): Call the renamed function.
33401 2017-01-23 Michael Matz <matz@suse.de>
33403 PR tree-optimization/78384
33404 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33406 2017-01-23 Richard Biener <rguenther@suse.de>
33408 PR tree-optimization/79186
33409 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33410 both incoming edges before moving an assert.
33412 2017-01-23 Martin Jambor <mjambor@suse.cz>
33414 * ipa-prop.c (load_from_param_1): Removed.
33415 (load_from_unmodified_param): Bits from load_from_param_1 put back
33417 (load_from_param): Removed.
33418 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33419 with stmt. Reverted back to use of load_from_unmodified_param.
33421 2017-01-23 Martin Jambor <mjambor@suse.cz>
33424 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33425 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33426 field a pointer to garbage collected vector, mark lattices and
33427 ipcp_orig_node with GTY((skip)).
33428 (ipa_get_param_count): Adjust to descriptors being a pointer.
33429 (ipa_get_param): Likewise.
33430 (ipa_get_type): Likewise.
33431 (ipa_get_param_move_cost): Likewise.
33432 (ipa_set_param_used): Likewise.
33433 (ipa_get_controlled_uses): Likewise.
33434 (ipa_set_controlled_uses): Likewise.
33435 (ipa_is_param_used): Likewise.
33436 (ipa_node_params_t): Move into garbage collector. New methods insert
33438 (ipa_node_params_sum): Annotate wth GTY(()).
33439 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33441 (ipa_load_from_parm_agg): Adjust declaration.
33442 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33443 * ipa-profile.c (ipa_profile): Likewise.
33444 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33445 (ipa_populate_param_decls): Make descriptors parameter garbage
33447 (ipa_dump_param): Adjust to descriptors being a pointer.
33448 (ipa_alloc_node_params): Likewise.
33449 (ipa_initialize_node_params): Likewise.
33450 (load_from_param_1): Make descriptors parameter garbage collected.
33451 (load_from_unmodified_param): Likewise.
33452 (load_from_param): Likewise.
33453 (ipa_load_from_parm_agg): Likewise.
33454 (ipa_node_params::~ipa_node_params): Removed.
33455 (ipa_free_all_node_params): Remove call to delete operator.
33456 (ipa_node_params_t::insert): New.
33457 (ipa_node_params_t::remove): Likewise.
33458 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33459 copy known_csts and known_contexts vectors.
33460 (ipa_read_node_info): Adjust to descriptors being a pointer.
33461 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33463 (ipcp_transform_function): Make descriptors variable garbage
33466 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33468 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33469 * config/i386/avx512dqintrin.h: Ditto.
33470 * config/i386/avx512fintrin.h: Ditto.
33471 * config/i386/i386.c: Handle new builtins.
33472 * config/i386/i386-builtin.def: Add new builtins.
33473 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33474 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33476 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33477 Martin Liska <mliska@suse.cz>
33479 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33480 * asan.c (asan_expand_poison_ifn): Support stores and use
33481 appropriate ASAN report function.
33482 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33483 * internal-fn.def (ASAN_POISON_USE): Declare.
33484 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33485 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33486 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33487 ASAN_POISON calls w/o LHS.
33488 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33489 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33490 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33491 * gimplify.c (asan_poison_variables): Add attribute
33492 use_after_scope_memory to variables that really needs to live
33494 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33495 having the attribute.
33497 2017-01-23 Martin Liska <mliska@suse.cz>
33499 * asan.c (create_asan_shadow_var): New function.
33500 (asan_expand_poison_ifn): Likewise.
33501 * asan.h (asan_expand_poison_ifn): New declaration.
33502 * internal-fn.c (expand_ASAN_POISON): Likewise.
33503 * internal-fn.def (ASAN_POISON): New builtin.
33504 * sanopt.c (pass_sanopt::execute): Expand
33505 asan_expand_poison_ifn.
33506 * tree-inline.c (copy_decl_for_dup_finish): Make function
33508 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33509 * tree-ssa.c (is_asan_mark_p): New function.
33510 (execute_update_addresses_taken): Rewrite local variables
33511 (identified just by use-after-scope as addressable) into SSA.
33513 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33515 * doc/install.texi (Specific): opensource.apple.com uses https
33516 now. Remove trailing slash.
33518 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33520 * README.Portability: Remove note on an Irix compatibility issue.
33522 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33524 * gcov.c (INCLUDE_ALGORITHM): Define.
33525 (INCLUDE_VECTOR): Define.
33526 No longer include <vector> and <algorithm> directly.
33528 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33530 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33532 * doc/invoke.texi (Code Gen Options): Ditto.
33534 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33537 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33539 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33541 rtl-optimization/79125
33542 * cprop.c (local_cprop_pass): Handle cases where we make an
33543 unconditional trap.
33545 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33549 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33550 read from, for big endian.
33552 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33554 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33555 register pauth builtins for LP64 only.
33557 2017-01-20 Marek Polacek <polacek@redhat.com>
33560 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33563 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33565 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33567 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33568 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33569 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33571 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33574 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33575 in big-endian mode when they are not a single duplicated value.
33577 2017-01-20 Richard Biener <rguenther@suse.de>
33579 * BASE-VER: Bump to 7.0.1.
33581 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33583 * omp-low.c (omplow_simd_context): New struct. Use it...
33584 (lower_rec_simd_input_clauses): ...here and...
33585 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33586 references to idx, lane, max_vf, is_simt.
33588 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33590 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33593 2017-01-20 Martin Jambor <mjambor@suse.cz>
33595 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33596 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33598 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33599 (GTFILES): Rename hsa.c to hsa-common.c.
33600 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33601 * hsa-dump.c: Likewise.
33602 * hsa-gen.c: Likewise.
33603 * hsa-regalloc.c: Likewise.
33604 * ipa-hsa.c: Likewise.
33605 * omp-expand.c: Likewise.
33606 * omp-low.c: Likewise.
33607 * toplev.c: Likewise.
33609 2017-01-20 Marek Polacek <polacek@redhat.com>
33612 * doc/invoke.texi: Document -Wduplicated-branches.
33613 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33614 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33615 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33616 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33617 return 0 only when not OEP_LEXICOGRAPHIC.
33618 (fold_build_cleanup_point_expr): Use the expression
33619 location when building CLEANUP_POINT_EXPR.
33620 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33621 * tree.c (add_expr): Handle error_mark_node.
33623 2017-01-20 Martin Liska <mliska@suse.cz>
33626 * tree-profile.c (init_ic_make_global_vars): Do not call
33628 (gimple_init_gcov_profiler): Likewise.
33630 2017-01-20 Martin Liska <mliska@suse.cz>
33633 * cgraph.h (maybe_create_reference): Remove argument and
33635 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33637 * ipa-cp.c (create_specialized_node): Likewise.
33638 * symtab.c (symtab_node::maybe_create_reference): Handle
33639 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33641 2017-01-20 Martin Liska <mliska@suse.cz>
33643 * read-rtl-function.c (function_reader::create_function): Use
33644 build_decl instread of build_decl_stat.
33646 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33648 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33649 * config/i386/avx512dqintrin.h: Ditto.
33650 * config/i386/avx512fintrin.h: Ditto.
33651 * config/i386/i386-builtin-types.def: Add new types.
33652 * config/i386/i386.c: Handle new types.
33653 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33654 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33655 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33656 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33657 (__builtin_ia32_kshiftridi): New.
33658 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33660 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33664 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33665 define to rs6000_init_stack_protect_guard.
33666 (rs6000_init_stack_protect_guard): New function.
33668 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33669 Yunqiang Su <yunqiang.su@imgtec.com>
33671 * config.gcc (supported_defaults): Add madd4.
33672 (with_madd4): Add validation.
33673 (all_defaults): Add madd4.
33674 * config/mips/mips.opt (mmadd4): New option.
33675 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33677 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33679 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33680 (ISA_HAS_FUSED_MADD4): Likewise.
33681 * doc/invoke.texi (-mmadd4): Document the new option.
33682 * doc/install.texi (--with-madd4): Document the new option.
33684 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33686 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33687 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33688 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33689 (aarch64_init_pauth_hint_builtins): New.
33690 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33691 (aarch64_expand_builtin): Expand new builtins.
33693 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33695 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33696 * combine-stack-adj.c (no_unhandled_cfa): Handle
33697 REG_CFA_TOGGLE_RA_MANGLE.
33698 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33699 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33700 info for return address signing.
33701 (aarch64_expand_epilogue): Likewise.
33703 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33705 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33706 * config/aarch64/aarch64-protos.h
33707 (aarch64_return_address_signing_enabled): New declaration.
33708 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33710 (aarch64_expand_prologue): Sign return address before it's pushed onto
33712 (aarch64_expand_epilogue): Authenticate return address fetched from
33714 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33715 (aarch64_attributes): New function attributes for "sign-return-address".
33716 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33717 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33718 ("*do_return"): Generate combined instructions according to key index.
33719 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33720 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33722 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33723 * config/aarch64/aarch64.opt (msign-return-address=): New.
33724 * doc/extend.texi (AArch64 Function Attributes): Documents
33725 "sign-return-address=".
33726 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33728 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33730 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33731 overall option summary.
33733 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33735 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33736 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33737 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33738 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33740 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33742 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33743 -mpower9-minmax by default for -mcpu=power9.
33744 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33745 128-bit floating point.
33747 2017-01-20 Alan Modra <amodra@gmail.com>
33749 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33750 optimizing for size.
33752 2017-01-20 Alan Modra <amodra@gmail.com>
33755 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33756 for strcmp and strncmp from corresponding builtin decl.
33758 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33760 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33761 instead of i386/rtems-64.h.
33762 * config/i386/rtems-64.h: Remove.
33764 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33768 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33770 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33772 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33774 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33775 Change int to HOST_WIDE_INT.
33776 * config/aarch64/aarch64-protos.h
33777 (aarch64_simd_gen_const_vector_dup): Likewise.
33778 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33780 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33782 * langhooks-def.h (lhd_type_for_size): New decl.
33783 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33784 * langhooks.c (lhd_type_for_size): New function, taken from
33787 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33789 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33790 define_bypass for CR latency.
33791 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33792 (power9-alu2): Add define_bypass for CR latency.
33794 (power9-mul): Update insn latency.
33795 (power9-mul-compare): Update insn latency, bypass latency and remove
33798 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33800 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33802 * config/aarch64/aarch64.md
33803 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33804 aarch64_nopcrelative_literal_loads.
33805 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33807 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33809 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33810 TARGET_LOONGSON_3A.
33811 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33813 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33816 * config.gcc (supported_defaults): Add lxc1-sxc1.
33817 (with_lxc1_sxc1): Add validation.
33818 (all_defaults): Add lxc1-sxc1.
33819 * config/mips/mips.opt (mlxc1-sxc1): New option.
33820 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33822 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33823 __mips_no_lxc1_sxc1.
33824 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33825 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33826 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33828 2017-01-19 Richard Biener <rguenther@suse.de>
33830 PR tree-optimization/72488
33831 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33832 sure to restore SSA info.
33833 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33835 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33837 PR rtl-optimization/79121
33838 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33839 of the inner type when shifting an extended value.
33841 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33844 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33845 interposable aliases.
33847 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33850 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33851 Use the evmergelohi instruction.
33852 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33853 (mov_sitf_e500_subreg8_2_be): Likewise.
33854 (mov_sitf_e500_subreg12_2_le): Likewise.
33855 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33856 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33857 (mov_sitf_e500_subreg8_2_le): Likewise.
33858 (mov_sitf_e500_subreg12_2_be): Likewise.
33860 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33862 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33863 attribute from vecsimple to vecperm.
33864 (altivec_vbpermq2): Likewise.
33866 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33869 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33871 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33872 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33873 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33874 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33875 case where N arg is SIZE_MAX.
33876 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33877 (cmpstrsi): Add pattern.
33879 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33881 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33882 __builtin_vec_revb builtins.
33883 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33884 built-in functions to support generation of the ISA 3.0 XXBR<x>
33885 vector byte reverse instructions.
33886 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33887 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33888 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33889 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33890 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33891 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33892 (P9V_BUILTIN_VEC_REVB): Likewise.
33893 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33894 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33895 (p9_xxbrq_v16qi): Likewise.
33896 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33897 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33898 (p9_xxbrh_v8hi): Likewise.
33899 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33900 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33901 vec_revb built-in functions.
33903 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33905 PR rtl-optimization/78952
33906 * config/i386/i386.md (any_extract): New code iterator.
33907 (*insvqi_2): Use any_extract for source operand.
33908 (*insvqi_3): Use any_shiftrt for source operand.
33910 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33912 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33914 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33916 2017-01-18 Matthias Klose <doko@ubuntu.com>
33918 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33920 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33922 * config/rs6000/altivec.h (vec_bperm): Change #define.
33923 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33924 (altivec_vbpermq2): New define_insn.
33925 (altivec_vbpermd): Likewise.
33926 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33927 function interface.
33928 (VBPERMD): Likewise.
33929 (VBPERM): New polymorphic function interface.
33930 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33931 Add entries for P9V_BUILTIN_VEC_VBPERM.
33932 * doc/extend.texi: Add interfaces for vec_bperm.
33934 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33936 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33937 first letter of error messages.
33938 (s390_resolve_overloaded_builtin): Likewise.
33939 * config/s390/s390.c (s390_expand_builtin): Likewise.
33940 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33941 (s390_valid_target_attribute_inner_p): Likewise.
33942 * config/s390/s390.md ("tabort"): Likewise.
33944 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33946 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33947 (ISA_AVOID_DIV_HILO): New macro.
33948 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33949 (ISA_HAS_DDIV): Likewise.
33951 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33953 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33955 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33957 * doc/invoke.texi (fabi-version): Spelling fix.
33959 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33962 * doc/invoke.texi (fabi-version): Mention mangling fix for
33965 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33968 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33970 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33973 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33974 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33976 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33978 (stack_protect_test): Ditto.
33979 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33980 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33982 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33983 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33984 -mstack-protector-guard-offset=.
33985 (RS/6000 and PowerPC Options): Ditto.
33987 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33989 * config/i386/i386.h (MASK_CLASS_P): New define.
33990 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33991 there are no registers from different register sets also when
33992 mask registers are used. Update function comment.
33993 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33994 to (*k/*r) and (*k/*km) alternatives.
33996 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33998 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33999 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34000 (EH_RETURN_HANDLER_RTX): New define.
34001 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34002 Force frame pointer in EH return functions.
34003 (aarch64_expand_epilogue): Add barrier for eh_return.
34004 (aarch64_final_eh_return_addr): Remove.
34005 (aarch64_eh_return_handler_rtx): New function.
34006 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34008 (aarch64_eh_return_handler_rtx): New prototype.
34010 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34012 * config/rs6000/altivec.h (vec_rlmi): New #define.
34013 (vec_vrlnm): Likewise.
34014 (vec_rlnm): Likewise.
34015 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34016 (UNSPEC_VRLNM): Likewise.
34017 (VIlong): New mode iterator.
34018 (altivec_vrl<VI_char>mi): New define_insn.
34019 (altivec_vrl<VI_char>nm): Likewise.
34020 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34022 (VRLDNM): Likewise.
34023 (RLNM): New polymorphic function entry.
34024 (VRLWMI): New monomorphic function entry.
34025 (VRLDMI): Likewise.
34026 (RLMI): New polymorphic function entry.
34027 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34028 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34029 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34032 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34035 * dwarf2out.c (field_byte_offset): Restore the
34036 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34037 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34038 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34041 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34044 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34046 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34049 * dwarf2out.c (add_data_member_location_attribute): For constant
34050 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34051 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34052 DW_AT_byte_size attributes.
34054 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34056 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34057 after forcing to constant memory when the code model is medium.
34059 2017-01-17 Julia Koval <julia.koval@intel.com>
34062 * config/i386/avx512fintrin.h
34063 (_mm512_i32gather_ps): Change __addr type to void const*.
34064 (_mm512_mask_i32gather_ps): Ditto.
34065 (_mm512_i32gather_pd): Ditto.
34066 (_mm512_mask_i32gather_pd): Ditto.
34067 (_mm512_i64gather_ps): Ditto.
34068 (_mm512_mask_i64gather_ps): Ditto.
34069 (_mm512_i64gather_pd): Ditto.
34070 (_mm512_mask_i64gather_pd): Ditto.
34071 (_mm512_i32gather_epi32): Ditto.
34072 (_mm512_mask_i32gather_epi32): Ditto.
34073 (_mm512_i32gather_epi64): Ditto.
34074 (_mm512_mask_i32gather_epi64): Ditto.
34075 (_mm512_i64gather_epi32): Ditto.
34076 (_mm512_mask_i64gather_epi32): Ditto.
34077 (_mm512_i64gather_epi64): Ditto.
34078 (_mm512_mask_i64gather_epi64): Ditto.
34079 (_mm512_i32scatter_ps): Change __addr type to void*.
34080 (_mm512_mask_i32scatter_ps): Ditto.
34081 (_mm512_i32scatter_pd): Ditto.
34082 (_mm512_mask_i32scatter_pd): Ditto.
34083 (_mm512_i64scatter_ps): Ditto.
34084 (_mm512_mask_i64scatter_ps): Ditto.
34085 (_mm512_i64scatter_pd): Ditto.
34086 (_mm512_mask_i64scatter_pd): Ditto.
34087 (_mm512_i32scatter_epi32): Ditto.
34088 (_mm512_mask_i32scatter_epi32): Ditto.
34089 (_mm512_i32scatter_epi64): Ditto.
34090 (_mm512_mask_i32scatter_epi64): Ditto.
34091 (_mm512_i64scatter_epi32): Ditto.
34092 (_mm512_mask_i64scatter_epi32): Ditto.
34093 (_mm512_i64scatter_epi64): Ditto.
34094 (_mm512_mask_i64scatter_epi64): Ditto.
34095 * config/i386/avx512pfintrin.h
34096 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34097 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34098 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34099 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34100 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34101 (_mm512_prefetch_i32scatter_ps): Ditto.
34102 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34103 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34104 (_mm512_prefetch_i64scatter_pd): Ditto.
34105 (_mm512_prefetch_i64scatter_ps): Ditto.
34106 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34107 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34108 * config/i386/avx512vlintrin.h
34109 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34110 (_mm_mmask_i32gather_ps): Ditto.
34111 (_mm256_mmask_i32gather_pd): Ditto.
34112 (_mm_mmask_i32gather_pd): Ditto.
34113 (_mm256_mmask_i64gather_ps): Ditto.
34114 (_mm_mmask_i64gather_ps): Ditto.
34115 (_mm256_mmask_i64gather_pd): Ditto.
34116 (_mm_mmask_i64gather_pd): Ditto.
34117 (_mm256_mmask_i32gather_epi32): Ditto.
34118 (_mm_mmask_i32gather_epi32): Ditto.
34119 (_mm256_mmask_i32gather_epi64): Ditto.
34120 (_mm_mmask_i32gather_epi64): Ditto.
34121 (_mm256_mmask_i64gather_epi32): Ditto.
34122 (_mm_mmask_i64gather_epi32): Ditto.
34123 (_mm256_mmask_i64gather_epi64): Ditto.
34124 (_mm_mmask_i64gather_epi64): Ditto.
34125 (_mm256_i32scatter_ps): Change __addr type to void*.
34126 (_mm256_mask_i32scatter_ps): Ditto.
34127 (_mm_i32scatter_ps): Ditto.
34128 (_mm_mask_i32scatter_ps): Ditto.
34129 (_mm256_i32scatter_pd): Ditto.
34130 (_mm256_mask_i32scatter_pd): Ditto.
34131 (_mm_i32scatter_pd): Ditto.
34132 (_mm_mask_i32scatter_pd): Ditto.
34133 (_mm256_i64scatter_ps): Ditto.
34134 (_mm256_mask_i64scatter_ps): Ditto.
34135 (_mm_i64scatter_ps): Ditto.
34136 (_mm_mask_i64scatter_ps): Ditto.
34137 (_mm256_i64scatter_pd): Ditto.
34138 (_mm256_mask_i64scatter_pd): Ditto.
34139 (_mm_i64scatter_pd): Ditto.
34140 (_mm_mask_i64scatter_pd): Ditto.
34141 (_mm256_i32scatter_epi32): Ditto.
34142 (_mm256_mask_i32scatter_epi32): Ditto.
34143 (_mm_i32scatter_epi32): Ditto.
34144 (_mm_mask_i32scatter_epi32): Ditto.
34145 (_mm256_i32scatter_epi64): Ditto.
34146 (_mm256_mask_i32scatter_epi64): Ditto.
34147 (_mm_i32scatter_epi64): Ditto.
34148 (_mm_mask_i32scatter_epi64): Ditto.
34149 (_mm256_i64scatter_epi32): Ditto.
34150 (_mm256_mask_i64scatter_epi32): Ditto.
34151 (_mm_i64scatter_epi32): Ditto.
34152 (_mm_mask_i64scatter_epi32): Ditto.
34153 (_mm256_i64scatter_epi64): Ditto.
34154 (_mm256_mask_i64scatter_epi64): Ditto.
34155 (_mm_i64scatter_epi64): Ditto.
34156 (_mm_mask_i64scatter_epi64): Ditto.
34157 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34158 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34159 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34160 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34161 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34162 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34163 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34164 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34165 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34166 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34167 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34168 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34169 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34170 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34171 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34172 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34173 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34174 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34175 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34176 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34177 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34178 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34179 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34180 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34181 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34182 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34183 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34184 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34185 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34186 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34187 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34188 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34189 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34190 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34191 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34192 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34193 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34194 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34195 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34196 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34197 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34198 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34199 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34200 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34201 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34202 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34203 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34204 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34205 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34206 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34207 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34208 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34209 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34210 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34211 definitions accordingly.
34213 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34214 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34217 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34220 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34223 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34224 hard regs for inner regno.
34226 2017-01-17 Martin Liska <mliska@suse.cz>
34229 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34230 assumption and add comment.
34232 2017-01-17 Nathan Sidwell <nathan@acm.org>
34234 * ipa-visibility.c (localize_node): New function, broken out of ...
34235 (function_and_variable_visibility): ... here. Call it.
34237 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34239 PR middle-end/77445
34240 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34241 correctly set frequency of oudgoing edge.
34242 (duplicate_thread_path): Fix profile updating.
34244 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34247 * configure.ac: Add GCC_BASE_VER.
34248 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34249 version from BASE-VER file.
34250 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34251 (gcc.o): Depend on $(BASEVER).
34252 * common.opt (dumpfullversion): New option.
34253 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34254 * doc/invoke.texi: Document -dumpfullversion.
34255 * doc/install.texi: Document --with-gcc-major-version-only.
34256 * configure: Regenerated.
34258 2017-01-17 Richard Biener <rguenther@suse.de>
34260 PR tree-optimization/71433
34261 * tree-vrp.c (register_new_assert_for): Merge same asserts
34262 on all incoming edges.
34263 (process_assert_insertions_for): Handle insertions at the
34266 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34268 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34269 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34271 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34274 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34277 2017-01-17 Alan Modra <amodra@gmail.com>
34280 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34281 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34282 symbolic stack limit when pic.
34284 2017-01-16 Martin Sebor <msebor@redhat.com>
34286 PR tree-optimization/78608
34287 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34289 2017-01-16 Jeff Law <law@redhat.com>
34292 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34293 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34294 for several include directories that may be relative to sysroot.
34295 * config/i386/x-mingw32 (gplus_includedir): Define.
34296 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34297 (native_system_includedir): Likewise.
34298 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34299 override if TARGET_SYSTEM_ROOT is defined.
34300 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34302 PR tree-optimization/79090
34303 PR tree-optimization/33562
34304 PR tree-optimization/61912
34305 PR tree-optimization/77485
34306 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34307 and computed trims into the dump file.
34309 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34311 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34313 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34316 * gimplify.c (gimplify_init_constructor): If want_value and
34317 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34321 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34322 sequence. Formatting fixes.
34323 (doloop_optimize): Formatting fixes.
34326 * gcc.c (debug_level_greater_than_spec_func): New function.
34327 (static_spec_functions): Add debug-level-gt spec function.
34328 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34330 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34331 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34332 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34333 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34334 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34335 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34337 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34339 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34340 QImode fixups to general and mask registers only.
34342 2017-01-16 Carl Love <cel@us.ibm.com>
34344 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34345 for built-in functions
34346 vector signed char vec_nabs (vector signed char)
34347 vector signed short vec_nabs (vector signed short)
34348 vector signed int vec_nabs (vector signed int)
34349 vector signed long long vec_nabs (vector signed long long)
34350 vector float vec_nabs (vector float)
34351 vector double vec_nabs (vector double)
34352 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34354 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34355 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34356 * doc/extend.texi: Update the documentation file for the new built-in
34359 2017-01-16 Martin Sebor <msebor@redhat.com>
34361 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34364 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34366 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34367 UNSPEC_VSX__XXSPLTD to require special splat handling.
34369 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34372 * system.h: Poison strndup.
34374 2017-01-16 Alan Modra <amodra@gmail.com>
34377 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34380 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34382 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34384 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34386 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34387 call recog here. Assert that INSN_CODE (insn) is non-negative.
34389 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34392 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34394 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34395 in the currently scheduled RTL fragment.
34397 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34399 PR rtl-optimization/78751
34400 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34403 2017-01-14 Jeff Law <law@redhat.com>
34405 PR tree-optimization/79090
34406 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34407 variable length stores.
34408 (compute_trims): Delete dead assignment to *trim_tail.
34409 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34412 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34414 PR rtl-optimization/78626
34415 PR rtl-optimization/78727
34416 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34417 of a block, and split such blocks after everything else is finished.
34419 2017-01-14 Alan Modra <amodra@gmail.com>
34422 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34423 target legitimate_combined_insn.
34424 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34425 (rs6000_legitimate_combined_insn): New function.
34426 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34428 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34429 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34430 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34432 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34434 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34436 2017-01-13 Jeff Law <law@redhat.com>
34438 PR tree-optimization/33562
34439 PR tree-optimization/61912
34440 PR tree-optimization/77485
34441 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34443 (delete_dead_assignment): Likewise.
34444 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34445 statement to delete_dead_call and delete_dead_assignment.
34447 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34450 * substring-locations.c (format_warning_va): Strengthen case 1 so
34451 that both endpoints of the substring must be within the format
34452 range for just the substring to be printed.
34454 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34456 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34457 * config/i386/i386.c (ix86_target_string): Add missing options
34458 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34459 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34460 flags_other and ix86_target_other to flags2_other. Display unknown
34462 (ix86_valid_target_attribute_inner_p): Add missing options and
34463 reorder options by implied ISAs, as in ix86_target_string.
34465 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34467 * hash-table.h (hash_table::too_empty_p): New function.
34468 (hash_table::expand): Use it.
34469 (hash_table::traverse): Likewise.
34470 (hash_table::empty_slot): Use sizeof (value_type) instead of
34471 sizeof (PTR) to convert bytes to elements. Shrink the table
34472 if the current size is excessive for the current number of
34475 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34477 * ira-costs.c (record_reg_classes): Break from the inner loop
34478 early once alt_fail is known to be true. Update outer loop
34479 handling accordingly.
34481 2017-01-13 Jeff Law <law@redhat.com>
34483 * tree-ssa-dse.c (decrement_count): New function.
34484 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34485 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34486 when we know the partially dead statement is a mem* function.
34488 PR tree-optimization/61912
34489 PR tree-optimization/77485
34490 * tree-ssa-dse.c: Include expr.h.
34491 (maybe_trim_constructor_store): New function.
34492 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34494 PR tree-optimization/33562
34495 PR tree-optimization/61912
34496 PR tree-optimization/77485
34497 * doc/invoke.texi: Document new dse-max-object-size param.
34498 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34499 * tree-ssa-dse.c: Include params.h.
34500 (dse_store_status): New enum.
34501 (initialize_ao_ref_for_dse): New, partially extracted from
34503 (valid_ao_ref_for_dse, normalize_ref): New.
34504 (setup_live_bytes_from_ref, compute_trims): Likewise.
34505 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34506 (maybe_trim_partially_dead_store): Likewise.
34507 (maybe_trim_complex_store): Likewise.
34508 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34509 Track what bytes live from the original store. Return tri-state
34510 for dead, partially dead or live.
34511 (dse_dom_walker): Add constructor, destructor and new private members.
34512 (delete_dead_call, delete_dead_assignment): New extracted from
34514 (dse_optimize_stmt): Make a member of dse_dom_walker.
34515 Use initialize_ao_ref_for_dse.
34517 PR tree-optimization/33562
34518 PR tree-optimization/61912
34519 PR tree-optimization/77485
34520 * sbitmap.h (bitmap_count_bits): Prototype.
34521 (bitmap_clear_range, bitmap_set_range): Likewise.
34522 * sbitmap.c (bitmap_clear_range): New function.
34523 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34525 2017-01-13 Martin Liska <mliska@suse.cz>
34528 * function.c (set_cfun): Add new argument force.
34529 * function.h (set_cfun): Likewise.
34530 * ipa-inline-transform.c (inline_call): Use the function when
34531 strict alising from is dropped for function we inline to.
34533 2017-01-13 Richard Biener <rguenther@suse.de>
34535 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34536 for dumping GIMPLE INTEGER_CSTs.
34538 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34540 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34541 to 201112L since C++17.
34543 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34546 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34547 if -fsanitize=kernel-address is present.
34549 2017-01-13 Richard Biener <rguenther@suse.de>
34551 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34552 as _Literal ( type ) number in case usual suffixes do not
34553 preserve all information.
34555 2017-01-13 Richard Biener <rguenther@suse.de>
34557 PR tree-optimization/77283
34558 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34559 and ssa-iterators.h.
34560 (is_feasible_trace): Implement a cost model based on joiner
34563 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34566 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34567 char or short to __float128/_Float128 directly.
34569 2017-01-12 Martin Sebor <msebor@redhat.com>
34571 to -Wformat-overflow.
34572 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34573 (min_bytes_remaining): Same.
34574 (get_string_length): Same.
34575 (format_string): Same.
34576 (format_directive): Same.
34578 (pass_sprintf_length::handle_gimple_call): Same.
34580 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34582 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34583 info.nowrite calls with no lhs that can't throw. Return bool
34584 whether gsi_remove has been called or not.
34585 (pass_sprintf_length::handle_gimple_call): Return bool whether
34586 try_substitute_return_value called gsi_remove. Formatting fix.
34587 (pass_sprintf_length::execute): Don't use gsi_remove if
34588 handle_gimple_call returned true.
34591 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34592 be removed due to side-effects, don't remove following barrier nor
34593 turn the successor edge into fallthru edge.
34595 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34598 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34599 element-reversing loads and stores as not swappable.
34601 2017-01-12 Nathan Sidwell <nathan@acm.org>
34602 Nicolai Stange <nicstange@gmail.com>
34604 * combine.c (try_combine): Don't ignore result of overlap checking
34605 loop. Combine overlap & asm check into single loop.
34607 2017-01-12 Richard Biener <rguenther@suse.de>
34609 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34610 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34612 2017-01-12 Richard Biener <rguenther@suse.de>
34614 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34615 and TS_TARGET_OPTION directly derive from TS_BASE.
34616 * tree-core.h (tree_optimization_option): Derive from tree_base.
34617 (tree_target_option): Likewise.
34619 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34621 * config/i386/i386.c (memory_address_length): Increase len
34622 only when rip_relative_addr_p returns false.
34624 2017-01-11 Julia Koval <julia.koval@intel.com>
34626 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34627 (OPTION_MASK_ISA_SGX_SET): New.
34628 (ix86_handle_option): Handle OPT_msgx.
34629 * config.gcc: Added sgxintrin.h.
34630 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34631 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34632 * config/i386/i386.c (ix86_target_string): Add -msgx.
34634 (ix86_option_override_internal): Handle new options.
34635 (ix86_valid_target_attribute_inner_p): Add sgx.
34636 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34637 * config/i386/i386.opt: Add msgx.
34638 * config/i386/sgxintrin.h: New file.
34639 * config/i386/x86intrin.h: Add sgxintrin.h.
34641 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34644 * fold-const.c (maybe_nonzero_address): Return 1 for function
34646 (tree_single_nonzero_warnv_p): Don't handle function local objects
34650 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34653 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34656 * opts.c: Include "spellcheck.h"
34657 (struct string_fragment): New struct.
34658 (struct edit_distance_traits<const string_fragment &>): New
34660 (get_closest_sanitizer_option): New function.
34661 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34663 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34665 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34667 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34668 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34669 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34670 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34671 for initial die_offset if dwarf_split_debug_info.
34672 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34673 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34675 (output_skeleton_debug_sections): Formatting fix. Use
34676 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34677 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34679 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34681 * config/arm/cortex-a53.md: Add bypasses for
34682 cortex_a53_r2f_cvt.
34683 (cortex_a53_r2f): Only use for transfers.
34684 (cortex_a53_f2r): Likewise.
34685 (cortex_a53_r2f_cvt): Add reservation for conversions.
34686 (cortex_a53_f2r_cvt): Likewise.
34688 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34690 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34691 to all inlined functions, change static to extern.
34693 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34696 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34698 (arm_assemble_integer): Likewise.
34700 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34702 * config.gcc: Use new awk script to check CPU, FPU and architecture
34703 parameters for --with-... options.
34704 * config/arm/parsecpu.awk: New file
34705 * config/arm/arm-cpus.in: New file.
34706 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34708 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34710 * config/arm/t-arm: Update dependency rules.
34711 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34712 of processing .def files.
34713 * config/arm/genopt.sh: Deleted.
34714 * config/arm/gentune.sh: Deleted.
34715 * config/arm/arm-cores.def: Deleted.
34716 * config/arm/arm-arches.def: Deleted.
34717 * config/arm/arm-fpus.def: Deleted.
34718 * config/arm/arm-tune.md: Regenerated.
34719 * config/arm/arm-tables.opt: Regenerated.
34720 * config/arm/arm-cpu.h: New generated file.
34721 * config/arm/arm-cpu-data.h: New generated file.
34722 * config/arm/arm-cpu-cdata.h: New generated file.
34724 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34727 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34729 (input_varpool_node): Unpack dynamically_initialized bit.
34731 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34733 PR rtl-optimization/79032
34734 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34735 the alignment of the adjusted memory reference against that of MODE,
34736 instead of the alignment of the original memory reference.
34738 2017-01-11 Martin Jambor <mjambor@suse.cz>
34740 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34742 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34743 decorated functions.
34745 2017-01-11 Richard Biener <rguenther@suse.de>
34747 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34748 set range/nonnull info for PHI results. Do not set it on
34749 stmts marked for removal.
34751 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34753 * expr.c (store_field): In the bitfield case, fetch the return value
34754 from the registers before applying a single big-endian adjustment.
34755 Always do a final load for a BLKmode value not larger than a word.
34757 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34760 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34761 that we correctly handle column numbers greater than
34762 LINE_MAP_MAX_COLUMN_NUMBER.
34764 2017-01-10 Martin Sebor <msebor@redhat.com>
34766 PR middle-end/78245
34767 * gimple-ssa-sprintf.c (get_destination_size): Call
34768 {init,fini}object_sizes.
34769 * tree-object-size.c (addr_object_size): Adjust.
34770 (pass_through_call): Adjust.
34771 (pass_object_sizes::execute): Adjust.
34772 * tree-object-size.h (fini_object_sizes): Declare.
34774 2017-01-10 Martin Sebor <msebor@redhat.com>
34776 PR tree-optimization/78775
34777 * builtins.c (get_size_range): Move...
34778 * calls.c: ...to here.
34779 (alloc_max_size): Accept zero argument.
34780 (operand_signed_p): Remove.
34781 (maybe_warn_alloc_args_overflow): Call get_size_range.
34782 * calls.h (get_size_range): Declare.
34784 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34786 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34787 from TI's devices.csv file as of September 2016.
34788 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34790 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34792 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34793 * doc/invoke.texi: Likewise.
34794 * doc/md.texi: Likewise.
34795 * doc/objc.texi: Likewise.
34797 2017-01-10 Joshua Conner <joshconner@google.com>
34799 * config/arm/fuchsia-elf.h: New file.
34800 * config/fuchsia.h: New file.
34801 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34802 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34804 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34806 2016-01-10 Richard Biener <rguenther@suse.de>
34808 PR tree-optimization/79034
34809 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34810 Propagate out degenerate PHIs in the joiner.
34812 2017-01-10 Martin Liska <mliska@suse.cz>
34814 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34815 (sort_congruence_classes_by_decl_uid): Likewise.
34816 (sort_congruence_class_groups_by_decl_uid): Likewise.
34817 (sem_item_optimizer::merge_classes): Sort class, groups in these
34818 classes and members in the groups by DECL_UID of declarations.
34819 This would make merge operations stable.
34821 2017-01-10 Martin Liska <mliska@suse.cz>
34823 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34824 usage of m_classes_vec.
34825 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34826 (sem_item_optimizer::get_group_by_hash): Likewise.
34827 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34828 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34829 (sem_item_optimizer::verify_classes): Likewise.
34830 (sem_item_optimizer::process_cong_reduction): Likewise.
34831 (sem_item_optimizer::dump_cong_classes): Likewise.
34832 (sem_item_optimizer::merge_classes): Likewise.
34833 * ipa-icf.h (congruence_class_hash): Rename from
34834 congruence_class_group_hash. Remove declaration of m_classes_vec.
34836 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34838 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34839 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34840 * config.gcc: Add avx512vpopcntdqintrin.h.
34841 * config/i386/avx512vpopcntdqintrin.h: New.
34842 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34843 * config/i386/i386-builtin-types.def: Add new types.
34844 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34845 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34846 __builtin_ia32_vpopcountq_v8di_mask): New.
34847 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34848 __AVX512VPOPCNTDQ__.
34849 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34850 (PTA_AVX512VPOPCNTDQ): Define.
34851 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34852 TARGET_AVX512VPOPCNTDQ_P): Define.
34853 * config/i386/i386.opt: Add mavx512vpopcntdq.
34854 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34855 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34857 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34859 PR middle-end/77484
34860 * predict.def (PRED_CALL): Set to 67.
34862 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34864 * expr.c (store_field): In the bitfield case, if the value comes from
34865 a function call and is of an aggregate type returned in registers, do
34866 not modify the field mode; extract the value in all cases if the mode
34867 is BLKmode and the size is not larger than a word.
34869 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34872 * config/i386/cpuid.h: Fix undefined behavior.
34874 2017-01-04 Jeff Law <law@redhat.com>
34876 PR tree-optimization/79007
34877 PR tree-optimization/67955
34878 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34879 conservative for pt.null when flag_non_call_exceptions is on.
34881 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34883 PR translation/79019
34884 PR translation/79020
34885 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34886 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34887 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34889 * config/avr/avr.opt (maccumulate-args): Likewise.
34890 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34891 * common.opt (freport-bug): Likewise.
34892 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34893 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34894 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34895 translatable string.
34896 * config/i386/i386.c (function_value_32): Likewise.
34897 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34898 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34900 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34901 * common/config/msp430/msp430-common.c (msp430_handle_option):
34903 * symtab.c (symtab_node::verify_base): Likewise.
34904 * opts.c (set_debug_level): Likewise.
34905 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34906 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34907 missing whitespace to translatable strings.
34908 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34909 * config/sh/superh.h: Likewise.
34910 * config/i386/xopintrin.h: Likewise.
34911 * config/i386/znver1.md: Likewise.
34912 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34913 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34914 * double-int.h (struct double_int): Likewise.
34915 * double-int.c (div_and_round_double): Likewise.
34916 * wide-int.cc: Likewise.
34917 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34918 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34919 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34920 (crossjumps_occurred): ... this.
34921 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34924 PR tree-optimization/78899
34925 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34926 returning bool return struct loop *, NULL for failure and the new
34928 (versionable_outer_loop_p): Don't version outer loop if it has
34929 dont_vectorized bit set.
34930 (tree_if_conversion): When versioning outer loop, ensure
34931 tree_if_conversion is performed also on the inner loop of the
34932 non-vectorizable outer loop copy.
34933 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34934 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34935 prevent vectorization of it.
34936 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34937 the outer loop vectorization of the non-scalar version is attempted
34938 before vectorization of the inner loop in scalar version. If
34939 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34940 vectorization of its inner loop.
34941 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34942 has 2 inner loops, rename also on edges from bb whose single pred
34943 is outer_loop->header. Fix typo in function comment.
34945 2017-01-09 Martin Sebor <msebor@redhat.com>
34948 * asan.c (asan_emit_stack_protection): Increase local buffer size
34949 to avoid snprintf truncation warning.
34951 2017-01-09 Andrew Pinski <apinski@cavium.com>
34953 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34954 to reference thunderx2t99 for the tuning structure
34955 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34957 (thunderx2t99_extra_costs): This.
34958 * config/aarch64/aarch64-tune.md: Regenerate.
34959 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34960 (vulcan_addrcost_table): This.
34961 (vulcan_regmove_cost): Rename to ...
34962 (thunderx2t99_regmove_cost): This.
34963 (vulcan_vector_cost): Rename to ...
34964 (thunderx2t99_vector_cost): this.
34965 (vulcan_branch_cost): Rename to ...
34966 (thunderx2t99_branch_cost): This.
34967 (vulcan_tunings): Rename to ...
34968 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34969 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34971 2017-01-09 Martin Jambor <mjambor@suse.cz>
34975 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34976 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34977 (propagate_vr_accross_jump_function): Use the above function for all
34978 value range computations for pass-through jump functions and type
34979 converasion from explicit value range values.
34980 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34981 parameters from TYPE_ARG_TYPES.
34982 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34983 (ipa_write_node_info): Stream type of the actual argument.
34984 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34986 2017-01-09 Martin Liska <mliska@suse.cz>
34989 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34990 (lookup_compiler): Do not show error message with have_E.
34992 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34994 PR tree-optimization/78938
34995 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34996 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34997 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34998 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35001 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35003 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35006 2017-01-09 Richard Biener <rguenther@suse.de>
35008 PR tree-optimization/78997
35009 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35010 name condition properly.
35012 2017-01-09 Richard Biener <rguenther@suse.de>
35015 * dwarf2out.c (is_cxx): New overload with context.
35016 (is_naming_typedef_decl): Use it.
35018 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35020 * invoke.texi (Option Summary): Correct spacing in option lists
35021 and add line breaks to fix over-long lines.
35023 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35025 PR middle-end/17660
35027 * extend.texi (Common Variable Attributes): Add xref to GCC
35028 Internals manual to explain mode attribute keywords.
35030 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35033 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35034 and Preprocessor Options.
35035 (Options for Linking): Document -pthread here....
35036 (RS/6000 and PowerPC Options): ...not here.
35037 (Solaris 2 Options): ...or here.
35038 * doc/cppopts.texi: Document -pthread.
35040 2017-01-08 Martin Sebor <msebor@redhat.com>
35042 PR middle-end/77708
35043 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35044 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35045 New member functions.
35046 (format_directive): Used them.
35048 (pass_sprintf_length::handle_gimple_call): Same.
35049 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35050 to avoid truncation for any argument.
35051 (extract_affine_mul): Same.
35052 * tree.c (get_file_function_name): Same.
35054 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35056 PR middle-end/77484
35057 * predict.def (PRED_INDIR_CALL): Set to 86.
35059 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35061 PR preprocessor/54124
35062 * doc/cppopts.texi: Reformat -d subtable to list the full name
35063 of the options. Add cross-reference to the docs for the general
35064 compiler -d options.
35065 * doc/invoke.texi (Developer Options): Add cross-reference to the
35066 preprocessor-specific -d option documentation.
35068 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35070 PR preprocessor/13498
35071 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35072 redudant material, and reflect new command-line options.
35073 (System Headers): Likewise.
35075 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35077 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35078 -isystem, and -idirafter. Copy-edit.
35079 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35080 default for -ftrack-macro-expansion. Delete obsolete and
35081 badly-formatted implementation details about -fdebug-cpp output.
35082 * doc/cppwarnopts.texi: Copy-edit.
35084 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35087 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35088 that the transition from a max line width >= 1<<10 to narrower
35089 lines works correctly.
35091 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35093 * doc/options.texi (PerFunction): New.
35094 * opt-functions.awk (switch_flags): Map both Optimization and
35095 PerFunction to CL_OPTIMIZATION.
35096 * opth-gen.awk: Test for PerFunction flag along with
35098 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35099 it only when the latter is present. Skip those that don't in
35100 the hash function generator.
35101 * common.opt (fvar-tracking): Mark as PerFunction instead of
35103 (fvar-tracking-assignments): Likewise.
35104 (fvar-tracking-assignments-toggle): Likewise.
35105 (fvar-tracking-uninit): Likewise.
35107 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35109 PR translation/79018
35110 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35113 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35116 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35117 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35118 TARGET_LONG_JUMP_TABLE_OFFSETS.
35119 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35120 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35121 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35122 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35123 * config/m68k/m68k.md (tablejump expander): Likewise.
35124 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35125 TARGET_LONG_JUMP_TABLE_OFFSETS.
35126 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35127 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35129 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35130 David Holsgrove <david.holsgrove@xilinx.com>
35132 * common/config/microblaze/microblaze-common.c
35133 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35134 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35136 * config/microblaze/microblaze.c (microblaze_must_save_register)
35137 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35139 (microblaze_eh_return): New function.
35140 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35141 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35142 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35143 * config/microblaze/microblaze.md (eh_return): New pattern.
35145 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35147 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35148 GCC_DIAGNOSTIC_STRINGIFY): Define.
35150 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35152 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35154 * config/arm/arm.md (<mcrr>): New.
35156 * config/arm/arm.c (arm_arch5te): New.
35157 (arm_option_override): Set arm_arch5te.
35158 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35160 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35161 (arm_mcrr_qualifiers): ... this. New.
35162 (MRRC_QUALIFIERS): Define to...
35163 (arm_mrrc_qualifiers): ... this. New.
35164 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35166 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35167 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35168 (MRRCI, mrrc, MRRC): New.
35169 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35170 VUNSPEC_MRRC2): New.
35172 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35174 * config/arm/arm.md (<mcr>): New.
35176 * config/arm/arm.c (arm_coproc_builtin_available): Add
35177 support for mcr, mrc, mcr2 and mrc2.
35178 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35179 (arm_mcr_qualifiers): ... this. New.
35180 (MRC_QUALIFIERS): Define to ...
35181 (arm_mrc_qualifiers): ... this. New.
35182 (MCR_QUALIFIERS): Define to ...
35183 (arm_mcr_qualifiers): ... this. New.
35184 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35186 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35187 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35188 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35189 VUNSPEC_MRC2): New.
35191 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35193 * config/arm/arm.md (*ldc): New.
35197 * config/arm/arm.c (arm_coproc_builtin_available): Add
35198 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35199 (arm_coproc_ldc_stc_legitimate_address): New.
35200 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35201 'qualifier_const_pointer'.
35202 (LDC_QUALIFIERS): Define to...
35203 (arm_ldc_qualifiers): ... this. New.
35204 (STC_QUALIFIERS): Define to...
35205 (arm_stc_qualifiers): ... this. New.
35206 * config/arm/arm-protos.h
35207 (arm_coproc_ldc_stc_legitimate_address): New.
35208 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35209 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35210 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35211 stc2, stcl, stc2l): New.
35212 * config/arm/constraints.md (Uz): New.
35213 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35214 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35215 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35216 VUNSPEC_STC2L): New.
35218 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35220 * config/arm/arm.md (<cdp>): New.
35221 * config/arm/arm.c (neon_const_bounds): Rename this ...
35222 (arm_const_bounds): ... this.
35223 (arm_coproc_builtin_available): New.
35224 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35225 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35226 (CDP_QUALIFIERS): Define to...
35227 (arm_cdp_qualifiers): ... this. New.
35229 (arm_expand_builtin_args): Add case for 6 arguments.
35230 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35231 (arm_const_bounds): ... this.
35232 (arm_coproc_builtin_available): New.
35233 * config/arm/arm_acle.h (__arm_cdp): New.
35235 * config/arm/arm_acle_builtins.def (cdp): New.
35237 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35238 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35239 'arm_const_bounds'.
35240 * config/arm/types.md (coproc): New.
35241 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35242 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35243 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35244 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35246 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35248 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35249 (UBINOP_QUALIFIERS): New.
35251 (acle_builtin_data): New. Change comment.
35252 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35253 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35254 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35255 arm_acle_builtins.def.
35256 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35257 (arm_init_acle_builtins): New.
35258 (CRC32_BUILTIN): Remove.
35259 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35260 crc32cb, crc32ch and crc32cw.
35261 (arm_init_crc32_builtins): Remove.
35262 (arm_init_builtins): Use arm_init_acle_builtins rather
35263 than arm_init_crc32_builtins.
35264 (arm_expand_acle_builtin): New.
35265 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35266 * config/arm/arm_acle_builtins.def: New.
35268 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35270 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35271 (arm_builtin_datum): ... this.
35272 (arm_init_neon_builtin): Rename to ...
35273 (arm_init_builtin): ... this. Add a new parameters PREFIX
35274 and USE_SIG_IN_NAME.
35275 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35276 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35277 'arm_builtin_datum'.
35278 (arm_init_vfp_builtins): Likewise.
35279 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35280 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35281 (arm_expand_neon_args): Rename to ...
35282 (arm_expand_builtin_args): ... this. Rename builtin_arg
35283 enum values and differentiate between ARG_BUILTIN_MEMORY
35284 and ARG_BUILTIN_NEON_MEMORY.
35285 (arm_expand_neon_builtin_1): Rename to ...
35286 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35287 values, arm_expand_builtin_args and add bool parameter NEON.
35288 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35289 (arm_expand_vfp_builtin): Likewise.
35290 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35292 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35294 PR middle-end/77484
35295 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35296 * predict.c (tree_estimate_probability_bb): Reverse direction of
35297 polymorphic call predictor.
35299 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35301 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35302 (determine_pass_name_match): ...this new function and...
35303 (should_skip_pass_p): ...this new function.
35305 2017-01-06 Nathan Sidwell <nathan@acm.org>
35307 * ipa-visibility.c (function_and_variable_visibility): Reformat
35308 comments and long lines. Remove extrneous if.
35309 * symtab.c (symtab_node::make_decl_local): Fix code format.
35310 (symtab_node::set_section_for_node): Fix comment typo.
35312 2017-01-06 Martin Liska <mliska@suse.cz>
35315 * lra-constraints.c: Rename invariant to lra_invariant.
35316 * predict.c (set_even_probabilities): Initialize e to NULL.
35318 2017-01-05 Martin Sebor <msebor@redhat.com>
35320 PR tree-optimization/78910
35321 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35322 (format_integer): Correct off-by-one error in the handling
35323 of precision with negative numbers in signed conversions..
35325 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35327 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35329 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35331 PR tree-optimization/71016
35332 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35333 factor_out_conditional_conversion. Formatting fix.
35334 (factor_out_conditional_conversion): Add cond_stmt argument.
35335 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35336 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35339 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35341 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35342 read-rtl-function.o, and selftest-rtl.o.
35343 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35344 (selftest::aarch64_test_loading_full_dump): New function.
35345 (selftest::aarch64_run_selftests): New function.
35346 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35347 selftest::aarch64_run_selftests.
35348 * config/i386/i386.c
35349 (selftest::ix86_test_loading_dump_fragment_1): New function.
35350 (selftest::ix86_test_loading_call_insn): New function.
35351 (selftest::ix86_test_loading_full_dump): New function.
35352 (selftest::ix86_test_loading_unspec): New function.
35353 (selftest::ix86_run_selftests): Call the new functions.
35354 * emit-rtl.c (maybe_set_max_label_num): New function.
35355 * emit-rtl.h (maybe_set_max_label_num): New decl.
35356 * function.c (instantiate_decls): Guard call to
35357 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35358 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35360 * gensupport.c (gen_reader::gen_reader): Pass "false"
35361 for new "compact" param of rtx_reader.
35362 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35363 rather than an empty string for NULL strings.
35364 * read-md.c: Potentially include config.h rather than bconfig.h.
35365 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35366 (have_error): New global, copied from errors.c.
35367 (md_reader::read_name): Rename to...
35368 (md_reader::read_name_1): ...this, adding "out_loc" param,
35369 and converting "missing name or number" to returning false, rather
35371 (md_reader::read_name): Reimplement in terms of read_name_1.
35372 (md_reader::read_name_or_nil): New function.
35373 (md_reader::read_string): Handle "(nil)" by returning NULL.
35374 (md_reader::md_reader): Add new param "compact".
35375 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35376 (md_reader::read_file): New method.
35377 * read-md.h (md_reader::md_reader): Add new param "compact".
35378 (md_reader::read_file): New method.
35379 (md_reader::is_compact): New accessor.
35380 (md_reader::read_name): Convert return type from void to file_location.
35381 (md_reader::read_name_or_nil): New decl.
35382 (md_reader::read_name_1): New decl.
35383 (md_reader::m_compact): New field.
35384 (noop_reader::noop_reader): Pass "false" for new "compact" param
35386 (rtx_reader::rtx_reader): Add new "compact" param.
35387 (rtx_reader::read_rtx_operand): Make virtual and convert return
35388 type from void to rtx.
35389 (rtx_reader::read_until): New decl.
35390 (rtx_reader::handle_any_trailing_information): New virtual function.
35391 (rtx_reader::postprocess): New virtual function.
35392 (rtx_reader::finalize_string): New virtual function.
35393 (rtx_reader::m_in_call_function_usage): New field.
35394 (rtx_reader::m_reuse_rtx_by_id): New field.
35395 * read-rtl-function.c: New file.
35396 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35397 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35398 (selftest::verify_three_block_rtl_cfg): New decl.
35399 * read-rtl-function.h: New file.
35400 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35401 For host, include function.h, memmodel.h, and emit-rtl.h.
35402 (one_time_initialization): New function.
35403 (struct compact_insn_name): New struct.
35404 (compact_insn_names): New array.
35405 (find_code): Handle insn codes in compact dumps.
35406 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35407 (bind_subst_iter_and_attr): Likewise.
35408 (add_condition_to_string): Likewise.
35409 (add_condition_to_rtx): Likewise.
35410 (apply_attribute_uses): Likewise.
35411 (add_current_iterators): Likewise.
35412 (apply_iterators): Likewise.
35413 (initialize_iterators): Guard usage of apply_subst_iterator with
35414 #ifdef GENERATOR_FILE.
35415 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35416 (md_reader::read_mapping): Likewise.
35417 (add_define_attr_for_define_subst): Likewise.
35418 (add_define_subst_attr): Likewise.
35419 (read_subst_mapping): Likewise.
35420 (check_code_iterator): Likewise.
35421 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35423 (one_time_initialization): New function.
35424 (rtx_reader::read_until): New method.
35425 (read_flags): New function.
35426 (parse_reg_note_name): New function.
35427 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35428 Handle reuse_rtx ids.
35429 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35430 Add parsing support for RTL dumps, mirroring the special-cases in
35431 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35432 values, and calling handle_any_trailing_information.
35433 (rtx_reader::read_rtx_operand): Convert return type from void
35434 to rtx, returning return_rtx. Handle case 'e'. Call
35435 finalize_string on XSTR and XTMPL fields.
35436 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35437 "(nil)" values were omitted. Call the postprocess vfunc on the
35439 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35440 class ctor. Initialize m_in_call_function_usage. Call
35441 one_time_initialization.
35442 * rtl-tests.c (selftest::test_uncond_jump): Call
35443 set_new_first_and_last_insn.
35444 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35445 * selftest-rtl.c: New file.
35446 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35447 (selftest::get_insn_by_uid): New decl.
35448 * selftest-run-tests.c (selftest::run_tests): Call
35449 read_rtl_function_c_tests.
35450 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35451 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35454 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35456 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35457 operands in a special way. Assert that pos+len <= mode precision.
35459 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35461 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35462 3 argument Alias with unlimited for the negative form.
35463 (fno-vect-cost-model): Removed.
35465 2017-01-05 Martin Liska <mliska@suse.cz>
35467 * hsa-gen.c (gen_hsa_divmod): New function.
35468 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35470 2017-01-05 Martin Liska <mliska@suse.cz>
35473 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35476 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35478 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35479 small constant length operands.
35481 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35483 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35484 between loop iterations.
35486 2017-01-05 Martin Liska <mliska@suse.cz>
35489 * gimplify.c (gimplify_decl_expr): Compare to
35490 asan_poisoned_variables instread of checking flags.
35491 (gimplify_target_expr): Likewise.
35492 (gimplify_expr): Likewise.
35493 (gimplify_function_tree): Conditionally initialize
35494 asan_poisoned_variables.
35496 2017-01-04 Jeff Law <law@redhat.com>
35498 PR tree-optimizatin/78812
35499 * rtl.h (contains_mem_rtx_p): Prototype.
35500 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35501 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35502 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35503 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35504 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35506 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35508 * input.c (assert_char_at_range): Default-initialize actual_range.
35510 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35512 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35513 to match the caller.
35515 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35517 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35518 insns after final jump in test to emit dummy move.
35520 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35522 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35523 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35525 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35527 * multiple_target.c (create_dispatcher_calls): Init e_next.
35528 * tree-ssa-loop-split.c (split_loop): Init border.
35529 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35532 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35537 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35538 (altivec_register_operand): Do not return true if the operand
35539 contains a SUBREG mixing SImode and SFmode.
35540 (vsx_register_operand): Likewise.
35541 (vsx_reg_sfsubreg_ok): New predicate.
35542 (vfloat_operand): Do not return true if the operand contains a
35543 SUBREG mixing SImode and SFmode.
35544 (vint_operand): Likewise.
35545 (vlogical_operand): Likewise.
35546 (gpc_reg_operand): Likewise.
35547 (int_reg_operand): Likewise.
35548 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35549 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35550 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35552 (rs6000_emit_move_si_sf_subreg): New helper function.
35553 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35554 fixup SUBREGs involving SImode and SFmode.
35555 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35556 numbers for the new peephole2 optimization.
35557 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35558 the GLIBC math library that do AND/IOR/XOR operations on single
35559 precision floating point.
35560 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35561 target macros to say whether we need to avoid SUBREGs mixing
35563 (TARGET_ALLOW_SF_SUBREG): Likewise.
35564 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35565 (UNSPEC_SI_FROM_SF): Likewise.
35566 (iorxor): Change spacing.
35567 (and_ior_xor): New iterator for AND, IOR, and XOR.
35568 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35569 (movdi_from_sf_zero_ext): Likewise.
35570 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35571 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35572 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35573 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35574 (fms<mode>4): Likewise.
35575 (fnma<mode>4): Likewise.
35576 (fnms<mode>4): Likewise.
35577 (nfma<mode>4): Likewise.
35578 (nfms<mode>4): Likewise.
35580 2017-01-04 Marek Polacek <polacek@redhat.com>
35583 * doc/invoke.texi: Document -Wpointer-compare.
35585 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35587 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35590 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35591 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35594 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35597 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35598 documentation of the powerpc_popcntb_ok attribute.
35599 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35600 code to issue warning messages if a requested CPU configuration is
35601 not supported by the binary (assembler and loader) toolchain.
35602 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35603 made to define a built-in function that has been disabled.
35604 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35605 made to define a built-in function that has been disabled.
35606 (altivec_init_builtins): Add comment explaining why definition
35607 of the DST built-in functions is not preceded by an assertion
35608 check. Add assertions to prevent ICE if attempts are made to
35609 define an altivec predicate or an abs* built-in function that has
35611 (htm_init_builtins): Add comment explaining why definition of the
35612 htm built-in functions is not preceded by an assertion check.
35614 2017-01-04 Jeff Law <law@redhat.com>
35616 PR tree-optimizatin/67955
35617 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35618 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35619 the points-to solution does not include pt_null. Use DECL_PT_UID
35622 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35624 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35625 Use gen_int_mode instead of gen_lopwart for const_int operands.
35627 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35629 PR tree-optimization/71563
35630 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35631 out of range value - has low bits known to be zero.
35633 2017-01-04 Alan Modra <amodra@gmail.com>
35635 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35636 * configure: Regenerate.
35637 * config.in: Regenerate.
35639 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35642 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35643 a substring of the message, but strcmp with the whole message. Ifdef
35644 ENABLE_NLS, translate the message first using dgettext.
35646 2017-01-03 Jeff Law <law@redhat.com>
35648 PR tree-optimizatin/78856
35649 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35650 (mark_threaded_blocks): Remove code to truncate thread paths that
35651 cross multiple loop headers. Instead invalidate the cached loop
35652 iteration information and handle case of a thread path walking
35653 into an irreducible region.
35655 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35658 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35659 assertions. Add support for doing the signbit if the IEEE 128-bit
35660 floating point value is in a GPR.
35661 * config/rs6000/rs6000.md (Fsignbit): Delete.
35662 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35663 Update the length attribute if the value is in a GPR.
35664 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35665 the sign or zero extension instruction, since the value is always 0/1.
35666 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35669 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35670 extracting SImode to a GPR register so that we can generate a
35671 store, limit the vector to be in a traditional Altivec register
35672 for the vextuwrx instruction.
35674 2017-01-03 Ian Lance Taylor <iant@google.com>
35676 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35678 2017-01-03 Martin Sebor <msebor@redhat.com>
35680 PR tree-optimization/78696
35681 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35682 precision. Use MPFR for %f for greater fidelity. Correct handling
35684 (pass_sprintf_length::compute_format_length): Set width and precision
35685 specified by asrerisk to void_node for vararg functions.
35686 (try_substitute_return_value): Adjust dump output.
35688 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35690 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35692 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35694 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35695 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35696 -mlra/-mno-lra was passed to the compiler.
35698 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35700 PR rtl-optimization/65618
35701 * emit-rtl.c (try_split): Move initialization of "before" and
35702 "after" to just before the call to emit_insn_after_setloc.
35704 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35706 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35708 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35710 * dwarf2out.c (gen_enumeration_type_die): When
35711 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35713 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35715 PR tree-optimization/78965
35716 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35717 Change first argument from const call_info & to call_info &. For %n
35718 set info.nowrite to false.
35720 PR middle-end/78901
35721 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35722 possibly throwing calls.
35724 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35725 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35726 and fns handling, rather than in a separate case SSA_NAME.
35728 2017-01-02 Jeff Law <law@redhat.com>
35730 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35731 fixes for first_period and second_period variables.
35733 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35736 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35737 (*insvqi_1): New insn pattern.
35738 (*insvqi_1_mem_rex64): Ditto.
35739 (*insvqi_2): Ditto.
35740 (*insvqi_3): Rename from *insvqi.
35742 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35744 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35746 * doc/cfg.texi (Edges): Remove reference to Java.
35747 (Maintaining the CFG): Ditto.
35749 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35751 PR middle-end/77674
35752 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35753 transparent aliases.
35755 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35757 PR middle-end/77484
35758 * predict.def (PRED_CALL): Update hitrate.
35759 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35760 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35761 into direct/indirect/polymorphic variants.
35763 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35765 Update copyright years.
35767 * gcc.c (process_command): Update copyright notice dates.
35768 * gcov-dump.c (print_version): Ditto.
35769 * gcov.c (print_version): Ditto.
35770 * gcov-tool.c (print_version): Ditto.
35771 * gengtype.c (create_file): Ditto.
35772 * doc/cpp.texi: Bump @copying's copyright year.
35773 * doc/cppinternals.texi: Ditto.
35774 * doc/gcc.texi: Ditto.
35775 * doc/gccint.texi: Ditto.
35776 * doc/gcov.texi: Ditto.
35777 * doc/install.texi: Ditto.
35778 * doc/invoke.texi: Ditto.
35780 Copyright (C) 2017 Free Software Foundation, Inc.
35782 Copying and distribution of this file, with or without modification,
35783 are permitted in any medium without royalty provided the copyright
35784 notice and this notice are preserved.