1 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
4 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
8 * vec.h (debug_helper): New function.
9 (DEFINE_DEBUG_VEC): New macro.
10 * hash-set.h (debug_helper): New function.
11 (DEFINE_DEBUG_HASH_SET): New macro.
12 * cfg.c (debug_slim (edge)): New function.
13 Call DEFINE_DEBUG_VEC for edges.
14 Call DEFINE_DEBUG_HASH_SET for edges.
15 * cfghooks.c (debug_slim (basic_block)): New function.
16 Call DEFINE_DEBUG_VEC for basic blocks.
17 Call DEFINE_DEBUG_HASH_SET for basic blocks.
18 * print-tree.c (debug_slim): New function to handle trees.
19 Call DEFINE_DEBUG_VEC for trees.
20 Call DEFINE_DEBUG_HASH_SET for trees.
21 (debug (vec<tree, va_gc>) &): Remove.
22 (debug (<vec<tree, va_gc>) *): Remove.
23 * print-rtl.c (debug_slim): New function to handle const_rtx.
24 Call DEFINE_DEBUG_VEC for rtx_def.
25 Call DEFINE_DEBUG_VEC for rtx_insn.
26 Call DEFINE_DEBUG_HASH_SET for rtx_def.
27 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
28 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
29 (debug (vec<rtx_insn *> *ptr): Remove.
30 (debug_insn_vector): Remove.
31 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
33 2017-11-20 Tom de Vries <tom@codesourcery.com>
35 PR rtl-optimization/82020
36 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
37 IF_THEN_ELSE condition.
39 2017-11-19 Jeff Law <law@redhat.com>
41 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
42 of degenerates resulting from ignoring an edge.
44 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
47 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
49 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
52 * profile-count.c (profile_count::to_sreal_scale): Fix return value
53 for uninitialied counts.
55 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
58 * tree-inline.c (estimate_num_insns): Set to 1 at least.
60 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
63 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane
66 2017-11-19 Tom de Vries <tom@codesourcery.com>
68 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
71 2017-11-19 Tom de Vries <tom@codesourcery.com>
73 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
76 2017-11-19 Tom de Vries <tom@codesourcery.com>
78 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
81 2017-11-19 Tom de Vries <tom@codesourcery.com>
83 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
85 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
87 2017-11-19 Tom de Vries <tom@codesourcery.com>
89 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
91 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
92 semicolon after MCORE_EXPORT_NAME call.
94 2017-11-19 Tom de Vries <tom@codesourcery.com>
97 * vmsdbgout.c (vmsdbgout_early_finish): New function.
98 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
100 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
102 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
103 (cgraph_edge::clone): Cleanup updating of profile.
104 * ipa-cp.c (update_profiling_info): Likewise.
105 * ipa-inline-transform.c (inline_transform): Likewise.
106 * ipa-inline.c (inline_small_functions): Add missing space to dump.
107 * ipa-split.c (execute_split_functions): Do not split when function
109 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
110 * profile-count.c (profile_count::dump): Add global0.
111 (profile_count::to_cgraph_frequency): Do not ICE when entry is
113 (profile_count::to_sreal_scale): Likewise.
114 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
115 (profile_count::combine_with_ipa_count): New function.
116 * profile-count.h (profile_guessed_global0adjusted): New.
117 (profile_count::adjusted_zero): New.
118 (profile_count::global0adjusted): New.
119 (profile_count::combine_with_ipa_count): New.
120 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
121 correct profile of return block of split functions.
122 (copy_cfg_body): Remove unused profile_count.
123 (copy_body): Likewise.
124 (expand_call_inline): Update.
125 (tree_function_versioning): Update.
127 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
129 * hash-set.h (hash_set::empty): New.
130 * tree-ssa-threadbackward.h: Delete.
131 * tree-ssa-threadbackward.c (class thread_jumps): New.
132 Move max_threaded_paths into class.
133 (fsm_find_thread_path): Remove arguments that are now in class.
134 (profitable_jump_thread_path): Rename to...
135 (thread_jumps::profitable_jump_thread_path): ...this.
136 (convert_and_register_jump_thread_path): Rename to...
137 (thread_jumps::convert_and_register_current_path): ...this.
138 (check_subpath_and_update_thread_path): Rename to...
139 (thread_jumps::check_subpath_and_update_thread_path): ...this.
140 (register_jump_thread_path_if_profitable): Rename to...
141 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
142 (handle_phi): Rename to...
143 (thread_jumps::handle_phi): ...this.
144 (handle_assignment): Rename to...
145 (thread_jumps::handle_assignment): ...this.
146 (fsm_find_control_statement_thread_paths): Rename to...
147 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
148 (find_jump_threads_backwards): Rename to...
149 (thread_jumps::find_jump_threads_backwards): ...this.
150 Initialize path local data.
151 (pass_thread_jumps::execute): Call find_jump_threads_backwards
152 from within thread_jumps class.
153 (pass_early_thread_jumps::execute): Same.
155 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
157 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
159 2017-11-17 Steve Ellcey <sellcey@cavium.com>
161 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
162 to canonical location.
164 2017-11-17 Steve Ellcey <sellcey@cavium.com>
167 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
169 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
171 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
173 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
174 rcrt1.o%s/grcrt1.o%s for -static-pie.
176 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
178 * i386.c (ix86_multiplication_cost, ix86_division_cost,
179 ix86_shift_rotate_cost): Break out from ...
180 (ix86_rtx_costs): ... here.
181 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
184 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
186 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
187 when entry block was promoted unlikely.
188 (estimate_bb_frequencies): Increase frequency scale.
189 * profile-count.h (profile_count): Export precision info.
191 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
193 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
194 disturbin profile of entry block.
196 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
198 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
201 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
203 * ipa-cp.c (update_profiling_info): Handle conversion to local
205 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
207 2017-11-17 Jeff Law <law@redhat.com>
209 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
210 from evrp_dom_walker class. Various methods moved into new class.
211 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
212 (evrp_range_analyzer::enter): New method.
213 (evrp_range_analyzer::leave): New method.
214 (evrp_dom_walker): Remove delegators no longer needed by this class.
215 Replace vr_values data member with evrp_range_analyzer
217 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
218 method extracted from evrp_dom_walker::before_dom_children.
219 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
220 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
222 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
223 Add private copy constructor and move assignment operators.
224 Privatize methods and class data where trivially possible.
225 (evrp_dom_walker::cleanup): New function, extracted from
226 execute_early_vrp. Simplify access to class data.
228 * vr-values.h (get_output_for_vrp): Prototype.
229 * vr-values.c (get_output_for_vrp): New function extracted from
230 vrp_visit_assignment_or_call and extract_range_from_stmt.
231 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
233 2017-11-17 Luis Machado <luis.machado@linaro.org>
235 * config/aarch64/aarch64.c
236 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
237 (qdf24xx_tunings) <autoprefetcher_model>: Set to
238 tune_params::AUTOPREFETCHER_WEAK.
240 2017-11-17 Tamar Christina <tamar.christina@arm.com>
243 * config/arm/arm.c (arm_valid_target_attribute_rec):
244 Parse "arch=" and "+<ext>".
245 (arm_valid_target_attribute_tree): Re-init global options.
246 (arm_option_override): Make non-static.
247 (arm_options_perform_arch_sanity_checks): Make errors fatal.
248 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
249 (__ARM_FEATURE_CRC32): Support undef.
250 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
251 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
253 2017-11-17 David Malcolm <dmalcolm@redhat.com>
255 * gdbinit.in (break-on-diagnostic): New command.
257 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
259 * config/i386/i386.c (ix86_expand_epilogue): Change simple
260 return to indirect jump for EH return if control-flow
261 protection is enabled. Change explicit 'false' argument in
262 pro_epilogue_adjust_stack with a value of
264 * config/i386/i386.md (simple_return_indirect_internal):
265 Remove SImode restriction to support 64-bit.
267 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
269 * combine.c (added_notes_insn): New.
270 (try_combine): Handle added_notes_insn like added_links_insn.
271 Rewrite return value code.
272 (distribute_notes): Set added_notes_insn to the earliest insn we added
275 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
277 PR rtl-optimization/82621
278 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
279 dest of one of those SETs is unused.
281 2017-11-17 Richard Biener <rguenther@suse.de>
284 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
285 * tree-pretty-print.c (dump_generic_node): Handle
286 annot_expr_parallel_kind.
287 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
288 * gimplify.c (gimple_boolify): Likewise.
290 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
292 * config.gcc (extra_headers): Add cet.h for x86 targets.
293 * config/i386/cet.h: New file.
294 * doc/install.texi: Add --enable-cet/--disable-cet.
296 2017-11-17 Richard Biener <rguenther@suse.de>
298 PR tree-optimization/83017
299 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
300 (gen_parallel_loop): Properly count iterations.
301 (parallelize_loops): Handle loop->can_be_parallel independent
302 of flag_loop_parallelize_all. Make static profitability test match
304 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
305 * invoke.texi (parloops-min-per-thread): Document.
307 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
309 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
310 upstreaming review comments.
312 2017-11-17 Tamar Christina <tamar.christina@arm.com>
314 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
315 with fast unaligned access.
316 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
318 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
320 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
321 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
323 2017-11-17 Richard Biener <rguenther@suse.de>
325 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
326 folding of references.
328 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
331 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
332 of replacing call to strncmp with corresponding call to strcmp when
335 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
337 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
338 option prefer-avx256 for skylake-avx512 configuration.
339 * config/i386/i386.c (ix86_option_override_internal): Ditto.
340 (get_builtin_code_for_version): Ditto.
342 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
343 Monk Chiang <sh.chiang04@gmail.com>
345 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
346 (FIXED_REGISTERS): Reserve more register numbers.
347 (CALL_USED_REGISTERS): Likewise.
348 (REG_ALLOC_ORDER): Likewise.
349 (REG_CLASS_CONTENTS): Likewise.
350 (REGISTER_NAMES): Likewise.
352 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
353 Kito Cheng <kito.cheng@gmail.com>
355 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
357 * config/nds32/iterators.md: Add vector mode iterators and attributes.
359 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
361 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
362 parameter list for vec_splats.
364 2017-11-16 Joseph Myers <joseph@codesourcery.com>
366 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
368 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
371 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
374 * doc/install.texi: Document incompatibility of Perl >=5.6.26
375 with the required version of automake 1.11.6.
377 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
379 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
381 (DU_C2_3_power9): Correct reservation combinations.
382 (FP_DIV_power9, VEC_DIV_power9): New.
383 (power9-alu): Split out rotate/shift...
384 (power9-rot): ...to here, correct dispatch resource.
385 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
387 (power9-fp): Correct latency.
388 (power9-sdiv): Add div/sqrt resource.
389 (power9-ddiv): Correct latency, add div/sqrt resource.
390 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
391 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
393 (power9-qpdiv, power9-qpmul): Adjust resource usage.
395 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
397 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
398 switch statement mapping KF built-ins to TF built-ins if we don't
399 have the proper ISA 3.0 assembler support.
401 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
403 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
404 (lower_emutls_function_body): Do not compute it.
406 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
408 * ipa-split.c (split_bb_info): Turn time to sreal.
409 (split_point): Likewise.
410 (dump_split_point): Likewise.
411 (fine_split_points): Likewise.
412 (execute_split_functions): Only zero split_bbs; turn time to sreals.
414 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
416 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
419 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
421 * predict.c (combine_predictions_for_bb): Preserve zero predicted
423 (expensive_function_p): Remove useless assert.
424 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
426 2017-11-16 Martin Sebor <msebor@redhat.com>
428 PR tree-optimization/82588
429 PR tree-optimization/82583
430 * tree-vrp.c (check_array_ref): Handle flexible array members,
431 string literals, and inner indices.
432 (search_for_addr_array): Add detail to diagnostics.
434 2017-11-16 Nathan Sidwell <nathan@acm.org>
438 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
439 (SET_DECL_ASSEMBLER_NAME): Forward to
440 overwrite_decl_assembler_name.
441 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
442 (overwrite_decl_assembler_name): Declare.
443 * tree.c (overwrite_decl_assembler_name): New.
444 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
445 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
446 (LANG_HOOKS_INITIALIZER): Add it.
447 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
448 * langhooks.c (lhd_set_decl_assembler_name): Use
449 SET_DECL_ASSEMBLER_NAME.
450 (lhd_overwrite_decl_assembler_name): Default implementation.
452 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
453 Jackson Woodruff <jackson.woodruff@arm.com>
455 PR tree-optimization/71026
456 * match.pd: Canonicalize constant multiplies in division.
458 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
460 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
461 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
462 Remove OPT_fomit_frame_pointer entry.
463 * common/config/arc/arc-common.c: Likewise.
464 * common/config/arm/arm-common.c: Likewise.
465 * common/config/avr/avr-common.c: Likewise.
466 * common/config/c6x/c6x-common.c: Likewise.
467 * common/config/cr16/cr16-common.c: Likewise.
468 * common/config/cris/cris-common.c: Likewise.
469 * common/config/epiphany/epiphany-common.c: Likewise.
470 * common/config/fr30/fr30-common.c: Likewise.
471 * common/config/frv/frv-common.c: Likewise.
472 * common/config/ia64/ia64-common.c: Likewise.
473 * common/config/iq2000/iq2000-common.c: Likewise.
474 * common/config/lm32/lm32-common.c: Likewise.
475 * common/config/m32r/m32r-common.c: Likewise.
476 * common/config/mcore/mcore-common.c: Likewise.
477 * common/config/microblaze/microblaze-common.c: Likewise.
478 * common/config/mips/mips-common.c: Likewise.
479 * common/config/mmix/mmix-common.c: Likewise.
480 * common/config/mn10300/mn10300-common.c: Likewise.
481 * common/config/nios2/nios2-common.c: Likewise.
482 * common/config/pa/pa-common.c: Likewise.
483 * common/config/pdp11/pdp11-common.c: Likewise.
484 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
485 * common/config/riscv/riscv-common.c: Likewise.
486 * common/config/rs6000/rs6000-common.c: Likewise.
487 * common/config/rx/rx-common.c: Likewise.
488 * common/config/s390/s390-common.c: Likewise.
489 * common/config/sh/sh-common.c: Likewise.
490 * common/config/sparc/sparc-common.c: Likewise.
491 * common/config/tilegx/tilegx-common.c: Likewise.
492 * common/config/tilepro/tilepro-common.c: Likewise.
493 * common/config/v850/v850-common.c: Likewise.
494 * common/config/visium/visium-common.c: Likewise.
495 * common/config/xstormy16/xstormy16-common.c: Likewise.
496 * common/config/xtensa/xtensa-common.c: Likewise.
497 * invoke.texi (-fomit-frame-pointer): Update documentation.
499 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
501 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
503 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
505 * cfg.c (scale_bbs_frequencies_int,
506 cale_bbs_frequencies_gcov_type): Remove.
507 * cfg.h (scale_bbs_frequencies_int,
508 cale_bbs_frequencies_gcov_type): Remove.
510 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
512 * tree-ssa-loop-manip.c
513 (scale_dominated_blocks_in_loop): Update to profile counts.
514 (tree_transform_and_unroll_loop): Likewise.
516 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
518 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
519 scale_bbs_frequencies_int.
521 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
523 * final.c (compute_alignments): Use counts rather than frequencies.
525 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
527 * cfgloopanal.c: Include sreal.h
528 (average_num_loop_insns): Use counts and sreal for accounting.
530 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
532 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
535 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
536 Kito Cheng <kito.cheng@gmail.com>
538 * config/nds32/constraints.md: Provide more constraints.
539 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
540 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
541 support constraints usage.
543 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
545 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
547 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
548 Kito Cheng <kito.cheng@gmail.com>
550 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
551 * config/nds32/nds32.opt: Refine the layout.
552 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
553 TARGET_EXT_STRING): Support new options.
554 * config/nds32/nds32.h: Likewise.
555 * config/nds32/nds32.md: Likewise.
556 * config/nds32/nds32-predicates.c: Likewise.
557 * config/nds32/constraints.md: Likewise.
558 * common/config/nds32/nds32-common.c: Likewise.
560 2017-11-16 Julia Koval <julia.koval@intel.com>
563 * config/i386/gfniintrin.h: Add sse check.
564 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
566 2017-11-16 Julia Koval <julia.koval@intel.com>
568 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
569 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
570 (ix86_handle_option): Handle -mavx512vbmi2.
571 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
572 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
573 * config/i386/i386-c.c (__AVX512VBMI2__): New.
574 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
575 (ix86_valid_target_attribute_inner_p): Ditto.
576 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
577 * config/i386/i386.opt (mavx512vbmi2): New option.
578 * doc/invoke.texi: Add new option.
580 [2017-11-16 Julia Koval <julia.koval@intel.com>
582 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
583 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
584 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
585 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
586 _mm512_gf2p8mul_epi8): New intrinsics.
587 * config/i386/i386-builtin-types.def
588 (V64QI_FTYPE_V64QI_V64QI): New type.
589 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
590 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
591 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
592 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
593 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
594 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
596 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
598 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
599 explicitly as a stream of bytes.
601 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
603 * config/rs6000/altivec.h (vec_xst_be): New #define.
604 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
605 and externalize from *altivec_vperm_<mode>_internal.
606 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
608 (XL_BE_V8HI): Likewise.
609 (XL_BE_V4SI): Likewise.
610 (XL_BE_V4SI): Likewise.
611 (XL_BE_V2DI): Likewise.
612 (XL_BE_V4SF): Likewise.
613 (XL_BE_V2DF): Likewise.
614 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
615 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
616 all array entries with these keys: VSX_BUILTIN_VEC_XL,
617 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
618 VSX_BUILTIN_VEC_XST_BE.
619 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
620 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
622 (altivec_init_builtins): Replace conditional calls to def_builtin
623 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
624 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
625 with unconditional calls. Remove calls to def_builtin for
626 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
627 __builtin_vec_xst_be.
628 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
629 to define_expand, and add alternate RTL generation for P8.
630 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
632 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
633 add alternate RTL generation for P8.
634 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
635 vsx_ld_elemrev_v16qi.
636 (vsx_st_elemrev_v8hi): Convert define_insn
637 to define_expand, and add alternate RTL generation for P8.
638 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
640 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
641 add alternate RTL generation for P8.
642 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
643 vsx_st_elemrev_v16qi.
645 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
648 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
649 TARGET_AVX512ER check.
650 (ix86_option_override_internal): Set MASK_VZEROUPPER if
651 neither -mzeroupper nor -mno-zeroupper is used and
652 TARGET_EMIT_VZEROUPPER is set.
653 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
654 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
656 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
658 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
659 folding of vector compares.
660 (fold_build_vec_cmp): New helper function.
661 (fold_compare_helper): New helper function.
662 (builtin_function_type): Add compare builtins to the list of functions
663 having unsigned arguments. Cosmetic updates to comment indentation.
664 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
665 the not+eq combination.
667 2017-11-15 Bin Cheng <bin.cheng@arm.com>
669 PR tree-optimization/82726
670 PR tree-optimization/70754
671 * tree-predcom.c (order_drefs_by_pos): New function.
672 (combine_chains): Move code setting has_max_use_after to...
673 (try_combine_chains): ...here. New parameter. Sort combined chains
674 according to position information.
675 (tree_predictive_commoning_loop): Update call to above function.
676 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
678 2017-11-15 Bin Cheng <bin.cheng@arm.com>
680 PR tree-optimization/82726
682 2017-01-23 Bin Cheng <bin.cheng@arm.com>
684 PR tree-optimization/70754
685 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
686 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
687 combined stmt before it if not NULL.
688 (combine_chains): Process refs reversely and compute dominance point
692 2017-02-23 Bin Cheng <bin.cheng@arm.com>
694 PR tree-optimization/79663
695 * tree-predcom.c (combine_chains): Process refs in reverse order
696 only for ZERO length chains, and add explaining comment.
698 2017-11-15 Tamar Christina <tamar.christina@arm.com>
700 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
701 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
702 * doc/invoke.texi (ARM Options): Add armv8.3-a.
704 2017-11-15 Tamar Christina <tamar.christina@arm.com>
706 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
708 2017-11-15 Martin Liska <mliska@suse.cz>
710 * tree-cfg.c (pass_warn_function_return::execute):
711 Compare warn_return_type for greater than zero.
713 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
717 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
718 to return true on Xeon and not on Xeon Phi.
719 (ix86_check_avx256_register): Changed to ...
720 (ix86_check_avx_upper_register): ... this. Add extra check for
721 VALID_AVX512F_REG_OR_XI_MODE.
722 (ix86_avx_u128_mode_needed): Changed
723 ix86_check_avx256_register to ix86_check_avx_upper_register.
724 (ix86_check_avx256_stores): Changed to ...
725 (ix86_check_avx_upper_stores): ... this. Changed
726 ix86_check_avx256_register to ix86_check_avx_upper_register.
727 (ix86_avx_u128_mode_after): Changed
728 avx_reg256_found to avx_upper_reg_found. Changed
729 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
730 (ix86_avx_u128_mode_entry): Changed
731 ix86_check_avx256_register to ix86_check_avx_upper_register.
732 (ix86_avx_u128_mode_exit): Ditto.
733 * config/i386/i386.h: (host_detect_local_cpu): New define.
735 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
737 * config/arm/xgene1.md (xgene1): Split into automatons
738 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
739 (xgene1_f_load): Adjust reservations and/or types.
740 (xgene1_f_store): Likewise.
741 (xgene1_load_pair): Likewise.
742 (xgene1_store_pair): Likewise.
743 (xgene1_fp_load1): Likewise.
744 (xgene1_load1): Likewise.
745 (xgene1_store1): Likewise.
746 (xgene1_move): Likewise.
747 (xgene1_alu): Likewise.
748 (xgene1_simd): Likewise.
749 (xgene1_bfm): Likewise.
750 (xgene1_neon_load1): Likewise.
751 (xgene1_neon_store1): Likewise.
752 (xgene1_neon_logic): Likewise.
753 (xgene1_neon_st1): Likewise.
754 (xgene1_neon_ld1r): Likewise.
755 (xgene1_alu_cond): Added.
756 (xgene1_shift_reg): Likwise.
757 (xgene1_bfx): Likewise.
758 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
760 2017-11-15 Jakub Jelinek <jakub@redhat.com>
763 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
765 (can_widen_mult_without_libcall): New function.
766 (expand_mul_overflow): If only checking unsigned mul overflow,
767 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
768 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
769 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
770 (expand_DIVMOD): Formatting fix.
771 * expmed.h (expand_mult): Add NO_LIBCALL argument.
772 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
773 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
775 PR tree-optimization/82977
776 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
777 constructed temporary to strlen_to_stridx.put.
779 2017-11-15 Martin Liska <mliska@suse.cz>
781 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
782 * configure: Regenerate.
784 2017-11-15 Martin Liska <mliska@suse.cz>
787 * config/sh/sh-mem.cc: Use proper probability for
790 2017-11-14 Jeff Law <law@redhat.com>
792 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
793 the red zone for stack_clash_protection_final_dynamic_probe targets
794 when the total dynamic stack size is zero bytes.
796 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
797 blocks is post order.
799 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
801 * dumpfile.h (TDF_COMPARE_DEBUG): New.
802 * final.c (rest_of_clean_state): Set it for the
803 -fcompare-debug dump.
804 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
805 class when TDF_COMPARE_DEBUG is set.
807 * dwarf2out.c (gen_producer_string): Discard
810 2017-11-15 Joseph Myers <joseph@codesourcery.com>
813 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
814 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
815 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
816 (__TGMATH_REAL_2_3): Remove macros.
817 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
818 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
820 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
822 (remquo): Define using __TGMATH_REAL_3.
824 2017-11-14 Jeff Law <law@redhat.com>
826 * vr-values.c: New file with contents extracted from tree-vrp.c.
827 * Makefile.in (OBJS): Add vr-values.o
828 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
829 (set_value_range, set_and_canonicalize_value_range): Likewise.
830 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
831 (value_range_constant_singleton, symbolic_range_p): Likewise.
832 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
833 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
834 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
835 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
836 (find_case_label_range, find_case_label_index): Likewise.
837 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
838 (range_int_cst_singleton_p, value_inside_range): Likewise.
839 (get_single_symbol): Likewise.
840 (switch_update): Move structure definition here.
841 (to_remove_edges, to_update_switch_stmts): Provide externs.
842 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
843 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
844 (vrp_val_is_min, set_value_range): Likewise.
845 (set_and_canonicalize_value_range, copy_value_range): Likewise.
846 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
847 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
848 (range_is_nonnull, range_int_cst_p): Likewwise.
849 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
850 (get_single_symbol, operand_less_p): Likewise
851 (compare_values_warnv, compare_values): Likewise.
852 (value_inside_range, value_range_constant_singleton): Likewise.
853 (zero_nonzero_bitgs_from_vr): Likewise.
854 (extract_range_from_binary_expr_1): Likewise.
855 (overflow_comparison_p): Likewise.
856 (to_remove_edges, to_update_switch_stmts): Likewise.
857 (find_case_label-index, find_case_label_range): Likewise.
858 (switch_update, set_value_range_to_nonnegative): Remove.
859 (set_value_range_to_truthvalue): Likewise.
860 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
861 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
862 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
863 (find_case_label_ranges, test_for_singularity): Likewise.
864 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
865 (x_vr_values): Move to its remaining use site.
867 2017-11-10 Jeff Law <law@redhat.com>
869 * vr-values.h (VR_INITIALIZER): Move #define here.
870 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
871 * Makefile.in (OBJS): Add tree-evrp.o
872 * tree-vrp.h (assert_info): Move structure definition here.
873 (set_value_range_to_varying): Prototype.
874 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
875 (infer_value_range, register_edge_assert_for): Likewise.
876 (stmt_interesting_for_vrp): Likewise.
877 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
878 (set_value_range_to_varying): No longer static.
879 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
880 (infer_value_range, register_edge_assert_for): Likewise.
882 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
884 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
885 generate the XXBRD instruction.
887 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
888 (rs6000_builtin_type_compatible): Treat _Float128 and long double
889 as being compatible if -mabi=ieeelongdouble.
890 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
891 to setup float128 built-ins with hardware support.
892 (BU_FLOAT128_HW_2): Likewise.
893 (BU_FLOAT128_HW_3): Likewise.
894 (BU_FLOAT128_HW_VSX_1): Likewise.
895 (BU_FLOAT128_HW_VSX_2): Likewise.
896 (scalar_extract_expq): Change float128 built-in functions to
897 accommodate having both KFmode and TFmode functions. Use the
898 KFmode variant as the default.
899 (scalar_extract_sigq): Likewise.
900 (scalar_test_neg_qp): Likewise.
901 (scalar_insert_exp_q): Likewise.
902 (scalar_insert_exp_qp): Likewise.
903 (scalar_test_data_class_qp): Likewise.
904 (sqrtf128_round_to_odd): Delete processing the round to odd
905 built-in functions as special built-in functions, and define them
906 as float128 built-ins. Use the KFmode variant as the default.
907 (truncf128_round_to_odd): Likewise.
908 (addf128_round_to_odd): Likewise.
909 (subf128_round_to_odd): Likewise.
910 (mulf128_round_to_odd): Likewise.
911 (divf128_round_to_odd): Likewise.
912 (fmaf128_round_to_odd): Likewise.
913 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
914 support for KFmode and TFmode xststdcqp calls.
915 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
916 point, switch the built-in handlers for the get/set float128
917 exponent, get float128 mantissa, float128 test built-ins, and the
918 float128 round to odd built-in functions. Eliminate creating the
919 float128 round to odd built-in functions as special built-ins.
920 (rs6000_init_builtins): Eliminate special creation of the float128
921 round to odd built-in functions.
922 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
923 function insns to support both TFmode and KFmode variants.
924 (xsxsigqp_<mode>): Likewise.
925 (xsiexpqpf_<mode>): Likewise.
926 (xsiexpqp_<mode>): Likewise.
927 (xststdcqp_<mode>): Likewise.
928 (xststdcnegqp_<mode>): Likewise.
929 (xststdcqp_<mode>): Likewise.
931 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
933 * tree-ssa-threadupdate.c (compute_path_counts): Remove
934 unused path_in_freq_ptr parameter.
935 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
937 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
939 * ipa-inline.c (edge_badness): Dump sreal frequency.
940 (compute_inlined_call_time): Match natural implementaiton ...
941 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
942 forgotten division by CGRAPH_FREQ_BASE.
944 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
946 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
947 Solaris 11. Update comment.
948 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
950 * config/sol2.h (STARTFILE_SPEC): Likewise.
951 * configure: Regenerate.
953 2017-11-14 Carl Love <cel@us.ibm.com>
955 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
956 le_ and be_ prefixes to swap* variables. Remove
957 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
960 2017-11-14 Jason Merrill <jason@redhat.com>
962 Support GTY((cache)) on hash_map.
963 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
964 (ggc_cache_remove): Override it instead of ggc_mx.
965 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
966 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
967 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
968 (simple_cache_map_traits): Override maybe_mx.
969 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
970 (hash_map): Friend gt_cleare_cache.
971 (gt_cleare_cache): New.
972 * tree.h (tree_cache_traits): New hash_map traits class.
973 (tree_cache_map): New typedef.
975 2017-11-14 Richard Biener <rguenther@suse.de>
977 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
978 paramter and handling.
979 (cleanup_control_flow_bb): Likewise.
980 (cleanup_control_flow_pre): New helper performing a DFS walk
981 to call cleanup_control_flow_bb in PRE order.
982 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
983 via cleanup_control_flow_pre.
985 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
987 * config/aarch64/aarch64-simd.md
988 (aarch64_simd_bsl<mode>_internal): Remove DImode.
989 (*aarch64_simd_bsl<mode>_alt): Likewise.
990 (aarch64_simd_bsldi_internal): New.
991 (aarch64_simd_bsldi_alt): Likewise.
993 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
995 * tracer.c (better_p): Do not compare frequencies.
996 * reg-stack.c (better_edge): Likewise.
997 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
1000 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1002 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
1003 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
1004 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1005 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1006 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1007 * predict.c (maybe_hot_frequency_p): Inline to ...
1008 (maybe_hot_count_p): ... here; rewrite to counts.
1009 (counts_to_freqs): Rename to ...
1010 (update_max_bb_count): ... this one.
1011 (expensive_function_p): Use counts.
1012 (estimate_bb_frequencies): Update.
1013 (rebuild_frequencies): Update.
1014 * predict.h (counts_to_freqs): Rename to ...
1015 (update_max_bb_count): ... this one.
1016 * profile.c (compute_branch_probabilities): Add debug info
1017 * tree-inline.c (expand_call_inline): Update debug info.
1018 (optimize_inline_calls): Use update_max_bb_count..
1019 (tree_function_versioning): Use update_max_bb_count..
1020 * value-prof.c (gimple_value_profile_transformations):
1021 Do not use update_max_bb_count.
1023 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1025 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1026 always use frequencies.
1028 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1030 * bb-reorder.c: Remove frequencies from comments.
1031 (better_edge_p): Use profile counts.
1032 (find_traces): Dump profile counts.
1033 (rotate_loop): Use profile counts.
1034 (find_traces_1_round): Likewise.
1035 (connect_better_edge_p): Use counts instead of probabilities for
1037 (copy_bb_p): Drop early check for non-0 frequency.
1038 (sanitize_hot_paths): Update comments.
1040 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1042 * ipa-split.c (struct split_point): Add count.
1043 (consider_split): Do not compute incoming frequency; compute incoming
1044 count and store it to split_point.
1045 (split_function): Set count of the call to split part correctly.
1047 2017-11-13 Carl Love <cel@us.ibm.com>
1049 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1051 2017-11-13 Tom Tromey <tom@tromey.com>
1053 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1055 2017-11-13 Carl Love <cel@us.ibm.com>
1057 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1058 Add support for builtins:
1059 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1060 vector {un,}signed {char,int,short},
1061 vector {un,}signed {char,int,short}) arguments.
1062 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1063 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1064 Add BU_P9V_AV_2 expansions for the builtins.
1065 * config/rs6000/altivec.h (vec_first_match_index,
1066 vec_first_mismatch_index, vec_first_match_or_eos_index,
1067 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1068 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1069 new extern declaration.
1070 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1071 * config/rs6000/vsx.md (first_match_index_<mode>,
1072 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1073 first_mismatch_or_eos_index_<mode>): Add define expand.
1074 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1075 * doc/extend.texi: Update the built-in documenation file for the new
1078 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1080 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1081 into the min/max operations for _Float<N> and _Float<N>X types.
1083 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1086 * dwarf2out.c (do_eh_frame): New static variable.
1087 (dwarf2out_begin_prologue): Set it.
1088 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1090 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1092 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1094 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1096 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1099 * debug.h (dwarf2out_do_eh_frame): Declare.
1100 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1101 (dwarf2out_do_frame): Use it.
1102 (dwarf2out_do_cfi_asm): Likewise.
1103 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1104 (dwarf2out_assembly_start): Likewise.
1105 (dwarf2out_begin_prologue): Fix comment.
1106 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1107 if the target needs either debug or unwind DWARF2 info.
1108 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1109 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1110 -ftrapping-math, -ftrapv and -fwrapv.
1112 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1114 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1115 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1116 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1117 (estimate_edge_size_and_time): Likewise.
1118 (ipa_merge_fn_summary_after_inlining): Likewise.
1119 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1120 (compute_uninlined_call_time): Use sreal_frequency.
1121 (compute_inlined_call_time): Likewise.
1122 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1123 * profile-count.c: Include sreal.h.
1124 (profile_count::to_sreal_scale): New.
1125 * profile-count.h: Forward declare sreal.
1126 (profile_count::to_sreal_scale): Declare.
1128 2017-11-13 Nathan Sidwell <nathan@acm.org>
1130 * diagnostic.c (maybe_line_and_column): New.
1131 (diagnostic_get_location_text): Use it.
1132 (diagnostic_report_current_module): Likewise.
1133 (test_diagnostic_get_location_text): Add tests.
1135 2017-11-13 Luis Machado <luis.machado@linaro.org>
1137 * doc/md.texi (Specifying processor pipeline description): Fix
1138 incorrect latency for the div instruction example.
1140 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1142 PR tree-optimization/78821
1143 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1144 that bit_not_p is the same.
1145 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1146 (split_group): Count precisely bit_not_p bits in each statement.
1147 (invert_op): New function.
1148 (imm_store_chain_info::output_merged_store): Use invert_op to
1149 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1150 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1152 2017-11-13 Martin Liska <mliska@suse.cz>
1154 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1155 (struct source_info): Likewise.
1156 (add_branch_counts): Likewise.
1157 (add_line_counts): Likewise.
1158 (function_summary): Likewise.
1159 (output_intermediate_line): Likewise.
1160 (generate_results): Likewise.
1162 2017-11-13 Martin Liska <mliska@suse.cz>
1164 * gcov.c (struct block_info): Remove typedef for block_t.
1165 (struct line_info): Likewise.
1166 (line_info::has_block): Likewise.
1167 (EXIT_BLOCK): Likewise.
1168 (unblock): Likewise.
1169 (circuit): Likewise.
1170 (get_cycles_count): Likewise.
1171 (process_file): Likewise.
1172 (read_graph_file): Likewise.
1173 (solve_flow_graph): Likewise.
1174 (find_exception_blocks): Likewise.
1175 (add_line_counts): Likewise.
1176 (accumulate_line_info): Likewise.
1177 (output_line_details): Likewise.
1179 2017-11-13 Martin Liska <mliska@suse.cz>
1181 * gcov.c (struct arc_info): Remove typedef for arc_t.
1182 (struct line_info): Likewise.
1183 (add_branch_counts): Likewise.
1184 (output_branch_count): Likewise.
1185 (function_info::~function_info): Likewise.
1186 (circuit): Likewise.
1187 (output_intermediate_line): Likewise.
1188 (read_graph_file): Likewise.
1189 (solve_flow_graph): Likewise.
1190 (find_exception_blocks): Likewise.
1191 (add_line_counts): Likewise.
1192 (accumulate_line_info): Likewise.
1193 (output_line_details): Likewise.
1194 (output_function_details): Likewise.
1196 2017-11-13 Martin Liska <mliska@suse.cz>
1198 * gcov.c (struct function_info): Remove typedef for function_t.
1199 (struct source_info): Likewise.
1200 (source_info::get_functions_at_location): Likewise.
1201 (solve_flow_graph): Likewise.
1202 (find_exception_blocks): Likewise.
1203 (add_line_counts): Likewise.
1204 (output_intermediate_file): Likewise.
1205 (process_file): Likewise.
1206 (generate_results): Likewise.
1207 (release_structures): Likewise.
1208 (read_graph_file): Likewise.
1209 (read_count_file): Likewise.
1210 (accumulate_line_counts): Likewise.
1211 (output_lines): Likewise.
1213 2017-11-13 Martin Liska <mliska@suse.cz>
1215 * gcov.c (function_info::function_info): Remove num_counts
1216 and add vector<gcov_type>.
1217 (function_info::~function_info): Use the vector.
1218 (process_file): Likewise.
1219 (read_graph_file): Likewise.
1220 (read_count_file): Likewise.
1221 (solve_flow_graph): Likewise.
1223 2017-11-13 Martin Liska <mliska@suse.cz>
1225 * gcov.c (function_info::is_artificial): New function.
1226 (process_file): Erase all artificial early.
1227 (generate_results): Skip as all artificial are already
1230 2017-11-13 Martin Liska <mliska@suse.cz>
1232 * gcov.c (read_graph_file): Store to global vector of functions.
1233 (read_count_file): Iterate the vector.
1234 (process_file): Likewise.
1235 (generate_results): Likewise.
1236 (release_structures): Likewise.
1238 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1240 PR tree-optimization/82954
1241 * gimple-ssa-store-merging.c
1242 (imm_store_chain_info::coalesce_immediate_stores): If
1243 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1245 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1247 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1248 Upddate call to ENDIAN_LANE_N.
1249 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1250 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1251 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1252 and use aarch64_endian_lane_rtx.
1253 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1255 2017-11-12 Tom de Vries <tom@codesourcery.com>
1257 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1259 2017-11-12 Tom de Vries <tom@codesourcery.com>
1261 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1263 2017-11-12 Tom de Vries <tom@codesourcery.com>
1265 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1267 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1268 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1269 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1271 2017-11-11 Martin Sebor <msebor@redhat.com>
1274 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1277 * prefic.c (translate_name): Replace strncpy with memcpy to
1278 avoid -Wstringop-truncation.
1280 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1282 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1284 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1286 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1288 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1290 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1293 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1295 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1296 enable generating XXBRH if the value is in a vector register.
1297 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1298 value is in a vector register.
1299 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1300 register to register bswap64's instead of doing the GPR sequence
1301 used on previous machines.
1302 (bswapdi2_xxbrd): New insn.
1303 (bswapdi2_reg): Disallow on ISA 3.0.
1304 (register to register bswap64 splitter): Do not split the insn on
1305 ISA 3.0 systems that use XXBRD.
1307 2017-11-10 Martin Sebor <msebor@redhat.com>
1310 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1312 (find_subframework_file): Same.
1314 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1316 * auto-profile.c (afdo_indirect_call): Drop frequency.
1317 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1318 (cgraph_node::create_edge): Drop frequency argument.
1319 (cgraph_node::create_indirect_edge): Drop frequency argument.
1320 (cgraph_edge::make_speculative): Drop frequency arguments.
1321 (cgraph_edge::resolve_speculation): Do not update frequencies
1322 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1323 (cgraph_node::dump): Check consistency in IPA mode.
1324 (cgraph_edge::maybe_hot_p): Use IPA counter.
1325 (cgraph_edge::verify_count_and_frequency): Rename to ...
1326 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1327 (cgraph_node::verify_node): Update.
1328 * cgraph.h (struct cgraph_edge): Drop frequency.
1329 (cgraph_edge::frequency): New function.
1330 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1332 (cgraph_edge::rebuild_edges): Likewise.
1333 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1334 (duplicate_thunk_for_node): Do not pass frequency.
1335 (cgraph_node::create_clone): Scale only counts.
1336 (cgraph_node::create_virtual_clone): Do not pass frequency.
1337 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1338 (cgraph_node::create_version_clone): Do not pass frequency.
1339 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1340 (cgraph_node::expand_thunk): Do not pass frequency.
1341 (cgraph_node::create_wrapper): Do not pass frequency.
1342 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1344 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1345 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1346 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1347 (gather_caller_stats): Use frequency function.
1348 (ipcp_cloning_candidate_p): Use frequency function.
1349 (ipcp_propagate_stage): Use frequency function.
1350 (get_info_about_necessary_edges): Use frequency function.
1351 (update_profiling_info): Update only IPA profile.
1352 (update_specialized_profile): Use frequency functoin.
1353 (perhaps_add_new_callers): Update only IPA profile.
1354 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1355 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1356 (dump_ipa_call_summary): Use frequency function.
1357 (estimate_edge_size_and_time): Use frequency function.
1358 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1359 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1360 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1361 (update_noncloned_counts): ... ths one; scale counts only.
1362 (clone_inlined_nodes): Do not scale frequency.
1363 (inline_call): Do not pass frequency.
1364 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1365 (compute_inlined_call_time): Use IPA profile.
1366 (want_inline_small_function_p): Use IPA profile.
1367 (want_inline_self_recursive_call_p): Use IPA profile.
1368 (edge_badness): Use IPA profile.
1369 (lookup_recursive_calls): Use IPA profile.
1370 (recursive_inlining): Do not pass frequency.
1371 (resolve_noninline_speculation): Do not update frequency.
1372 (inline_small_functions): Collect max of IPA profile.
1373 (dump_overall_stats): Dump IPA porfile.
1374 (dump_inline_stats): Dump IPA porfile.
1375 (ipa_inline): Collect IPA stats.
1376 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1377 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1378 (ipa_propagate_frequency): Use frequency function.
1379 (ipa_profile): Cleanup.
1380 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1381 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1382 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1383 (input_node): Do not stream frequency.
1384 (input_edge): Do not stream frequency.
1385 (merge_profile_summaries): Scale only IPA profiles.
1386 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1387 * predict.c (drop_profile): Do not recompute frequency.
1388 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1389 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1390 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1391 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1392 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1393 * tree-inline.c (copy_bb): Do not scale frequency.
1394 (expand_call_inline): Do not scale frequency.
1395 (tree_function_versioning): Do not scale frequency.
1396 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1398 2017-11-10 Julia Koval <julia.koval@intel.com>
1400 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1401 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1402 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1403 (_mm256_maskz_gf2p8affine_epi64_epi8)
1404 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1405 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1406 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1407 (__builtin_ia32_vgf2p8affineqb_v32qi)
1408 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1409 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1411 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1415 (arm_option_override): Refactor.
1416 (arm_option_reconfigure_globals): New.
1417 (arm_options_perform_arch_sanity_checks): New.
1418 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1420 (arm_options_perform_arch_sanity_checks): Likewise
1422 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1424 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1425 (power9-qpmul): New.
1426 * rs6000/rs6000.md ("type" attr): Add qmul.
1427 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1428 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1429 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1431 2017-11-10 Martin Sebor <msebor@redhat.com>
1434 * builtins.c (compute_objsize): Handle arrays that
1435 compute_builtin_object_size likes to fail for. Make extern.
1436 * builtins.h (compute_objsize): Declare.
1437 (check_strncpy_sizes): New function.
1438 (expand_builtin_strncpy): Call check_strncpy_sizes.
1439 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1440 -Wstringop-truncation.
1441 (gimple_fold_builtin_strncat): Same.
1442 * gimple.c (gimple_build_call_from_tree): Set call location.
1443 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1444 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1445 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1446 (handle_builtin_strlen): Use strlen_to_stridx.
1447 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1449 Use strlen_to_stridx.
1450 (pass_strlen::execute): Release strlen_to_stridx.
1451 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1452 (-Wstringop-truncation): Document new option.
1454 2017-11-10 Martin Liska <mliska@suse.cz>
1456 PR gcov-profile/82702
1457 * gcov.c (main): Handle intermediate files in a different
1459 (get_gcov_intermediate_filename): New function.
1460 (output_gcov_file): Remove support of intermediate files.
1461 (generate_results): Allocate intermediate file.
1462 (release_structures): Clean-up properly fn_end.
1463 (output_intermediate_file): Start iterating with line 1.
1465 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1467 PR tree-optimization/82929
1468 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1469 ops_swapped_p non-static data member.
1470 (store_immediate_info::store_immediate_info): Clear it.
1471 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1472 ops set ops_swapped_p.
1473 (count_multiple_uses): Handle ops_swapped_p.
1475 2017-11-10 Martin Liska <mliska@suse.cz>
1477 * coverage.c (coverage_init): Stream information about
1478 support of has_unexecuted_blocks.
1479 * doc/gcov.texi: Document that.
1480 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1481 * gcov.c (read_graph_file): Likewise.
1482 (output_line_beginning): Fix a small issue with
1485 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1487 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1488 reference of trivial component.
1490 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1493 * gimple-ssa-store-merging.c
1494 (pass_store_merging::terminate_all_aliasing_chains): For
1495 gimple_store_p stmts also call refs_output_dependent_p.
1497 PR rtl-optimization/82913
1498 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1501 2017-11-09 Jeff Law <law@redhat.com>
1503 * vr-values.h: New file with vr_values class.
1504 * tree-vrp.c: Include vr-values.h
1505 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1506 data objects into the vr_values class.
1507 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1508 (get_value_range): Make it a member function within vr_values class.
1509 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1510 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1511 (op_with_constant_singleton_value_range): Likewise.
1512 (extract_range_for_var_from_comparison_expr): Likewise.
1513 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1514 (extract_range_from_binary_expr): Likewise.
1515 (extract_range_from_unary_expr): Likewise.
1516 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1517 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1518 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1519 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1520 (compare_name_with_value, compare_names): Likewise.
1521 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1522 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1523 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1524 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1525 (extract_range_from_phi_node): Likewise.
1526 (simplify_truth_ops_using_ranges): Likewise.
1527 (simplify_div_or_mod_using_ranges): Likewise.
1528 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1529 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1530 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1531 (simplify_float_conversion_using_ranges): Likewise.
1532 (simplify_internal_call_using_ranges): Likewise.
1533 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1534 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1535 poitner via x_vr_values for calls into gimple folder.
1536 (vrp_initialize_lattice): Make this the vr_values ctor.
1537 (vrp_free_lattice): Make this the vr_values dtor.
1538 (set_vr_value): New function.
1539 (class vrp_prop): Add vr_values data member. Add various member
1540 functions as well as member functions that delegate to vr_values.
1541 (check_array_ref): Make a member function within vrp_prop class.
1542 (search_for_addr_array, vrp_initialize): Likewise.
1543 (vrp_finalize): Likewise. Revamp to avoid direct access to
1544 vr_value, values_propagated, etc.
1545 (check_array_bounds): Extract vrp_prop class instance pointer from
1546 walk info structure. Use it to call member functions.
1547 (check_all_array_refs): Make a member function within vrp_prop class.
1548 Smuggle class instance pointer via walk info structure.
1549 (x_vr_values): New local static.
1550 (vrp_valueize): Use x_vr_values to get class instance.
1551 (vr_valueize_1): Likewise.
1552 (class vrp_folder): Add vr_values data member. Add various member
1553 functions as well as member functions that delegate to vr_values.
1554 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1555 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1556 class instance from vr_values. Use it to call member functions.
1557 (vrp_dom_walker): Add vr_values data member.
1558 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1559 instance via x_vr_values.
1560 (identify_jump_threads): Accept vr_values as argument. Store
1561 it into the walker structure.
1562 (evrp_dom_walker): Add vr_values class data member. Add various
1564 (evrp_dom_walker::try_find_new_range): Use vr_values data
1565 member to access the memory allocator.
1566 (evrp_dom_walker::before_dom_children): Store vr_values class
1567 instance into the vrp_folder class.
1568 (evrp_dom_walker::push_value_range): Rework to avoid direct
1569 access to num_vr_values and vr_value.
1570 (evrp_dom_walker::pop_value_range): Likewise.
1571 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1572 Use vr_values to get to dump_all_value_ranges member function.
1573 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1574 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1575 Pass vr_values class instance down to identify_jump_threads.
1576 Remove call to vrp_free_lattice.
1577 (debug_all_value_ranges): Remove.
1579 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1580 (vrp_folder): Likewise.
1582 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1583 Get it from the existing bitmap instead.
1584 (vrp_intersect_ranges_1): Likewise.
1586 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1588 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1590 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1591 set bit_not_p to it.
1592 (imm_store_chain_info::coalesce_immediate_stores): Break group
1593 if bit_not_p is different.
1594 (count_multiple_uses, split_group,
1595 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1596 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1597 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1598 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1599 to store_immediate_info ctor.
1601 2017-11-09 Jim Wilson <jimw@sifive.com>
1603 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1604 (scan_prog_file): Likewise.
1606 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1608 * bb-reorder.c (max_entry_frequency): Remove.
1609 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1610 connect_traces, push_to_next_round_p): Remove prototypes.
1611 (find_traces_1_round): Use counts only.
1612 (push_to_next_round_p): Likewise.
1613 (find_traces): Likewise.
1614 (rotate_loop): Likewise.
1615 (find_traces_1_round): Likewise.
1616 (connect_traces): Likewise.
1617 (edge_order): Likewise.
1619 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1621 * config/arm/arm.c (output_return_instruction): Add comments to
1622 indicate requirement for cmse_nonsecure_entry return to account
1623 for the size of clearing instruction output here.
1624 (thumb_exit): Likewise.
1625 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1626 return in hardfloat mode.
1628 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1630 * config/rs6000/rs6000.c (machine_function): Add a bool,
1631 "toc_is_wrapped_separately".
1632 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1633 if it wasn't explicitly set or unset, we are optimizing for speed, and
1634 doing separate shrink-wrapping.
1635 (rs6000_get_separate_components): Enable the TOC component if
1636 saving the TOC register in the prologue.
1637 (rs6000_components_for_bb): Handle the TOC component.
1638 (rs6000_emit_prologue_components): Store the TOC register where needed.
1639 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1640 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1642 2017-11-09 Martin Jambor <mjambor@suse.cz>
1644 * ipa-param-manipulation.c: New file.
1645 * ipa-param-manipulation.h: Likewise.
1646 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1647 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1648 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1649 (ipa_parm_adjustment): Likewise.
1650 (ipa_parm_adjustment_vec): Likewise.
1651 (ipa_get_vector_of_formal_parms): Moved declaration to
1652 ipa-param-manipulation.h.
1653 (ipa_get_vector_of_formal_parm_types): Likewise.
1654 (ipa_modify_formal_parameters): Likewise.
1655 (ipa_modify_call_arguments): Likewise.
1656 (ipa_combine_adjustments): Likewise.
1657 (ipa_dump_param_adjustments): Likewise.
1658 (ipa_modify_expr): Likewise.
1659 (ipa_get_adjustment_candidate): Likewise.
1660 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1661 ipa-param-manipulation.c.
1662 (ipa_get_vector_of_formal_parm_types): Likewise.
1663 (ipa_modify_formal_parameters): Likewise.
1664 (ipa_modify_call_arguments): Likewise.
1665 (ipa_modify_expr): Likewise.
1666 (get_ssa_base_param): Likewise.
1667 (ipa_get_adjustment_candidate): Likewise.
1668 (index_in_adjustments_multiple_times_p): Likewise.
1669 (ipa_combine_adjustments): Likewise.
1670 (ipa_dump_param_adjustments): Likewise.
1671 * tree-sra.c: Also include ipa-param-manipulation.h
1672 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1675 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1676 Alan Hayward <alan.hayward@arm.com>
1677 David Sherwood <david.sherwood@arm.com>
1679 * doc/sourcebuild.texi (vect_masked_store): Document.
1681 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1682 Alan Hayward <alan.hayward@arm.com>
1683 David Sherwood <david.sherwood@arm.com>
1685 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1687 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1688 Alan Hayward <alan.hayward@arm.com>
1689 David Sherwood <david.sherwood@arm.com>
1691 * doc/sourcebuild.texi (vect_variable_length): Document.
1693 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1694 Alan Hayward <alan.hayward@arm.com>
1695 David Sherwood <david.sherwood@arm.com>
1697 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1699 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1700 Alan Hayward <alan.hayward@arm.com>
1701 David Sherwood <david.sherwood@arm.com>
1703 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1705 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1706 Alan Hayward <alan.hayward@arm.com>
1707 David Sherwood <david.sherwood@arm.com>
1709 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1710 previously undocumented selectors.
1711 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1713 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1715 * doc/rtl.texi (const_vector): Say that elements can be
1716 const_wide_ints too.
1717 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1718 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1719 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1720 * optabs.c (expand_vector_broadcast): Likewise.
1722 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1723 Alan Hayward <alan.hayward@arm.com>
1724 David Sherwood <david.sherwood@arm.com>
1726 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1727 scaled index even if the unscaled address was invalid.
1728 Don't increase the complexity of using a scale in that case.
1730 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1731 Alan Hayward <alan.hayward@arm.com>
1732 David Sherwood <david.sherwood@arm.com>
1734 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1735 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1736 UNITS_PER_WORD bytes.
1737 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1738 based on REGMODE_NATURAL_SIZE of the inner mode.
1739 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1740 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1741 * expmed.c (lowpart_bit_field_p): Divide the value up into
1742 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1743 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1744 whether something is likely to occupy more than one register.
1746 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1749 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1750 profile_count::adjust_for_ipa_scaling.
1751 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1752 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1754 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1756 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1758 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1759 (split_group): Add total_orig and total_new arguments, estimate the
1760 number of statements related to the store group without store merging
1761 and with store merging.
1762 (imm_store_chain_info::output_merged_store): Adjust split_group
1763 callers, punt if estimated number of statements with store merging
1764 is not smaller than estimated number of statements without it.
1766 (handled_load): Remove has_single_use checks.
1767 (pass_store_merging::process_store): Likewise.
1769 2017-11-09 Richard Biener <rguenther@suse.de>
1771 PR tree-optimization/82902
1772 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1774 2017-11-09 Martin Liska <mliska@suse.cz>
1777 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1780 2017-11-09 Martin Liska <mliska@suse.cz>
1782 PR tree-optimization/82669
1783 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1785 2017-11-09 Martin Liska <mliska@suse.cz>
1787 PR gcov-profile/48463
1788 * coverage.c (coverage_begin_function): Output also end locus
1789 of a function and information whether the function is
1791 * gcov-dump.c (tag_function): Parse and print the information.
1792 * gcov.c (INCLUDE_MAP): Add include.
1793 (INCLUDE_SET): Likewise.
1794 (struct line_info): Move earlier in the source file because
1795 of vector<line_info> in function_info structure.
1796 (line_info::line_info): Likewise.
1797 (line_info::has_block): Likewise.
1798 (struct source_info): Add new member index.
1799 (source_info::get_functions_at_location): New function.
1800 (function_info::group_line_p): New function.
1801 (output_intermediate_line): New function.
1802 (output_intermediate_file): Use the mentioned function.
1803 (struct function_start): New.
1804 (struct function_start_pair_hash): Likewise.
1805 (process_file): Add code that identifies group functions.
1806 Assign lines either to global or function scope.
1807 (generate_results): Skip artificial functions.
1808 (find_source): Assign index for each source file.
1809 (read_graph_file): Read new flag artificial and end_line.
1810 (add_line_counts): Assign it either to global of function scope.
1811 (accumulate_line_counts): Isolate core of the function to
1812 accumulate_line_info and call it for both function and global
1814 (accumulate_line_info): New function.
1815 (output_line_beginning): Fix GNU coding style.
1816 (print_source_line): New function.
1817 (output_line_details): Likewise.
1818 (output_function_details): Likewise.
1819 (output_lines): Iterate both source (global) scope and function
1821 (struct function_line_start_cmp): New class.
1822 * doc/gcov.texi: Reflect changes in documentation.
1824 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1827 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1828 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1829 and similarly for not instead of neg.
1831 2017-11-08 Andi Kleen <ak@linux.intel.com>
1833 * config/i386/i386.opt: Add -mforce-indirect-call.
1834 * config/i386/predicates.md: Check for flag_force_indirect_call.
1835 * doc/invoke.texi: Document -mforce-indirect-call
1837 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1839 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1841 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1842 config/riscv/riscv.c (predict.h): New include.
1843 (riscv_slow_unaligned_access_p): No longer static.
1844 (riscv_block_move_straight): Add require.
1845 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1847 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1850 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1851 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1852 nonimmediate_operand predicate for operand 1 instead of
1855 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1857 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1859 * config/aarch64/constraints.md (Uml): New constraint.
1860 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1863 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1865 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1866 of two vec_duplicates into a vec_concat.
1868 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1870 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1871 Simplify vec_merge of vec_duplicate and vec_concat.
1872 * config/aarch64/constraints.md (Utq): New constraint.
1873 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1876 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1878 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1879 Simplify vec_merge of vec_duplicate and const_vector.
1880 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1882 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1883 mode iterator. Update predicate on operand 1 to
1884 handle non-const_vec constants. Delete constraints.
1885 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1887 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1889 PR tree-optimization/78821
1890 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1892 (store_operand_info::store_operand_info): Initialize it to false.
1893 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1894 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1895 store in the group, and if chain_info is non-NULL, to ignore altogether
1897 (compatible_load_p): Fail if bit_not_p does not match.
1898 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1899 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1900 (pass_store_merging::process_store): Adjust
1901 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1902 call terminate_all_aliasing_chains newly when adding a store into
1903 a chain with non-NULL chain_info.
1905 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1907 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1909 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1911 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1913 (aarch64_layout_frame): Initialise emit_frame_chain.
1914 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1915 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1917 2017-11-08 Martin Liska <mliska@suse.cz>
1919 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1920 of gimple_call_internal_p.
1922 2017-11-07 Tom de Vries <tom@codesourcery.com>
1924 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1926 2017-11-07 Tom de Vries <tom@codesourcery.com>
1928 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1931 2017-11-08 Martin Liska <mliska@suse.cz>
1934 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1936 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1938 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1940 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1943 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1944 (*cmp<mode>_ccz_1): New insn with $k alternative.
1947 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1948 RTX_COMM_COMPARE as commutative as well.
1949 (ix86_binary_operator_ok): Formatting fix.
1950 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1951 *<code><mode>3<mask_name><round_saeonly_name>,
1952 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1953 *<s>mul<mode>3_highpart<mask_name>,
1954 *vec_widen_umult_even_v16si<mask_name>,
1955 *vec_widen_umult_even_v8si<mask_name>,
1956 *vec_widen_umult_even_v4si<mask_name>,
1957 *vec_widen_smult_even_v16si<mask_name>,
1958 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1959 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1960 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1961 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1962 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1963 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1964 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1965 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1966 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1967 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1968 ix86_binary_operator_ok. Formatting fixes.
1969 (*<plusminus_insn><mode>3<mask_name><round_name>,
1970 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1973 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1975 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1976 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1977 eq and ne if TARGET_ISEL.
1978 (cmp): New code_iterator.
1979 (UNS, UNSU_, UNSIK): New code_attrs.
1980 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1981 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1983 ("eq<mode>3"): ... this.
1984 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1986 ("ne<mode>3"): ... this.
1988 2017-11-07 Julia Koval <julia.koval@intel.com>
1991 * common/config/i386/i386-common.c
1992 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1993 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1994 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1995 * config/i386/i386.opt: Ditto.
1996 * config/i386/i386.c (ix86_target_string): Ditto.
1997 (ix86_option_override_internal): Ditto.
1998 (ix86_init_mpx_builtins): Move MPX to args2.
1999 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
2000 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2001 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2002 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2003 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2004 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2005 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2007 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2010 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2011 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2012 (zero-extendsidi peephole2): Remove peephole.
2014 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2017 * stor-layout.c: Include attribs.h.
2018 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2019 explicit lookup of "aligned" attribute.
2021 2017-11-07 Andrew Waterman <andrew@sifive.com>
2023 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2024 (riscv_expand_block_move): Likewise.
2025 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2027 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2028 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2029 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2030 (riscv_adjust_block_mem): Likewise.
2031 (riscv_block_move_loop): Likewise.
2032 (riscv_expand_block_move): Likewise.
2033 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2035 2017-11-07 Michael Clark <michaeljclark@mac.com>
2037 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2038 (MUSL_DYNAMIC_LINKER): Likewise.
2040 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2042 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2043 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2046 2017-11-07 Richard Biener <rguenther@suse.de>
2048 * match.pd: Fix build.
2050 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2051 Jackson Woodruff <jackson.woodruff@arm.com>
2053 PR tree-optimization/71026
2054 * match.pd: Canonicalize negate in division.
2056 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2059 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2061 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2063 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2064 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2066 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2068 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2069 non-scalar integral types.
2070 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2071 (-(A-B), -(~A)): New transformations.
2073 2017-11-07 Tom de Vries <tom@codesourcery.com>
2075 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2076 semicolon after "do {} while (0)".
2077 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2078 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2079 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2080 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2081 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2083 2017-11-07 Tom de Vries <tom@codesourcery.com>
2085 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2086 after "do {} while (0)".
2087 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2088 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2089 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2090 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2091 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2093 2017-11-07 Tom de Vries <tom@codesourcery.com>
2096 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2098 (arm_rtx_costs_internal): Add missing semicolon after
2099 HANDLE_NARROW_SHIFT_ARITH call.
2101 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2103 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2104 disable isel if it was not set explicitly.
2106 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2108 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
2110 (add_type_attribute) likewise.
2112 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2114 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2115 of struct ix86_frame.
2116 (ix86_initial_elimination_offset): Likewise.
2117 (ix86_expand_split_stack_prologue): Likewise.
2119 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2121 * tree-vrp.h (enum value_range_type): Update stale comment.
2123 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2124 Alan Hayward <alan.hayward@arm.com>
2125 David Sherwood <david.sherwood@arm.com>
2127 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2128 (aarch64_expand_vec_perm_const): Take the number of units too.
2129 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2130 (aarch64_expand_vec_perm_const): Likewise.
2131 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2132 (vec_perm<mode>): Update accordingly.
2134 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2135 Alan Hayward <alan.hayward@arm.com>
2136 David Sherwood <david.sherwood@arm.com>
2138 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2139 Take the number of units too.
2140 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2141 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2142 but check for a vector mode before rather than after the call.
2143 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2144 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2145 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2146 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2147 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2148 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2149 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2150 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2151 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2152 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2153 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2154 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2155 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2156 (aarch64_sqdmull2_n<mode>): Update accordingly.
2158 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2159 Alan Hayward <alan.hayward@arm.com>
2160 David Sherwood <david.sherwood@arm.com>
2162 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2163 the number of units too.
2164 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2165 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2166 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2167 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2168 (vec_store_lanesxi<mode>): Update accordingly.
2170 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2171 Alan Hayward <alan.hayward@arm.com>
2172 David Sherwood <david.sherwood@arm.com>
2174 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2175 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2176 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2177 of units rather than the mode.
2178 * config/aarch64/iterators.md (nunits): New mode attribute.
2179 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2180 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2181 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2182 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2183 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2184 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2185 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2186 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2187 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2188 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2189 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2190 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2191 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2192 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2193 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2194 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2195 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2196 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2197 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2198 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2199 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2200 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2201 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2202 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2203 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2204 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2205 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2206 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2207 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2208 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2209 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2210 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2211 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2212 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2213 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2214 (aarch64_simd_vec_setv2di): Likewise.
2216 2017-11-06 Carl Love <cel@us.ibm.com>
2218 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2220 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2221 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2223 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2225 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2226 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2227 Add power 8 macro expansions.
2228 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2229 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2230 power 8 instructions. (VSX_XXBR): Add iterator.
2232 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2234 * config/arm/arm.md (predicable_short_it): Change default to "no",
2235 improve documentation, remove uses that are identical to the default.
2236 (enabled_for_depr_it): Rename to enabled_for_short_it.
2237 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2238 Remove default uses.
2239 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2240 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2241 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2242 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2244 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2247 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2248 float128 helper macros, which are no longer used after deleting
2249 the old 'q' built-in functions, and moving the round to odd
2250 built-in functions to being special built-in functions.
2251 (BU_FLOAT128_2): Likewise.
2252 (BU_FLOAT128_1_HW): Likewise.
2253 (BU_FLOAT128_2_HW): Likewise.
2254 (BU_FLOAT128_3_HW): Likewise.
2255 (FABSQ): Delete old 'q' built-in functions.
2256 (COPYSIGNQ): Likewise.
2257 (SQRTF128_ODD): Move round to odd built-in functions to be
2258 special built-in functions, so that we can handle
2259 -mabi=ieeelongdouble.
2260 (TRUNCF128_ODD): Likewise.
2261 (ADDF128_ODD): Likewise.
2262 (SUBF128_ODD): Likewise.
2263 (MULF128_ODD): Likewise.
2264 (DIVF128_ODD): Likewise.
2265 (FMAF128_ODD): Likewise.
2266 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2267 built-in names to 'f128'.
2268 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2269 old 'q' built-in functions, as the machine independent code for
2270 'f128' built-in functions handles this.
2271 (rs6000_expand_builtin): Add expansion for float128 round to odd
2272 functions, keying off on -mabi=ieeelongdouble of whether to use
2273 the KFmode or TFmode variant.
2274 (rs6000_init_builtins): Initialize the _Float128 round to odd
2276 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2277 _Float128 'q' built-in functions are now mapped into the new
2278 'f128' built-in functions.
2280 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2282 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2283 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2285 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2288 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2289 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2290 * toplev.c: Include "ipa-fnsummary.h".
2291 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2293 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2295 PR tree-optimization/82838
2296 * gimple-ssa-store-merging.c
2297 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2298 on a separate gimple_seq which is then appended to seq.
2300 2017-11-06 Jeff Law <law@redhat.com>
2303 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2304 (get_probe_interval): New functions.
2305 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2306 (ix86_adjust_stack_and_probe): Likewise.
2307 (output_adjust_stack_and_probe): Likewise.
2308 (ix86_emit_probe_stack_range): Likewise.
2309 (ix86_expand_prologue): Likewise.
2311 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2313 PR tree-optimization/82816
2314 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2315 if the modes of the two types are the same.
2316 (convert_plusminus_to_widen): Likewise.
2318 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2320 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2322 (usadv16qi): New define_expand.
2323 (usadv8hi): New define_expand.
2325 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2328 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2330 (inline_transform): Likewise.
2331 * predict.c (counts_to_freqs): Remove useless conditional.
2332 * profile-count.h (profile_count::apply_scale): Move sanity check.
2333 * tree-inline.c (copy_bb): Always scale.
2334 (copy_cfg_body): Likewise.
2336 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2339 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2341 (*cmp_ite0): Add enabled_for_depr_it attribute.
2342 (*cmp_ite1): Likewise.
2344 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2346 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2349 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2351 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2352 return the value by pointer.
2353 (extract_range_from_multiplicative_op_1): Update accordingly.
2354 Return as soon as an operation fails.
2356 2017-11-05 Tom de Vries <tom@codesourcery.com>
2359 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2360 (DEF_SANITIZER_BUILTIN): ... here.
2361 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2362 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2364 2017-11-05 Tom de Vries <tom@codesourcery.com>
2367 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2369 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2370 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2371 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2373 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2374 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2375 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2377 2017-11-05 Tom de Vries <tom@codesourcery.com>
2380 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2383 2017-11-04 Michael Clark <michaeljclark@mac.com>
2385 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2386 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2388 (*addsi3_extended): Likewise.
2389 (*addsi3_extended2): Likewise.
2390 (<optab>si3): Likewise.
2391 (<optab>di3): Likewise.
2392 (<optab><mode>3): Likewise.
2393 (<*optabe>si3_internal): Likewise.
2394 (zero_extendqi<SUPERQI:mode>2): Likewise.
2395 (*add<mode>hi3): Likewise.
2396 (*xor<mode>hi3): Likewise.
2397 (<optab>di3): Likewise.
2398 (*<optab>si3_extend): Likewise.
2399 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2400 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2401 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2403 2017-11-04 Andrew Waterman <andrew@sifive.com>
2405 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2406 TARGET_STRICT_ALIGN based upon -mtune argument.
2408 2017-11-04 Andrew Waterman <andrew@sifive.com>
2410 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2412 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2414 config/i386/i386.c (choose_basereg): Use optional scratch
2415 register and add assertion.
2416 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2417 needed, and don't allocate stack.
2418 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2419 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2420 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2422 2017-11-03 Jeff Law <law@redhat.com>
2424 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2425 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2426 to probe at the start of a noreturn function.
2428 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2430 PR tree-optimization/78821
2431 * gimple-ssa-store-merging.c: Update the file comment.
2432 (MAX_STORE_ALIAS_CHECKS): Define.
2433 (struct store_operand_info): New type.
2434 (store_operand_info::store_operand_info): New constructor.
2435 (struct store_immediate_info): Add rhs_code and ops data members.
2436 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2437 and op1r arguments to the ctor, initialize corresponding data members.
2438 (struct merged_store_group): Add load_align_base and load_align
2440 (merged_store_group::merged_store_group): Initialize them.
2441 (merged_store_group::do_merge): Update them.
2442 (merged_store_group::apply_stores): Pick the constant for
2443 encode_tree_to_bitpos from one of the two operands, or skip
2444 encode_tree_to_bitpos if neither operand is a constant.
2445 (class pass_store_merging): Add process_store method decl. Remove
2446 bool argument from terminate_all_aliasing_chains method decl.
2447 (pass_store_merging::terminate_all_aliasing_chains): Remove
2448 var_offset_p argument and corresponding handling.
2449 (stmts_may_clobber_ref_p): New function.
2450 (compatible_load_p): New function.
2451 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2452 if there is overlap and rhs_code is not INTEGER_CST. For
2453 non-overlapping stores terminate group if rhs is not mergeable.
2454 (get_alias_type_for_stmts): Change first argument from
2455 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2456 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2457 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2459 (get_location_for_stmts): Change first argument from
2460 auto_vec<gimple *> & to vec<gimple *> &.
2461 (struct split_store): Remove orig_stmts data member, add orig_stores.
2462 (split_store::split_store): Create orig_stores rather than orig_stmts.
2463 (find_constituent_stmts): Renamed to ...
2464 (find_constituent_stores): ... this. Change second argument from
2465 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2466 to info structures rather than the statements.
2467 (split_group): Rename ALLOW_UNALIGNED argument to
2468 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2469 it. Adjust find_constituent_stores caller.
2470 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2471 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2472 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2473 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2474 (mem_valid_for_store_merging): New function.
2475 (handled_load): New function.
2476 (pass_store_merging::process_store): New method.
2477 (pass_store_merging::execute): Use process_store method. Adjust
2478 terminate_all_aliasing_chains caller.
2480 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2482 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2483 Return true for more constants, symbols and label references.
2484 (aarch64_valid_floating_const): Remove unused function.
2486 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2489 * config/aarch64/aarch64.c (aarch64_layout_frame):
2490 Undo forcing of LR at bottom of frame.
2492 2017-11-03 Jeff Law <law@redhat.com>
2495 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2496 for int_registers_saved.
2498 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2499 extracted from tree-ssa-dom.c.
2500 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2501 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2502 (record_equivalences_from_incoming_edge): Add additional argument
2503 to single_pred_edge_ignoring_loop_edges call.
2504 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2505 (uncprop_dom_walker::before_dom_children): Add additional argument
2506 to single_pred_edge_ignoring_loop_edges call.
2507 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2508 single_pred_edge_ignoring_loop_edges rather than open coding.
2509 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2511 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2513 * match.pd (-(-A)): Rewrite.
2515 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2517 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2518 (rs6000_emit_int_cmove): New declaration.
2519 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2520 (rs6000_emit_sISEL): Delete.
2521 (rs6000_emit_int_cmove): Make non-static.
2522 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2523 instead of rs6000_emit_sISEL.
2525 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2527 * asan.c (create_cond_insert_point): Maintain profile.
2528 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2530 * basic-block.h (struct basic_block_def): Remove frequency.
2531 (EDGE_FREQUENCY): Use to_frequency
2532 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2534 (find_traces): Update to use to_frequency.
2535 (find_traces_1_round): Likewise; use only IPA counts.
2536 (bb_to_key): Likewise.
2537 (connect_traces): Use IPA counts only.
2538 (copy_bb_p): Update to use to_frequency.
2539 (fix_up_crossing_landing_pad): Likewise.
2540 (sanitize_hot_paths): Likewise.
2541 * bt-load.c (basic_block_freq): Likewise.
2542 * cfg.c (init_flow): Set count_max to uninitialized.
2543 (check_bb_profile): Remove frequencies; check counts.
2544 (dump_bb_info): Do not dump frequencies.
2545 (update_bb_profile_for_threading): Update counts only.
2546 (scale_bbs_frequencies_int): Likewise.
2547 (MAX_SAFE_MULTIPLIER): Remove.
2548 (scale_bbs_frequencies_gcov_type): Update counts only.
2549 (scale_bbs_frequencies_profile_count): Update counts only.
2550 (scale_bbs_frequencies): Update counts only.
2551 * cfg.h (struct control_flow_graph): Add count-max.
2552 (update_bb_profile_for_threading): Update prototype.
2553 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2554 (find_many_sub_basic_blocks): Likewise.
2555 * cfgcleanup.c (try_forward_edges): Likewise.
2556 (try_crossjump_to_edge): Likewise.
2557 * cfgexpand.c (expand_gimple_cond): Likewise.
2558 (expand_gimple_tailcall): Likewise.
2559 (construct_init_block): Likewise.
2560 (construct_exit_block): Likewise.
2561 * cfghooks.c (verify_flow_info): Check consistency of counts.
2562 (dump_bb_for_graph): Do not dump frequencies.
2563 (split_block_1): Do not update frequencies.
2564 (split_edge): Do not update frequencies.
2565 (make_forwarder_block): Do not update frequencies.
2566 (duplicate_block): Do not update frequencies.
2567 (account_profile_record): Do not update frequencies.
2568 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2569 for global heuristics.
2570 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2571 (expected_loop_iterations_unbounded): Use counts only.
2572 * cfgloopmanip.c (scale_loop_profile): Simplify.
2573 (create_empty_loop_on_edge): Simplify
2575 (duplicate_loop_to_header_edge): Simplify
2576 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2577 (update_br_prob_note): Take care of removing note when profile
2579 (relink_block_chain): Do not dump frequency.
2580 (rtl_account_profile_record): Use to_frequency.
2581 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2582 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2583 (cgraph_update_edges_for_call_stmt_node): Likewise.
2584 (cgraph_edge::verify_count_and_frequency): Update.
2585 (cgraph_node::verify_node): Temporarily disable frequency verification.
2586 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2587 to_cgraph_frequency.
2588 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2589 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2591 (cgraph_node::expand_thunk): Update profile.
2592 * except.c (dw2_build_landing_pads): Do not update frequency.
2593 * final.c (compute_alignments): Use to_frequency.
2594 (dump_basic_block_info): Do not dump frequency.
2595 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2596 (dump_gimple_bb_header): Do not dump frequency.
2597 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2599 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2600 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2601 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2602 (init_before_recovery): Do not update frequency.
2603 (sched_create_recovery_edges): Do not update frequency.
2604 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2605 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2606 (ipa_cp_c_finalize): Set max_count to uninitialized.
2607 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2608 (param_change_prob): Use counts.
2609 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2611 * ipa-split.c (consider_split): Use to_frequency.
2612 (split_function): Use to_frequency.
2613 * ira-build.c (loop_compare_func): Likewise.
2614 (mark_loops_for_removal): Likewise.
2615 (mark_all_loops_for_removal): Likewise.
2616 * loop-doloop.c (doloop_modify): Do not update frequency.
2617 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2619 * lto-streamer-in.c (input_function): Update count_max.
2620 * omp-expand.c (expand_omp_taskreg): Update count_max.
2621 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2622 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2623 (maybe_hot_count_p): Use ipa counts only.
2624 (maybe_hot_bb_p): Simplify.
2625 (maybe_hot_edge_p): Simplify.
2626 (probably_never_executed): Do not take frequency argument.
2627 (probably_never_executed_bb_p): Do not pass frequency.
2628 (probably_never_executed_edge_p): Likewise.
2629 (combine_predictions_for_bb): Check that profile is nonzero.
2630 (propagate_freq): Do not set frequency.
2631 (drop_profile): Simplify.
2632 (counts_to_freqs): Simplify.
2633 (expensive_function_p): Use to_frequency.
2634 (propagate_unlikely_bbs_forward): Simplify.
2635 (determine_unlikely_bbs): Simplify.
2636 (estimate_bb_frequencies): Add hack to silence graphite issues.
2637 (compute_function_frequency): Use ipa counts.
2638 (pass_profile::execute): Update.
2639 (rebuild_frequencies): Use counts only.
2640 (force_edge_cold): Use counts only.
2641 * profile-count.c (profile_count::dump): Dump new count types.
2642 (profile_count::differs_from_p): Check compatiblity.
2643 (profile_count::to_frequency): New function.
2644 (profile_count::to_cgraph_frequency): New function.
2645 * profile-count.h (struct function): Declare.
2646 (enum profile_quality): Add profile_guessed_local and
2647 profile_guessed_global0.
2648 (class profile_proability): Decrease number of bits to 29;
2649 update from_reg_br_prob_note and to_reg_br_prob_note.
2650 (class profile_count: Update comment; decrease number of bits
2651 to 61. Check compatibility.
2652 (profile_count::compatible_p): New private member function.
2653 (profile_count::ipa_p): New member function.
2654 (profile_count::operator<): Handle global zero correctly.
2655 (profile_count::operator>): Handle global zero correctly.
2656 (profile_count::operator<=): Handle global zero correctly.
2657 (profile_count::operator>=): Handle global zero correctly.
2658 (profile_count::nonzero_p): New member function.
2659 (profile_count::force_nonzero): New member function.
2660 (profile_count::max): New member function.
2661 (profile_count::apply_scale): Handle IPA scalling.
2662 (profile_count::guessed_local): New member function.
2663 (profile_count::global0): New member function.
2664 (profile_count::ipa): New member function.
2665 (profile_count::to_frequency): Declare.
2666 (profile_count::to_cgraph_frequency): Declare.
2667 * profile.c (OVERLAP_BASE): Delete.
2668 (compute_frequency_overlap): Delete.
2669 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2670 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2671 * sched-ebb.c (rank): Use counts only.
2672 * shrink-wrap.c (handle_simple_exit): Use counts only.
2673 (try_shrink_wrapping): Use counts only.
2674 (place_prologue_for_one_component): Use counts only.
2675 * tracer.c (find_best_predecessor): Use to_frequency.
2676 (find_trace): Use to_frequency.
2677 (tail_duplicate): Use to_frequency.
2678 * trans-mem.c (expand_transaction): Do not update frequency.
2679 * tree-call-cdce.c: Do not update frequency.
2680 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2681 (gimple_merge_blocks): Likewise.
2682 (gimple_split_edge): Likewise.
2683 (gimple_duplicate_sese_region): Likewise.
2684 (gimple_duplicate_sese_tail): Likewise.
2685 (move_sese_region_to_fn): Likewise.
2686 (gimple_account_profile_record): Likewise.
2687 (insert_cond_bb): Likewise.
2688 * tree-complex.c (expand_complex_div_wide): Likewise.
2689 * tree-eh.c (lower_resx): Update profile.
2690 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2692 (initialize_cfun): Do not initialize frequencies
2693 (freqs_to_counts): Delete.
2694 (copy_cfg_body): Ignore count parameter.
2695 (copy_body): Update.
2696 (expand_call_inline): Update count_max.
2697 (optimize_inline_calls): Update count_max.
2698 (tree_function_versioning): Update count_max.
2699 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2700 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2702 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2703 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2704 (try_peel_loop): Likewise.
2705 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2707 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2708 (tree_transform_and_unroll_loop): Do not use frequencies
2709 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2710 Use reliable prediction only.
2711 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2712 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2713 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2714 probability scaling.
2715 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2716 not update frequency
2717 (any_remaining_duplicated_blocks): Likewise.
2718 (update_profile): Likewise.
2719 (estimated_freqs_path): Delete.
2720 (freqs_to_counts_path): Delete.
2721 (clear_counts_path): Delete.
2722 (ssa_fix_duplicate_block_edges): Likewise.
2723 (duplicate_thread_path): Likewise.
2724 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2725 * tree-tailcall.c (decrease_profile): Do not update frequency.
2726 (eliminate_tail_call): Likewise.
2727 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2728 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2729 (optimize_mask_stores): Likewise.
2730 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2731 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2732 (ubsan_expand_ptr_ifn): Update profile.
2733 * value-prof.c (gimple_ic): Simplify.
2734 * value-prof.h (gimple_ic): Update prototype.
2735 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2736 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2738 (want_inline_self_recursive_call_p): Likewise.
2739 (resolve_noninline_speculation): Only cummulate defined counts.
2740 (inline_small_functions): Use nonzero_p.
2741 (ipa_inline): Do not access freed node.
2743 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2745 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2746 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2748 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2750 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2751 non-legitimate address.
2753 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2755 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2756 (*lt0_<mode>di, *lt0_<mode>si): New.
2758 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2760 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2761 TARGET_PAIRED_FLOAT.
2763 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2764 Jim Wilson <jim.wilson@linaro.org>
2766 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2767 * config/aarch64/aarch64-tune.md: Regenerated.
2768 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2769 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2771 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2773 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2774 (arc_expand_prologue): Restore blink for millicode.
2775 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2777 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2780 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2781 gen_vec_duplicate after forcing the scalar into a register.
2783 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2785 * combine (try_combine): Print the insns input to try_combine to the
2788 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2791 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2792 Remove second argument from aarch64_process_target_attr call.
2793 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2795 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2797 (aarch64_handle_attr_arch): Remove second argument.
2798 (aarch64_handle_attr_cpu): Ditto.
2799 (aarch64_handle_attr_tune): Ditto.
2800 (aarch64_handle_attr_isa_flags): Ditto.
2801 (aarch64_process_one_target_attr): Ditto.
2802 (aarch64_process_target_attr): Ditto.
2803 (aarch64_option_valid_attribute_p): Remove second argument.
2804 on aarch64_process_target_attr call.
2806 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2808 * diagnostic.c: Include "selftest-diagnostic.h".
2809 (selftest::assert_location_text): New function.
2810 (selftest::test_diagnostic_get_location_text): New function.
2811 (selftest::diagnostic_c_tests): Call it.
2813 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2815 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2816 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2817 (class selftest::test_diagnostic_context): Move to...
2818 * selftest-diagnostic.c: New file.
2819 * selftest-diagnostic.h: New file.
2821 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2823 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2824 offset range for FT32B.
2825 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2826 * config/ft32/ft32.md: Add TARGET_NOPM.
2827 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2828 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2830 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2832 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2834 2017-11-02 Jeff Law <law@redhat.com>
2836 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2837 virtual keyword on FINAL OVERRIDE members.
2839 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2840 virtuals together. Add virtual destructor.
2841 (substitute_and_fold_engine): Similarly.
2843 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2845 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2847 2017-11-02 Richard Biener <rguenther@suse.de>
2849 PR tree-optimization/82795
2850 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2852 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2854 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2856 (gcc_GAS_CHECK_FEATURE): Remove.
2857 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2858 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2859 * configure: Regenerate.
2861 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2863 * config/arc/arc.c (hwloop_optimize): Account for empty
2866 2017-11-02 Richard Biener <rguenther@suse.de>
2869 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2870 Truncate ARRAY_REF index and element size.
2872 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2874 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2876 2017-11-01 Jeff Law <law@redhat.com>
2878 * tree-ssa-ccp.c (ccp_folder): New class derived from
2879 substitute_and_fold_engine.
2880 (ccp_folder::get_value): New member function.
2881 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2882 (ccp_fold_stmt): Remove prototype.
2883 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2884 * tree-ssa-copy.c (copy_folder): New class derived from
2885 substitute_and_fold_engine.
2886 (copy_folder::get_value): Renamed from get_value.
2887 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2888 * tree-vrp.c (vrp_folder): New class derived from
2889 substitute_and_fold_engine.
2890 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2891 (vrp_folder::get_value): New member function.
2892 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2893 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2894 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2895 provide a class interface to folder/substitute routines.
2896 (ssa_prop_fold_stmt_fn): Remove typedef.
2897 (ssa_prop_get_value_fn): Likewise.
2898 (subsitute_and_fold): Remove prototype.
2899 (replace_uses_in): Likewise.
2900 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2901 Renamed from replace_uses_in. Call the virtual member function
2902 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2903 (substitute_and_fold_dom_walker): Remove initialization of
2904 data member entries for calbacks. Add substitute_and_fold_engine
2905 member and initialize it.
2906 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2907 member functions for get_value, replace_phi_args_in c
2908 replace_uses_in, and fold_stmt calls.
2909 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2910 substitute_and_fold. Remove assert. Update ctor call.
2912 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2913 (ssa_prop_visit_phi_fn): Likewise.
2914 (class ssa_propagation_engine): New class to provide an interface
2916 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2918 (ssa_prop_visit_phi): Likewise.
2919 (ssa_propagation_engine::simulate_stmt): Moved into class.
2920 Call visit_phi/visit_stmt from the class rather than via
2921 file scoped static variables.
2922 (ssa_propagation_engine::simulate_block): Moved into class.
2923 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2924 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2925 set file scoped statics for the visit_stmt/visit_phi callbacks.
2926 * tree-complex.c (complex_propagate): New class derived from
2927 ssa_propagation_engine.
2928 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2929 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2930 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2932 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2933 ssa_propagation_engine.
2934 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2935 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2936 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2937 * tree-ssa-copy.c (copy_prop): New class derived from
2938 ssa_propagation_engine.
2939 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2940 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2941 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2942 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2943 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2944 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2945 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2947 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2949 PR rtl-optimization/82778
2950 PR rtl-optimization/82597
2951 * compare-elim.c (struct comparison): Add in_a_setter field.
2952 (find_comparison_dom_walker::before_dom_children): Remove killed
2953 bitmap and df_simulate_find_defs call, instead walk the defs.
2954 Compute last_setter and initialize in_a_setter. Merge definitions
2955 with first initialization for a few variables.
2956 (try_validate_parallel): Use insn_invalid_p instead of
2957 recog_memoized. Return insn rather than just the pattern.
2958 (try_merge_compare): Fix up comment. Don't uselessly test if
2959 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2961 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2964 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2965 Alan Hayward <alan.hayward@arm.com>
2966 David Sherwood <david.sherwood@arm.com>
2968 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2969 aarch64_hard_regno_nregs to get the number of registers
2972 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2973 Alan Hayward <alan.hayward@arm.com>
2974 David Sherwood <david.sherwood@arm.com>
2976 * config/aarch64/constraints.md (Upl): Rename to...
2978 * config/aarch64/aarch64.md
2979 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2982 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2983 Alan Hayward <alan.hayward@arm.com>
2984 David Sherwood <david.sherwood@arm.com>
2986 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2987 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2990 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2991 Alan Hayward <alan.hayward@arm.com>
2992 David Sherwood <david.sherwood@arm.com>
2994 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2995 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2996 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2998 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2999 of a single element.
3000 * config/aarch64/iterators.md: Add a comment above the permute
3001 unspecs to say that they are generated directly by
3002 aarch64_expand_vec_perm_const.
3003 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3005 2017-11-01 Nathan Sidwell <nathan@acm.org>
3007 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3009 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3011 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3012 and medany code models, and describe what they do.
3014 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3016 Revert accidental duplicate:
3018 * combine.c (can_change_dest_mode): Reject changes in
3019 REGMODE_NATURAL_SIZE.
3021 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3023 PR rtl-optimization/64682
3024 PR rtl-optimization/69567
3025 PR rtl-optimization/69737
3026 PR rtl-optimization/82683
3027 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3028 register mentioned in the note, drop the note, unless it came from I3,
3029 in which case it should go to I3 again.
3031 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3033 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3034 and return false if not.
3035 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3037 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3039 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3040 range is known to be empty.
3042 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3043 Alan Hayward <alan.hayward@arm.com>
3044 David Sherwood <david.sherwood@arm.com>
3046 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3047 and CONST_VECTOR_NUNITS instead of computing the number of units from
3048 the byte sizes of the vector and element.
3049 (simplify_binary_operation_1): Likewise.
3050 (simplify_const_binary_operation): Likewise.
3051 (simplify_ternary_operation): Likewise.
3053 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3054 Alan Hayward <alan.hayward@arm.com>
3055 David Sherwood <david.sherwood@arm.com>
3057 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3058 (int_mem_offset): ...this new function.
3059 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3060 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3061 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3064 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3065 Alan Hayward <alan.hayward@arm.com>
3066 David Sherwood <david.sherwood@arm.com>
3068 * lower-subreg.c (interesting_mode_p): New function.
3069 (compute_costs, find_decomposable_subregs, decompose_register)
3070 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3071 (resolve_clobber, dump_choices): Use it.
3073 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3074 Alan Hayward <alan.hayward@arm.com>
3075 David Sherwood <david.sherwood@arm.com>
3077 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3079 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3080 Alan Hayward <alan.hayward@arm.com>
3081 David Sherwood <david.sherwood@arm.com>
3083 * alias.c (find_base_value, find_base_term): Only process integer
3084 truncations. Check the precision rather than the size.
3086 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3087 Alan Hayward <alan.hayward@arm.com>
3088 David Sherwood <david.sherwood@arm.com>
3090 * machmode.h (is_narrower_int_mode): New function
3091 * optabs.c (expand_float, expand_fix): Use it.
3092 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3094 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3095 Alan Hayward <alan.hayward@arm.com>
3096 David Sherwood <david.sherwood@arm.com>
3098 * rtl.h (narrower_subreg_mode): New function.
3099 * ira-color.c (update_costs_from_allocno): Use it.
3101 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3102 Alan Hayward <alan.hayward@arm.com>
3103 David Sherwood <david.sherwood@arm.com>
3105 * optabs-query.h (convert_optab_p): New function, split out from...
3106 (convert_optab_handler): ...here.
3107 (widening_optab_handler): Delete.
3108 (find_widening_optab_handler): Remove permit_non_widening parameter.
3109 (find_widening_optab_handler_and_mode): Likewise. Provide an
3110 override that operates on mode class wrappers.
3111 * optabs-query.c (widening_optab_handler): Delete.
3112 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3113 parameter. Assert that the two modes are the same class and that
3114 the "from" mode is narrower than the "to" mode. Use
3115 convert_optab_handler instead of widening_optab_handler.
3116 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3117 instead of widening_optab_handler.
3118 * expr.c (expand_expr_real_2): Update calls to
3119 find_widening_optab_handler.
3120 * optabs.c (expand_widen_pattern_expr): Likewise.
3121 (expand_binop_directly): Take the insn_code as a parameter.
3122 (expand_binop): Only call find_widening_optab_handler for
3123 conversion optabs; use optab_handler otherwise. Update calls
3124 to find_widening_optab_handler and expand_binop_directly.
3125 Use convert_optab_handler instead of widening_optab_handler.
3126 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3127 find_widening_optab_handler and use scalar_mode rather than
3129 (convert_plusminus_to_widen): Likewise.
3131 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3132 Alan Hayward <alan.hayward@arm.com>
3133 David Sherwood <david.sherwood@arm.com>
3135 * machmode.h (fixed_size_mode): New class.
3136 * rtl.h (get_pool_mode): Return fixed_size_mode.
3137 * gengtype.c (main): Add fixed_size_mode.
3138 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3139 (get_raw_arg_mode): Likewise.
3140 * doc/tm.texi: Regenerate.
3141 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3142 * targhooks.c (default_get_reg_raw_mode): Likewise.
3143 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3144 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3145 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3146 (msp430_get_raw_result_mode): Likewise.
3147 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3148 * dbxout.c (dbxout_parms): Require fixed-size modes.
3149 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3150 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3151 * omp-low.c (lower_oacc_reductions): Likewise.
3152 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3153 (simplify_subreg): Update accordingly.
3154 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3155 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3156 that aren't fixed-size.
3157 (get_pool_mode): Return a fixed_size_mode.
3158 (output_constant_pool_2): Take a fixed_size_mode.
3160 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3161 Alan Hayward <alan.hayward@arm.com>
3162 David Sherwood <david.sherwood@arm.com>
3164 * doc/rtl.texi (vec_series): Document.
3165 (const): Say that the operand can be a vec_series.
3166 * rtl.def (VEC_SERIES): New rtx code.
3167 * rtl.h (const_vec_series_p_1): Declare.
3168 (const_vec_series_p): New function.
3169 * emit-rtl.h (gen_const_vec_series): Declare.
3170 (gen_vec_series): Likewise.
3171 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3172 (gen_vec_series): Likewise.
3173 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3174 * simplify-rtx.c (simplify_unary_operation): Handle negations
3176 (simplify_binary_operation_series): New function.
3177 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3178 (test_vector_ops_series): New function.
3179 (test_vector_ops): Call it.
3180 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3181 gen_const_vec_series.
3182 (altivec_lvsr): Likewise.
3183 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3185 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3186 Alan Hayward <alan.hayward@arm.com>
3187 David Sherwood <david.sherwood@arm.com>
3189 * doc/rtl.texi (const): Update description of address constants.
3190 Say that vector constants are allowed too.
3191 * common.md (E, F): Use CONSTANT_P instead of checking for
3193 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3194 checking for CONST_VECTOR.
3195 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3197 * expr.c (expand_expr_real_2): Check for vector modes instead
3198 of checking for CONST_VECTOR.
3199 * rtl.h (const_vec_p): New function.
3200 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3201 (unwrap_const_vec_duplicate): Handle them here too.
3203 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3204 David Malcolm <dmalcolm@redhat.com>
3205 Alan Hayward <alan.hayward@arm.com>
3206 David Sherwood <david.sherwood@arm.com>
3208 * rtl.h (vec_duplicate_p): New function.
3209 * selftest-rtl.c (assert_rtx_eq_at): New function.
3210 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3211 (assert_rtx_eq_at): Declare.
3212 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3213 * selftest-run-tests.c (selftest::run_tests): Call it.
3214 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3215 (simplify_unary_operation_1): Recursively handle vector duplicates.
3216 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3218 (simplify_subreg): Handle subregs of vector duplicates.
3219 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3220 (selftest::simplify_rtx_c_tests): New functions.
3222 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3223 Alan Hayward <alan.hayward@arm.com>
3224 David Sherwood <david.sherwood@arm.com>
3226 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3227 (gen_vec_duplicate): Likewise.
3228 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3230 (gen_const_vector): ...here.
3231 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3232 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3233 whose elements are all equal.
3234 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3235 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3236 (simplify_relational_operation): Likewise.
3237 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3239 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3240 (aarch64_expand_vector_init): Likewise.
3241 * config/arm/arm.c (neon_vdup_constant): Likewise.
3242 (neon_expand_vector_init): Likewise.
3243 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3244 (arm_block_set_unaligned_vect): Likewise.
3245 (arm_block_set_aligned_vect): Likewise.
3246 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3247 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3248 (expand_vec_perm_even_odd_pack): Likewise.
3249 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3250 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3251 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3252 gen_const_vec_duplicate.
3253 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3254 * config/mips/mips.c (mips_gen_const_int_vector): Use
3255 gen_const_vec_duplicate.
3256 (mips_expand_vector_init): Use CONST0_RTX.
3257 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3258 (define_split): Use gen_const_vec_duplicate.
3259 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3260 (define_split): Use gen_const_vec_duplicate.
3261 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3262 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3263 * config/spu/spu.c (spu_const): Likewise.
3265 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3266 Alan Hayward <alan.hayward@arm.com>
3267 David Sherwood <david.sherwood@arm.com>
3269 * combine.c (can_change_dest_mode): Reject changes in
3270 REGMODE_NATURAL_SIZE.
3272 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3274 * configure.ac (--enable-libssp): New.
3275 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3276 trying to determine target-specific default. Adjust indentation.
3277 * configure: Regenerated.
3278 * doc/install.texi (Configuration): Expand --disable-libssp
3281 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3283 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3286 2017-10-31 Martin Jambor <mjambor@suse.cz>
3289 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3291 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3293 * auto-profile.c (autofdo_source_profile::read): Use
3294 UNKNOWN_LOCATION rather than 0.
3295 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3296 (warning_at): ...this overload.
3297 (warning_at_rich_loc_n): Rename to...
3298 (warning_n): ...this overload.
3299 (error_at_rich_loc): Rename to...
3300 (error_at): ...this overload.
3301 (pedwarn_at_rich_loc): Rename to...
3302 (pedwarn): ...this overload.
3303 (permerror_at_rich_loc): Rename to...
3304 (permerror): ...this overload.
3305 (inform_at_rich_loc): Rename to...
3306 (inform): ...this overload.
3307 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3308 (diagnostic_n_impl_richloc): Rename to...
3309 (diagnostic_n_impl): ...this rich_location *-based decl.
3310 (inform_at_rich_loc): Rename to...
3311 (inform): ...this, and add an assertion.
3312 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3313 (warning_at_rich_loc): Rename to...
3314 (warning_at): ...this, and add an assertion.
3315 (warning_at_rich_loc_n): Rename to...
3316 (warning_n): ...this, and add an assertion.
3317 (warning_n): Update location_t-based implementation for removal of
3318 location_t-based diagnostic_n_impl.
3319 (pedwarn_at_rich_loc): Rename to...
3320 (pedwarn): ...this, and add an assertion.
3321 (permerror_at_rich_loc): Rename to...
3322 (permerror): ...this, and add an assertion.
3323 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3324 (error_at_rich_loc): Rename to...
3325 (error_at): ...this, and add an assertion.
3326 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3327 (driver::do_spec_on_infiles): Likewise.
3328 * substring-locations.c (format_warning_va): Update for renaming
3329 of inform_at_rich_loc.
3331 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3333 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3334 _Float<N>X built-in functions so that the variant without the
3335 "__builtin_" prefix is only enabled for the GNU C and Objective C
3336 languages when they are in non-strict ANSI/ISO mode.
3337 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3338 * target.def (floatn_builtin_p): Add a target hook to control
3339 whether _Float<N> and _Float<N>X built-in functions without the
3340 "__builtin_" prefix are enabled, and return true for C and
3341 Objective C in the default hook. Include langhooks.h in
3343 * targhooks.h (default_floatn_builtin_p): Likewise.
3344 * targhooks.c (default_floatn_builtin_p): Likewise.
3345 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3346 floatn_builtin_p target hook.
3347 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3349 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3350 Eric Botcazou <ebotcazou@adacore.com>
3352 PR rtl-optimization/81803
3353 * lra-constraints.c (curr_insn_transform): Also reload the whole
3354 register for a strict subreg no wider than a word if this is for
3355 a WORD_REGISTER_OPERATIONS target.
3357 2017-10-31 Jason Merrill <jason@redhat.com>
3359 * gdbinit.in: Skip over inlines from timevar.h.
3361 2017-10-31 Martin Liska <mliska@suse.cz>
3363 * doc/gcov.texi: Document new option.
3364 * gcov.c (print_usage): Likewise print it.
3365 (process_args): Support the argument.
3366 (format_count): New function.
3367 (format_gcov): Use the function.
3369 2017-10-31 Martin Liska <mliska@suse.cz>
3371 * gcov.c (struct name_map): do not use typedef.
3372 Define operator== and operator<.
3373 (name_search): Remove.
3374 (name_sort): Remove.
3375 (main): Do not allocate names.
3376 (process_file): Add vertical space.
3377 (generate_results): Use std::find.
3378 (release_structures): Do not release memory.
3379 (find_source): Use std::find.
3381 2017-10-31 Martin Liska <mliska@suse.cz>
3383 * gcov.c (struct line_info): Remove it's typedef.
3384 (line_info::line_info): Add proper ctor.
3385 (line_info::has_block): Do not use a typedef.
3386 (struct source_info): Do not use typedef.
3387 (circuit): Likewise.
3388 (get_cycles_count): Likewise.
3389 (output_intermediate_file): Iterate via vector iterator.
3390 (add_line_counts): Use std::vector methods.
3391 (accumulate_line_counts): Likewise.
3392 (output_lines): Likewise.
3394 2017-10-31 Martin Liska <mliska@suse.cz>
3396 * gcov.c (struct source_info): Remove typedef.
3397 (source_info::source_info): Add proper ctor.
3398 (accumulate_line_counts): Use struct, not it's typedef.
3399 (output_gcov_file): Likewise.
3400 (output_lines): Likewise.
3401 (main): Do not allocate an array.
3402 (output_intermediate_file): Use size of vector container.
3403 (process_file): Resize the vector.
3404 (generate_results): Do not preallocate, use newly added vector
3406 (release_structures): Do not release sources.
3407 (find_source): Use vector methods.
3408 (add_line_counts): Do not use typedef.
3410 2017-10-31 Martin Liska <mliska@suse.cz>
3412 * doc/gcov.texi: Document that.
3413 * gcov.c (add_line_counts): Mark lines with a non-executed
3415 (output_line_beginning): Handle such lines.
3416 (output_lines): Pass new argument.
3417 (output_intermediate_file): Print it in intermediate format.
3419 2017-10-31 Martin Liska <mliska@suse.cz>
3421 * color-macros.h: New file.
3422 * diagnostic-color.c: Factor out color related to macros to
3424 * doc/gcov.texi: Document -k option.
3425 * gcov.c (INCLUDE_STRING): Include string.h.
3426 (print_usage): Add -k option.
3427 (process_args): Parse it.
3428 (pad_count_string): New function.
3429 (output_line_beginning): Likewise.
3430 (DEFAULT_LINE_START): New macro.
3431 (output_lines): Support color output.
3433 2017-10-31 Martin Liska <mliska@suse.cz>
3435 PR gcov-profile/82633
3436 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3437 their interaction with GCOV infrastructure.
3438 * configure.ac: Add -fkeep-{inline,static}-functions to
3440 * configure: Regenerate.
3442 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3445 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3447 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3450 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3451 into a register if it does not fit into an immediate offset field.
3453 2017-10-31 Olivier Hainque <hainque@adacore.com>
3455 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3457 2017-10-31 Julia Koval <julia.koval@intel.com>
3459 * config.gcc: Add gfniintrin.h.
3460 * config/i386/gfniintrin.h: New.
3461 * config/i386/i386-builtin-types.def
3462 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3463 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3464 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3465 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3466 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3467 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3468 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3469 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3470 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3471 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3472 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3473 * config/i386/immintrin.h: Include gfniintrin.h.
3474 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3476 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3478 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3480 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3482 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3483 (arm_ashldi3_1bit): Remove pattern.
3484 (ashrdi3): Remove shift by 1 expansion.
3485 (arm_ashrdi3_1bit): Remove pattern.
3486 (lshrdi3): Remove shift by 1 expansion.
3487 (arm_lshrdi3_1bit): Remove pattern.
3488 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3489 cost of ashldi3 by 1.
3490 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3491 (<shift>di3_neon): Likewise.
3493 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3495 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3496 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3497 and (*aarch64_simd_mov<VQ:mode>).
3498 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3499 pattern alternative.
3500 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3501 attributes to match pattern alternative.
3503 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3505 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3506 * config/rs6000/emmintrin.h: New file.
3507 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3509 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3511 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3512 * (movdi_vfp_cortexa8): Remove pattern.
3514 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3516 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3518 (Specific, alpha*-dec-osf5.1): Remove.
3519 (Specific, mips-sgi-irix5): Remove.
3520 (Specific, mips-sgi-irix6): Remove.
3522 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3525 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3526 arguments to clear_bit_region_be.
3528 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3530 * gimplify.c: Include memmodel.h.
3532 2017-10-30 Martin Jambor <mjambor@suse.cz>
3534 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3535 condition whether loop should be executed at all.
3537 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3539 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3540 gimple folding of vec_madd() intrinsics.
3541 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3542 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3543 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3545 2017-10-30 Richard Biener <rguenther@suse.de>
3547 PR tree-optimization/82762
3549 2017-10-23 Richard Biener <rguenther@suse.de>
3551 PR tree-optimization/82129
3553 2017-08-01 Richard Biener <rguenther@suse.de>
3555 PR tree-optimization/81181
3556 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3557 (compute_antic): ... end of iteration here.
3559 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3561 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3562 -std=iso9899:2017 and -std=gnu17.
3563 * doc/standards.texi (C Language): Document C17 support.
3564 * doc/cpp.texi (Overview): Mention -std=c17.
3565 (Standard Predefined Macros): Document C11 and C17 values of
3566 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3567 * doc/extend.texi (Inline): Do not list individual options for
3568 standards newer than C99.
3569 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3571 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3574 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3576 * asan.c (asan_finish_file): Align asan globals array by shadow
3579 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3582 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3583 (struct store_immediate_info): Add bitregion_start and bitregion_end
3585 (store_immediate_info::store_immediate_info): Add brs and bre
3586 arguments and initialize bitregion_{start,end} from those.
3587 (struct merged_store_group): Add bitregion_start, bitregion_end,
3588 align_base and mask fields. Drop unnecessary struct keyword from
3589 struct store_immediate_info. Add do_merge method.
3590 (clear_bit_region_be): Use memset instead of loop storing zeros.
3591 (merged_store_group::do_merge): New method.
3592 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3593 stores as long as the surrounding bitregions have no gaps.
3594 (merged_store_group::merge_overlapping): Use do_merge.
3595 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3596 is byte aligned, rather than requiring that start and width are
3597 byte aligned. Drop unnecessary struct keyword from
3598 struct store_immediate_info. Allocate and populate also mask array.
3599 Make start of the arrays relative to bitregion_start rather than
3600 start and size them according to bitregion_{end,start} difference.
3601 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3602 struct store_immediate_info.
3603 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3604 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3605 struct keyword from struct store_immediate_info.
3606 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3607 between stores as long as the surrounding bitregions have no gaps.
3609 (struct split_store): Add orig non-static data member.
3610 (split_store::split_store): Initialize orig to false.
3611 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3612 if there is exactly a single original stmt. Change stmts argument
3613 to pointer from reference, if NULL, don't push anything to it. Add
3614 first argument, use it to optimize skipping over orig stmts that
3615 are known to be before bitpos already. Simplify.
3616 (split_group): Return unsigned int count how many stores are or
3617 would be needed rather than a bool. Add allow_unaligned argument.
3618 Change split_stores argument from reference to pointer, if NULL,
3619 only do a dry run computing how many stores would be produced.
3620 Rewritten algorithm to use both alignment and misalign if
3621 !allow_unaligned and handle bitfield stores with gaps.
3622 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3623 from bitregion_start instead of start. Compute allow_unaligned
3624 here, if true, do 2 split_group dry runs to compute which one
3625 produces fewer stores and prefer aligned if equal. Punt if
3626 new count is bigger or equal than original before emitting any
3627 statements, rather than during that. Remove no longer needed
3628 new_ssa_names tracking. Replace num_stmts with
3629 split_stores.length (). Use 32-bit stack allocated entries
3630 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3631 if possible. Handle bitfields with gaps.
3632 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3633 Compute bitregion_{start,end} for the stores and construct
3634 store_immediate_info with that. Formatting fixes.
3636 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3639 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3640 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3642 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3644 * gimplify.c: Include tm_p.h.
3646 * common.opt (gcoff): Re-add as ignored option.
3647 (gcoff1, gcoff2, gcoff3): Likewise.
3649 * Makefile.in (OBJS): Delete sdbout.o.
3650 (GTFILES): Delete $(srcdir)/sdbout.c.
3651 * debug.h: Delete sdb_debug_hooks.
3652 * final.c: Delete sdbout.h include.
3653 (final_scan_insn): Delete SDB_DEBUG check.
3654 (rest_of_clean_state): Likewise.
3655 * output.h: Delete sdb_begin_function_line.
3658 * toplev.c: Delete sdbout.h include.
3659 (process_options): Delete SDB_DEBUG check.
3660 * tree-core.h (tree_type_common): Delete pointer field of
3662 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3663 TYPE_SYMTAB_POINTER.
3664 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3665 (TYPE_SYMTAB_IS_POINTER): Delete.
3666 (TYPE_SYMTAB_IS_DIE): Renumber.
3667 * xcoffout.c: Refer to former sdbout.c file.
3668 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3670 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3671 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3672 (Debugging Options): Delete -gcoff.
3673 (-gstabs): Delete SDB reference.
3675 (-gcoff@var{level}): Delete.
3676 * doc/passes.texi (Debugging information output): Delete SDB and
3677 sdbout.c references.
3678 * doc/tm.texi: Regenerate.
3679 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3680 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3682 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3683 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3684 (SDB_DEBUGGING_INFO): Delete.
3685 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3686 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3687 * target.def (output_source_filename): Delete COFF reference.
3689 * common.opt (gcoff): Delete.
3690 (gxcoff+): Update Negative chain.
3691 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3693 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3694 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3695 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3697 (expand_function_start): Change sdb reference to past tense.
3698 (expand_function_end): Change sdb reference to past tense.
3699 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3700 * opts.c (debug_type_names): Delete coff entry.
3701 (common_handle_option): Delete OPT_gcoff case.
3702 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3704 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3705 * config/cris/cris.h: Delete SDB reference in comment.
3706 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3707 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3708 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3709 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3711 (ix86_expand_prologue): Likewise.
3712 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3713 * config/ia64/ia64.h: Likewise.
3714 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3715 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3717 * config/mmix/mmix.h: Likewise.
3718 * config/nds32/nds32.c: Likewise.
3719 * config/stormy/storym16.h: Likewise.
3720 * config/visium/visium.h: Likewise.
3721 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3723 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3725 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3726 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3727 FRAME_POINTER_REGNUM point at high end of local var area.
3729 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3731 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3732 Move comment around. Do not reset best_edge for a copiable
3733 destination if the copy would cause a partition change.
3734 (better_edge_p): Remove redundant check.
3736 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3738 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3740 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3742 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3743 for math functions that have _Float<N> and _Float<N>X variants.
3744 (mathfn_built_in_2): Add support for math functions that have
3745 _Float<N> and _Float<N>X variants.
3746 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3747 (expand_builtin_mathfn_ternary): Add support for fma with
3748 _Float<N> and _Float<N>X variants.
3749 (expand_builtin): Likewise.
3750 (fold_builtin_3): Likewise.
3751 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3752 create math function _Float<N> and _Float<N>X variants as external
3754 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3755 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3756 the __builtin_ prefix and if not strict ansi, without the prefix.
3757 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3758 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3759 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3760 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3761 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3762 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3763 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3764 function signatures for fma _Float<N> and _Float<N>X variants.
3765 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3766 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3767 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3768 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3769 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3770 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3771 * gencfn-macros.c (print_case_cfn): Add support for math functions
3772 that have _Float<N> and _Float<N>X variants.
3773 (print_define_operator_list): Likewise.
3774 (fltfn_suffixes): Likewise.
3776 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3777 for math functions that have _Float<N> and _Float<N>X variants.
3778 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3779 and _Float<N>X variants.
3780 (COPYSIGN): Likewise.
3783 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3784 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3786 (integer_valued_read_call_p): Likewise.
3787 * fold-const-call.c (fold_const_call_ss): Likewise.
3788 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3789 _Float<N> and _Float<N>X variants.
3790 (fold_const_call_ssss): Add support for fma _Float<N> and
3791 _Float<N>X variants.
3792 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3793 support for copysign and fma _Float<N> and _Float<N>X variants.
3794 (backprop::process_builtin_call_use): Likewise.
3795 * tree-call-cdce.c (can_test_argument_range); Add support for
3796 sqrt _Float<N> and _Float<N>X variants.
3797 (edom_only_function): Likewise.
3798 (get_no_error_domain): Likewise.
3799 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3800 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3801 copysign _Float<N> and _Float<N>X variants.
3802 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3803 handled by machine independent code.
3804 (FMAF128): Likewise.
3805 * doc/cpp.texi (Common Predefined Macros): Document defining
3806 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3807 fma _Float<N> and _Float<N>X variants.
3809 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3812 * config/i386/i386-modes.def (CCFPU): Remove definition.
3813 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3814 (ix86_cc_modes_compatible): Ditto.
3815 (ix86_expand_carry_flag_compare): Ditto.
3816 (ix86_expand_int_movcc): Ditto.
3817 (ix86_expand_int_addcc): Ditto.
3818 (ix86_reverse_condition): Ditto.
3819 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3820 Return true/false for unordered/ordered fp comparisons.
3821 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3822 (ix86_prepare_fp_compare_args): Update for rename.
3823 (ix86_expand_fp_compare): Update for rename. Generate unordered
3824 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3825 (ix86_expand_sse_compare_and_jump): Ditto.
3826 * config/i386/predicates.md (fcmov_comparison_operator):
3827 Remove CCFPU mode handling.
3828 (ix86_comparison_operator): Ditto.
3829 (ix86_carry_flag_operator): Ditto.
3830 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3831 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3832 (*cmpu<mode>_cc_i387): Ditto.
3833 (FPCMP): Remove mode iterator.
3834 (unord): Remove mode attribute.
3835 (unord_subst): New define_subst transformation
3836 (unord): New define_subst attribute.
3838 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3839 (*cmpi<unord>xf_i387): Ditto.
3840 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3841 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3842 using unord_subst transformation.
3843 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3844 (round_saeonly): Also handle CCFP mode.
3845 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3846 Remove UNSPEC_SAHF unspec handling.
3848 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3850 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3852 2017-10-27 Jeff Law <law@redhat.com>
3854 * gimple-ssa-sprintf.c: Include domwalk.h.
3855 (class sprintf_dom_walker): New class, derived from dom_walker.
3856 (sprintf_dom_walker::before_dom_children): New function.
3857 (struct call_info): Moved into sprintf_dom_walker class
3858 (compute_formath_length, handle_gimple_call): Likewise.
3859 (sprintf_length::execute): Call the dominator walker rather
3860 than walking the statements.
3862 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3863 gimple statement locations.
3864 (check_array_bounds): Corresponding changes. Get the statement's
3865 location directly from wi->stmt.
3867 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3870 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3872 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3874 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3875 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3877 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3880 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3881 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3882 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3883 maybe_get_pool_constant.
3884 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3887 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3889 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3890 2.26 caveat. Update gas and gld versions.
3891 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3894 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3896 * cgraph.h (set_malloc_flag): Declare.
3897 * cgraph.c (set_malloc_flag_1): New function.
3898 (set_malloc_flag): Likewise.
3899 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3900 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3902 (read_ipa_call_summary): Add support for reading is_return_callee.
3903 (write_ipa_call_summary): Stream is_return_callee.
3904 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3905 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3906 ipa-prop.h, ipa-fnsummary.h.
3907 (pure_const_names): Change to static.
3908 (malloc_state_e): Define.
3909 (malloc_state_names): Define.
3910 (funct_state_d): Add field malloc_state.
3911 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3912 (check_retval_uses): New function.
3913 (malloc_candidate_p): Likewise.
3914 (analyze_function): Add support for malloc attribute.
3915 (pure_const_write_summary): Stream malloc_state.
3916 (pure_const_read_summary): Add support for reading malloc_state.
3917 (dump_malloc_lattice): New function.
3918 (propagate_malloc): New function.
3919 (warn_function_malloc): New function.
3920 (ipa_pure_const::execute): Call propagate_malloc and
3921 ipa_free_fn_summary.
3922 (pass_local_pure_const::execute): Add support for malloc attribute.
3923 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3924 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3926 2017-10-27 Martin Liska <mliska@suse.cz>
3928 PR gcov-profile/82457
3929 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3930 for fork-like functions to be properly instrumented.
3932 2017-10-27 Richard Biener <rguenther@suse.de>
3935 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3936 info when we redirected EH.
3938 2017-10-26 Michael Collison <michael.collison@arm.com>
3940 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3942 (<optab>_trunchf<GPI:mode>2: New pattern.
3943 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3944 * config/aarch64/iterators.md (wv): New mode attribute.
3945 (vf, VF): New mode attributes.
3946 (vgp, VGP): New mode attributes.
3947 (s): Update attribute with SImode and DImode prefixes.
3949 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3951 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3952 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3953 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3954 (nios2_option_overide): Initialize it. Don't allow R0-relative
3955 addressing with PIC.
3956 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3957 (nios2_symbolic_constant_p): Likewise.
3958 (nios2_legitimate_address_p): Likewise.
3959 (nios2_r0rel_section_name_p): New.
3960 (nios2_symbol_ref_in_r0rel_data_p): New.
3961 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3962 (r0rel_constant_p): New.
3963 (nios2_print_operand_address): Handle r0rel_constant_p.
3964 (nios2_cdx_narrow_form_p): Likewise.
3965 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3966 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3967 (Nios II Options): Document -mr0rel-sec.
3969 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3971 * config/nios2/nios2.c: Include xregex.h.
3972 (nios2_gprel_sec_regex): New.
3973 (nios2_option_overide): Initialize it. Don't allow GP-relative
3974 addressing with PIC.
3975 (nios2_small_section_name_p): Check for regex match.
3976 * config/nios2/nios2.opt (mgprel-sec=): New option.
3977 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3978 (Nios II Options): Document -mgprel-sec.
3980 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3982 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3984 2017-10-26 Tom de Vries <tom@codesourcery.com>
3986 PR tree-optimization/82707
3987 * gimple.c (gimple_copy): Fix unsharing of
3988 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3990 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3992 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3993 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3994 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3995 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3996 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3997 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3998 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3999 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
4000 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
4001 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
4002 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
4003 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
4004 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4005 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4006 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4007 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4008 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4010 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4012 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4014 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4015 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4016 warning to rs6000.c. Remove the Undocumented flag, since it has
4018 (-mabi=ibmlongdouble): Likewise.
4019 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4020 already set, set the default format for long double.
4021 (rs6000_debug_reg_global): Print whether long double is IBM or
4023 (rs6000_option_override_internal): Rework setting long double
4024 format. Only warn if the user is changing the long double default
4025 and they did not use -Wno-psabi.
4026 * doc/invoke.texi (PowerPC options): Update the documentation for
4027 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4029 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4030 Alan Hayward <alan.hayward@arm.com>
4031 David Sherwood <david.sherwood@arm.com>
4033 * rtl.h (wider_subreg_mode): New function.
4034 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4035 rather than an unsigned int *.
4036 * ira-color.c (regno_max_ref_width): Replace with...
4037 (regno_max_ref_mode): ...this new variable.
4038 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4039 Use wider_subreg_mode.
4040 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4041 rather than an unsigned int *.
4042 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4043 (process_alt_operands): Likewise.
4044 (invariant_p): Likewise.
4045 * lra-spills.c (assign_mem_slot): Likewise.
4046 (add_pseudo_to_slot): Likewise.
4047 * lra.c (collect_non_operand_hard_regs): Likewise.
4048 (add_regs_to_insn_regno_info): Likewise.
4049 * reload1.c (regno_max_ref_width): Replace with...
4050 (regno_max_ref_mode): ...this new variable.
4051 (reload): Update accordingly. Update call to
4052 ira_sort_regnos_for_alter_reg.
4053 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4054 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4055 (scan_paradoxical_subregs): Likewise.
4057 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4059 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4060 (aarch64_frame): Add emit_frame_chain boolean.
4061 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4062 Move eh_return case to aarch64_layout_frame.
4063 (aarch64_layout_frame): Initialize emit_frame_chain.
4064 (aarch64_expand_prologue): Use emit_frame_chain.
4066 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4068 * config/aarch64/aarch64.c (aarch64_layout_frame):
4069 Ensure LR is always stored at the bottom of the callee-saves.
4070 Remove rarely used frame layout which saves callee-saves at top of
4071 frame, so the store of LR can be used as a valid probe in all cases.
4073 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4075 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4076 Improve unaligned TImode/TFmode base/offset split.
4078 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4079 Alan Hayward <alan.hayward@arm.com>
4080 David Sherwood <david.sherwood@arm.com>
4082 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4083 * combine.c (find_single_use_1): Likewise.
4084 (expand_field_assignment): Likewise.
4085 (move_deaths): Likewise.
4086 * lra-constraints.c (simplify_operand_subreg): Likewise.
4087 (curr_insn_transform): Likewise.
4088 * lra.c (collect_non_operand_hard_regs): Likewise.
4089 (add_regs_to_insn_regno_info): Likewise.
4090 * rtlanal.c (reg_referenced_p): Likewise.
4091 (covers_regno_no_parallel_p): Likewise.
4093 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4095 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4096 Don't print any bits outside the precision of the value.
4097 * wide-int.cc (test_printing): Add some new tests.
4099 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4101 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4102 supports -xbrace_comment option.
4103 * configure: Regenerate.
4104 * config.in: Regenerate.
4105 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4106 (ASM_CPU_SPEC): Use it.
4108 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4110 * target.def (static_rtx_alignment): New hook.
4111 * targhooks.h (default_static_rtx_alignment): Declare.
4112 * targhooks.c (default_static_rtx_alignment): New function.
4113 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4114 * doc/tm.texi: Regenerate.
4115 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4116 instead of targetm.constant_alignment. Remove call to
4118 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4119 (cris_preferred_mininum_alignment): New function, split out from...
4120 (cris_constant_alignment): ...here.
4121 (cris_static_rtx_alignment): New function.
4122 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4124 (ix86_constant_alignment): ...here.
4125 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4126 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4127 (mmix_static_rtx_alignment): New function.
4128 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4129 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4131 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4134 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4135 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4137 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4139 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4140 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4141 mark as a sign-extending load.
4142 (local_pic_load_u): Define.
4144 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4147 * fold-const.c (operand_equal_for_comparison_p): Also return true
4148 if ARG0 is a simple variant of ARG1 with narrower precision.
4149 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4151 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4153 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4155 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4156 scatter_static, scatter_per_elt.
4157 * x86-tune-costs.h: Add new cost entries.
4159 2017-10-25 Richard Biener <rguenther@suse.de>
4161 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4162 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4163 class pass_fre): Move to ...
4164 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4165 class pass_fre): ... here and adjust for statistics.
4167 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4170 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4171 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4172 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4173 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4176 2017-10-25 Richard Biener <rguenther@suse.de>
4178 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4179 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4180 eliminate_push_avail, eliminate_insert): Move inside...
4181 (class eliminate_dom_walker): ... this class in preparation
4183 (fini_eliminate): Remove by merging with ...
4184 (eliminate): ... this function. Adjust for class changes.
4185 (pass_pre::execute): Remove fini_eliminate call.
4186 (pass_fre::execute): Likewise.
4188 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4191 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4192 (VPERMI2, VPERMI2I): New mode iterators.
4193 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4194 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4196 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4197 mode iterator. Remove 3 old define_insn patterns.
4198 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4199 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4200 VPERMI2 mode iterator, remove the other two expanders.
4201 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4202 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4203 instructions, remove the other two patterns.
4204 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4205 mode iterator, remove the other two patterns.
4206 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4207 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4208 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4209 and adjust argument order accordingly.
4210 (ix86_expand_vec_perm): Adjust caller.
4211 (expand_vec_perm_1): Likewise.
4212 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4213 (expand_vec_perm_vpermt2_vpshub2): ... this.
4214 (ix86_expand_vec_perm_const_1): Adjust caller.
4215 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4218 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4219 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4220 (vec_shl_<mode>): Remove unused expander.
4221 (avx512bw_<shift_insn><mode>3): New define_insn.
4222 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4223 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4225 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4228 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4231 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4233 PR rtl-optimization/82396
4234 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4235 (autopref_multipass_init): Simplify initialization.
4236 (autopref_rank_data): Simplify sort order.
4237 * gcc/sched-int.h (autopref_multipass_data_): Remove
4238 multi_mem_insn_p, min_offset and max_offset.
4240 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4243 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4244 Check special value of flag_omit_frame_pointer.
4245 (aarch64_can_eliminate): Likewise.
4246 (aarch64_override_options_after_change_1): Simplify handling of
4247 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4249 2017-10-24 Richard Biener <rguenther@suse.de>
4251 PR tree-optimization/82697
4252 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4253 zero for conditional load and unconditional store.
4255 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4257 * doc/install.texi: Document bootstrap-cet.
4259 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4262 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4263 ENDBR instruction at function entrance if function is only
4266 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4269 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4270 patterns to better describe from which operation the CF is computed.
4271 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4272 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4273 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4274 is 0, use _0 suffixed expanders instead of emitting a comparison
4277 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4279 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4280 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4282 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4285 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4286 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4287 * loop-iv.c (iv_get_reaching_def): Likewise.
4288 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4289 variable is promoted and the partition contains undefined values.
4291 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4293 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4295 (nios2_address_cost): Define.
4296 (nios2_legitimize_address): Recognize (exp + constant) directly.
4297 (TARGET_ADDRESS_COST): Define.
4299 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4301 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4302 (nios2_symbolic_memory_operand_p): Declare.
4303 (nios2_split_large_constant): Declare.
4304 (nios2_split_symbolic_memory_operand): Declare.
4305 * config/nios2/nios2.c: Adjust includes.
4306 (nios2_symbolic_constant_allowed): New.
4307 (nios2_symbolic_constant_p): New.
4308 (nios2_plus_symbolic_constant_p): New.
4309 (nios2_valid_addr_expr_p): Recognize addresses involving
4311 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4312 (nios2_symbolic_memory_operand_p): New.
4313 (nios2_large_constant_p): New.
4314 (nios2_split_large_constant): New.
4315 (nios2_split_plus_large_constant): New.
4316 (nios2_split_symbolic_memory_operand): New.
4317 (nios2_legitimize_address): Code refactoring. Handle addresses
4318 involving symbolic constants.
4319 (nios2_emit_move_sequence): Likewise.
4320 (nios2_print_operand): Improve error output.
4321 (nios2_print_operand_address): Handle LO_SUM.
4322 (nios2_cdx_narrow_form_p): Likewise.
4323 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4324 operands involving symbolic constants.
4325 (movhi_internal, movsi_internal): Likewise.
4326 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4327 (extendhisi2, extendqi<mode>2): Likewise.
4329 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4331 * tree-pass.h (PROP_rtl_split_insns): Define.
4332 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4334 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4336 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4338 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4341 * target.def (const_not_ok_for_debug_p): Default to
4342 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4343 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4344 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4345 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4346 which targetm.const_not_ok_for_debug_p returned true.
4347 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4349 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4351 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4352 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4353 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4355 (ix86_const_not_ok_for_debug_p): New function.
4356 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4357 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4359 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4362 * system.h: Conditionally include "unique-ptr.h" if
4363 INCLUDE_UNIQUE_PTR is defined.
4364 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4365 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4367 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4369 * config/rl78/rl78.md: New define_expand "subdi3".
4371 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4374 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4375 DF_REF_INSN if DF_REF_INSN_INFO is false.
4377 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4379 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4380 xmm_move instead of sse_move.
4381 (sse_store_index): New function.
4382 (ix86_register_move_cost): Be more sensible about mismatch stall;
4383 model AVX moves correctly; make difference between sse->integer and
4385 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4386 moves; make difference between SSE and AVX.
4387 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4388 and zmm_move. Increase size of sse load and store tables;
4389 add unaligned load and store tables; add ssemmx_to_integer.
4390 * x86-tune-costs.h: Update all entries according to real
4391 move latencies from Agner Fog's manual and chip documentation.
4393 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4396 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4397 * config/i386/constraints.md (Wf): New constraint.
4398 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4399 (cmp<dwi>_doubleword): Removed.
4400 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4401 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4402 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4403 expand with cmp<dwi>_doubleword. For LTU and GEU use
4404 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4406 * common.opt (gcolumn-info): Enable by default.
4407 * doc/invoke.texi (gcolumn-info): Document new default.
4409 2017-10-23 Richard Biener <rguenther@suse.de>
4411 PR tree-optimization/82672
4412 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4413 Fold the stmt if we propagated into it.
4415 2017-10-23 Richard Biener <rguenther@suse.de>
4417 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4418 (bitmap_remove_expr_from_set): ... this. All callers call this
4419 for non-constant values.
4420 (bitmap_set_subtract): Rename to...
4421 (bitmap_set_subtract_expressions): ... this. Adjust and
4423 (bitmap_set_contains_value): Remove superfluous check.
4424 (bitmap_set_replace_value): Inline into single caller ...
4425 (bitmap_value_replace_in_set): ... here and simplify.
4426 (dependent_clean): Merge into ...
4427 (clean): ... this using an overload. Adjust.
4428 (prune_clobbered_mems): Adjust.
4429 (compute_antic_aux): Likewise.
4430 (compute_partial_antic_aux): Likewise.
4432 2017-10-23 Richard Biener <rguenther@suse.de>
4434 PR tree-optimization/82129
4436 2017-08-01 Richard Biener <rguenther@suse.de>
4438 PR tree-optimization/81181
4439 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4440 (compute_antic): ... end of iteration here.
4442 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4444 * target.def (starting_frame_offset): New hook.
4445 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4446 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4447 * doc/tm.texi.in: Regenerate.
4448 * hooks.h (hook_hwi_void_0): Declare.
4449 * hooks.c (hook_hwi_void_0): New function.
4450 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4451 STARTING_FRAME_OFFSET.
4452 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4453 * reload1.c (reload): Likewise.
4454 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4455 instead of STARTING_FRAME_OFFSET.
4456 * function.c (try_fit_stack_local): Likewise.
4457 (assign_stack_local_1): Likewise
4458 (instantiate_virtual_regs): Likewise.
4459 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4460 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4461 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4462 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4463 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4464 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4465 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4466 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4467 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4468 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4469 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4470 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4471 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4472 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4473 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4474 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4475 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4476 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4477 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4478 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4479 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4480 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4481 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4482 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4483 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4484 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4485 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4486 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4487 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4488 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4489 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4490 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4491 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4492 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4493 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4494 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4495 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4496 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4497 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4498 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4499 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4500 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4501 return a HOST_WIDE_INT.
4502 (avr_builtin_setjmp_frame_value): Use it instead of
4503 STARTING_FRAME_OFFSET.
4504 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4505 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4506 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4508 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4509 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4510 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4511 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4512 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4513 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4514 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4515 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4516 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4517 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4518 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4519 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4521 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4522 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4523 * config/mips/mips.c (mips_compute_frame_info): Refer to
4524 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4525 (mips_starting_frame_offset): New function.
4526 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4527 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4528 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4529 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4530 and return a HOST_WIDE_INT.
4531 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4532 (mmix_initial_elimination_offset): Refer to
4533 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4534 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4535 * config/pa/pa.c (pa_starting_frame_offset): New function.
4536 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4537 (pa_expand_prologue): Likewise.
4538 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4539 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4540 !FRAME_GROWS_DOWNWARD handling to...
4541 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4542 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4543 !FRAME_GROWS_DOWNWARD handling to...
4544 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4545 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4546 !FRAME_GROWS_DOWNWARD handling to...
4547 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4548 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4550 (rs6000_starting_frame_offset): New function.
4551 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4552 !FRAME_GROWS_DOWNWARD handling to...
4553 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4554 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4555 !FRAME_GROWS_DOWNWARD handling to...
4556 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4557 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4558 !FRAME_GROWS_DOWNWARD handling to...
4559 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4560 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4561 (rs6000_starting_frame_offset): New function.
4562 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4563 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4564 * config/vax/vax.c (vax_starting_frame_offset): New function.
4565 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4566 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4567 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4568 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4569 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4570 * system.h (STARTING_FRAME_OFFSET): Poison.
4572 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4574 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4575 SCALAR_TYPE_MODE instead of TYPE_MODE.
4577 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4578 Alan Hayward <alan.hayward@arm.com>
4579 David Sherwood <david.sherwood@arm.com>
4581 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4583 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4584 Alan Hayward <alan.hayward@arm.com>
4585 David Sherwood <david.sherwood@arm.com>
4587 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4589 2017-10-23 Richard Biener <rguenther@suse.de>
4591 PR tree-optimization/82129
4592 * tree-ssa-pre.c (bitmap_set_and): Remove.
4593 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4594 canonicalizing expressions in the set to those with lowest
4595 ID rather than taking that from the first edge.
4597 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4599 * combine.c (rtx_equal_for_field_assignment_p): Use
4600 byte_lowpart_offset.
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 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4607 to a SUBREG_PROMOTED_VAR.
4609 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4610 Alan Hayward <alan.hayward@arm.com>
4611 David Sherwood <david.sherwood@arm.com>
4613 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4614 (expand_debug_source_expr): Likewise.
4615 * combine.c (combine_simplify_rtx): Likewise.
4616 * cse.c (fold_rtx): Likewise.
4617 * optabs.c (expand_float): Likewise.
4618 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4619 (simplify_binary_operation_1): Likewise.
4621 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4622 Alan Hayward <alan.hayward@arm.com>
4623 David Sherwood <david.sherwood@arm.com>
4625 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4626 (record_promoted_value): Likewise.
4627 * expr.c (expand_expr_real_2): Likewise.
4628 * ree.c (update_reg_equal_equiv_notes): Likewise.
4629 (combine_set_extension): Likewise.
4630 * rtlanal.c (low_bitmask_len): Likewise.
4631 * simplify-rtx.c (neg_const_int): Likewise.
4632 (simplify_binary_operation_1): Likewise.
4634 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4635 Alan Hayward <alan.hayward@arm.com>
4636 David Sherwood <david.sherwood@arm.com>
4638 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4639 * regcprop.c (maybe_mode_change): Likewise.
4640 * reload1.c (alter_reg): Likewise.
4642 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4644 * inchash.h (inchash::hash::add_wide_int): New function.
4645 * lto-streamer-out.c (hash_tree): Use it.
4647 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4649 * inchash.h (inchash::hash::add_wide_int): Rename to...
4650 (inchash::hash::add_hwi): ...this.
4651 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4652 (polymorphic_call_target_hasher::hash): Likewise.
4653 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4654 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4655 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4656 * lto-streamer-out.c (hash_tree): Likewise.
4657 * optc-save-gen.awk: Likewise.
4658 * tree.c (add_expr): Likewise.
4660 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4663 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4664 for ordered inequality comparisons even with TARGET_IEEE_FP.
4666 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4669 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4670 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4671 Expand with cmp<dwi>_doubleword.
4673 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4675 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4676 List CET intrinsics.
4677 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4678 specific to -fcf-protection option.
4680 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4682 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4683 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4684 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4685 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4686 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4687 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4688 (extra_objs): Add cet.o for Linux/x86 targets.
4689 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4690 * config/i386/cet.c: New file.
4691 * config/i386/cetintrin.h: Likewise.
4692 * config/i386/t-cet: Likewise.
4693 * config/i386/cpuid.h (bit_SHSTK): New.
4694 (bit_IBT): Likewise.
4695 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4696 pass IBT and SHSTK bits.
4697 * config/i386/i386-builtin-types.def
4698 (VOID_FTYPE_UNSIGNED_PVOID): New.
4699 (VOID_FTYPE_UINT64_PVOID): Likewise.
4700 * config/i386/i386-builtin.def: Add CET intrinsics.
4701 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4702 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4703 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4704 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4706 * config/i386/i386.c (rest_of_insert_endbranch): New.
4707 (pass_data_insert_endbranch): Likewise.
4708 (pass_insert_endbranch): Likewise.
4709 (make_pass_insert_endbranch): Likewise.
4710 (ix86_notrack_prefixed_insn_p): Likewise.
4711 (ix86_target_string): Add -mibt, -mshstk flags.
4712 (ix86_option_override_internal): Add flag_cf_protection
4714 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4715 (ix86_print_operand): Add 'notrack' prefix output.
4716 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4717 (ix86_expand_builtin): Expand CET intrinsics.
4718 (x86_output_mi_thunk): Add 'endbranch' instruction.
4719 * config/i386/i386.h (TARGET_IBT): New.
4720 (TARGET_IBT_P): Likewise.
4721 (TARGET_SHSTK): Likewise.
4722 (TARGET_SHSTK_P): Likewise.
4723 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4724 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4725 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4726 (builtin_setjmp_setup): New pattern.
4727 (builtin_longjmp): Likewise.
4728 (rdssp<mode>): Likewise.
4729 (incssp<mode>): Likewise.
4730 (saveprevssp): Likewise.
4731 (rstorssp): Likewise.
4732 (wrss<mode>): Likewise.
4733 (wruss<mode>): Likewise.
4734 (setssbsy): Likewise.
4735 (clrssbsy): Likewise.
4736 (nop_endbr): Likewise.
4737 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4739 * config/i386/immintrin.h: Include <cetintrin.h>.
4740 * config/i386/linux-common.h
4741 (file_end_indicate_exec_stack_and_cet): New prototype.
4742 (TARGET_ASM_FILE_END): New.
4744 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4746 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4747 latencies instead of having separate table; make difference between
4748 integer and float costs.
4749 * i386.h (processor_costs): Remove scalar_stmt_cost,
4750 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4751 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4753 * x86-tune-costs.h: Remove entries which has been removed in
4754 procesor_costs from all tables; make cond_taken_branch_cost
4755 and cond_not_taken_branch_cost COST_N_INSNS based.
4757 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4759 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4761 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4763 * config/i386/i386.md (isa): Remove fma_avx512f.
4764 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4765 <avx512>_fmadd_<mode>_mask3<round_name>,
4766 <avx512>_fmsub_<mode>_mask<round_name>,
4767 <avx512>_fmsub_<mode>_mask3<round_name>,
4768 <avx512>_fnmadd_<mode>_mask<round_name>,
4769 <avx512>_fnmadd_<mode>_mask3<round_name>,
4770 <avx512>_fnmsub_<mode>_mask<round_name>,
4771 <avx512>_fnmsub_<mode>_mask3<round_name>,
4772 <avx512>_fmaddsub_<mode>_mask<round_name>,
4773 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4774 <avx512>_fmsubadd_<mode>_mask<round_name>,
4775 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4776 (*vec_widen_umult_even_v16si<mask_name>,
4777 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4778 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4780 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4782 * extend.texi: Add 'nocf_check' documentation.
4783 * gimple.texi: Add second parameter to
4784 gimple_build_call_from_tree.
4785 * invoke.texi: Add -fcf-protection documentation.
4786 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4788 2017-10-20 Richard Biener <rguenther@suse.de>
4790 PR tree-optimization/82473
4791 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4792 the largest input type.
4794 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4796 * c-attribs.c (handle_nocf_check_attribute): New function.
4797 (c_common_attribute_table): Add 'nocf_check' handling.
4798 * gimple-parser.c: Add second argument NULL to
4799 gimple_build_call_from_tree.
4800 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4801 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4803 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4804 * common.opt: Add fcf-protection flag.
4805 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4806 * flag-types.h: Add enum cf_protection_level.
4807 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4808 Add 'nocf_check' attribute propagation to gimple call.
4809 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4810 (gimple_build_call_from_tree): Update prototype.
4811 (gimple_call_nocf_check_p): New function.
4812 (gimple_call_set_nocf_check): Likewise.
4813 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4814 * ipa-icf.c: Add nocf_check attribute in statement hash.
4815 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4816 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4817 * toplev.c (process_options): Add flag_cf_protection handling.
4819 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4821 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4823 2017-10-20 Richard Biener <rguenther@suse.de>
4825 PR tree-optimization/82603
4826 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4827 remove false predicated stores.
4829 2017-10-20 Richard Biener <rguenther@suse.de>
4831 * graphite-isl-ast-to-gimple.c
4832 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4833 Remove return value and simplify, dump copied stmt after lhs
4835 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4836 Reduce dump verbosity.
4837 (gsi_insert_earliest): Likewise.
4838 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4839 * graphite.c (print_global_statistics): Adjust dumping.
4840 (print_graphite_scop_statistics): Likewise.
4841 (print_graphite_statistics): Do not dump loops here.
4842 (graphite_transform_loops): But here.
4844 2017-10-20 Nicolas Roche <roche@adacore.com>
4846 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4847 * configure: Regenerate.
4849 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4852 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4853 functions when optimizing replace GIMPLE_RETURN stmts with
4854 calls to __builtin_unreachable ().
4857 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4858 for -fsanitize=thread link of executables.
4859 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4860 link of executables.
4863 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4865 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4866 define_insns for logical vector shifts to use VI248_AVX512BW
4867 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4868 condition, useless isa and prefix attributes. Change the first
4869 2 of these define_insns to ...
4870 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4871 define_insn for avx512vl.
4872 (<shift_insn><mode>3): ... and this, new define_insn without
4873 masking for non-avx512vl.
4876 * config/i386/sse.md (*andnot<mode>3,
4877 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4878 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4879 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4880 not applied use empty suffix even for TARGET_AVX512VL.
4881 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4882 is applied, supply evex,evex or evex,evex,evex instead of just
4885 2017-10-20 Julia Koval <julia.koval@intel.com>
4887 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4888 (OPTION_MASK_ISA_GFNI_UNSET): New.
4889 (ix86_handle_option): Handle OPT_mgfni.
4890 * config/i386/cpuid.h (bit_GFNI): New.
4891 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4892 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4893 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4894 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4895 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4896 * config/i386/i386.opt: Add mgfni.
4898 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4900 * config/msp430/msp430.c (msp430_option_override): Disable
4901 -fdelete-null-pointer-checks.
4902 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4904 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4906 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4907 of x87 and SSE instructions.
4909 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4911 * asan.c (create_cond_insert_point): Do not update edge count.
4912 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4913 (afdo_propagate_circuit): Likewise.
4914 (afdo_calculate_branch_prob): Likewise.
4915 (afdo_annotate_cfg): Likewise.
4916 * basic-block.h (struct edge_def): Remove count.
4917 (edge_def::count): New accessor.
4918 * bb-reorder.c (rotate_loop): Update.
4919 (find_traces_1_round): Update.
4920 (connect_traces): Update.
4921 (sanitize_hot_paths): Update.
4922 * cfg.c (unchecked_make_edge): Update.
4923 (make_single_succ_edge): Update.
4924 (check_bb_profile): Update.
4925 (dump_edge_info): Update.
4926 (update_bb_profile_for_threading): Update.
4927 (scale_bbs_frequencies_int): Update.
4928 (scale_bbs_frequencies_gcov_type): Update.
4929 (scale_bbs_frequencies_profile_count): Update.
4930 (scale_bbs_frequencies): Update.
4931 * cfganal.c (connect_infinite_loops_to_exit): Update.
4932 * cfgbuild.c (compute_outgoing_frequencies): Update.
4933 (find_many_sub_basic_blocks): Update.
4934 * cfgcleanup.c (try_forward_edges): Update.
4935 (try_crossjump_to_edge): Update
4936 * cfgexpand.c (expand_gimple_cond): Update
4937 (expand_gimple_tailcall): Update
4938 (construct_exit_block): Update
4939 * cfghooks.c (verify_flow_info): Update
4940 (redirect_edge_succ_nodup): Update
4941 (split_edge): Update
4942 (make_forwarder_block): Update
4943 (duplicate_block): Update
4944 (account_profile_record): Update
4945 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4946 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4947 * cfgloopmanip.c (scale_loop_profile): Update.
4949 (lv_adjust_loop_entry_edge): Update.
4950 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4951 (force_nonfallthru_and_redirect): Update.
4952 (purge_dead_edges): Update.
4953 (rtl_flow_call_edges_add): Update.
4954 * cgraphunit.c (init_lowered_empty_function): Update.
4955 (cgraph_node::expand_thunk): Update.
4956 * gimple-pretty-print.c (dump_probability): Update.
4957 (dump_edge_probability): Update.
4958 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4959 * haifa-sched.c (sched_create_recovery_edges): Update.
4960 * hsa-gen.c (convert_switch_statements): Update.
4961 * ifcvt.c (dead_or_predicable): Update.
4962 * ipa-inline-transform.c (inline_transform): Update.
4963 * ipa-split.c (split_function): Update.
4964 * ipa-utils.c (ipa_merge_profiles): Update.
4965 * loop-doloop.c (add_test): Update.
4966 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4967 * lto-streamer-in.c (input_cfg): Update.
4968 (input_function): Update.
4969 * lto-streamer-out.c (output_cfg): Update.
4970 * modulo-sched.c (sms_schedule): Update.
4971 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4972 * predict.c (maybe_hot_edge_p): Update.
4973 (unlikely_executed_edge_p): Update.
4974 (probably_never_executed_edge_p): Update.
4975 (dump_prediction): Update.
4976 (drop_profile): Update.
4977 (propagate_unlikely_bbs_forward): Update.
4978 (determine_unlikely_bbs): Update.
4979 (force_edge_cold): Update.
4980 * profile.c (compute_branch_probabilities): Update.
4981 * reg-stack.c (better_edge): Update.
4982 * shrink-wrap.c (handle_simple_exit): Update.
4983 * tracer.c (better_p): Update.
4984 * trans-mem.c (expand_transaction): Update.
4985 (split_bb_make_tm_edge): Update.
4986 * tree-call-cdce.c: Update.
4987 * tree-cfg.c (gimple_find_sub_bbs): Update.
4988 (gimple_split_edge): Update.
4989 (gimple_duplicate_sese_region): Update.
4990 (gimple_duplicate_sese_tail): Update.
4991 (gimple_flow_call_edges_add): Update.
4992 (insert_cond_bb): Update.
4993 (execute_fixup_cfg): Update.
4994 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4995 * tree-complex.c (expand_complex_div_wide): Update.
4996 * tree-eh.c (lower_resx): Update.
4997 (unsplit_eh): Update.
4998 (cleanup_empty_eh_move_lp): Update.
4999 * tree-inline.c (copy_edges_for_bb): Update.
5000 (freqs_to_counts): Update.
5001 (copy_cfg_body): Update.
5002 * tree-ssa-dce.c (remove_dead_stmt): Update.
5003 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
5004 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5005 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5006 (unloop_loops): Update.
5007 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5008 * tree-ssa-loop-split.c (connect_loops): Update.
5009 (split_loop): Update.
5010 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5011 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5012 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5013 * tree-ssa-reassoc.c (branch_fixup): Update.
5014 * tree-ssa-tail-merge.c (replace_block_by): Update.
5015 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5016 (compute_path_counts): Update.
5017 (update_profile): Update.
5018 (recompute_probabilities): Update.
5019 (update_joiner_offpath_counts): Update.
5020 (estimated_freqs_path): Update.
5021 (freqs_to_counts_path): Update.
5022 (clear_counts_path): Update.
5023 (ssa_fix_duplicate_block_edges): Update.
5024 (duplicate_thread_path): Update.
5025 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5026 (case_bit_test_cmp): Update.
5027 (collect_switch_conv_info): Update.
5028 (gen_inbound_check): Update.
5029 (do_jump_if_equal): Update.
5030 (emit_cmp_and_jump_insns): Update.
5031 * tree-tailcall.c (decrease_profile): Update.
5032 (eliminate_tail_call): Update.
5033 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5034 (vect_do_peeling): Update.
5035 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5036 * ubsan.c (ubsan_expand_null_ifn): Update.
5037 (ubsan_expand_ptr_ifn): Update.
5038 * value-prof.c (gimple_divmod_fixed_value): Update.
5039 (gimple_mod_pow2): Update.
5040 (gimple_mod_subtract): Update.
5041 (gimple_ic): Update.
5042 (gimple_stringop_fixed_value): Update.
5044 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5047 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5049 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5051 PR rtl-optimization/82395
5052 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5053 based on non_spilled_static_chain_regno_p.
5055 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5057 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5058 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5059 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5061 2017-10-19 Martin Sebor <msebor@redhat.com>
5063 PR tree-optimization/82596
5064 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5066 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5068 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5069 (get_mem_refs_of_builtin_call): Likewise.
5070 * builtins.c (expand_builtin_apply): Adjust call to
5071 allocate_dynamic_stack_space.
5072 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5073 the third argument to allocate_dynamic_stack_space, otherwise -1.
5074 (expand_builtin): Deal with all alloca variants.
5075 (is_inexpensive_builtin): Likewise.
5076 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5077 * calls.c (special_function_p): Deal with all alloca variants.
5078 (initialize_argument_information): Adjust call to
5079 allocate_dynamic_stack_space.
5080 (expand_call): Likewise.
5081 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5082 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5083 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5084 use it for the stack usage computation.
5085 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5086 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5087 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5088 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5089 (in_loop_p): Remove first argument and useless check.
5090 (pass_walloca::execute): Remove useless test and adjust call to above.
5091 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5092 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5093 (gimplify_call_expr): Deal with all alloca variants.
5094 * hsa-gen.c (gen_hsa_alloca): Likewise.
5095 (gen_hsa_insns_for_call): Likewise.
5096 * ipa-pure-const.c (special_builtin_state): Likewise.
5097 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5098 * tree-object-size.c (alloc_object_size): Likewise.
5099 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5100 (call_may_clobber_ref_p_1): Likewise.
5101 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5102 (ccp_fold_stmt): Likewise.
5103 (optimize_stack_restore): Likewise.
5104 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5105 (mark_all_reaching_defs_necessary_1): Likewise.
5106 (propagate_necessity): Likewise.
5107 (eliminate_unnecessary_stmts): Likewise.
5108 * tree.c (build_common_builtin_nodes): Build
5109 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5110 (build_alloca_call_expr): New function.
5111 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5112 (CASE_BUILT_IN_ALLOCA): Likewise.
5113 (build_alloca_call_expr): Declare.
5114 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5116 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5119 * dwarf2out.c (new_die_raw): New static inline function.
5120 (new_die): Use it to create the DIE.
5121 (add_AT_external_die_ref): Likewise.
5122 (clone_die): Likewise.
5123 (clone_as_declaration): Likewise.
5124 (dwarf2out_vms_debug_main_pointer): Likewise.
5125 (base_type_die): Likewise. Remove early return for corner cases.
5126 Do not call add_pubtype on the DIE here.
5127 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5128 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5129 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5130 native order exists for base types, attach the DIE manually and call
5131 add_pubtype on it. Do not equate a reverse order DIE to the type.
5133 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5135 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5136 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5138 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5140 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5142 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5143 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5144 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5145 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5146 * opts.c (sanitizer_opts): Add builtin.
5147 * ubsan.c (instrument_builtin): New function.
5148 (pass_ubsan::execute): Call it.
5149 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5150 * doc/invoke.texi: Document -fsanitize=builtin.
5152 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5153 builtins, store max (log2 (align), 0) into uchar field instead of
5154 align into uptr field.
5155 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5156 store uchar 0 field instead of uptr 0 field.
5157 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5158 instead of passing one address of struct with 2 locations pass
5159 two addresses of structs with 1 location each.
5160 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5161 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5162 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5163 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5164 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5165 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5166 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5167 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5169 2017-10-19 Martin Liska <mliska@suse.cz>
5172 * file-find.c (remove_prefix): Remove.
5173 * file-find.h (remove_prefix): Likewise.
5174 * gcc-ar.c: Remove smartness of lookup.
5176 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5178 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5179 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5180 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5182 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5185 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5186 (setcc + and to xor + setcc): New peephole2.
5188 2017-10-19 Tom de Vries <tom@codesourcery.com>
5190 * doc/sourcebuild.texi (Test Directives, Variants of
5191 dg-require-support): Add dg-require-stack-size.
5193 2017-10-19 Martin Liska <mliska@suse.cz>
5196 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5197 that have a large alignment.
5198 (gimplify_target_expr): Likewise.
5200 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5202 PR rtl-optimization/82602
5203 * ira.c (rtx_moveable_p): Return false for volatile asm.
5205 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5208 * config/i386/i386-modes.def (CCGZ): New CC mode.
5209 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5210 * config/i386/predicates.md (ix86_comparison_operator):
5212 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5213 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5214 with double-word subtraction.
5215 (put_condition_code): Handle CCGZmode.
5217 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5219 * wide-int.cc (debug (const wide_int &)): New.
5220 (debug (const wide_int *)): New.
5221 (debug (const widest_int &)): New.
5222 (debug (const widest_int *)): New.
5224 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5227 * lra-constraints.c (curr_insn_transform): Use non-input operand
5228 instead of output one for matched reload.
5230 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5232 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5233 (tree-ssa-loop-ivopts.h): New header file.
5234 (struct builtin_info): New fields.
5235 (classify_builtin_1): Compute and record base and offset parts for
5236 memset builtin partition by calling strip_offset.
5237 (offset_cmp, fuse_memset_builtins): New functions.
5238 (finalize_partitions): Fuse adjacent memset partitions by calling
5240 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5241 Expose the interface.
5242 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5244 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5246 PR tree-optimization/82574
5247 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5248 that data reference must be executed exactly once per iteration
5249 against the outermost loop in nest.
5250 (classify_partition): Update call to above function.
5252 2017-10-18 Richard Biener <rguenther@suse.de>
5254 PR tree-optimization/82591
5255 * graphite.c (graphite_transform_loops): Move code gen message
5257 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5258 Here. Handle scop_to_isl_ast failing.
5259 (scop_to_isl_ast): Limit the number of ISL operations.
5261 2017-10-18 Richard Biener <rguenther@suse.de>
5263 * graphite-isl-ast-to-gimple.c
5264 (translate_isl_ast_to_gimple::set_rename): Simplify.
5265 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5266 (graphite_copy_stmts_from_block): ... here.
5267 (copy_bb_and_scalar_dependences): Simplify.
5268 (add_parameters_to_ivs_params): Canonicalize.
5269 (generate_entry_out_of_ssa_copies): Simplify.
5270 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5271 by passing in ISL dimension.
5272 (parameter_index_in_region_1): Rename to ...
5273 (parameter_index_in_region): ... this.
5274 (extract_affine): Adjust assert, pass down parameter index.
5275 (add_param_constraints): Use range-info when available.
5276 (build_scop_context): Adjust.
5277 * sese.c (new_sese_info): Adjust.
5278 (free_sese_info): Likewise.
5279 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5280 Remove unused typedefs.
5281 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5283 2017-10-18 Martin Liska <mliska@suse.cz>
5285 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5287 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5289 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5290 (s390_sched_init): Do not reset s390_sched_state if we entered the
5291 current basic block via a fallthru edge and all others are unlikely.
5293 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5295 * config/s390/s390.c (NUM_SIDES): New variable.
5296 (LONGRUNNING_THRESHOLD): New variable.
5297 (LATENCY_FACTOR): New variable.
5298 (s390_sched_score): Decrease score for long-running instructions on
5300 (s390_sched_variable_issue): Perform bookkeeping for long-running
5303 2017-10-18 Richard Biener <rguenther@suse.de>
5305 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5306 Simplify with removal of the parameter rename map.
5307 (set_rename): Likewise.
5308 (should_copy_to_new_region): Likewise.
5309 (graphite_copy_stmts_from_block): Likewise.
5310 (copy_bb_and_scalar_dependences): Remove initialization of
5311 unused copied_bb_map.
5313 (copy_internal_parameters): Likewise.
5314 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5315 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5316 Use INTEGRAL_TYPE_P.
5317 (parameter_index_in_region_1): Rename to ...
5318 (assign_parameter_index_in_region): ... this. Assert we have
5319 a parameter we handle.
5320 (scan_tree_for_params): Adjust.
5321 * sese.h (parameter_rename_map_t): Remove.
5322 (struct sese_info_t): Remove unused parameter_rename_map and
5323 copied_bb_map members.
5324 * sese.c (new_sese_info): Adjust.
5325 (free_sese_info): Likewise.
5327 2017-10-18 Martin Liska <mliska@suse.cz>
5330 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5331 on an abnormal edge.
5333 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5335 * doc/invoke.texi (ffunction-sections and fdata-sections):
5338 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5340 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5341 the use statement can throw internally.
5343 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5345 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5346 any RTX present on the RHS of a SET.
5347 * compare-elim.c (try_eliminate_compare): Restore comment.
5349 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5351 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5352 may be also called on tcc_type nodes.
5353 * langhooks.c (lhd_tree_size): Likewise.
5355 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5357 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5358 format_warning_at_substring.
5359 (maybe_warn): Convert source_range * param to a location_t. Pass
5360 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5361 (format_directive): Remove code to extract source_ranges and
5362 source_range * in favor of just a location_t.
5363 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5365 * substring-locations.c (format_warning_va): Convert
5366 source_range * param to a location_t.
5367 (format_warning_at_substring): Likewise.
5368 * substring-locations.h (format_warning_va): Likewise.
5369 (format_warning_at_substring): Likewise.
5371 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5373 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5375 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5376 and scatter/gather ops.
5378 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5379 vec_gather_load and vec_scatter_store.
5380 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5381 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5382 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5383 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5384 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5385 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5387 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5389 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5390 Detect FTST instruction and handle its register pops. Only pop
5391 second operand if can_pop_second_op is true.
5392 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5393 set can_pop_second_op to false in the compare_for_stack_reg call.
5395 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5396 output_fp_compare for stack register operands.
5397 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5398 instructions here. Do not emit stack register pops here. Assert
5399 that FCOMPP pops next to top stack register. Rewrite function.
5401 2017-10-17 Nathan Sidwell <nathan@acm.org>
5404 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5405 use DECL_ASSEMBLER_NAME_RAW.
5408 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5411 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5412 Wilco Dijkstra <wilco.dijkstra@arm.com>
5414 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5415 converstion to Pmode from the buf_addr.
5417 2017-10-17 Richard Biener <rguenther@suse.de>
5419 * graphite-dependences.c (scop_get_reads_and_writes): Change
5420 output parameters to references.
5422 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5424 PR 71026/tree-optimization
5425 * fold-const.c (distribute_real_division): Removed.
5426 (fold_binary_loc): Remove calls to distribute_real_divison.
5428 2017-10-17 Richard Biener <rguenther@suse.de>
5430 * graphite-scop-detection.c
5431 (scop_detection::stmt_has_simple_data_refs_p): Always use
5432 the full nest as region.
5433 (try_generate_gimple_bb): Likewise.
5434 * sese.c (scalar_evolution_in_region): Simplify now that
5435 SCEV can handle instantiation in regions.
5436 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5437 in the non-loop part of a function if requested.
5439 2017-10-17 Richard Biener <rguenther@suse.de>
5441 PR tree-optimization/82563
5442 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5444 (graphite_regenerate_ast_isl): Call it.
5445 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5447 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5449 PR tree-optimization/82549
5450 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5451 Formatting fixes. Instead of calling make_bit_field_ref with negative
5454 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5456 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5457 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5458 _mm_maskz_reduce_ss): New.
5459 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5460 __builtin_ia32_reducess_mask): Ditto..
5461 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5462 * config/i386/sse.md (reduces<mode>): Renamed to ...
5463 (reduces<mode><mask_scalar_name>): ... this.
5464 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5466 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5467 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5469 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5471 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5472 * selftest-run-tests.c (selftest::run_tests): Call
5473 selftest::unique_ptr_tests_cc_tests.
5474 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5475 * unique-ptr-tests.cc: New file.
5477 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5480 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5482 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5483 (make_hard_regno_born, make_hard_regno_dead): Update
5484 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5486 2017-10-16 Jeff Law <law@redhat.com>
5488 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5490 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5492 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5494 2017-10-16 Olivier Hainque <hainque@adacore.com>
5496 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5497 with_cpu if we were configured for an e500v2 target cpu name.
5499 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5501 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5502 * doc/invoke.texi: Document +nodsp as a valid extension for
5505 2017-10-16 Martin Liska <mliska@suse.cz>
5507 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5508 (test_set_range): Likewise.
5509 (test_range_functions): Rename to ...
5510 (test_bit_in_range): ... this.
5511 (sbitmap_c_tests): Add new test.
5513 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5515 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5517 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5518 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5520 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5522 * config/aarch64/aarch64-builtins.c
5523 (aarch64_types_quadopu_lane_qualifiers): New.
5524 (TYPES_QUADOPU_LANE): New.
5525 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5526 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5527 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5528 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5529 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5530 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5531 (Vdottype, DOTPROD): New.
5532 (sur): Add SDOT and UDOT.
5534 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5536 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5537 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5538 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5540 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5541 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5542 Enable TARGET_DOTPROD.
5543 (cortex-a75.cortex-a55): Likewise.
5544 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5546 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5548 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5549 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5551 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5553 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5554 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5555 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5556 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5557 * config/arm/types.md (neon_dot, neon_dot_q): New.
5558 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5560 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5562 * config/arm/arm.h (TARGET_DOTPROD): New.
5563 * config/arm/arm.c (arm_arch_dotprod): New.
5564 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5565 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5566 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5567 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5568 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5569 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5570 * doc/invoke.texi (armv8.2-a): Document dotprod
5572 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5574 * i386.c (ix86_vec_cost): New function.
5575 (ix86_rtx_costs): Handle vector operations better.
5576 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5577 * x86-tune-costs.h: Add new costs to all tables.
5579 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5581 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5583 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5584 divsd, sqrtss and sqrtsd
5585 * x86-tune-costs.h: Add new entries to all costs.
5586 (znver1_cost): Fix to match real instruction latencies.
5588 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5589 Michael Collison <michael.collison@arm.com>
5591 * compare-elim.c: Include emit-rtl.h.
5592 (can_merge_compare_into_arith): New function.
5593 (try_validate_parallel): Likewise.
5594 (try_merge_compare): Likewise.
5595 (try_eliminate_compare): Call the above when no previous clobber
5597 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5600 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5604 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5605 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5609 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5610 to be any operand_equal_p operands. For & (B - 1) require
5611 B to be power of 2. Recognize
5612 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5614 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5617 * optabs.c (expand_memory_blockage): Fix call of
5618 targetm.have_memory_blockage.
5620 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5623 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5624 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5625 objects to extra_objs instead of overwriting it.
5627 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5629 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5630 Use any_fp_register_operand as operand[3] predicate. Simplify
5631 equality test for operands[2] and operands[4] memory location.
5632 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5633 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5634 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5635 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5636 any_fp_register_operand as operand[1] predicate. Simplify
5637 equality test for operands[0] and operands[3] memory location.
5638 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5639 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5640 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5642 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5644 * target-insns.def: Add memory_blockage.
5645 * optabs.c (expand_memory_blockage): New function.
5646 (expand_asm_memory_barrier): Rename ...
5647 (expand_asm_memory_blockage): ... to this.
5648 (expand_mem_thread_fence): Call expand_memory_blockage
5649 instead of expand_asm_memory_barrier.
5650 (expand_mem_singnal_fence): Ditto.
5651 (expand_atomic_load): Ditto.
5652 (expand_atomic_store): Ditto.
5653 * doc/md.texi (Standard Pattern Names For Generation):
5654 Document memory_blockage instruction pattern.
5656 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5658 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5659 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5660 * config/rl78/rl78.md: New define_expand "adddi3".
5662 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5664 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5667 2017-10-13 Jeff Law <law@redhat.com>
5669 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5671 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5674 * internal-fn.c (expand_mul_overflow): If both operands have
5675 the same highpart of -1 or 0 and the topmost bit of lowpart
5676 is different, overflow is if res <= 0 rather than res < 0.
5678 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5680 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5681 TARGET_P9_VECTOR code for unaligned_load case.
5683 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5685 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5687 2017-10-13 Nathan Sidwell <nathan@acm.org>
5689 * tree-core.h (tree_contains_struct): Make bool.
5690 * tree.c (tree_contains_struct): Likewise.
5691 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5692 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5693 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5694 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5695 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5697 2017-10-13 Richard Biener <rguenther@suse.de>
5699 * graphite-isl-ast-to-gimple.c
5700 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5701 parameters and dominance check.
5702 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5703 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5704 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5705 Do not update SSA form here or do intermediate IL verification.
5706 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5707 (graphite_initialize): Remove check on the number of loops in
5708 the function and inline into graphite_transform_loops.
5709 (graphite_finalize): Inline into graphite_transform_loops.
5710 (graphite_transform_loops): Perform SSA update and IL verification
5712 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5714 2017-10-13 Richard Biener <rguenther@suse.de>
5716 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5717 graphite_expression_type_precision): Avoid global constructor
5719 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5720 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5721 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5722 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5723 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5724 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5726 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5729 * config/i386/i386.h (ix86_red_zone_size): New.
5730 * config/i386/i386.md (push peephole2s): Replace
5731 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5733 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5734 Alan Hayward <alan.hayward@arm.com>
5735 David Sherwood <david.sherwood@arm.com>
5737 * combine.c (can_change_dest_mode): Reject changes in
5738 REGMODE_NATURAL_SIZE.
5740 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5741 Alan Hayward <alan.hayward@arm.com>
5742 David Sherwood <david.sherwood@arm.com>
5744 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5745 (expand_debug_source_expr): Likewise.
5746 * combine.c (combine_simplify_rtx): Likewise.
5747 * cse.c (fold_rtx): Likewise.
5748 * fwprop.c (canonicalize_address): Likewise.
5749 * targhooks.c (default_shift_truncation_mask): Likewise.
5751 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5752 Alan Hayward <alan.hayward@arm.com>
5753 David Sherwood <david.sherwood@arm.com>
5755 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5756 (widened_mode): Likewise.
5757 (expand_unop): Likewise.
5758 * ree.c (transform_ifelse): Likewise.
5759 (merge_def_and_ext): Likewise.
5760 (combine_reaching_defs): Likewise.
5761 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5763 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5764 Alan Hayward <alan.hayward@arm.com>
5765 David Sherwood <david.sherwood@arm.com>
5767 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5768 * combine.c (gen_lowpart_for_combine): Likewise.
5769 * dwarf2out.c (rtl_for_decl_location): Likewise.
5770 * final.c (alter_subreg): Likewise.
5771 * rtlhooks.c (gen_lowpart_general): Likewise.
5772 (gen_lowpart_if_possible): Likewise.
5774 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5775 Alan Hayward <alan.hayward@arm.com>
5776 David Sherwood <david.sherwood@arm.com>
5778 * calls.c (expand_call): Use subreg_lowpart_offset.
5779 * cse.c (cse_insn): Likewise.
5780 * regcprop.c (copy_value): Likewise.
5781 (copyprop_hardreg_forward_1): Likewise.
5783 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5786 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5787 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5788 =Q constraints to +Q and into insn condition add check
5789 that operands[0] and operands[1] are equal.
5790 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5791 =Q constraints to +Q and into insn condition add check
5792 that operands[0] is equal to either operands[1] or operands[2].
5795 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5796 instead of handling MINUS_EXPR twice (once for each argument),
5797 canonicalize operand order and handle just once, use rtype where
5798 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5801 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5802 any values of __C while still being pattern recognizable as a simple
5805 2017-10-13 Richard Biener <rguenther@suse.de>
5807 PR tree-optimization/82451
5809 2017-10-02 Richard Biener <rguenther@suse.de>
5811 PR tree-optimization/82355
5812 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5813 a mapping for the enclosing loop but avoid generating one for
5815 (copy_bb_and_scalar_dependences): Remove premature codegen
5816 error on PHIs in blocks duplicated into multiple places.
5817 * graphite-scop-detection.c
5818 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5819 in the region use it as loop and nest to analyze the DR in.
5820 (try_generate_gimple_bb): Likewise.
5821 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5822 (add_loop_constraints): For blocks in a loop not in the region
5823 create a dimension with a single iteration.
5824 * sese.h (gbb_loop_at_index): Remove assert.
5826 * cfgloop.c (loop_preheader_edge): For the loop tree root
5827 return the single successor of the entry block.
5828 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5829 Reset the SCEV hashtable and niters.
5830 * graphite-scop-detection.c
5831 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5832 assert that we only have POLYNOMIAL_CHREC that vary in loops
5833 contained in the region.
5834 (scop_detection::graphite_can_represent_expr): Adjust.
5835 (scop_detection::stmt_has_simple_data_refs_p): For loops
5836 not in the region set loop to NULL. The nest is now the
5837 entry edge to the region.
5838 (try_generate_gimple_bb): Likewise.
5839 * sese.c (scalar_evolution_in_region): Adjust for
5840 instantiate_scev change.
5841 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5842 Make nest parameter the edge into the region.
5843 (create_data_ref): Likewise.
5844 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5845 entry edge into a region and adjust instantiate_scev calls.
5846 (create_data_ref): Likewise.
5847 (graphite_find_data_references_in_stmt): Likewise.
5848 (find_data_references_in_stmt): Pass the loop preheader edge
5849 from the nest argument.
5850 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5851 parameter the edge into the region.
5852 (instantiate_parameters): Use the loop preheader edge as entry.
5853 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5855 (get_instantiated_value_entry): Make instantiate_below parameter
5856 the edge into the region.
5857 (instantiate_scev_name): Likewise. Adjust dominance checks,
5858 when we cannot use loop-based instantiation instantiate by
5859 walking use-def chains.
5860 (instantiate_scev_poly): Adjust.
5861 (instantiate_scev_binary): Likewise.
5862 (instantiate_scev_convert): Likewise.
5863 (instantiate_scev_not): Likewise.
5864 (instantiate_array_ref): Remove.
5865 (instantiate_scev_3): Likewise.
5866 (instantiate_scev_2): Likewise.
5867 (instantiate_scev_1): Likewise.
5868 (instantiate_scev_r): Do not blindly handle N-operand trees.
5869 Do not instantiate array-refs. Handle all constants and invariants.
5870 (instantiate_scev): Make instantiate_below parameter
5871 the edge into the region.
5872 (resolve_mixers): Use the loop preheader edge for the region
5873 parameter to instantiate_scev_r.
5874 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5876 2017-10-13 Richard Biener <rguenther@suse.de>
5878 PR tree-optimization/82525
5879 * graphite-isl-ast-to-gimple.c
5880 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5882 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5883 Fail code generation when we cannot represent the isl integer.
5884 (binary_op_to_tree): Elide modulo operations that are no-ops
5885 in the type we code generate. Remove now superfluous code
5888 2017-10-13 Richard Biener <rguenther@suse.de>
5890 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5891 (scop_detection::harmful_loop_in_region): Remove premature
5892 IV type restriction.
5893 (scop_detection::graphite_can_represent_scev): We can handle
5894 pointer IVs just fine.
5896 2017-10-13 Alan Modra <amodra@gmail.com>
5898 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5899 "Clobbers and Scratch Registers". Add paragraph on
5900 alternative to clobbers for scratch registers and OpenBLAS
5903 2017-10-13 Alan Modra <amodra@gmail.com>
5905 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5906 example of a memory input for a string of known length. Move
5907 commentary out of table. Add a number of new examples
5908 covering array memory inputs.
5910 2017-10-12 Martin Liska <mliska@suse.cz>
5912 PR tree-optimization/82493
5913 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5914 (test_range_functions): New function.
5915 (sbitmap_c_tests): Likewise.
5916 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5917 * selftest.h (sbitmap_c_tests): New function.
5919 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5921 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5923 * config/rs6000/amo.h: Fix spacing issue.
5925 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5928 * config/i386/i386.md (*ashl<mode>3_mask_1,
5929 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5930 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5933 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5935 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5936 (profile_probability): Set max_probability
5937 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5939 * profile-count.c (profile_probability::differs_from_p): Do not
5940 rely on max_probaiblity == 10000
5942 2017-10-12 Jeff Law <law@redhat.com>
5944 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5947 2017-10-12 Martin Sebor <msebor@redhat.com>
5951 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5952 (handle_alias_pairs): Call it.
5953 * common.opt (-Wattribute-alias): New option.
5954 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5955 * doc/invoke.texi (-Wattribute-alias): Document.
5957 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5960 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5962 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5964 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5965 (make_hard_regno_born, make_hard_regno_dead): Update
5966 bb_killed_pseudos and bb_gen_pseudos.
5968 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5970 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5972 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5974 * config/alpha/alpha.c (alpha_split_conditional_move):
5975 Use std::swap instead of manually swapping.
5976 (alpha_stdarg_optimize_hook): Ditto.
5977 (alpha_canonicalize_comparison): Ditto.
5979 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5981 * tree-loop-distribution.c (struct builtin_info): New struct.
5982 (struct partition): Refactor fields into struct builtin_info.
5983 (partition_free): Free struct builtin_info.
5984 (build_size_arg_loc, build_addr_arg_loc): Delete.
5985 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5986 information from struct builtin_info.
5987 (find_single_drs): New function refactored from classify_partition.
5988 Also moved builtin validity checks to this function.
5989 (compute_access_range, alloc_builtin): New functions.
5990 (classify_builtin_st, classify_builtin_ldst): New functions.
5991 (classify_partition): Refactor code into functions find_single_drs,
5992 classify_builtin_st and classify_builtin_ldst.
5993 (distribute_loop): Don't do runtime alias check when distributing
5995 (find_seed_stmts_for_distribution): New function.
5996 (pass_loop_distribution::execute): Refactor code finding seed
5997 stmts into above function. Support distribution for the innermost
5998 two-level loop nest. Adjust dump information.
6000 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6002 * tree-loop-distribution.c: Adjust the general comment.
6003 (NUM_PARTITION_THRESHOLD): New macro.
6004 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6005 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6006 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6007 in call to build_partition_graph.
6008 (finalize_partitions): New parameter. Make loop distribution more
6009 conservative by fusing more partitions.
6010 (distribute_loop): Don't do runtime alias check in case of loop nest
6012 (find_seed_stmts_for_distribution): New function.
6013 (prepare_perfect_loop_nest): New function.
6014 (pass_loop_distribution::execute): Refactor code finding seed stmts
6015 and loop nest into above functions. Support loop nest distribution.
6016 Adjust dump information accordingly.
6018 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6020 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6021 and set PTYPE_SEQUENTIAL for merged partition.
6023 2017-10-12 Richard Biener <rguenther@suse.de>
6025 PR tree-optimization/69728
6027 2017-09-19 Richard Biener <rguenther@suse.de>
6029 PR tree-optimization/69728
6030 * graphite-sese-to-poly.c (schedule_error): New global.
6031 (add_loop_schedule): Handle empty domain by failing the
6033 (build_original_schedule): Handle schedule_error.
6035 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6036 domain by returning an unchanged schedule.
6038 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6040 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6041 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6043 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6045 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6048 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6051 * expr.c (store_field): Don't optimize away bitsize == 0 store
6052 from CALL_EXPR with addressable return type.
6054 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6056 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6057 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6058 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6059 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6060 TARGET_ISEL instead of TARGET_ISEL<sel>.
6062 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6064 * config/rs6000/rs6000.c
6065 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6067 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6069 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6071 (reg_or_cint_operand): Fix comment.
6072 (reg_or_zero_operand): New predicate.
6073 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6074 * config/rs6000/rs6000.c (output_isel): Delete.
6075 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6076 instead of reg_or_cint_operand. Output instruction directly (not via
6078 (isel_unsigned_<mode>): Ditto.
6079 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6080 gpc_reg_operand. Add an instruction alternative for this. Output
6081 instruction directly.
6082 (*isel_reversed_unsigned_<mode>): Ditto.
6084 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6086 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6087 (TARGET_CANONICALIZE_COMPARISON): Define.
6089 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6092 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6093 Check whether the dest is REG before adding REG_EQUIV note.
6095 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6098 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6100 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6101 (make_hard_regno_born, make_hard_regno_dead): Update
6102 bb_killed_pseudos and bb_gen_pseudos.
6104 2017-10-11 Nathan Sidwell <nathan@acm.org>
6106 * incpath.h (enum incpath_kind): Name enum, prefix values.
6107 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6108 * incpath.c (heads, tails): Use INC_MAX.
6109 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6110 (merge_include_chains, split_quote_chain,
6111 register_include_chains): Update incpath_kind names.
6112 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6113 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6115 (add_framework_path, darwin_register_objc_includes): Likewise.
6116 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6118 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6120 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6121 Do not use float_operator operator predicate.
6122 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6123 * config/i386/predicates.md (float_operator): Remove predicate.
6125 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6127 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6128 (*jccxf_i387): Ditto.
6129 (*jcc<mode>_i387): Ditto.
6130 (*jccu<mode>_i387): Ditto.
6131 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6132 (*jcc_*_i387 splitters): Remove.
6133 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6134 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6135 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6138 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6140 * profile-count.h (slow_safe_scale_64bit): New function.
6141 (safe_scale_64bit): New inline.
6142 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6143 * profile-count.c: Include wide-int.h
6144 (slow_safe_scale_64bit): New.
6146 2017-10-11 Nathan Sidwell <nathan@acm.org>
6148 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6149 HAS_DECL_ASSEMBLER_NAME_P.
6150 * gimple-expr.c (gimple_decl_printable_name: Check
6151 HAS_DECL_ASSEMBLER_NAME_P too.
6152 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6153 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6154 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6155 * passes.c (rest_of_decl_compilation): Check
6156 HAS_DECL_ASSEMBLER_NAME_P too.
6157 * recog.c (verify_changes): Likewise.
6158 * tree-pretty-print.c (dump_decl_name): Likewise.
6159 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6161 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6162 (SET_DECL_ASSEMBLER_NAME): Use it.
6163 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6164 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6165 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6167 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6169 * config.gcc (i386, x86_64): Add extra objects.
6170 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6171 (ix86_min_insn_size): Declare.
6172 (ix86_issue_rate): Declare.
6173 (ix86_adjust_cost): Declare.
6174 (ia32_multipass_dfa_lookahead): Declare.
6175 (ix86_macro_fusion_p): Declare.
6176 (ix86_macro_fusion_pair_p): Declare.
6177 (ix86_bd_has_dispatch): Declare.
6178 (ix86_bd_do_dispatch): Declare.
6179 (ix86_core2i7_init_hooks): Declare.
6180 (ix86_atom_sched_reorder): Declare.
6181 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6182 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6183 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6184 (rip_relative_addr_p): Rename to ...
6185 (ix86_rip_relative_addr_p): ... this one; export.
6186 (memory_address_length): Update.
6187 (ix86_issue_rate): Move to x86-tune-sched.c.
6188 (ix86_flags_dependent): Move to x86-tune-sched.c.
6189 (ix86_agi_dependent): Move to x86-tune-sched.c.
6190 (exact_dependency_1): Move to x86-tune-sched.c.
6191 (exact_store_load_dependency): Move to x86-tune-sched.c.
6192 (ix86_adjust_cost): Move to x86-tune-sched.c.
6193 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6194 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6195 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6196 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6197 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6198 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6199 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6200 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6201 (min_insn_size): Rename to ...
6202 (ix86_min_insn_size): ... this one; export.
6203 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6204 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6205 (core2i7_first_cycle_multipass_backtrack): Move to
6206 x86-tune-sched-core.c.
6207 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6208 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6209 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6210 (ix86_avoid_jump_mispredicts): Update.
6211 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6212 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6213 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6214 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6215 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6216 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6217 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6218 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6219 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6220 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6221 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6222 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6223 (BIG): Move to ix86-tune-sched-bd.c.
6224 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6225 (enum insn_path): Move to ix86-tune-sched-bd.c.
6226 (get_mem_group): Move to ix86-tune-sched-bd.c.
6227 (is_cmp): Move to ix86-tune-sched-bd.c.
6228 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6229 (is_branch): Move to ix86-tune-sched-bd.c.
6230 (is_prefetch): Move to ix86-tune-sched-bd.c.
6231 (init_window): Move to ix86-tune-sched-bd.c.
6232 (allocate_window): Move to ix86-tune-sched-bd.c.
6233 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6234 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6235 (process_end_window): Move to ix86-tune-sched-bd.c.
6236 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6237 (find_constant): Move to ix86-tune-sched-bd.c.
6238 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6239 (has_immediate): Move to ix86-tune-sched-bd.c.
6240 (get_insn_path): Move to ix86-tune-sched-bd.c.
6241 (get_insn_group): Move to ix86-tune-sched-bd.c.
6242 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6243 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6244 (add_insn_window): Move to ix86-tune-sched-bd.c.
6245 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6246 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6247 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6248 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6249 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6250 (do_dispatch): Move to ix86-tune-sched-bd.c.
6251 (has_dispatch): Move to ix86-tune-sched-bd.c.
6252 * config/i386/t-i386: Add new object files.
6253 * config/i386/x86-tune-costs.h: New file.
6254 * config/i386/x86-tune-sched-atom.c: New file.
6255 * config/i386/x86-tune-sched-bd.c: New file.
6256 * config/i386/x86-tune-sched-core.c: New file.
6257 * config/i386/x86-tune-sched.c: New file.
6259 2017-10-11 Liu Hao <lh_mouse@126.com>
6261 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6262 the generic version below instead.
6263 (should_colorize): Recognize Windows consoles as terminals
6265 * pretty-print.c [__MINGW32__] (write_all): New function.
6266 [__MINGW32__] (find_esc_head): Likewise.
6267 [__MINGW32__] (find_esc_terminator): Likewise.
6268 [__MINGW32__] (eat_esc_sequence): Likewise.
6269 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6271 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6274 2017-10-11 Richard Biener <rguenther@suse.de>
6276 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6277 Properly call analyze_scalar_evolution with the loop of the stmt.
6279 2017-10-11 Richard Biener <rguenther@suse.de>
6281 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6282 * tree-core.h (tree_base): Add chrec_var union member.
6283 * tree.h (CHREC_VAR): Remove.
6284 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6285 * tree-chrec.h (build_polynomial_chrec): Adjust.
6286 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6287 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6289 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6291 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6292 * match.pd: ... here.
6293 ((T) X == (T) Y): Relax condition.
6295 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6297 PR tree-optimization/82472
6298 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6300 (break_alias_scc_partitions): Update postorder number.
6302 2017-10-11 Martin Liska <mliska@suse.cz>
6305 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6307 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6309 2017-10-11 Martin Liska <mliska@suse.cz>
6311 * print-rtl.c (print_insn): Move declaration of idbuf
6312 to same scope as name.
6314 2017-10-11 Martin Liska <mliska@suse.cz>
6319 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6322 2017-10-11 Martin Liska <mliska@suse.cz>
6325 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6328 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6330 * config.gcc (default_gnu_indirect_function): Default to yes for
6331 arm*-*-linux* with glibc.
6333 2017-10-11 Richard Biener <rguenther@suse.de>
6335 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6336 default-defs and types we do not want to analyze.
6337 (interpret_loop_phi): Replace unreachable code with an assert.
6338 (compute_scalar_evolution_in_loop): Remove and inline ...
6339 (analyze_scalar_evolution_1): ... here, replacing condition with
6340 what makes the intent clearer. Remove handling of cases
6341 get_scalar_evolution now handles.
6343 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6345 PR rtl-optimization/81434
6346 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6347 comment for main loop. In sched_group_found if, also add checks for
6348 pass and min_cost_group.
6350 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6352 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6353 (rs6000_insn_cost): New function.
6354 * config/rs6000/rs6000.md (cost): New attribute.
6356 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6357 H.J. Lu <hongjiu.lu@intel.com>
6361 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6362 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6363 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6364 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6365 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6366 OPTION_MASK_ISA_MMX and some other bit set require both
6367 mmx and the ISAs without the mmx bit.
6368 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6369 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6370 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6371 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6372 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6373 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6374 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6375 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6376 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6377 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6378 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6379 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6380 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6381 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6382 Add OPTION_MASK_ISA_MMX.
6384 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6386 * config.gcc (armv7*-*-freebsd*): New target.
6387 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6389 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6391 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6392 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6395 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6397 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6398 a template parameter.
6399 (WIDE_INT_REF_FOR): Update accordingly.
6400 * tree.h (wi::int_traits <const_tree>): Delete.
6401 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6402 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6403 (wi::tree_to_wide_ref): New typedef.
6404 (wi::to_wide): New function.
6405 * calls.c (get_size_range): Use wi::to_wide when operating on
6407 * cgraph.c (cgraph_node::create_thunk): Likewise.
6408 * config/i386/i386.c (ix86_data_alignment): Likewise.
6409 (ix86_local_alignment): Likewise.
6410 * dbxout.c (stabstr_O): Likewise.
6411 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6412 * expr.c (const_vector_from_tree): Likewise.
6413 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6414 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6415 (fold_negate_expr_1, int_const_binop_1, const_binop)
6416 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6417 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6418 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6419 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6420 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6421 (fold_not_const, round_up_loc): Likewise.
6422 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6423 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6424 (alloca_call_type): Likewise.
6425 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6426 * godump.c (go_output_typedef): Likewise.
6427 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6428 * internal-fn.c (get_min_precision): Likewise.
6429 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6430 * ipa-polymorphic-call.c
6431 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6432 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6433 (ipa_modify_call_arguments): Likewise.
6434 * match.pd: Likewise.
6435 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6436 * print-tree.c (print_node_brief, print_node): Likewise.
6437 * stmt.c (expand_case): Likewise.
6438 * stor-layout.c (layout_type): Likewise.
6439 * tree-affine.c (tree_to_aff_combination): Likewise.
6440 * tree-cfg.c (group_case_labels_stmt): Likewise.
6441 * tree-data-ref.c (dr_analyze_indices): Likewise.
6442 (prune_runtime_alias_test_list): Likewise.
6443 * tree-dump.c (dequeue_and_dump): Likewise.
6444 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6445 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6446 * tree-pretty-print.c (dump_generic_node): Likewise.
6447 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6448 (simple_iv_with_niters): Likewise.
6449 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6450 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6451 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6452 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6453 (refine_value_range_using_guard, number_of_iterations_ne_max)
6454 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6455 (get_cst_init_from_scev, record_nonwrapping_iv)
6456 (scev_var_range_cant_overflow): Likewise.
6457 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6458 * tree-ssa-pre.c (compute_avail): Likewise.
6459 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6460 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6461 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6462 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6463 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6464 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6465 (dump_case_nodes, try_switch_expansion): Likewise.
6466 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6467 (vect_do_peeling): Likewise.
6468 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6469 * tree-vect-stmts.c (vectorizable_load): Likewise.
6470 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6471 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6472 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6473 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6474 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6475 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6476 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6477 (evrp_dom_walker::before_dom_children): Likewise.
6478 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6479 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6480 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6481 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6482 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6483 (get_range_pos_neg): Likewise.
6484 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6485 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6486 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6487 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6488 * config/avr/avr.c (avr_fold_builtin): Likewise.
6489 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6490 * config/msp430/msp430.c (msp430_attr): Likewise.
6491 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6492 * config/powerpcspe/powerpcspe-c.c
6493 (altivec_resolve_overloaded_builtin): Likewise.
6494 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6495 (rs6000_expand_ternop_builtin): Likewise.
6496 * config/rs6000/rs6000-c.c
6497 (altivec_resolve_overloaded_builtin): Likewise.
6498 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6499 (rs6000_expand_ternop_builtin): Likewise.
6500 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6502 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6504 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6505 when copying loop nest with only one inner loop.
6507 2017-10-10 Richard Biener <rguenther@suse.de>
6509 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6510 blocks if SCEV is active.
6511 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6513 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6514 (scev_initialize): Assert we are not yet initialized.
6516 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6518 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6519 inner loop's exit stmt by making it always exit the loop, otherwise
6520 we would generate an infinite empty loop.
6522 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6524 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6525 renaming variables in new preheader if it's deleted.
6527 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6529 * tree-loop-distribution.c (struct partition): Remove unused field
6530 loops of the structure.
6531 (partition_alloc, partition_free): Ditto.
6532 (build_rdg_partition_for_vertex): Ditto.
6534 2017-10-09 Jeff Law <law@redhat.com>
6536 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6537 return type to match prototype and documentation.
6539 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6541 * config/rs6000/rs6000.c (processor_costs): Move to ...
6542 * config/rs6000/rs6000.h: ... here.
6543 (rs6000_cost): Declare.
6545 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6547 * except.c (setjmp_fn): New global variable.
6548 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6549 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6550 if DONT_USE_BUILTIN_SETJMP is defined.
6552 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6554 * target.def (insn_cost): New hook.
6555 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6556 * doc/tm.texi: Regenerate.
6557 * rtlanal.c (insn_cost): Use the new hook.
6559 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6561 * combine.c (combine_validate_cost): Compute the new insn_cost,
6562 not just pattern_cost.
6563 (try_combine): Adjust comment.
6565 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6567 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6569 * combine.c (uid_insn_cost): Adjust comment.
6570 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6572 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6573 * dse.c (find_shift_sequence): Ditto.
6574 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6575 (bb_valid_for_noce_process_p): Use pattern_cost.
6576 * rtl.h (insn_rtx_cost): Delete.
6577 (pattern_cost): New prototype.
6578 (insn_cost): New prototype.
6579 * rtlanal.c (insn_rtx_cost): Rename to...
6580 (pattern_cost): ... this.
6583 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6585 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6586 (*jcc<mode>_0_r_i387): Ditto.
6587 (*jccxf_r_i387): Ditto.
6588 (*jcc<mode>_r_i387): Ditto.
6589 (*jccu<mode>_r_i387): Ditto.
6590 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6591 (*jcc): Rename from *jcc_1.
6593 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6595 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6596 deferred rescans after the lvx/stvx recombination pre-pass.
6598 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6600 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6601 memory operation instruction support.
6602 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6603 (rs6000-ibm-aix[789]*): Likewise.
6604 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6605 Document new functions.
6607 2017-10-09 Richard Biener <rguenther@suse.de>
6609 PR tree-optimization/82397
6610 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6611 equality only for semantically equal trees.
6613 2017-10-09 Richard Biener <rguenther@suse.de>
6615 PR tree-optimization/82449
6616 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6617 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6618 allow constant addresses.
6619 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6622 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6624 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6627 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6630 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6633 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6636 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6638 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6641 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6642 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6644 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6646 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6647 (WI_BINARY_PREDICATE_RESULT): Likewise.
6648 (wi::binary_traits::operator_result): New type.
6649 (wi::binary_traits::predicate_result): Likewise.
6650 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6651 (generic_wide_int::operator==, generic_wide_int::operator!=)
6652 (generic_wide_int::operator&, generic_wide_int::and_not)
6653 (generic_wide_int::operator|, generic_wide_int::or_not)
6654 (generic_wide_int::operator^, generic_wide_int::operator+
6655 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6657 (operator~, unary operator-, operator==, operator!=, operator&)
6658 (operator|, operator^, operator+, binary operator-, operator*): New
6660 * expr.c (get_inner_reference): Use wi::bit_and_not.
6661 * fold-const.c (fold_binary_loc): Likewise.
6662 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6663 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6664 (bit_value_binop): Likewise.
6665 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6666 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6667 (extract_range_from_binary_expr_1): Likewise.
6668 (masked_increment): Likewise.
6669 (simplify_bit_ops_using_ranges): Likewise.
6671 2017-10-09 Martin Jambor <mjambor@suse.cz>
6674 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6675 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6676 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6678 (hsa_fixup_mov_insn_type): New function.
6679 (hsa_op_with_type::get_in_type): Use it.
6680 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6681 immediates in an assert.
6682 (hsa_op_with_type::extend_int_to_32bit): New method.
6683 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6684 types. Convert to dest type if necessary.
6685 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6686 (reg_for_gimple_ssa): Pass false as min32int to
6687 hsa_type_for_scalar_tree_type.
6688 (gen_hsa_addr): Fixup type when creating addresable temporary.
6689 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6690 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6691 necessary. Call hsa_fixup_mov_insn_type.
6692 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6693 extend operands and convert to dest type if necessary.
6694 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6695 to dest type if necessary.
6696 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6697 if conversion nt necessary and size matches.
6698 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6699 to dest type if necessary.
6700 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6701 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6703 (gen_hsa_clrsb): Likewise.
6704 (gen_hsa_ffs): Likewise.
6705 (gen_hsa_divmod): Extend operands and convert to dest type if
6707 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6709 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6711 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6712 Remove empty default arguments. Use a brace block as output
6714 (conditional return): Ditto.
6716 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6717 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6719 (group_ending_nop): Ditto.
6720 (doloop_end): Ditto.
6721 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6722 (splitters for those): Ditto.
6724 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6726 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6727 a conditional jump (and the compare for it) so that pc_rtx is the
6729 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6730 for the deleted and renamed ctr<mode>_internal[234] patterns.
6731 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6732 Delete second conditional return pattern.
6733 (ctr<mode>_internal2): Delete this second bdnz pattern.
6734 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6735 (ctr<mode>_internal4): Delete this second bdz pattern.
6737 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6739 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6740 (always_initialized_rtx_for_ssa_name_p): New predicate.
6741 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6742 (finish_out_of_ssa): Free new field of SA.
6743 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6744 * tree-ssa-coalesce.c: Include tree-ssa.h.
6745 (get_parm_default_def_partitions): Remove extern keyword.
6746 (get_undefined_value_partitions): New function.
6747 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6748 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6750 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6752 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6754 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6756 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6758 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6759 for targets that preffer 128bit.
6761 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6763 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6765 2017-10-08 Olivier Hainque <hainque@adacore.com>
6767 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6768 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6770 (thumb_set_return_address): Likewise.
6772 2017-10-08 Olivier Hainque <hainque@adacore.com>
6774 * common/config/arm/arm-common.c (arm_except_unwind_info):
6775 Handle DWARF2_UNWIND_INFO.
6777 2017-10-07 Michael Collison <michael.collison@arm.com>
6779 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6782 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6784 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6785 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6786 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6787 defined, force the creation of a new block for a dispatch label.
6789 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6791 * invoke.texi (Wsuggest-attribute=cold): Document.
6792 * common.opt (Wsuggest-attribute=cold): New
6793 * ipa-pure-const.c (warn_function_cold): New function.
6794 * predict.c (compute_function_frequency): Use it.
6795 * predict.h (warn_function_cold): Declare.
6797 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6799 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6802 2017-10-06 Martin Liska <mliska@suse.cz>
6804 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6805 keyword for member functions.
6806 (struct sanopt_tree_couple): New struct.
6807 (struct sanopt_tree_couple_hash): New function.
6808 (struct sanopt_ctx): Add new hash_map.
6809 (has_dominating_ubsan_ptr_check): New function.
6810 (record_ubsan_ptr_check_stmt): Likewise.
6811 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6812 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6813 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6815 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6818 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6819 aarch64_simd_valid_immediate on CONST_VECTORs.
6820 (aarch64_reg_or_bic_imm): Likewise.
6822 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6824 PR rtl-optimization/82396
6825 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6827 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6829 * graphite-dependences.c (scop_get_reads): Move code to...
6830 (scop_get_must_writes): Move code to...
6831 (scop_get_may_writes): Move code to...
6832 (scop_get_reads_and_writes): ... here.
6833 (scop_get_dependences): Call scop_get_reads_and_writes.
6835 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6837 PR tree-optimization/82434
6838 * fold-const.h (can_native_encode_type_p,
6839 can_native_encode_string_p): Remove.
6840 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6841 don't encode anything, just return what would be otherwise returned.
6842 (native_encode_fixed, native_encode_complex, native_encode_vector):
6844 (native_encode_string): Likewise. Inline by hand
6845 can_native_encode_string_p.
6846 (can_native_encode_type_p): Remove.
6847 (can_native_encode_string_p): Remove.
6848 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6849 STRING_CSTs using can_native_encode_string_p, test all
6850 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6851 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6852 argument from native_encode_expr.
6853 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6854 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6857 2017-10-06 Richard Biener <rguenther@suse.de>
6859 PR tree-optimization/82397
6860 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6861 operand_equal_p but rely on data_ref_compare_tree for detecting
6863 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6864 to match up with dr_group_sort_cmp.
6866 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6869 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6871 * config/s390/s390-builtin-types.def: Regenerate.
6873 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6876 * config/s390/s390-builtin-types.def: Regenerate.
6877 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6878 Change flag from B_VXE to B_VX.
6879 (s390_vec_min_dbl): Remove B_VXE flag.
6881 2017-10-06 Richard Biener <rguenther@suse.de>
6883 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6884 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6885 translate_isl_ast_to_gimple::is_valid_rename,
6886 translate_isl_ast_to_gimple::get_rename,
6887 translate_isl_ast_to_gimple::get_def_bb_for_const,
6888 translate_isl_ast_to_gimple::get_new_name,
6889 translate_isl_ast_to_gimple::collect_all_ssa_names,
6890 translate_isl_ast_to_gimple::copy_loop_phi_args,
6891 translate_isl_ast_to_gimple::collect_all_ssa_names,
6892 translate_isl_ast_to_gimple::copy_loop_phi_args,
6893 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6894 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6895 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6896 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6897 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6898 translate_isl_ast_to_gimple::copy_cond_phi_args,
6899 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6900 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6901 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6902 translate_isl_ast_to_gimple::rename_uses,
6903 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6904 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6905 (set_rename_for_each_def): Likewise.
6906 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6907 here. Handle rewriting SCEV analyzable uses here.
6908 (copy_bb_and_scalar_dependences): Generate code for PHI
6910 (graphite_regenerate_ast_isl): Adjust.
6911 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6912 (add_write, add_read): New functions.
6913 (build_cross_bb_scalars_def): Use it and simplify.
6914 (build_cross_bb_scalars_use): Likewise.
6915 (graphite_find_cross_bb_scalar_vars): Inline into...
6916 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6917 simulating out-of-SSA. Compute liveout and add dependencies.
6918 (build_scops): Force an empty entry block.
6919 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6921 (sese_build_liveouts): Declare.
6922 (sese_trivially_empty_bb_p): Likewise.
6923 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6924 compute liveout and debug_liveout.
6925 (sese_bad_liveouts_use): Remove.
6926 (sese_reset_debug_liveouts_bb): Likewise.
6927 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6928 (sese_build_liveouts): Build liveout and debug_liveout and store
6930 (new_sese_info): Adjust.
6931 (free_sese_info): Likewise.
6932 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6933 do not build liveout here.
6934 (move_sese_in_condition): Adjust region entry.
6935 (scev_analyzable_p): Match up with chrec_apply requirements.
6936 (sese_trivially_empty_bb_p): New.
6937 * tree-into-ssa.c (get_reaching_def): Properly support generating
6938 default-defs for incremental rewrite of anonymous names.
6940 2017-10-06 Richard Biener <rguenther@suse.de>
6942 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6943 precision do not perform modulo reduction.
6945 2017-10-06 Richard Biener <rguenther@suse.de>
6947 PR tree-optimization/82436
6948 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6949 conservatively choose the vectorization factor when checking
6950 whether we can perform the required load permutation.
6951 (vect_transform_slp_perm_load): Assert when we may not fail.
6953 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6955 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6956 message for incompatible -msdata=* and -mcall-* options.
6958 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6960 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6961 rate for post-reload scheduling.
6963 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6965 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6967 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6969 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6970 to improve monte carlo in scimark.
6972 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6974 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6975 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6976 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6977 pentium4_cost, nocona_cost): Set reassociation width to 1.
6978 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6979 width to 2 for fp operations and 1 otherwise.
6980 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6982 (atom_cost): Set reassociation width to 2.
6983 (slm_cost, generic_cost): Set fp reassociation width
6984 to 2 and 1 otherwise.
6985 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6986 (core_cost): Set fp reassociation width to 4 and vector to 2.
6987 (ix86_reassociation_width): Rewrite using cost table; special case
6988 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6989 and TARGET_AVX128_OPTIMAL.
6990 * config/i386/i386.h (processor_costs): Add
6991 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6992 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6993 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6994 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6995 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6996 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6998 2017-10-05 Nathan Sidwell <nathan@acm.org>
7000 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
7002 2017-10-05 Tamar Christina <tamar.christina@arm.com>
7004 * config/arm/arm.c (arm_test_fpu_data): New.
7005 (arm_run_selftests): Call arm_test_fpu_data.
7007 2017-10-04 Nathan Sidwell <nathan@acm.org>
7009 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7011 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7013 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7015 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7016 check type for aarch64_simd_valid_immediate.
7017 (aarch64_output_simd_mov_immediate): Update prototype.
7018 (aarch64_simd_valid_immediate): Update prototype.
7019 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7020 support for ORR-immediate.
7021 (and<mode>3): modified pattern to add support for BIC-immediate.
7022 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7023 now checks for valid immediate for BIC and ORR based on new enum
7025 (aarch64_output_simd_mov_immediate): Function now used to output
7026 BIC/ORR imm as well based on new enum argument.
7027 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7029 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7030 (aarch64_reg_or_bic_imm): Likewise.
7032 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7034 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7035 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7037 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7041 PR rtl-optimization/82396
7042 * haifa-sched.c (autopref_multipass_init): Simplify
7044 (autopref_rank_data): Simplify sort order.
7045 * sched-int.h (autopref_multipass_data_): Remove
7046 multi_mem_insn_p, min_offset and max_offset.
7048 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7050 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7052 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7054 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7057 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7059 * doc/sourcebuild.texi: Document vect_long_mult.
7061 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7063 PR tree-optimization/82413
7064 * fold-const.c (build_range_check): Use widest_int when comparing
7065 the maximum ETYPE value with HIGH.
7067 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7069 PR rtl-optimization/82396
7070 * haifa-sched.c (autopref_multipass_init): Simplify
7072 (autopref_rank_data): Simplify sort order.
7073 * sched-int.h (autopref_multipass_data_): Remove
7074 multi_mem_insn_p, min_offset and max_offset.
7076 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7078 PR tree-optimization/82381
7079 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7080 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7083 PR tree-optimization/82374
7084 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7085 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7086 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7087 current_function_decl to the new decl.
7089 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7091 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7092 helper macro for IEEE float128 hardware built-in functions.
7093 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7095 (TRUNCF128_ODD): Likewise.
7096 (ADDF128_ODD): Likewise.
7097 (SUBF128_ODD): Likewise.
7098 (MULF128_ODD): Likewise.
7099 (DIVF128_ODD): Likewise.
7100 (FMAF128_ODD): Likewise.
7101 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7102 UNSPEC_TRUNC_ROUND_TO_ODD.
7103 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7104 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7105 floating point round to odd instructions.
7106 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7107 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7108 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7109 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7110 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7111 (trunc<mode>sf2_hw): Change the truncate with round to odd
7112 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7113 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7114 to odd hardware instructions.
7115 (sub<mode>3_odd): Likewise.
7116 (mul<mode>3_odd): Likewise.
7117 (div<mode>3_odd): Likewise.
7118 (sqrt<mode>2_odd): Likewise.
7119 (fma<mode>4_odd): Likewise.
7120 (fms<mode>4_odd): Likewise.
7121 (nfma<mode>4_odd): Likewise.
7122 (nfms<mode>4_odd): Likewise.
7123 (trunc<mode>df2_odd): Change the truncate with round to odd
7124 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7126 * doc/extend.texi (PowerPC built-in functions): Update documentation
7127 for existing IEEE float128-bit built-in functions. Add built-in
7128 functions that generate the IEEE 128-bit floating point round to
7131 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7133 PR rtl-optimization/77729
7134 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7135 to (X&(C1&~C2))|C2 transformations.
7137 2017-10-03 Martin Jambor <mjambor@suse.cz>
7139 PR tree-optimization/82363
7140 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7141 mismatch, mark lacc written regardless of racc.
7143 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7145 PR tree-optimization/82381
7146 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7147 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7148 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7149 fallthrough into reassoc_stmt_dominates_stmt_p.
7152 * combine.c (combine_instructions): Don't combine in unreachable
7155 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7158 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7159 function to not use the have_cpu variable. Do not set cpu_index,
7160 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7162 (rs6000_valid_attribute_p): Remove duplicate initializations of
7163 old_optimize and func_optimize.
7164 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7165 (rs6000_activate_target_options): Make global.
7166 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7169 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7171 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7172 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7173 Set *poffset to 0 and *psize and *pmax_size to -1 if
7174 *poffset + *psize overflows in HOST_WIDE_INT.
7176 PR tree-optimization/82387
7177 PR tree-optimization/82388
7178 PR tree-optimization/82389
7179 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7180 instead of live_bytes non-NULL.
7182 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7185 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7188 2017-10-02 Richard Biener <rguenther@suse.de>
7190 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7191 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7192 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7194 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7196 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7197 requested precision matches the type's.
7198 * calls.c (alloc_max_size): Calculate the new candidate size as
7199 a widest_int and use wi::to_widest when comparing it with the
7200 current candidate size.
7201 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7202 zero rather than integer_zero_node.
7203 * match.pd: Check for a no-op conversion before using wi::add
7204 rather than after. Use tree_to_uhwi when summing small shift
7205 counts into an unsigned int.
7207 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7208 Alan Hayward <alan.hayward@arm.com>
7209 David Sherwood <david.sherwood@arm.com>
7212 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7213 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7214 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7215 POINTER_AND_FP_REGS.
7217 2017-10-02 Richard Biener <rguenther@suse.de>
7219 PR tree-optimization/82355
7220 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7221 a mapping for the enclosing loop but avoid generating one for
7223 (copy_bb_and_scalar_dependences): Remove premature codegen
7224 error on PHIs in blocks duplicated into multiple places.
7225 * graphite-scop-detection.c
7226 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7227 in the region use it as loop and nest to analyze the DR in.
7228 (try_generate_gimple_bb): Likewise.
7229 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7230 (add_loop_constraints): For blocks in a loop not in the region
7231 create a dimension with a single iteration.
7232 * sese.h (gbb_loop_at_index): Remove assert.
7234 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7236 * omp-expand.c (adjust_context_scope): New function.
7237 (expand_parallel_call): Call adjust_context_scope.
7239 2017-10-01 Jeff Law <law@redhat.com>
7241 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7242 dom_opt_dom_walker class with direct access to private members.
7243 Add comments. Call test_for_singularity.
7244 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7245 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7246 m_dummy_cond anymore.
7247 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7249 (pass_dominator:execute): Build the dummy_cond here and pass it
7250 to the dom_opt_dom_walker ctor.
7251 (test_for_singularity): New function.
7253 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7254 Maya Rashish <coypu@sdf.org>
7256 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7257 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7258 (alpha*-*-netbsd*) Use nbsd_tm_file.
7259 (arm*-*-netbsdelf*) Likewise.
7260 (i[34567]86-*-netbsdelf*) Likewise.
7261 (x86_64-*-netbsd*) Likewise.
7262 (mips*-*-netbsd*) Likewise.
7263 (powerpc-*-netbsd*) Likewise.
7264 (sh*-*-netbsd*) Likewise.
7265 (sparc-*-netbsdelf*) Likewise.
7266 (sparc64-*-netbsd*) Likewise.
7267 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7269 (vax-*-netbsdelf*) Likewise.
7270 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7271 (UINT_FAST8_TYPE) Likewise.
7272 (INT_FAST16_TYPE) Check CHAR_FAST16.
7273 (UINT_FAST16_TYPE) Likewise.
7275 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7278 * config/i386/i386.md
7279 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7280 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7281 *divmodsi4_zext_2): New define_insn_and_split.
7282 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7283 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7284 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7285 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7286 New define_insn_and_split.
7287 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7288 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7289 operands[1] having DImode when mode is SImode.
7291 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7292 always SImode for DIV and MOD in REG_EQUAL notes.
7294 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7297 * match.pd: Fix handling of NaNs in pattern.
7299 2017-09-29 Jeff Law <law@redhat.com>
7301 * sbitmap.c (bitmap_bit_in_range_p): New function.
7302 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7303 * tree-ssa-dse.c (live_bytes_read): New function.
7304 (dse_classify_store): Ignore reads of dead bytes.
7306 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7307 typos and whitespace errors.
7308 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7309 * config/s390/s390.c (s390_emit_prologue): Likewise.
7311 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7314 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7315 with a symbol for LRA.
7317 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7319 PR rtl-optimization/82338
7320 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7322 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7324 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7325 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7326 (qsort_chk): Declare.
7327 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7328 (qsort_chk): New function.
7330 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7332 PR tree-optimization/82337
7333 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7334 phi definition if the PHI result appears in an abnormal PHI.
7335 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7336 basis appears in an abnormal PHI.
7338 2017-09-29 Richard Biener <rguenther@suse.de>
7340 * graphite-isl-ast-to-gimple.c
7341 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7342 (binary_op_to_tree): Use it.
7343 (get_rename_from_scev): Likewise.
7344 (copy_loop_phi_nodes): Likewise.
7345 (copy_bb_and_scalar_dependences): Likewise.
7346 (translate_pending_phi_nodes): Likewise.
7348 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7351 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7352 for movabsq $(i32 << shift), r64.
7354 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7356 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7357 index when encoding %esp as %rsp to avoid 0x67 prefix.
7359 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7361 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7362 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7364 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7366 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7367 Extensions with more than 16 double VFP registers.
7368 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7369 to_clear_mask and all code related to it. Replace the remaining
7370 entry by a sbitmap and adapt code accordingly.
7372 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7374 * brig-builtins.def: Change pure attributes to const.
7376 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7378 * config.gcc (default_gnu_indirect_function): Default to yes for
7379 sparc*-*-linux* with glibc.
7381 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7383 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7384 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7385 when creating .init_array and .fini_array sections with priority
7388 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7391 * config/aarch64/aarch64.c
7392 (aarch64_builtin_support_vector_misalignment): Always return false
7393 when misalignment is unknown.
7395 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7397 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7398 this function to return false if the definition used by the swap
7399 instruction is artificial, or if the memory address from which the
7400 constant value is loaded is not represented by a base address held
7401 in a register or if the base address register is a frame or stack
7402 pointer. Additionally, return false if the base address of the
7403 loaded constant is a SYMBOL_REF but is not considered to be a
7405 (replace_swapped_load_constant): New function.
7406 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7407 loaded constant vector with a load of a swapped constant vector.
7409 2017-09-27 Carl Love <cel@us.ibm.com>
7411 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7412 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7413 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7416 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7418 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7419 first, always call autopref_rank_data otherwise.
7421 2017-09-27 Richard Biener <rguenther@suse.de>
7423 * graphite-scop-detection.c (find_scop_parameters): Move
7424 loop bound handling ...
7425 (gather_bbs::before_dom_children): ... here, avoiding the need
7426 to build scop_info->loop_nest.
7427 (record_loop_in_sese): Remove.
7428 * sese.h (sese_info_t::loop_nest): Remove.
7429 * sese.c (new_sese_info): Do not allocate loop_nest.
7430 (free_sese_info): Do not free loop_nest.
7432 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7435 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7436 lhs from calls if the lhs has addressable type.
7438 2017-09-27 Richard Biener <rguenther@suse.de>
7440 * graphite.h (scop::max_alias_set): New member.
7441 * graphite-scop-detection.c: Remove references to non-existing
7442 --param in comments.
7443 (build_alias_sets): Record the maximum alias set used for drs.
7444 (build_scops): Support zero as unlimited for
7445 --param graphite-max-arrays-per-scop.
7446 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7448 (build_poly_sr_1): ... here. Compute alias set based on the
7449 maximum alias set used for drs rather than
7450 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7452 2017-09-27 Richard Biener <rguenther@suse.de>
7454 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7455 --param loop-block-tile-size=0 to disable tiling.
7457 2017-09-27 Richard Biener <rguenther@suse.de>
7459 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7460 (graphite-max-nb-scop-params): Document special value zero.
7461 * domwalk.h (dom_walker::STOP): New symbolical constant.
7462 (dom_walker::dom_walker): Add optional parameter for bb to
7464 (dom_walker::~dom_walker): Declare.
7465 (dom_walker::before_dom_children): Document STOP return value.
7466 (dom_walker::m_user_bb_to_rpo): New member.
7467 (dom_walker::m_bb_to_rpo): Likewise.
7468 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7469 mapping here if not provided by the user.
7470 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7471 provided by the user.
7472 (dom_walker::STOP): Define.
7473 (dom_walker::walk): Do not compute bb to RPO mapping here.
7474 Support STOP return value from before_dom_children to stop
7476 * graphite-optimize-isl.c (optimize_isl): If the schedule
7477 is the same still generate code if -fgraphite-identity
7478 or -floop-parallelize-all are given.
7479 * graphite-scop-detection.c: Include cfganal.h.
7480 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7482 (gather_bbs::before_dom_children): Return STOP for BBs
7484 (build_scops): Compute bb to RPO mapping and pass it to
7485 the domwalk. Treat --param graphite-max-nb-scop-params=0
7486 as not limiting the number of params.
7487 * graphite.c (graphite_initialize): Remove limit on the
7488 number of basic-blocks in a function.
7489 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7490 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7491 default value of 10.
7493 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7495 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7496 Adjust code to eliminate needing to do the shift right 32-bits
7497 operation after XSCVDPSPN.
7499 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7501 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7502 ((X / Y) != 0 -> X >= Y): Likewise.
7504 2017-09-26 Carl Love <cel@us.ibm.com>
7506 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7507 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7508 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7509 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7510 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7511 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7512 definitions and overloading.
7513 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7514 statement for P9V_BUILTIN_XST_LEN_R.
7515 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7516 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7517 define_expand and define_insn for the instructions and builtins.
7518 * doc/extend.texi: Update the built-in documentation file for the new
7520 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7521 define_insn for the instructions
7523 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7526 * gcc/config/netbsd-protos.h: New file.
7527 * gcc/config/netbsd.c: New file.
7528 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7529 * gcc/config/t-netbsd: New file.
7530 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7531 (tmake_file) Add t-netbsd.
7532 (extra_objs) Add netbsd.o.
7534 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7538 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7540 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7542 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7543 sign extension from a vector register to a GPR by doing a 32-bit
7544 direct move and then an EXTSW.
7545 (extendsi<mode>2 splitter): Likewise.
7546 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7547 right or vector extract after doing XSCVDPSPN. Use
7548 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7550 (movdi_from_sf_zero_ext): Likewise.
7551 (reload_gpr_from_vsxsf): Likewise.
7552 (p8_mfvsrd_4_disf): Delete, no longer used.
7553 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7554 then needing to move the SFmode to a GPR to use the XSCVDPSP
7555 instruction instead of FRSP and XSCVDPSPN.
7556 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7557 it is adjacent to the other XSCVSPDP insns.
7558 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7559 SFmode to be in traditional Altivec registers.
7560 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7561 (vsx_xscvspdpn): Likewise.
7562 (vsx_xscvspdpn_scalar): Likewise.
7564 2017-09-26 Martin Jambor <mjambor@suse.cz>
7566 * tree-sra.c (compare_access_positions): Put integral types first,
7567 stabilize sorting of integral types, remove conditions putting
7568 non-full-precision integers last.
7569 (sort_and_splice_var_accesses): Disable scalarization if a
7570 non-integert would be represented by a non-full-precision integer.
7572 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7574 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7575 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7576 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7577 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7578 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7579 conditionals inside the function instead of around it. Call
7580 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7581 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7583 2017-09-26 Richard Biener <rguenther@suse.de>
7585 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7587 (scop_detection::build_scop_breadth): ... this. Removed.
7588 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7589 (scop_detection::harmful_stmt_in_bb): Likewise.
7590 (scop_detection::graphite_can_represent_stmt): Likewise.
7591 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7592 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7593 (scop_detection::can_represent_loop_1): ... this. Removed.
7594 (scop_detection::harmful_loop_in_region): Simplify after inlining
7595 the above and remove more quadraticness.
7596 (build_scops): Adjust.
7597 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7600 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7603 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7604 REGNO (base) == SP_REG if base is a REG.
7607 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7608 if it is different SSA_NAME.
7609 (optimize_range_tests_cmp_bitwise): New function.
7610 (optimize_range_tests): Call it.
7612 2017-09-26 Richard Biener <rguenther@suse.de>
7614 PR tree-optimization/82321
7615 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7616 for the def being inside the loop.
7618 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7620 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7622 * config/s390/s390-builtins.def: Fix constraint on op4.
7624 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7626 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7627 independent expanders.
7628 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7629 ("vec_ordered", "vec_unordered"): New expanders.
7631 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7633 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7636 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7638 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7639 vec_unpacks_lo_v16qi.
7640 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7642 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7644 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7645 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7646 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7648 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7650 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7652 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7654 ("*vec_slb<mode>"): New insn pattern.
7655 ("vec_shr_<mode>"): New expander.
7656 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7657 and force the shift count operand to V16QImode.
7658 ("vec_srb<mode>"): Set shift count mode to V16QI.
7660 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7662 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7663 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7664 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7666 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7669 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7671 2017-09-26 Richard Biener <rguenther@suse.de>
7673 PR tree-optimization/82320
7674 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7677 2017-09-25 Jeff Law <law@redhat.com>
7679 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7680 prototype for new argument.
7681 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7682 mostly extracted from rs6000_emit_allocate_stack.
7683 (rs6000_emit_probe_stack_range_stack_clash): New function.
7684 (rs6000_emit_allocate_stack): Call
7685 rs6000_emit_probe_stack_range_stack_clash as needed.
7686 (rs6000_emit_probe_stack_range): Add additional argument
7687 to call to gen_probe_stack_range{si,di}.
7688 (output_probe_stack_range): New.
7689 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7690 (output_probe_stack_range_stack_clash): New.
7691 (rs6000_emit_prologue): Emit notes into dump file as requested.
7692 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7693 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7694 Add additional operand and pass it to output_probe_stack_range.
7696 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7698 PR tree-optimization/82163
7699 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7700 (checking_verify_loop_closed_ssa): New parameter.
7701 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7702 (check_loop_closed_ssa_stmt): Delete.
7703 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7704 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7705 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7708 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7710 * brig-builtins.def: Treat HSAIL barrier builtins as
7711 setjmp/longjump style functions.
7713 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7715 * target.def (constant_alignment): New hook.
7716 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7717 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7718 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7719 * doc/tm.texi: Regenerate.
7720 * targhooks.h (default_constant_alignment): Declare.
7721 (constant_alignment_word_strings): Likewise.
7722 * targhooks.c (default_constant_alignment): New function.
7723 (constant_alignment_word_strings): Likewise.
7724 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7725 instead of CONSTANT_ALIGNMENT.
7726 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7727 (force_const_mem): Likewise.
7728 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7729 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7730 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7731 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7732 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7734 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7735 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7736 constant_alignment_word_strings.
7737 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7738 (CONSTANT_ALIGNMENT): Likewise.
7739 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7740 (arm_constant_alignment): New function.
7741 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7742 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7743 constant_alignment_word_strings.
7744 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7745 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7746 constant_alignment_word_strings.
7747 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7748 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7749 (cris_constant_alignment): New function.
7750 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7751 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7752 (epiphany_constant_alignment): New function.
7753 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7754 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7755 constant_alignment_word_strings.
7756 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7757 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7758 constant_alignment_word_strings.
7759 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7760 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7761 constant_alignment_word_strings.
7762 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7763 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7764 * config/i386/i386.c (ix86_constant_alignment): Make static.
7765 Use the same interface as the target hook.
7766 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7767 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7768 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7769 constant_alignment_word_strings.
7770 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7771 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7772 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7773 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7774 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7775 constant_alignment_word_strings.
7776 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7777 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7778 constant_alignment_word_strings.
7779 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7780 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7781 constant_alignment_word_strings.
7782 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7783 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7785 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7786 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7787 * config/mips/mips.c (mips_constant_alignment): New function.
7788 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7789 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7790 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7791 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7792 (mmix_constant_alignment): Make static. Use the same interface
7794 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7795 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7796 constant_alignment_word_strings.
7797 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7798 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7799 constant_alignment_word_strings.
7800 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7801 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7802 constant_alignment_word_strings.
7803 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7804 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7805 (rs6000_constant_alignment): New function.
7806 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7807 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7808 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7809 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7810 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7811 (rs6000_constant_alignment): New function.
7812 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7813 * config/s390/s390.c (s390_constant_alignment): New function.
7814 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7815 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7816 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7817 constant_alignment_word_strings.
7818 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7819 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7820 (sparc_constant_alignment): New function.
7821 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7822 * config/spu/spu.c (spu_constant_alignment): New function.
7823 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7824 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7825 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7826 constant_alignment_word_strings.
7827 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7828 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7829 constant_alignment_word_strings.
7830 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7831 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7832 constant_alignment_word_strings.
7833 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7834 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7835 (visium_constant_alignment): New function.
7836 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7837 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7838 (xtensa_constant_alignment): New function.
7839 * system.h (CONSTANT_ALIGNMENT): Poison.
7841 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7843 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7844 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7845 (rs6000_builtin_valid_without_lhs): New helper function.
7846 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7847 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7849 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7851 * target.h (vec_perm_indices): Use unsigned short rather than
7853 (auto_vec_perm_indices): Likewise.
7854 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7855 Use unsigned int rather than unsigned char.
7856 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7858 2017-09-25 Richard Biener <rguenther@suse.de>
7860 * cfgloop.h (sort_sibling_loops): Declare.
7861 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7862 (sort_sibling_loops): New function sorting the sibling loop list
7864 * graphite.c (graphite_transform_loops): Sort sibling loops.
7866 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7868 * target.def (vec_perm_const_ok): Change sel parameter to
7870 * optabs-query.c (can_vec_perm_p): Update accordingly.
7871 * doc/tm.texi: Regenerate.
7872 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7873 auto_vec_perm_indices and remove separate nelt field.
7874 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7875 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7876 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7877 (aarch64_expand_vec_perm_const): Update accordingly.
7878 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7879 to vec_perm_indices.
7880 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7881 auto_vec_perm_indices and remove separate nelt field.
7882 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7883 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7884 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7886 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7887 to vec_perm_indices.
7888 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7889 sel to vec_perm_indices.
7890 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7891 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7892 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7894 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7897 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7900 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7901 on the FUNCTION_DECL function context if it has a DIE that is a
7904 2017-09-25 Richard Biener <rguenther@suse.de>
7906 PR tree-optimization/82285
7907 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7910 2017-09-25 Tom de Vries <tom@codesourcery.com>
7914 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7917 2017-09-25 Richard Biener <rguenther@suse.de>
7919 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7920 ISL errors other than isl_error_quota happen. Dump if the
7921 schedule is the same.
7922 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7923 errors instead of aborting inside ISL.
7925 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7928 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7929 of libgcc_eh for m64.
7930 * config/i386/darwin64.h: Likewise.
7932 2017-09-25 Richard Biener <rguenther@suse.de>
7935 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7936 attribute for incomplete types nor twice for complete ones.
7938 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7941 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7942 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7945 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7948 * config/i386/i386.opt (mprefer-avx256): Use
7949 ix86_target_flags variable.
7950 * config/i386/i386.c (ix86_target_string): Move
7951 -mprefer-avx256 to flag2_opts.
7953 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7956 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7957 and x != -1 | y != -1 into (x & y) != -1.
7959 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7961 * config.gcc: Add new case statement to set
7962 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7963 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7964 s390x-*-linux* case statements. Added aarch64 to the list of
7965 supported architectures.
7967 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7969 PR tree-optimization/82289
7970 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7971 STMT_VINFO_RELEVANT_P.
7973 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7974 Alan Hayward <alan.hayward@arm.com>
7975 David Sherwood <david.sherwood@arm.com>
7977 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7978 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7979 (extract_range_from_binary_expr_1): Don't call
7980 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7982 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7983 Alan Hayward <alan.hayward@arm.com>
7984 David Sherwood <david.sherwood@arm.com>
7986 * target.def (preferred_vector_alignment): New hook.
7987 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7989 * doc/tm.texi: Regenerate.
7990 * targhooks.h (default_preferred_vector_alignment): Declare.
7991 * targhooks.c (default_preferred_vector_alignment): New function.
7992 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7994 (DR_TARGET_ALIGNMENT): New macro.
7995 (aligned_access_p): Update commentary.
7996 (vect_known_alignment_in_bytes): New function.
7997 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7999 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
8000 Calculate the misalignment based on the target alignment rather than
8002 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
8003 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
8004 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8005 the target alignment, rather than masking the element misalignment
8006 with the number of elements in a vector. Also use the target
8007 alignment when calculating the maximum number of peels.
8008 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8009 instead of TYPE_ALIGN_UNIT.
8010 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8011 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8012 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8013 (vect_create_data_ref_ptr): Likewise.
8014 (vect_setup_realignment): Realign by ANDing with
8015 -DR_TARGET_MISALIGNMENT.
8016 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8017 the number of peels based on DR_TARGET_ALIGNMENT.
8018 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8019 with the guaranteed alignment boundary when deciding whether
8021 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8022 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8023 (ensure_base_align): Remove stmt_info parameter. Get the
8024 target base alignment from DR_TARGET_ALIGNMENT.
8025 (vectorizable_store): Update call accordingly. Interpret
8026 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8028 (vectorizable_load): Likewise.
8030 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8031 Alan Hayward <alan.hayward@arm.com>
8032 David Sherwood <david.sherwood@arm.com>
8034 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8035 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8036 (vect_enhance_data_refs_alignment): Likewise.
8038 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8040 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8041 error. Only quit immediately if parsing is complete.
8042 (BEGIN): Initialize fatal_err and parse_done.
8043 (begin fpu, end fpu): Check number of arguments.
8044 (begin arch, end arch): Likewise.
8045 (begin cpu, end cpu): Likewise.
8046 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8047 (optalias): Likewise.
8049 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8051 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8052 * config/arm/arm-isa.h: Delete. Move definitions to ...
8053 * arm-cpus.in: ... here. Use new feature and fgroup values.
8054 * config/arm/arm.c (arm_option_override): Use lower case for feature
8056 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8057 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8058 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8060 (print_isa_bits_for): New function.
8061 (gen_isa): New function.
8062 (gen_comm_data): Use print_isa_bits_for.
8063 (define feature): New keyword.
8064 (define fgroup): New keyword.
8065 * config/arm/t-arm (TM_H): Remove.
8066 (GTM_H): Add arm-isa.h.
8067 (arm-isa.h): Add rule to generate file.
8068 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8069 case for feature bit names.
8071 2017-09-22 Richard Biener <rguenther@suse.de>
8073 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8075 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8076 print of no dependency loops ...
8077 * graphite.c (graphite_transform_loops): ... here.
8078 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8080 (same_close_phi_node, remove_duplicate_close_phi,
8081 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8082 (canonicalize_loop_closed_ssa): ... here and simplify.
8083 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8084 (optimize_isl): Use dump_printf_loc to tell when we stopped
8085 optimizing because of an ISL timeout.
8087 2017-09-22 Richard Biener <rguenther@suse.de>
8089 PR tree-optimization/82291
8090 * tree-if-conv.c (predicate_mem_writes): Make sure to
8091 remove writes in blocks predicated with false.
8093 2017-09-22 Richard Biener <rguenther@suse.de>
8095 * sese.c: Include cfganal.h.
8096 (if_region_set_false_region): Remove.
8097 (create_if_region_on_edge): Likewise.
8098 (move_sese_in_condition): Re-implement without destroying
8101 2017-09-22 Richard Biener <rguenther@suse.de>
8103 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8104 Verify both BBs contain loop PHI nodes before dispatching to
8106 (graphite_regenerate_ast_isl): Do not recompute dominators,
8107 do not verify three times. Restructure for clarity.
8108 * graphite-scop-detection.c (same_close_phi_node,
8109 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8110 defined_in_loop_p, canonicalize_loop_closed_ssa,
8111 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8112 checking and SSA rewrite, move to ...
8113 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8114 (graphite_initialize): Do not pass in ctx, do not reset the
8115 SCEV cache, compute only dominators.
8116 (graphite_transform_loops): Allocate ISL ctx after
8117 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8118 Maintain post-dominators only around build_scops.
8119 * sese.c (if_region_set_false_region): Make static. Free
8120 and recompute dominators.
8121 (move_sese_in_condition): Assert we don't get called with
8122 post-dominators computed.
8123 * sese.h (if_region_set_false_region): Remove.
8125 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8127 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8128 mode attribute for TARGET_AVX512VL.
8130 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8132 * config/i386/i386.opt (mprefer-avx256): New option.
8133 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8135 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8136 for TARGET_PREFER_AVX256.
8137 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8139 2017-09-21 Jeff Law <law@redhat.com>
8141 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8142 Fix dump output if the only stack space is for pushed registers.
8144 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8146 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8149 2017-09-21 Martin Sebor <msebor@redhat.com>
8152 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8153 code (in C++) or code that triggers warnings.
8155 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8157 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8159 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8161 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8162 * sched-rgn.c: Ditto.
8163 * sel-sched-ir.c: Ditto.
8165 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8167 * toplev.h (set_random_seed): Adjust return type.
8168 * toplev.c (init_local_tick): Move eager initialization of random_seed
8169 to get_random_seed. Adjust comment.
8170 (init_random_seed): Inline to get_random_seed, delete.
8171 (get_random_seed): Initialize random_seed lazily.
8172 (set_random_seed): Do not return previous value.
8173 (print_switch_value): Do not call get_random_seed.
8175 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8177 * cgraph.c (delete_function_version): New, broken out from...
8178 (cgraph_node::delete_function_version): ...here. Rename to
8179 cgraph_node::delete_function_version_by_decl. Update all uses.
8180 (cgraph_node::remove): Call delete_function_version.
8182 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8185 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8186 VAR_DECLs to which addressable non-volatile parameters are mapped
8187 and for id->retvar after the return value assignment. Clear
8188 id->retval and id->retbnd after inlining.
8190 2017-09-21 Richard Biener <rguenther@suse.de>
8192 PR tree-optimization/82276
8193 PR tree-optimization/82244
8194 * tree-vrp.c (build_assert_expr_for): Set
8195 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8197 (remove_range_assertions): Revert earlier change.
8199 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8202 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8204 2017-09-21 Richard Biener <rguenther@suse.de>
8206 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8207 Restore valid IL after code generation errors.
8208 * graphite.c (graphite_transform_loops): Diagnose code
8209 generation issues as MSG_MISSED_OPTIMIZATION and continue
8210 with processing SCOPs.
8212 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8213 Alan Hayward <alan.hayward@arm.com>
8214 David Sherwood <david.sherwood@arm.com>
8216 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8217 rather than choosing between plus_constant and gen_rtx_<CODE>.
8218 * expr.c (emit_push_insn): Likewise.
8219 (expand_expr_real_2): Likewise.
8221 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8222 Alan Hayward <alan.hayward@arm.com>
8223 David Sherwood <david.sherwood@arm.com>
8225 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8227 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8228 Alan Hayward <alan.hayward@arm.com>
8229 David Sherwood <david.sherwood@arm.com>
8231 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8232 calling tree_to_uhwi.
8234 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8235 Alan Hayward <alan.hayward@arm.com>
8236 David Sherwood <david.sherwood@arm.com>
8238 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8239 INTEGER_CST rather than a negative test for ADDR_EXPR.
8241 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8242 Alan Hayward <alan.hayward@arm.com>
8243 David Sherwood <david.sherwood@arm.com>
8245 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8246 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8248 2017-09-21 Richard Biener <rguenther@suse.de>
8250 PR tree-optimization/71351
8251 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8252 graphite_create_new_loop_guard): Remove, fold remaining parts
8254 (translate_isl_ast_node_for): ... here and simplify.
8256 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8259 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8260 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8261 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8262 alternative always use QI mode, for -Os imov (=R,R) alternative
8263 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8266 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8267 Jeff Law <law@redhat.com>
8269 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8270 (allocate_stack_space): New function, partially extracted from
8272 (s390_emit_prologue): Track offset to most recent stack probe.
8273 Code to allocate space moved into allocate_stack_space.
8274 Dump actions when no stack is allocated.
8275 (s390_prologue_plus_offset): New function.
8276 (s390_emit_stack_probe): Likewise.
8278 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8280 * common.opt (Wa, Wl, Wp, g, gz=): Add
8282 (gno-column-info): Remove.
8283 (gcolumn-info): Drop RejectNegative.
8285 (gno-record-gcc-switches): Remove.
8286 (grecord-gcc-switches): Drop RejectNegative.
8287 (gno-split-dwarf): Remove.
8288 (gsplit-dwarf): Drop RejectNegative.
8289 (gno-strict-dwarf): Remove.
8290 (gstrict-dwarf): Drop RejectNegative.
8291 * config/darwin.opt (gfull, gused): Add RejectNegative.
8292 * dwarf2out.c (gen_producer_string): Drop
8293 gno-record-gcc-switches handler.
8294 * optc-gen.awk: Add g to prefixes with negative forms.
8295 * opts-common.c (remapping_prefix_p): New.
8296 (find_opt): Check it.
8297 (generate_canonical_option): Test g prefix.
8298 (option_map): Add -gno- mapping.
8299 (add_misspelling_candidates): Check remapping_prefix_p.
8301 2017-09-20 Jeff Law <law@redhat.com>
8303 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8304 thinko in stack clash protection support.
8306 * explow.c (compute_stack_clash_protection_loop_data): Use
8307 CONST_INT_P instead of explicit test. Verify object is a
8308 CONST_INT_P before looking at INTVAL.
8309 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8310 instead of explicit test.
8312 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8315 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8316 address instead of to r1 and r11.
8318 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8320 * config.gcc: Support "knm".
8321 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8322 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8324 * config/i386/i386.c (m_KNM): Define.
8325 (processor_target_table): Add "knm".
8327 (ix86_option_override_internal): Add "knm".
8328 (ix86_issue_rate): Add PROCESSOR_KNM.
8329 (ix86_adjust_cost): Ditto.
8330 (ia32_multipass_dfa_lookahead): Ditto.
8331 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8332 (fold_builtin_cpu): Add M_INTEL_KNM.
8333 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8334 (processor_type): Add PROCESSOR_KNM.
8335 * config/i386/x86-tune.def: Add m_KNM.
8336 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8338 2017-09-20 Richard Biener <rguenther@suse.de>
8340 PR tree-optimization/80213
8341 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8342 are allowed in empty BBs as well.
8343 (canonicalize_loop_closed_ssa): Also look for other complex
8345 (scop_detection::get_sese): Include the loop-closed PHI block
8347 (scop_detection::merge_sese): Remove code adding extra blocks.
8348 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8349 (build_scops): Assert the final returned scop is invalid.
8351 2017-09-20 Richard Biener <rguenther@suse.de>
8353 PR tree-optimization/82264
8354 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8356 (vn_phi_lookup): Likewise.
8357 (vn_phi_insert): Likewise.
8359 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8361 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8362 that fits into uhwi or shwi, add DW_AT_const_value regardless
8363 of early_dwarf without going through RTL, using add_AT_unsigned
8366 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8367 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8368 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8369 (DEBUG_MACRO_SECTION): Likewise.
8370 (DEBUG_LTO_MACRO_SECTION): Likewise.
8371 (DEBUG_STR_DWO_SECTION): Likewise.
8372 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8373 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8374 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8375 (DEBUG_LTO_LINE_STR_SECTION): Define.
8376 (init_sections_and_labels): Initialize debug_line_str_section
8377 variable. Initialize debug_loc_section for -gdwarf-5 to
8378 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8380 2017-09-20 Richard Biener <rguenther@suse.de>
8382 * graphite-sese-to-poly.c (extract_affine): Properly handle
8383 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8385 2017-09-20 Richard Biener <rguenther@suse.de>
8387 PR tree-optimization/81373
8388 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8389 Force SESE live-out defs to be handled even if they are
8392 2017-09-19 Jeff Law <law@redhat.com>
8394 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8395 nothing for stack adjustments with REG_STACK_CHECK.
8396 * sched-deps.c (parse_add_or_inc): Reject insns with
8397 REG_STACK_CHECK from dependency breaking.
8398 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8399 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8400 * reg-notes.def (STACK_CHECK): New note.
8402 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8403 (ix86_expand_prologue): Dump stack clash info as needed.
8404 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8406 * function.c (dump_stack_clash_frame_info): New function.
8407 * function.h (dump_stack_clash_frame_info): Prototype.
8408 (enum stack_clash_probes): New enum.
8410 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8411 flag_stack_clash_protection.
8412 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8413 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8414 (arm_frame_pointer_required): Likewise.
8415 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8416 (ia64_expand_prologue): Likewise.
8417 * config/mips/mips.c (mips_expand_prologue): Likewise.
8418 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8419 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8420 (sparc_flat_expand_prologue): Likewise.
8421 * config/spu/spu.c (spu_expand_prologue): Likewise.
8423 * explow.c: Include "params.h".
8424 (anti_adjust_stack_and_probe_stack_clash): New function.
8425 (get_stack_check_protect): Likewise.
8426 (compute_stack_clash_protection_loop_data): Likewise.
8427 (emit_stack_clash_protection_loop_start): Likewise.
8428 (emit_stack_clash_protection_loop_end): Likewise.
8429 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8430 Use anti_adjust_stack_and_probe_stack_clash.
8431 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8432 (emit_stack_clash_protection_loop_start): Likewise.
8433 (emit_stack_clash_protection_loop_end): Likewise.
8434 * rtl.h (get_stack_check_protect): Prototype.
8435 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8436 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8437 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8439 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8441 * doc/tm.texi: Rebuilt.
8442 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8443 get_stack_check_protect.
8444 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8445 * config/arm/arm.c (arm_expand_prologue): Likewise.
8446 (arm_frame_pointer_required): Likewise.
8447 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8448 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8449 * config/mips/mips.c (mips_expand_prologue): Likewise.
8450 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8451 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8452 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8453 (sparc_flat_expand_prologue): Likewise.
8455 * common.opt (-fstack-clash-protection): New option.
8456 * flag-types.h (enum stack_check_type): Note difference between
8457 -fstack-check= and -fstack-clash-protection.
8458 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8459 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8460 * toplev.c (process_options): Issue warnings/errors for cases
8461 not handled with -fstack-clash-protection.
8462 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8463 (-fstack-check): Note additional problem with -fstack-check=generic.
8464 Note that -fstack-check is primarily for Ada and refer users
8465 to -fstack-clash-protection for stack-clash-protection.
8466 Document new params for stack clash protection.
8468 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8470 * config/i386/i386.c (ix86_split_long_move): Do not handle
8471 address used for LEA in a special way.
8473 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8475 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8477 2017-09-19 Martin Sebor <msebor@redhat.com>
8480 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8481 of incompatible types.
8483 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8485 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8486 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8487 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8488 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8490 2017-09-19 Richard Biener <rguenther@suse.de>
8492 PR tree-optimization/82244
8493 * tree-vrp.c (remove_range_assertions): Do not propagate
8494 a constant to abnormals but replace the assert with a copy.
8496 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8498 PR rtl-optimization/57878
8499 PR rtl-optimization/68988
8500 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8501 avoidance test involving non_reload_pseudos. Move frequency test
8502 below the general fragmentation avoidance test.
8504 2017-09-19 Richard Biener <rguenther@suse.de>
8506 PR tree-optimization/69728
8507 * graphite-sese-to-poly.c (schedule_error): New global.
8508 (add_loop_schedule): Handle empty domain by failing the
8510 (build_original_schedule): Handle schedule_error.
8512 2017-09-19 Richard Biener <rguenther@suse.de>
8514 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8515 Do not iterate to sibling loops but only to siblings of inner
8518 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8521 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8522 that the registers are different.
8524 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8526 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8527 to processor_model and "amdfam17h" to arch_names_table.
8528 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8530 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8533 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8535 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8536 Alan Hayward <alan.hayward@arm.com>
8537 David Sherwood <david.sherwood@arm.com>
8539 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8541 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8543 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8544 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8545 vect_schedule_slp_instance.
8546 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8547 Update call to vect_slp_analyze_node_operations. Simplify return
8549 (vect_slp_analyze_bb_1): Update call accordingly.
8550 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8551 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8552 (vect_schedule_slp): Update call accordingly.
8554 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8555 Alan Hayward <alan.hayward@arm.com>
8556 David Sherwood <david.sherwood@arm.com>
8558 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8559 with types that aren't in fact scalar.
8561 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8563 * tree-vect-slp.c (vect_record_max_nunits): New function,
8565 (vect_build_slp_tree_1): ...here.
8566 (vect_build_slp_tree_2): Call it for phis too.
8568 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8570 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8571 to vect_get_vec_def_for_operand when getting the mask operand.
8573 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8574 Alan Hayward <alan.hayward@arm.com>
8575 David Sherwood <david.sherwood@arm.com>
8577 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8580 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8581 Alan Hayward <alan.hayward@arm.com>
8582 David Sherwood <david.sherwood@arm.com>
8584 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8585 calculation for vector booleans.
8587 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8588 Alan Hayward <alan.hayward@arm.com>
8589 David Sherwood <david.sherwood@arm.com>
8591 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8593 (vect_transform_stmt): ...here.
8594 (vect_analyze_stmt): Use it instead of calling
8595 vectorizable_live_operation directly.
8597 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8599 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8600 non-SIMT targets in acc vector loops.
8602 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8604 * configure.ac: Add arc and check if assembler supports gdwarf2.
8605 * configure: Regenerate.
8607 2017-09-18 Richard Biener <rguenther@suse.de>
8609 PR tree-optimization/82220
8610 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8611 epilogue niters from the min_profitable_iters compute.
8613 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8616 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8617 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8618 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8620 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8623 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8624 switching to a new text section.
8626 2017-09-18 Richard Biener <rguenther@suse.de>
8628 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8630 (build_alias_set): Reject aliases with no access function.
8632 2017-09-18 Richard Biener <rguenther@suse.de>
8634 PR tree-optimization/79622
8635 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8637 (build_cross_bb_scalars_use): Likewise.
8639 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8641 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8643 2017-09-18 Alan Modra <amodra@gmail.com>
8646 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8647 stack_pointer_rtx for count 0. Update comments. Break up
8648 large rtl expression.
8650 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8652 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8653 Increase to 20 bytes.
8654 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8655 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8656 or avx version of the stub.
8658 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8661 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8662 compute the minimum stack alignment. Also update preferred stack
8663 boundary for leaf functions.
8665 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8667 PR tree-optimization/82228
8668 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8671 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8673 * common/config/nds32/nds32-common.c
8674 (nds32_option_optimization_table): Refine formatting.
8675 (nds32_option_optimization_table): Use -fsched-pressure and
8676 -fomit-frame-pointer for specific optimization level.
8678 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8680 * config/nds32/nds32.c: Refine formatting and comments.
8681 * config/nds32/nds32.h: Likewise.
8682 * config/nds32/nds32.md: Likewise.
8683 * config/nds32/nds32-cost.c: Likewise.
8684 * config/nds32/nds32-isr.c: Likewise.
8685 * config/nds32/nds32-md-auxiliary.c: Likewise.
8686 * config/nds32/nds32-multiple.md: Likewise.
8687 * config/nds32/nds32-predicates.c: Likewise.
8689 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8690 Jakub Jelinek <jakub@redhat.com>
8692 Add support for -std=c++2a.
8693 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8695 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8697 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8700 * doc/extend.texi (Common Function Attributes): Add
8701 references to ARM, AArch64, and S/390 specific attributes.
8702 (Function Specific Option Pragmas): Add AArch64 and S/390
8703 to list of back ends that support the target pragma.
8705 2017-09-15 Nathan Sidwell <nathan@acm.org>
8707 * doc/standards.texi: Fix C++17 description. Update URLs for
8710 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8712 * common.opt (Wcast-align=strict): New warning option.
8713 * doc/invoke.texi: Document -Wcast-align=strict.
8715 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8717 * cgraph.h (cgraph_thunk_info): Add comments.
8718 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8719 assert for VIRTUAL_* arguments stricter.
8721 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8723 PR tree-optimization/71026
8724 * match.pd: Move RDIV patterns from fold-const.c
8725 * fold-const.c (distribute_real_division): Removed.
8726 (fold_binary_loc): Remove calls to distribute_real_divison.
8728 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8730 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8731 c++1z and gnu++1z as deprecated. Change other references to
8732 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8733 Change -Wc++1z-compat to -Wc++17-compat.
8734 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8735 * dwarf2out.c (highest_c_language): Handle C++17.
8736 (gen_compile_unit_die): Likewise.
8738 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8740 PR rtl-optimization/82192
8741 * combine.c (make_extraction): Don't look through non-paradoxical
8742 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8745 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8746 Alan Hayward <alan.hayward@arm.com>
8747 David Sherwood <david.sherwood@arm.com>
8749 * target.def (function_arg_offset): New hook.
8750 * targhooks.h (default_function_arg_offset): Declare.
8751 * targhooks.c (default_function_arg_offset): New function.
8752 * function.c (locate_and_pad_parm): Use
8753 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8754 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8755 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8756 * doc/tm.texi: Regenerate.
8757 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8758 * config/spu/spu.c (spu_function_arg_offset): New function.
8759 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8760 * system.h (FUNCTION_ARG_OFFSET): Poison.
8762 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8763 Alan Hayard <alan.hayward@arm.com>
8764 David Sherwood <david.sherwood@arm.com>
8766 * target.def (truly_noop_truncation): New hook.
8767 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8768 than TRULY_NOOP_TRUNCATION.
8769 * hooks.h (hook_bool_uint_uint_true): Declare.
8770 * hooks.c (hook_bool_uint_uint_true): New function.
8771 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8772 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8773 * doc/tm.texi: Regenerate.
8774 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8775 rather than TRULY_NOOP_TRUNCATION in comments.
8776 (simplify_comparison): Likewise.
8777 (record_truncated_value): Likewise.
8778 * expmed.c (extract_bit_field_1): Likewise.
8779 (extract_split_bit_field): Likewise.
8780 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8781 instead of TRULY_NOOP_TRUNCATION.
8782 * function.c (assign_parm_setup_block): Likewise.
8783 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8784 * rtlhooks.c: Include target.h.
8785 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8786 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8787 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8788 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8789 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8790 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8791 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8792 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8793 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8794 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8795 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8796 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8797 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8798 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8799 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8800 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8801 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8802 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8803 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8804 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8805 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8806 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8807 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8808 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8809 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8810 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8811 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8812 rather than TRULY_NOOP_TRUNCATION in comments.
8813 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8814 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8815 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8816 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8817 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8818 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8819 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8820 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8821 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8822 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8823 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8824 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8825 rather than TRULY_NOOP_TRUNCATION in comments.
8826 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8827 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8828 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8829 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8830 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8831 TRULY_NOOP_TRUNCATION condition.
8832 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8833 (TRULY_NOOP_TRUNCATION): Delete.
8834 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8835 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8836 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8837 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8838 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8839 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8840 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8841 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8842 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8843 rather than TRULY_NOOP_TRUNCATION in comments.
8844 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8845 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8846 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8847 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8848 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8849 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8851 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8854 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8855 (*cmp_ior): Likewise.
8856 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8857 (*ior_scc_scc_cmp): Likewise.
8858 (*and_scc_scc): Likewise.
8859 (*and_scc_scc_cmp): Likewise.
8861 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8862 Alan Hayard <alan.hayward@arm.com>
8863 David Sherwood <david.sherwood@arm.com>
8865 * target.def (can_change_mode_class): New hook.
8866 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8867 (hard_regno_nregs): Likewise.
8868 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8869 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8870 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8871 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8872 (LOAD_EXTEND_OP): Update accordingly.
8873 * doc/tm.texi: Regenerate.
8874 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8875 CANNOT_CHANGE_MODE_CLASS.
8876 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8877 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8878 * combine.c (simplify_set): Update accordingly.
8879 * emit-rtl.c (validate_subreg): Likewise.
8880 * recog.c (general_operand): Likewise.
8881 * regcprop.c (mode_change_ok): Likewise.
8882 * reload1.c (choose_reload_regs): Likewise.
8883 (inherit_piecemeal_p): Likewise.
8884 * rtlanal.c (simplify_subreg_regno): Likewise.
8885 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8886 instead of CANNOT_CHANGE_MODE_CLASS.
8887 (reload_cse_simplify_operands): Likewise.
8888 * reload.c (push_reload): Use targetm.can_change_mode_class
8889 instead of CANNOT_CHANGE_MODE_CLASS.
8890 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8891 REG_CANNOT_CHANGE_MODE_P.
8892 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8893 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8894 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8895 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8896 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8897 (arm_can_change_mode_class): New function.
8898 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8899 than CANNOT_CHANGE_MODE_CLASS in comments.
8900 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8901 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8902 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8903 (ix86_can_change_mode_class): ...this new function, inverting the
8904 sense of the return value.
8905 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8906 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8907 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8908 (ia64_can_change_mode_class): New function.
8909 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8910 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8911 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8912 (m32c_can_change_mode_class): ...this new function, inverting the
8913 sense of the return value.
8914 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8915 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8916 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8917 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8918 (mips_can_change_mode_class): ...this new function, inverting the
8919 sense of the return value.
8920 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8921 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8922 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8923 (msp430_can_change_mode_class): New function.
8924 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8925 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8926 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8927 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8928 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8929 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8930 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8931 (pa_cannot_change_mode_class): Replace with...
8932 (pa_can_change_mode_class): ...this new function, inverting the
8933 sense of the return value.
8934 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8935 than CANNOT_CHANGE_MODE_CLASS in comments.
8936 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8937 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8938 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8939 (pdp11_cannot_change_mode_class): Replace with...
8940 (pdp11_can_change_mode_class): ...this new function, inverting the
8941 sense of the return value.
8942 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8943 * config/powerpcspe/powerpcspe-protos.h
8944 (rs6000_cannot_change_mode_class_ptr): Delete.
8945 * config/powerpcspe/powerpcspe.c
8946 (rs6000_cannot_change_mode_class_ptr): Delete.
8947 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8948 (rs6000_option_override_internal): Assign to
8949 targetm.can_change_mode_class instead of
8950 rs6000_cannot_change_mode_class_ptr.
8951 (rs6000_cannot_change_mode_class): Replace with...
8952 (rs6000_can_change_mode_class): ...this new function, inverting the
8953 sense of the return value.
8954 (rs6000_debug_cannot_change_mode_class): Replace with...
8955 (rs6000_debug_can_change_mode_class): ...this new function.
8956 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8957 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8958 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8959 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8960 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8962 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8963 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8964 (rs6000_option_override_internal): Assign to
8965 targetm.can_change_mode_class instead of
8966 rs6000_cannot_change_mode_class_ptr.
8967 (rs6000_cannot_change_mode_class): Replace with...
8968 (rs6000_can_change_mode_class): ...this new function, inverting the
8969 sense of the return value.
8970 (rs6000_debug_cannot_change_mode_class): Replace with...
8971 (rs6000_debug_can_change_mode_class): ...this new function.
8972 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8973 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8974 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8975 (s390_can_change_mode_class): ...this new function, inverting the
8976 sense of the return value.
8977 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8978 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8979 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8980 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8981 (sh_cannot_change_mode_class): Replace with...
8982 (sh_can_change_mode_class): ...this new function, inverting the
8983 sense of the return value.
8984 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8985 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8986 (sparc_can_change_mode_class): New function.
8987 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8988 * config/spu/spu.c (spu_can_change_mode_class): New function.
8989 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8990 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8991 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8992 (visium_can_change_mode_class): New function.
8993 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8995 2017-09-15 Richard Biener <rguenther@suse.de>
8997 PR tree-optimization/82217
8998 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8999 but not undefined case.
9001 2017-09-15 Jakub Jelinek <jakub@redhat.com>
9004 * postreload.c (reload_cse_simplify_operands): Skip
9005 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9007 2017-09-15 Richard Biener <rguenther@suse.de>
9009 PR tree-optimization/68823
9010 * graphite-scop-detection.c (build_alias_set): If we have a
9011 possible dependence check whether we can handle them by just
9012 looking at the DRs DR_ACCESS_FNs.
9013 (build_scops): If build_alias_set fails, fail the SCOP.
9015 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9017 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9018 to support float128 built-in functions that require the ISA 3.0
9020 (BU_FLOAT128_3_HW): Likewise.
9021 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9023 (FMAF128): Likewise.
9025 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9026 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9027 floating point instructions.
9028 (rs6000_invalid_builtin): Likewise.
9029 (rs6000_builtin_mask_names): Likewise.
9030 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9031 (RS6000_BTM_FLOAT128_HW): Likewise.
9032 (RS6000_BTM_COMMON): Likewise.
9033 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9035 * doc/extend.texi (RS/6000 built-in functions): Document the
9036 IEEE 128-bit floating point square root and fused multiply-add
9039 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9041 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9042 reg (r2) isn't in the set of registers defined in the prologue.
9044 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9045 Alan Hayward <alan.hayward@arm.com>
9046 David Sherwood <david.sherwood@arm.com>
9048 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9049 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9050 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9051 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9052 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9054 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9055 max_vectorization_factor.
9056 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9058 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9059 Alan Hayward <alan.hayward@arm.com>
9060 David Sherwood <david.sherwood@arm.com>
9062 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9063 (vect_worthwhile_without_simd_p): Declare.
9064 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9065 (vectorizable_reduction): Use it.
9066 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9067 (vectorizable_operation): Likewise.
9069 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9070 Alan Hayward <alan.hayward@arm.com>
9071 David Sherwood <david.sherwood@arm.com>
9073 * tree-vectorizer.h (vect_get_num_copies): New function.
9074 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9075 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9076 (vectorizable_induction): Likewise.
9077 (vectorizable_live_operation): Likewise.
9078 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9079 (vectorizable_bswap): Likewise.
9080 (vectorizable_call): Likewise.
9081 (vectorizable_conversion): Likewise.
9082 (vectorizable_assignment): Likewise.
9083 (vectorizable_shift): Likewise.
9084 (vectorizable_operation): Likewise.
9085 (vectorizable_store): Likewise.
9086 (vectorizable_load): Likewise.
9087 (vectorizable_condition): Likewise.
9088 (vectorizable_comparison): Likewise.
9089 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9091 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9092 Alan Hayward <alan.hayward@arm.com>
9093 David Sherwood <david.sherwood@arm.com>
9095 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9096 of vect_init_vector.
9098 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9099 Alan Hayward <alan.hayward@arm.com>
9100 David Sherwood <david.sherwood@arm.com>
9102 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9103 an inline wrapper that provides a location.
9104 (gimple_build_vector): Likewise.
9105 * gimple-fold.c (gimple_build_vector_from_val): New function.
9106 (gimple_build_vector): Likewise.
9107 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9108 functions to build the initial value. Always return a gimple value.
9109 (get_initial_defs_for_reduction): Likewise. Only compute
9111 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9112 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9113 (vectorizable_induction): Use gimple_build_vector rather than
9116 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9117 Alan Hayward <alan.hayward@arm.com>
9118 David Sherwood <david.sherwood@arm.com>
9120 * target.h (vec_perm_indices): New typedef.
9121 (auto_vec_perm_indices): Likewise.
9122 * optabs-query.h: Include target.h
9123 (can_vec_perm_p): Take a vec_perm_indices *.
9124 * optabs-query.c (can_vec_perm_p): Likewise.
9125 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9126 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9127 * tree-vect-generic.c (lower_vec_perm): Likewise.
9128 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9129 (vect_grouped_load_supported): Likewise.
9130 (vect_shift_permute_load_chain): Likewise.
9131 (vect_permute_store_chain): Use auto_vec_perm_indices.
9132 (vect_permute_load_chain): Likewise.
9133 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9134 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9135 Update uses of can_vec_perm_p.
9136 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9137 mode with a number of elements. Take a vec_perm_indices *.
9138 (vect_create_epilog_for_reduction): Update accordingly.
9139 Use auto_vec_perm_indices.
9140 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9141 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9142 (vect_transform_slp_perm_load): Likewise.
9143 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9144 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9145 (vect_gen_perm_mask_checked): Likewise.
9146 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9147 (vect_gen_perm_mask_checked): Likewise.
9148 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9149 (vectorizable_store): Likewise.
9150 (vectorizable_load): Likewise.
9151 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9152 (vectorizable_bswap): Likewise.
9154 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9155 Alan Hayward <alan.hayward@arm.com>
9156 David Sherwood <david.sherwood@arm.com>
9158 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9159 * tree.c (build_vector): Likewise.
9160 (build_vector_from_ctor): Update accordingly.
9161 (build_vector_from_val): Likewise.
9162 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9163 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9164 * tree-vect-generic.c (add_rshift): Likewise.
9165 (expand_vector_divmod): Likewise.
9166 (optimize_vector_constructor): Likewise.
9167 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9168 (vect_transform_slp_perm_load): Likewise.
9169 (vect_schedule_slp_instance): Likewise.
9170 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9171 (vectorizable_call): Likewise.
9172 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9173 * expmed.c (make_tree): Likewise.
9174 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9175 a vector passed to build_vector.
9176 (fold_convert_const): Likewise.
9177 (exact_inverse): Likewise.
9178 (fold_ternary_loc): Likewise.
9179 (fold_relational_const): Likewise.
9180 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9181 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9182 (const_unop): Likewise. Store the reduction accumulator in a
9183 variable rather than an array.
9184 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9185 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9186 the new vector, rather than constructing it after the input arrays.
9187 (native_interpret_vector): Use auto_vec<tree> when building
9188 a vector passed to build_vector. Add elements in order.
9189 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9190 auto_vec<tree> when building a vector passed to build_vector.
9191 (vect_create_epilog_for_reduction): Likewise.
9192 (vectorizable_induction): Likewise.
9193 (get_initial_def_for_reduction): Likewise. Fix indentation of
9195 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9197 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9198 passed to build_vector.
9200 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9201 Alan Hayward <alan.hayward@arm.com>
9202 David Sherwood <david.sherwood@arm.com>
9204 * tree-core.h (tree_base::u): Add an "nelts" field.
9205 (tree_vector): Use VECTOR_CST_NELTS as the length.
9206 * tree.c (tree_size): Likewise.
9207 (make_vector): Initialize VECTOR_CST_NELTS.
9208 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9209 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9210 TYPE_VECTOR_SUBPARTS.
9211 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9212 as the number of units, setting it from VECTOR_CST_NELTS.
9213 (const_vector_from_tree): Likewise.
9214 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9215 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9216 (fold_negate_expr_1): Likewise.
9217 (fold_convert_const): Likewise.
9218 (const_binop): Likewise. Differentiate the number of output and
9220 (const_unop): Likewise.
9221 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9222 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9223 in cases that did the opposite.
9225 2017-09-14 Richard Biener <rguenther@suse.de>
9227 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9230 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9232 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9234 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9237 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9238 if and where to split a bb, except for splitting before debug insn
9239 sequences followed by non-label real insn. Delete debug insns
9240 in between basic blocks.
9242 * combine.c (make_compound_operation_int): Formatting fixes.
9244 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9245 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9246 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9247 * config/sol2.h (LINK_EH_SPEC): Likewise.
9248 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9249 * config/s390/linux.h (LINK_SPEC): Likewise.
9250 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9251 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9252 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9253 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9254 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9255 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9256 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9257 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9259 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9261 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9263 (ENDFILE_LINUX_SPEC): Likewise.
9264 (LINK_EH_SPEC): Likewise.
9265 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9266 (LINK_OS_LINUX_SPEC32): Likewise.
9267 (LINK_OS_LINUX_SPEC64): Likewise.
9268 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9269 (LINK_OS_LINUX_SPEC): Likewise.
9271 2017-09-13 Martin Liska <mliska@suse.cz>
9274 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9275 CASE_HIGH is NULL_TREE.
9277 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9278 Alan Hayward <alan.hayward@arm.com>
9279 David Sherwood <david.sherwood@arm.com>
9281 * target.def (secondary_memory_needed): New hook.
9282 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9283 instead of SECONDARY_MEMORY_NEEDED.
9284 (secondary_memory_needed_mode): Likewise.
9285 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9286 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9287 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9288 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9289 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9290 * doc/tm.texi: Regenerate.
9291 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9292 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9293 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9294 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9295 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9296 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9297 mode argument first and change the reg_class arguments to reg_class_t.
9298 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9299 Make static. Update the call to inline_secondary_memory_needed.
9300 (ix86_register_move_cost): Update the call to
9301 inline_secondary_memory_needed.
9302 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9303 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9305 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9306 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9308 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9309 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9310 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9311 and match hook interface. Add comment from mips.h.
9312 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9313 * config/mmix/mmix.md (truncdfsf2): Refer to
9314 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9316 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9317 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9318 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9319 (pa_secondary_memory_needed): New function.
9320 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9321 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9322 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9323 (pdp11_secondary_memory_needed): Make static and match hook interface.
9324 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9325 * config/powerpcspe/powerpcspe-protos.h
9326 (rs6000_secondary_memory_needed_ptr): Delete.
9327 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9329 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9330 (rs6000_option_override_internal): Assign to
9331 targetm.secondary_memory_needed rather than
9332 rs6000_secondary_memory_needed_ptr.
9333 (rs6000_secondary_memory_needed): Match hook interface.
9334 (rs6000_debug_secondary_memory_needed): Likewise.
9335 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9336 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9337 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9338 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9339 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9340 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9342 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9343 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9344 (rs6000_option_override_internal): Assign to
9345 targetm.secondary_memory_needed rather than
9346 rs6000_secondary_memory_needed_ptr.
9347 (rs6000_secondary_memory_needed): Match hook interface.
9348 (rs6000_debug_secondary_memory_needed): Likewise.
9349 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9350 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9351 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9352 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9353 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9354 (sparc_secondary_memory_needed): New function.
9355 * lra-constraints.c (check_and_process_move): Refer to
9356 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9358 (curr_insn_transform): Likewise.
9359 (process_alt_operands): Use targetm.secondary_memory_needed
9360 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9361 (check_secondary_memory_needed_p): Likewise.
9362 (choose_split_class): Likewise.
9363 * reload.c: Unconditionally include code that was previously
9364 conditional on SECONDARY_MEMORY_NEEDED.
9365 (push_secondary_reload): Use targetm.secondary_memory_needed
9366 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9367 (push_reload): Likewise.
9368 * reload1.c: Unconditionally include code that was previously
9369 conditional on SECONDARY_MEMORY_NEEDED.
9370 (choose_reload_regs): Use targetm.secondary_memory_needed
9371 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9372 (gen_reload): Likewise.
9373 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9375 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9376 Alan Hayward <alan.hayward@arm.com>
9377 David Sherwood <david.sherwood@arm.com>
9379 * target.def (secondary_memory_needed_mode): New hook:
9380 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9381 * targhooks.h (default_secondary_memory_needed_mode): New function.
9382 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9383 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9384 * doc/tm.texi: Regenerate.
9385 * lra-constraints.c (check_and_process_move): Use
9386 targetm.secondary_memory_needed_mode instead of
9387 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9388 (curr_insn_transform): Likewise.
9389 * reload.c (get_secondary_mem): Likewise.
9390 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9391 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9393 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9394 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9395 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9396 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9397 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9399 * config/powerpcspe/powerpcspe-protos.h
9400 (rs6000_secondary_memory_needed_mode): Delete.
9401 * config/powerpcspe/powerpcspe.c
9402 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9403 (rs6000_secondary_memory_needed_mode): Make static.
9404 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9405 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9407 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9409 (rs6000_secondary_memory_needed_mode): Make static.
9410 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9411 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9412 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9413 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9414 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9416 (sparc_secondary_memory_needed_mode): New function.
9417 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9419 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9421 * config/aarch64/constraints.md (Umq): New constraint.
9422 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9424 (mov<mode>): Update condition.
9426 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9428 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9429 when bitposition is the same.
9431 2017-09-13 Richard Biener <rguenther@suse.de>
9433 * dwarf2out.c (output_die_symbol): Remove.
9434 (output_die): Do not output a DIEs symbol.
9436 2017-09-13 Richard Biener <rguenther@suse.de>
9439 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9440 default-def to avoid breaking iterator update with the weird
9441 interaction with cgraph_update_edges_for_call_stmt_node.
9443 2017-09-13 Richard Biener <rguenther@suse.de>
9445 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9446 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9447 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9448 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9450 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9452 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9453 Disable pc relative literal load irrespective of
9454 TARGET_FIX_ERR_A53_84341 for default.
9456 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9458 * config/sparc/sparc.c (output_return): Output the source location of
9459 the insn in the delay slot, if any.
9460 (output_sibcall): Likewise.
9462 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9465 * common.opt (-static-pie): New alias.
9466 (shared): Negate static-pie.
9467 (-no-pie): Update help text.
9469 (static-pie): New option.
9470 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9471 -static-pie support.
9472 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9473 (LINK_EH_SPEC): Likewise.
9474 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9475 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9476 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9477 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9478 (init_gcc_specs): Likewise.
9479 (init_spec): Likewise.
9480 (display_help): Update help message for -pie.
9481 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9484 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9486 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9487 (movdi_aarch64): Likewise.
9488 (movti_aarch64): Likewise.
9490 2017-09-12 Simon Wright <simon@pushface.org>
9493 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9494 calculation of the minor version, always output as 0.
9496 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9499 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9500 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9501 on it early, rather than manual conversion late. For
9502 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9503 instead of performing manual conversion.
9505 2017-09-12 Carl Love <cel@us.ibm.com>
9507 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9508 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9510 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9511 VMULOSW): Add definitions.
9512 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9513 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9514 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9515 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9516 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9518 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9520 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9522 (movti_aarch64): Likewise.
9523 (movdf_aarch64): Likewise.
9524 (movtf_aarch64): Likewise.
9525 (load_pairdi): Likewise.
9526 (store_pairdi): Likewise.
9527 (load_pairdf): Likewise.
9528 (store_pairdf): Likewise.
9529 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9530 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9531 (ldr_got_small_<mode>): Likewise.
9532 (ldr_got_small_28k_<mode>): Likewise.
9533 (ldr_got_tiny): Likewise.
9534 * config/aarch64/iterators.md (ldst_sz): New.
9535 (ldpstp_sz): Likewise.
9536 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9538 (thunderx_load): Split load_8 to load_16.
9539 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9541 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9542 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9543 (xgene1_store_pair): Split store_8 to store_16.
9544 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9545 (falkor_st_0_st_sd): Split store_8 to store_16.
9547 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9549 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9550 and store1/2/3/4 to store_4/8/12/16.
9551 * config/aarch64/aarch64.md: Update for rename.
9552 * config/arm/arm.md: Likewise.: Likewise.
9553 * config/arm/arm.c: Likewise.
9554 * config/arm/thumb1.md: Likewise.
9555 * config/arm/thumb2.md: Likewise.
9556 * config/arm/vfp.md: Likewise.
9557 * config/arm/arm-generic.md: Likewise.
9558 * config/arm/arm1020e.md: Likewise.
9559 * config/arm/arm1026ejs.md: Likewise.
9560 * config/arm/arm1136jfs.md: Likewise.
9561 * config/arm/arm926ejs.md: Likewise.
9562 * config/arm/cortex-a15.md: Likewise.
9563 * config/arm/cortex-a17.md: Likewise.
9564 * config/arm/cortex-a5.md: Likewise.
9565 * config/arm/cortex-a53.md: Likewise.
9566 * config/arm/cortex-a57.md: Likewise.
9567 * config/arm/cortex-a7.md: Likewise.
9568 * config/arm/cortex-a8.md: Likewise.
9569 * config/arm/cortex-a9.md: Likewise.
9570 * config/arm/cortex-m4.md: Likewise.
9571 * config/arm/cortex-m7.md: Likewise.
9572 * config/arm/cortex-r4.md: Likewise.
9573 * config/arm/exynos-m1.md: Likewise.
9574 * config/arm/fa526.md: Likewise.
9575 * config/arm/fa606te.md: Likewise.
9576 * config/arm/fa626te.md: Likewise.
9577 * config/arm/fa726te.md: Likewise.
9578 * config/arm/fmp626.md: Likewise.
9579 * config/arm/iwmmxt.md: Likewise.
9580 * config/arm/ldmstm.md: Likewise.
9581 * config/arm/marvell-pj4.md: Likewise.
9582 * config/arm/xgene1.md: Likewise.
9583 * config/aarch64/thunderx.md: Likewise.
9584 * config/aarch64/thunderx2t99.md: Likewise.
9585 * config/aarch64/falkor.md: Likewise.
9587 2017-09-12 Martin Liska <mliska@suse.cz>
9589 * attribs.c (private_lookup_attribute): New function.
9590 * attribs.h (private_lookup_attribute): Declared here.
9591 (lookup_attribute): Called from this place.
9593 2017-09-12 Richard Biener <rguenther@suse.de>
9595 PR tree-optimization/82157
9596 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9597 stmts with side-effects.
9599 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9600 Alan Hayward <alan.hayward@arm.com>
9601 David Sherwood <david.sherwood@arm.com>
9603 * target.def (hard_regno_nregs): New hook.
9604 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9605 * targhooks.h (default_hard_regno_nregs): Declare.
9606 * targhooks.c (default_hard_regno_nregs): New function.
9607 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9608 (TARGET_HARD_REGNO_NREGS): ...this hook.
9609 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9610 (CLASS_MAX_NREGS): Likewise.
9611 * doc/tm.texi: Regenerate.
9612 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9613 instead of HARD_REGNO_NREGS.
9614 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9615 HARD_REGNO_NREGS in the comment.
9616 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9617 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9618 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9619 Return an unsigned int.
9620 (TARGET_HARD_REGNO_NREGS): Redefine.
9621 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9622 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9623 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9624 (arc_hard_regno_nregs): New function.
9625 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9626 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9627 (arm_hard_regno_nregs): New function.
9628 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9629 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9630 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9631 (TARGET_HARD_REGNO_NREGS): Redefine.
9632 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9633 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9634 (HARD_REGNO_NREGS): Delete.
9635 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9636 (cr16_hard_regno_nregs): New function.
9637 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9638 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9639 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9640 (cris_hard_regno_nregs): New function.
9641 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9642 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9643 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9644 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9645 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9646 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9647 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9648 (frv_hard_regno_nregs): Make static. Take and return an
9650 (frv_class_max_nregs): Remove outdated copy of documentation.
9651 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9652 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9653 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9654 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9655 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9656 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9657 (TARGET_HARD_REGNO_NREGS): Redefine.
9658 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9659 (CLASS_MAX_NREGS): Update comment.
9660 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9661 (ia64_hard_regno_nregs): New function.
9662 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9663 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9664 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9665 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9666 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9668 (m32c_hard_regno_nregs): Likewise. Make static.
9669 (TARGET_HARD_REGNO_NREGS): Redefine.
9670 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9671 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9672 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9673 (m68k_hard_regno_nregs): New function.
9674 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9675 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9676 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9677 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9678 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9679 Take and return an unsigned int.
9680 (TARGET_HARD_REGNO_NREGS): Redefine.
9681 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9682 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9683 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9684 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9685 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9686 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9687 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9688 (msp430_hard_regno_nregs): Make static. Take and return an
9690 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9691 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9692 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9693 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9694 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9695 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9696 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9697 (TARGET_HARD_REGNO_NREGS): Redefine.
9698 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9699 (PA_HARD_REGNO_NREGS): ...this.
9700 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9701 (PA_HARD_REGNO_NREGS): ...this.
9702 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9703 (pa_hard_regno_nregs): New function.
9704 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9705 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9706 (pdp11_hard_regno_nregs): New function.
9707 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9708 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9709 (rs6000_hard_regno_nregs_hook): New function.
9710 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9711 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9712 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9713 Take and return an unsigned int. Move earlier in file.
9714 (TARGET_HARD_REGNO_NREGS): Redefine.
9715 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9716 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9717 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9718 (rl78_hard_regno_nregs): Make static. Take and return an
9720 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9721 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9722 (rs6000_hard_regno_nregs_hook): New function.
9723 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9724 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9725 (TARGET_HARD_REGNO_NREGS): Redefine.
9726 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9727 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9728 instead of HARD_REGNO_NREGS.
9729 (s390_hard_regno_nregs): New function.
9730 (s390_hard_regno_mode_ok): Add comment from s390.h.
9731 (TARGET_HARD_REGNO_NREGS): Redefine.
9732 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9733 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9734 (sh_hard_regno_nregs): New function.
9735 (sh_pass_in_reg_p): Use it.
9736 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9737 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9738 (sparc_hard_regno_nregs): New function.
9739 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9740 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9741 (spu_function_arg_advance): Use it, supplying a valid register number.
9742 (TARGET_HARD_REGNO_NREGS): Redefine.
9743 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9744 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9745 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9746 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9747 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9748 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9749 (CLASS_MAX_NREGS): Remove copy of old documentation.
9750 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9751 (visium_hard_regno_nregs): New function.
9752 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9753 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9754 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9755 xtensa_hard_regno_nregs): New function.
9756 * system.h (HARD_REGNO_NREGS): Poison.
9758 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9760 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9761 hard_regno_nregs instead of HARD_REGNO_NREGS.
9762 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9763 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9764 (c6x_expand_epilogue): Likewise.
9765 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9766 (frv_read_iacc_argument): Likewise.
9767 * config/sh/sh.c: Include regs.h.
9768 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9769 (regs_used): Likewise.
9770 (output_stack_adjust): Likewise.
9771 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9772 * expmed.c: Include regs.h.
9773 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9774 * ree.c: Include regs.h.
9775 (combine_reaching_defs): Use hard_regno_nregs instead of
9777 (add_removable_extension): Likewise.
9779 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9781 * regs.h (hard_regno_nregs): Turn into a function.
9782 (end_hard_regno): Update accordingly.
9783 * caller-save.c (setup_save_areas): Likewise.
9784 (save_call_clobbered_regs): Likewise.
9785 (replace_reg_with_saved_mem): Likewise.
9786 (insert_restore): Likewise.
9787 (insert_save): Likewise.
9788 * combine.c (can_change_dest_mode): Likewise.
9789 (move_deaths): Likewise.
9790 (distribute_notes): Likewise.
9791 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9792 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9793 (rs6000_split_multireg_move): Likewise.
9794 (rs6000_register_move_cost): Likewise.
9795 (rs6000_memory_move_cost): Likewise.
9796 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9797 (rs6000_split_multireg_move): Likewise.
9798 (rs6000_register_move_cost): Likewise.
9799 (rs6000_memory_move_cost): Likewise.
9800 * cselib.c (cselib_reset_table): Likewise.
9801 (cselib_lookup_1): Likewise.
9802 * emit-rtl.c (set_mode_and_regno): Likewise.
9803 * function.c (aggregate_value_p): Likewise.
9804 * ira-color.c (setup_profitable_hard_regs): Likewise.
9805 (check_hard_reg_p): Likewise.
9806 (calculate_saved_nregs): Likewise.
9807 (assign_hard_reg): Likewise.
9808 (improve_allocation): Likewise.
9809 (calculate_spill_cost): Likewise.
9810 * ira-emit.c (modify_move_list): Likewise.
9811 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9812 (ira_hard_reg_in_set_p): Likewise.
9813 * ira.c (setup_reg_mode_hard_regset): Likewise.
9814 (clarify_prohibited_class_mode_regs): Likewise.
9815 (check_allocation): Likewise.
9816 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9817 (lra_setup_reg_renumber): Likewise.
9818 (setup_try_hard_regno_pseudos): Likewise.
9819 (spill_for): Likewise.
9820 (assign_hard_regno): Likewise.
9821 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9822 * lra-constraints.c (in_class_p): Likewise.
9823 (lra_constraint_offset): Likewise.
9824 (simplify_operand_subreg): Likewise.
9825 (lra_constraints): Likewise.
9826 (split_reg): Likewise.
9827 (split_if_necessary): Likewise.
9828 (invariant_p): Likewise.
9829 (inherit_in_ebb): Likewise.
9830 * lra-lives.c (process_bb_lives): Likewise.
9831 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9832 (get_hard_regs): Likewise.
9833 (do_remat): Likewise.
9834 * lra-spills.c (assign_spill_hard_regs): Likewise.
9835 * mode-switching.c (create_pre_exit): Likewise.
9836 * postreload.c (reload_combine_recognize_pattern): Likewise.
9837 * recog.c (peep2_find_free_register): Likewise.
9838 * regcprop.c (kill_value_regno): Likewise.
9839 (set_value_regno): Likewise.
9840 (copy_value): Likewise.
9841 (maybe_mode_change): Likewise.
9842 (find_oldest_value_reg): Likewise.
9843 (copyprop_hardreg_forward_1): Likewise.
9844 * regrename.c (check_new_reg_p): Likewise.
9845 (regrename_do_replace): Likewise.
9846 * reload.c (push_reload): Likewise.
9847 (combine_reloads): Likewise.
9848 (find_dummy_reload): Likewise.
9849 (operands_match_p): Likewise.
9850 (find_reloads): Likewise.
9851 (find_equiv_reg): Likewise.
9852 (reload_adjust_reg_for_mode): Likewise.
9853 * reload1.c (count_pseudo): Likewise.
9854 (count_spilled_pseudo): Likewise.
9855 (find_reg): Likewise.
9856 (clear_reload_reg_in_use): Likewise.
9857 (free_for_value_p): Likewise.
9858 (allocate_reload_reg): Likewise.
9859 (choose_reload_regs): Likewise.
9860 (reload_adjust_reg_for_temp): Likewise.
9861 (emit_reload_insns): Likewise.
9862 (delete_output_reload): Likewise.
9863 * rtlanal.c (subreg_get_info): Likewise.
9864 * sched-deps.c (sched_analyze_reg): Likewise.
9865 * sel-sched.c (init_regs_for_mode): Likewise.
9866 (mark_unavailable_hard_regs): Likewise.
9867 (choose_best_reg_1): Likewise.
9868 (verify_target_availability): Likewise.
9869 * valtrack.c (dead_debug_insert_temp): Likewise.
9870 * var-tracking.c (track_loc_p): Likewise.
9871 (emit_note_insn_var_location): Likewise.
9872 * varasm.c (make_decl_rtl): Likewise.
9873 * reginfo.c (choose_hard_reg_mode): Likewise.
9874 (init_reg_modes_target): Refer directly to
9875 this_target_regs->x_hard_regno_nregs.
9877 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9879 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9880 instead of hard_regno_nregs.
9882 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9884 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9885 end_hard_regno instead of hard_regno_nregs.
9886 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9887 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9888 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9889 * ira-color.c (improve_allocation): Likewise.
9890 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9891 * lra-lives.c (mark_regno_live): Likewise.
9892 (mark_regno_dead): Likewise.
9893 * lra-remat.c (operand_to_remat): Likewise.
9894 * lra.c (collect_non_operand_hard_regs): Likewise.
9895 * postreload.c (reload_combine_note_store): Likewise.
9896 (move2add_valid_value_p): Likewise.
9897 * reload.c (regno_clobbered_p): Likewise.
9899 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9901 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9903 * config/v850/v850.c (v850_reorg): Likewise.
9904 * reload.c (refers_to_regno_for_reload_p): Likewise.
9905 (find_equiv_reg): Likewise.
9906 * reload1.c (reload_reg_reaches_end_p): Likewise.
9908 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9910 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9912 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9913 * config/arm/arm.c (output_move_neon): Likewise.
9914 (arm_attr_length_move_neon): Likewise.
9915 (neon_split_vcombine): Likewise.
9916 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9917 (c6x_mark_reg_written): Likewise.
9918 (c6x_dwarf_register_span): Likewise.
9919 * config/i386/i386.c (ix86_save_reg): Likewise.
9920 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9921 (rws_access_reg): Likewise.
9922 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9923 * mode-switching.c (create_pre_exit): Likewise.
9924 * ree.c (combine_reaching_defs): Likewise.
9925 (add_removable_extension): Likewise.
9926 * regcprop.c (find_oldest_value_reg): Likewise.
9927 (copyprop_hardreg_forward_1): Likewise.
9928 * reload.c (reload_inner_reg_of_subreg): Likewise.
9929 (push_reload): Likewise.
9930 (combine_reloads): Likewise.
9931 (find_dummy_reload): Likewise.
9932 (reload_adjust_reg_for_mode): Likewise.
9933 * reload1.c (find_reload_regs): Likewise.
9934 (forget_old_reloads_1): Likewise.
9935 (reload_reg_free_for_value_p): Likewise.
9936 (reload_adjust_reg_for_temp): Likewise.
9937 (emit_reload_insns): Likewise.
9938 (delete_output_reload): Likewise.
9939 * sel-sched.c (choose_best_reg_1): Likewise.
9940 (choose_best_pseudo_reg): Likewise.
9942 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9943 Alan Hayward <alan.hayward@arm.com>
9944 David Sherwood <david.sherwood@arm.com>
9946 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9947 * target.def (slow_unaligned_access): New hook.
9948 * targhooks.h (default_slow_unaligned_access): Declare.
9949 * targhooks.c (default_slow_unaligned_access): New function.
9950 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9951 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9952 * doc/tm.texi: Regenerate.
9953 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9954 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9955 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9957 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9958 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9960 (rs6000_slow_unaligned_access): New function.
9961 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9962 (expand_block_compare): Likewise.
9963 (expand_strn_compare): Likewise.
9964 (rs6000_rtx_costs): Likewise.
9965 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9966 (riscv_slow_unaligned_access): Likewise.
9967 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9968 (riscv_slow_unaligned_access_p): ...this and make static.
9969 (riscv_option_override): Update accordingly.
9970 (riscv_slow_unaligned_access): New function.
9971 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9972 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9973 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9974 (rs6000_slow_unaligned_access): New function.
9975 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9976 (rs6000_rtx_costs): Likewise.
9977 * config/rs6000/rs6000-string.c (expand_block_compare)
9978 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9979 of SLOW_UNALIGNED_ACCESS.
9980 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9981 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9982 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9983 of SLOW_UNALIGNED_ACCESS.
9984 * expmed.c (simple_mem_bitfield_p): Likewise.
9985 * expr.c (alignment_for_piecewise_move): Likewise.
9986 (emit_group_load_1): Likewise.
9987 (emit_group_store): Likewise.
9988 (copy_blkmode_from_reg): Likewise.
9989 (emit_push_insn): Likewise.
9990 (expand_assignment): Likewise.
9991 (store_field): Likewise.
9992 (expand_expr_real_1): Likewise.
9993 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9994 * lra-constraints.c (simplify_operand_subreg): Likewise.
9995 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9996 * gimple-ssa-store-merging.c: Likewise in block comment at start
9998 * tree-ssa-strlen.c: Include target.h.
9999 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
10000 of SLOW_UNALIGNED_ACCESS.
10001 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
10003 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10005 PR rtl-optimization/82185
10006 * expmed.c (emit_store_flag_int): Only test tem if it has been
10009 2017-09-12 Richard Biener <rguenther@suse.de>
10011 PR middle-end/82149
10012 * match.pd ((FTYPE) N CMP CST): Fix typo.
10014 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10016 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10018 (mips_near_type_p): Add 'short_call' attribute as a synonym
10020 * doc/extend.texi (short_call): Document new function attribute.
10022 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10025 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10026 assertion check that in the condition.
10027 (get_atomic_generic_size): Likewise. Before testing if parameter
10028 has pointer type, if it has array type, call for C++
10029 default_conversion to perform array-to-pointer conversion.
10031 2017-09-12 Richard Biener <rguenther@suse.de>
10033 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10034 for operations we cannot scalarize.
10036 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10038 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10039 vectors heap vectors. Clean up comments.
10040 Make visited_bbs a reference.
10041 (profitable_jump_thread_path): Make GC
10042 vectors heap vectors. Clean up comments.
10044 (convert_and_register_jump_thread_path): Make GC vectors heap
10046 (check_subpath_and_update_thread_path): Same. Clean up comments.
10047 Make visited_bbs a reference.
10048 (handle_phi): Abstract common code to to
10049 register_jump_thread_path_if_profitable.
10050 Rename VAR_BB to DEF_BB.
10052 Make GC vectors heap vectors.
10053 Make visited_bbs a reference.
10054 (handle_assignment): Same.
10055 (register_jump_thread_path_if_profitable): New.
10056 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10058 Make GC vectors heap vectors. Clean up comments.
10059 Make visited_bbs a reference.
10060 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10061 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10064 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10067 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10068 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10070 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10072 Revert r251800 and r251799.
10074 2017-09-11 Martin Jambor <mjambor@suse.cz>
10077 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10078 arguments in advance.
10079 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10080 use it to find predecessor edges.
10081 (naive_outof_ssa): Collect vector of predecessors.
10083 2017-09-08 Jason Merrill <jason@redhat.com>
10085 PR c++/70029 - ICE with ref-qualifier and -flto
10086 * langhooks.h (struct lang_hooks_for_types): Add
10087 copy_lang_qualifiers.
10088 * attribs.c (build_type_attribute_qual_variant): Use it.
10089 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10091 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10092 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10094 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10097 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10098 (*mulsi3_sp64): New instruction.
10099 (mulsi3): New expander.
10101 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10103 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10105 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10107 * sancov.c: Include memmodel.h.
10109 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10112 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10115 2017-09-07 Carl Love <cel@us.ibm.com>
10117 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10118 the sldi instruction.
10120 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10122 * sancov.c: Include tm_p.h.
10124 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10127 * output.h (switch_to_other_text_partition): New declaration.
10128 * varasm.c (switch_to_other_text_partition): New function.
10129 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10130 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10131 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10132 to the other text partition before emitting LCL label and switch back
10133 after emitting the word after it.
10135 2017-09-07 Richard Biener <rguenther@suse.de>
10137 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10138 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10139 critical edges here, after loop init.
10140 (pass_data_pre): Remove PROP_no_crit_edges flags.
10141 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10142 for valueization of call args to avoid leaking VN_TOP.
10143 (visit_use): Assert we do not visit default defs.
10144 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10145 Use error_mark_node to more easily detect leaking VN_TOP.
10146 All default-defs are varying, not VN_TOP. Mark them visited.
10147 (run_scc_vn): Make code match comment.
10149 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10151 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10152 OPTION_MASK_FLOAT128_KEYWORD.
10153 (POWERPC_MASKS): Likewise.
10154 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10155 support for the -mfloat128-type option, and make -mfloat128
10156 default on PowerPC Linux systems. Define or undefine
10157 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10158 Define __float128 to be __ieee128 if IEEE 128-bit support is
10159 enabled, or undefine it.
10160 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10161 Delete defining __FLOAT128_TYPE__.
10162 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10163 -mfloat128-type option and make -mfloat128 default on PowerPC
10165 (TARGET_FLOAT128_TYPE): Likewise.
10166 (-mfloat128-type): Likewise.
10167 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10168 Delete the -mfloat128-type option and make -mfloat128 default on
10169 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10170 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10171 128-bit floating point is enabled. Change tests from using
10172 -mfloat128-type to -mfloat128.
10173 (rs6000_mangle_type): Use the correct mangling for the __float128
10174 type even if normal long double is restricted to 64-bits.
10175 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10177 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10178 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10179 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10181 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10182 documentation for -mfloat128.
10184 2017-09-06 Olivier Hainque <hainque@adacore.com>
10186 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10188 2017-09-06 Wish Wu <wishwu007@gmail.com>
10189 Jakub Jelinek <jakub@redhat.com>
10191 * asan.c (initialize_sanitizer_builtins): Add
10192 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10193 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10194 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10195 BT_FN_VOID_UINT64_PTR variables.
10196 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10197 (BT_FN_VOID_UINT16_UINT16): Likewise.
10198 (BT_FN_VOID_UINT32_UINT32): Likewise.
10199 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10200 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10201 (BT_FN_VOID_UINT64_PTR): Likewise.
10202 * common.opt (flag_sanitize_coverage): New variable.
10203 (fsanitize-coverage=trace-pc): Remove.
10204 (fsanitize-coverage=): Add.
10205 * flag-types.h (enum sanitize_coverage_code): New enum.
10206 * fold-const.c (fold_range_test): Disable non-short-circuit
10207 optimization if flag_sanitize_coverage.
10208 (fold_truth_andor): Likewise.
10209 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10210 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10211 (coverage_sanitizer_opts): New array.
10212 (get_closest_sanitizer_option): Add OPTS argument, handle also
10213 OPT_fsanitize_coverage_.
10214 (parse_sanitizer_options): Adjusted to also handle
10215 OPT_fsanitize_coverage_.
10216 (common_handle_option): Add OPT_fsanitize_coverage_.
10217 * sancov.c (instrument_comparison, instrument_switch): New function.
10218 (sancov_pass): Add trace-cmp support.
10219 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10220 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10221 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10222 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10223 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10224 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10225 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10226 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10227 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10228 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10230 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10232 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10233 error. Only quit immediately if parsing is complete.
10234 (BEGIN): Initialize fatal_err and parse_done.
10235 (begin fpu, end fpu): Check number of arguments.
10236 (begin arch, end arch): Likewise.
10237 (begin cpu, end cpu): Likewise.
10238 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10239 (optalias): Likewise.
10241 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10243 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10244 * config/arm/arm-isa.h: Delete. Move definitions to ...
10245 * arm-cpus.in: ... here. Use new feature and fgroup values.
10246 * config/arm/arm.c (arm_option_override): Use lower case for feature
10248 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10249 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10250 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10252 (print_isa_bits_for): New function.
10253 (gen_isa): New function.
10254 (gen_comm_data): Use print_isa_bits_for.
10255 (define feature): New keyword.
10256 (define fgroup): New keyword.
10257 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10258 (arm-isa.h): Add rule to generate file.
10259 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10260 case for feature bit names.
10262 2017-09-06 Richard Biener <rguenther@suse.de>
10264 * tree-ssa-pre.c (NECESSARY): Remove.
10265 (create_expression_by_pieces): Do not touch pass-local flags.
10266 (insert_into_preds_of_block): Likewise.
10267 (do_pre_regular_insertion): Likewise.
10268 (eliminate_insert): Likewise.
10269 (eliminate_dom_walker::before_dom_children): Likewise.
10270 (fini_eliminate): Do not look at inserted_exprs.
10271 (mark_operand_necessary): Remove.
10272 (remove_dead_inserted_code): Replace with simple work-list
10273 algorithm based on inserted_exprs and SSA uses.
10274 (pass_pre::execute): Re-order fini_eliminate and
10275 remove_dead_inserted_code.
10277 2017-09-06 Olivier Hainque <hainque@adacore.com>
10279 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10280 for VxWorks 7. Adjust surrounding comments.
10282 2017-09-06 Richard Biener <rguenther@suse.de>
10284 * gimple-ssa-strength-reduction.c
10285 (find_candidates_dom_walker::before_dom_children): Also allow
10288 2017-09-06 Richard Biener <rguenther@suse.de>
10290 PR tree-optimization/82108
10291 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10292 for gap in the non-permutation SLP case.
10294 2017-09-06 Martin Jambor <mjambor@suse.cz>
10296 PR tree-optimization/82078
10297 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10298 add_access_to_work_queue...
10299 (build_accesses_from_assign): ...here.
10300 (propagate_all_subaccesses): Make sure racc is the group
10301 representative, if there is one.
10303 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10305 PR middle-end/82095
10306 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10309 2017-09-06 Richard Biener <rguenther@suse.de>
10311 * gimple-ssa-strength-reduction.c
10312 (find_candidates_dom_walker::before_doom_children): Use a
10313 type and not a mode check.
10315 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10318 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10320 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10322 2017-09-05 Jeff Law <law@redhat.com>
10324 PR tree-optimization/64910
10325 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10326 cases where we have 3 or more operands.
10328 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10330 PR middle-end/81768
10331 * omp-low.c (lower_omp_for): Recompute tree invariant if
10332 gimple_omp_for_initial/final is ADDR_EXPR.
10334 PR middle-end/81768
10335 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10336 into gimple val before gimplification fo the COND_EXPR.
10338 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10340 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10341 REGION_COPY argument.
10342 (thread_through_all_blocks): Remove unused argument to
10343 duplicate_thread_path.
10345 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10346 Alan Hayward <alan.hayward@arm.com>
10347 David Sherwood <david.sherwood@arm.com>
10349 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10350 Take a scalar_mode rather than a machine_mode.
10351 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10352 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10353 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10354 (aarch64_gen_adjusted_ldpstp): Likewise.
10355 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10357 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10358 Alan Hayward <alan.hayward@arm.com>
10359 David Sherwood <david.sherwood@arm.com>
10361 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10362 Take a scalar_int_mode instead of a machine_mode.
10363 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10364 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10365 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10366 (aarch64_simd_attr_length_rglist): Delete.
10367 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10368 a scalar_int_mode instead of a machine_mode.
10369 (aarch64_add_offset): Likewise.
10370 (aarch64_internal_mov_immediate): Likewise
10371 (aarch64_add_constant_internal): Likewise.
10372 (aarch64_add_constant): Likewise.
10373 (aarch64_movw_imm): Likewise.
10374 (aarch64_rtx_arith_op_extract_p): Likewise.
10375 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10376 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10377 Remove assert that the mode isn't a vector.
10378 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10379 (aarch64_expand_mov_immediate): Update calls after above changes.
10380 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10381 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10382 (aarch64_move_imm): Likewise.
10383 (aarch64_can_const_movi_rtx_p): Likewise.
10384 (aarch64_strip_extend): Likewise.
10385 (aarch64_extr_rtx_p): Likewise.
10386 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10387 a CONST_INT when the mode parameter is VOIDmode.
10388 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10390 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10392 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10393 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10394 (bitwise_type_for_mode): Update accordingly.
10396 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10398 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10399 * stor-layout.c (mode_for_size_tree): Likewise.
10400 (mode_for_array): Update accordingly.
10401 (layout_decl): Likewise.
10402 (compute_record_mode): Likewise. Only set the mode once.
10404 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10406 * target.def (get_mask_mode): Change return type to opt_mode.
10408 * doc/tm.texi: Regenerate.
10409 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10410 * targhooks.c (default_get_mask_mode): Likewise.
10411 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10412 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10413 targetm.get_mask_mode.
10414 * tree.c (build_truth_vector_type): Likewise.
10416 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10418 * machmode.h (mode_for_vector): Return an opt_mode.
10419 * stor-layout.c (mode_for_vector): Likewise.
10420 (mode_for_int_vector): Update accordingly.
10421 (layout_type): Likewise.
10422 * config/i386/i386.c (emit_memmov): Likewise.
10423 (ix86_expand_set_or_movmem): Likewise.
10424 (ix86_expand_vector_init): Likewise.
10425 (ix86_get_mask_mode): Likewise.
10426 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10428 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10429 * expmed.c (extract_bit_field_1): Likewise.
10430 * expr.c (expand_expr_real_2): Likewise.
10431 * optabs-query.c (can_vec_perm_p): Likewise.
10432 (can_vec_mask_load_store_p): Likewise.
10433 * optabs.c (expand_vec_perm): Likewise.
10434 * targhooks.c (default_get_mask_mode): Likewise.
10435 * tree-vect-stmts.c (vectorizable_store): Likewise.
10436 (vectorizable_load): Likewise.
10437 (get_vectype_for_scalar_type_and_size): Likewise.
10439 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10441 * machmode.h (mode_for_int_vector): New function.
10442 * stor-layout.c (mode_for_int_vector): Likewise.
10443 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10444 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10445 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10446 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10447 (s390_expand_vcond): Likewise.
10449 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10451 * machmode.h (opt_machine_mode): New type.
10452 (opt_mode<T>): Allow construction from anything that can be
10454 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10455 (mode_for_size): Return an opt_machine_mode.
10456 * stor-layout.c (mode_for_size): Likewise.
10457 (mode_for_size_tree): Update call accordingly.
10458 (bitwise_mode_for_mode): Likewise.
10459 (make_fract_type): Likewise.
10460 (make_accum_type): Likewise.
10461 * caller-save.c (replace_reg_with_saved_mem): Update call
10463 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10464 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10465 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10466 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10467 * expmed.c (extract_bit_field_1): Likewise.
10468 * reload.c (get_secondary_mem): Likewise.
10469 * varasm.c (assemble_integer): Likewise.
10470 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10473 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10475 * machmode.h (decimal_float_mode_for_size): New function.
10476 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10477 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10478 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10479 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10480 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10481 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10483 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10485 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10486 (get_builtin_sync_mode): Likewise.
10487 (expand_ifn_atomic_compare_exchange): Likewise.
10488 (expand_builtin_atomic_clear): Likewise.
10489 (expand_builtin_atomic_test_and_set): Likewise.
10490 (fold_builtin_atomic_always_lock_free): Likewise.
10491 * calls.c (compute_argument_addresses): Likewise.
10492 (emit_library_call_value_1): Likewise.
10493 (store_one_arg): Likewise.
10494 * combine.c (combine_instructions): Likewise.
10495 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10496 * config/arm/arm.c (arm_function_value): Likewise.
10497 (aapcs_allocate_return_reg): Likewise.
10498 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10499 * config/i386/i386.c (construct_container): Likewise.
10500 (ix86_gimplify_va_arg): Likewise.
10501 (ix86_expand_sse_cmp): Likewise.
10502 (emit_memmov): Likewise.
10503 (emit_memset): Likewise.
10504 (expand_small_movmem_or_setmem): Likewise.
10505 (ix86_expand_pextr): Likewise.
10506 (ix86_expand_pinsr): Likewise.
10507 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10508 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10510 * config/mips/mips.c (mips_function_value_1) Likewise.
10511 (mips_block_move_straight): Likewise.
10512 (mips_expand_ins_as_unaligned_store): Likewise.
10513 * config/powerpcspe/powerpcspe.c
10514 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10515 (rs6000_darwin64_record_arg_flush): Likewise.
10516 * config/rs6000/rs6000.c
10517 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10518 (rs6000_darwin64_record_arg_flush): Likewise.
10519 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10520 (sparc_function_value_1): Likewise.
10521 * config/spu/spu.c (adjust_operand): Likewise.
10522 (spu_emit_branch_or_set): Likewise.
10523 (arith_immediate_p): Likewise.
10524 * emit-rtl.c (gen_lowpart_common): Likewise.
10525 * expr.c (expand_expr_real_1): Likewise.
10526 * function.c (assign_parm_setup_block): Likewise.
10527 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10528 * reload1.c (alter_reg): Likewise.
10529 * stor-layout.c (mode_for_vector): Likewise.
10530 (layout_type): Likewise.
10532 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10534 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10535 (spu_convert_move): Likewise.
10536 * lower-subreg.c (resolve_simple_move): Likewise.
10538 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10541 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10542 define_insn to a define_expand.
10543 (altivec_vsum2sws_direct): New define_insn.
10544 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10546 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10548 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10551 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10556 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10557 realign stack if stack alignment needed is less than incoming
10560 2017-09-05 Marek Polacek <polacek@redhat.com>
10563 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10566 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10568 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10570 2017-09-05 Richard Biener <rguenther@suse.de>
10572 PR tree-optimization/82084
10573 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10575 2017-09-05 Richard Biener <rguenther@suse.de>
10577 PR tree-optimization/82102
10578 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10580 2017-09-05 Martin Liska <mliska@suse.cz>
10582 PR tree-optimization/82032
10583 * tree-cfg.c (generate_range_test): New function.
10584 * tree-cfg.h (generate_range_test): Declared here.
10585 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10586 (cleanup_control_expr_graph): Use it.
10587 * tree-switch-conversion.c (try_switch_expansion): Remove
10589 (emit_case_nodes): Use generate_range_test.
10591 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10594 * config/i386/i386.md (*<btsc><mode>_mask): Add
10595 TARGET_USE_BT to insn constraint.
10596 (*btr<mode>_mask): Ditto.
10598 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10600 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10601 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10603 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10606 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10607 TARGET_NEON and TARGET_IWMMXT.
10608 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10609 TARGET_NEON and TARGET_IWMMXT.
10610 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10612 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10614 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10615 (ix86_rewrite_tls_address): Ditto.
10616 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10617 (ix86_rewrite_tls_address_1): Ditto.
10618 (ix86_rewrite_tls_address): Ditto.
10619 * config/i386/predicates.md (tls_address_pattern): New predicate.
10620 * config/i386/i386.md (TLS address splitter): New splitter.
10622 2017-09-04 Richard Biener <rguenther@suse.de>
10624 PR tree-optimization/82084
10625 * fold-const.h (can_native_encode_string_p): Declare.
10626 * fold-const.c (can_native_encode_string_p): Factor out from ...
10627 (native_encode_string): ... here.
10628 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10629 vectorizing stores from constants we later cannot handle.
10631 2017-09-04 Marek Polacek <polacek@redhat.com>
10634 * doc/invoke.texi: Update -Wtautological-compare documentation.
10636 2017-09-04 Jeff Law <law@redhat.com>
10638 PR tree-optimization/64910
10639 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10640 swap the first and last operand if the last is a constant.
10642 2017-09-04 Marek Polacek <polacek@redhat.com>
10645 * convert.c (do_narrow): When sanitizing signed integer overflows,
10646 bail out for signed types.
10647 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10649 2017-09-04 Richard Biener <rguenther@suse.de>
10651 PR tree-optimization/82060
10652 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10653 Move devirtualization after stmt folding and before EH/AB/noreturn
10654 cleanup to get the stmt refs canonicalized. Use a bool instead
10655 of gimple_modified_p since that doesn't work for NOPs. Schedule
10656 NOPs generated by folding for removal.
10658 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10659 Alan Hayward <alan.hayward@arm.com>
10660 David Sherwood <david.sherwood@arm.com>
10662 * coretypes.h (pad_direction): New enum.
10663 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10664 (FUNCTION_ARG_PADDING): Likewise.
10665 * target.def (function_arg_padding): New hook.
10666 * targhooks.h (default_function_arg_padding): Declare.
10667 * targhooks.c (default_function_arg_padding): New function.
10668 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10669 (TARGET_FUNCTION_ARG_PADDING): ...this.
10670 * doc/tm.texi: Regenerate.
10671 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10672 instead of direction.
10673 (compute_argument_addresses): Likewise.
10674 (load_register_parameters): Likewise.
10675 (emit_library_call_value_1): Likewise.
10676 (store_one_arg): Use targetm.calls.function_arg_padding instead
10677 of FUNCTION_ARG_PADDING.
10678 (must_pass_in_stack_var_size_or_pad): Likewise.
10679 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10680 (emit_group_store): Likewise.
10681 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10682 instead of FUNCTION_ARG_PADDING.
10683 (emit_push_insn): Likewise, and propagate enum change throughout
10685 * function.h (direction): Delete.
10686 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10688 * function.c (assign_parm_find_stack_rtl): Likewise.
10689 (assign_parm_setup_block_p): Likewise.
10690 (assign_parm_setup_block): Likewise.
10691 (gimplify_parameters): Likewise.
10692 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10693 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10695 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10696 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10697 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10698 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10699 (aarch64_function_arg_padding): ...this new function.
10700 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10701 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10702 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10703 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10704 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10705 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10706 (arm_pad_arg_upward): Replace with...
10707 (arm_function_arg_padding): ...this new function.
10708 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10710 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10711 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10712 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10713 (ia64_hpux_function_arg_padding): Replace with...
10714 (ia64_function_arg_padding): ...this new function. Use pad_direction
10715 instead of direction. Check for TARGET_HPUX.
10716 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10717 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10718 (iq2000_function_arg_padding): New function.
10719 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10720 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10721 (mips_function_arg_padding): ...this new function.
10722 (mips_pad_reg_upward): Update accordingly.
10723 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10724 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10725 targetm.calls.function_arg_padding.
10726 (FUNCTION_ARG_PADDING): Delete.
10727 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10728 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10729 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10730 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10731 (nios2_block_reg_padding): Return pad_direction instead of direction.
10732 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10733 instead of direction.
10734 (nios2_function_arg_padding): Likewise. Make static.
10735 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10736 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10737 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10738 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10739 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10740 (pa_function_arg_padding): Make static. Return pad_direction instead
10742 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10743 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10744 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10745 instead of direction. Use targetm.calls.function_arg_padding.
10746 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10747 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10748 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10749 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10750 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10752 (function_arg_padding): Rename to...
10753 (rs6000_function_arg_padding): ...this. Make static. Return
10754 pad_direction instead of direction.
10755 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10756 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10757 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10758 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10759 instead of direction. Use targetm.calls.function_arg_padding.
10760 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10761 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10762 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10763 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10764 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10765 (function_arg_padding): Rename to...
10766 (rs6000_function_arg_padding): ...this. Make static. Return
10767 pad_direction instead of direction.
10768 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10769 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10770 * config/s390/s390.c (s390_function_arg_padding): New function.
10771 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10772 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10773 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10774 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10775 (function_arg_padding): Rename to...
10776 (sparc_function_arg_padding): ...this. Make static. Return
10777 pad_direction instead of direction.
10778 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10779 * config/spu/spu.c (spu_function_arg_padding): New function.
10780 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10781 * system.h (FUNCTION_ARG_PADDING): Poison.
10783 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10784 Alan Hayward <alan.hayward@arm.com>
10785 David Sherwood <david.sherwood@arm.com>
10787 * target.def (modes_tieable_p): New hook.
10788 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10789 (TARGET_MODES_TIEABLE_P): ...this.
10790 * doc/tm.texi.in: Regenerate.
10791 * hooks.h (hook_bool_mode_mode_true): Declare.
10792 * hooks.c (hook_bool_mode_mode_true): New function.
10793 * combine.c (subst): Use targetm.modes_tieable_p instead of
10795 * dse.c (find_shift_sequence): Likewise.
10796 * expmed.c (extract_low_bits): Likewise.
10797 * lower-subreg.c: Include target.h.
10798 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10800 * rtlanal.c (rtx_cost): Likewise.
10801 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10802 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10803 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10804 (TARGET_MODES_TIEABLE_P): Redefine.
10805 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10806 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10807 (TARGET_MODES_TIEABLE_P): Redefine.
10808 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10809 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10810 (arc_modes_tieable_p): New function.
10811 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10812 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10813 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10814 (arm_modes_tieable_p): Make static.
10815 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10816 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10817 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10818 (TARGET_MODES_TIEABLE_P): Redefine.
10819 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10820 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10821 (TARGET_MODES_TIEABLE_P): Redefine.
10822 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10823 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10824 (cr16_modes_tieable_p): New function.
10825 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10826 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10827 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10828 (TRULY_NOOP_TRUNCATION): Update comment.
10829 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10830 (TRULY_NOOP_TRUNCATION): Update comment.
10831 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10832 (frv_modes_tieable_p): New function.
10833 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10834 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10835 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10836 (TARGET_MODES_TIEABLE_P): Redefine.
10837 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10838 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10839 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10840 (TARGET_MODES_TIEABLE_P): Redefine.
10841 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10842 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10843 (ia64_modes_tieable_p): New function.
10844 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10845 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10846 (iq2000_modes_tieable_p): New function.
10847 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10848 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10849 (lm32_modes_tieable_p): New function.
10850 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10851 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10852 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10853 (TARGET_MODES_TIEABLE_P): Redefine.
10854 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10855 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10856 (m32r_modes_tieable_p): New function.
10857 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10858 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10859 (m68k_modes_tieable_p): New function.
10860 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10861 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10862 (mcore_modes_tieable_p): New function.
10863 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10864 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10866 (TARGET_MODES_TIEABLE_P): Redefine.
10867 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10868 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10869 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10870 (TARGET_MODES_TIEABLE_P): Redefine.
10871 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10872 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10873 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10874 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10875 (mn10300_modes_tieable_p): ...this and make static.
10876 (TARGET_MODES_TIEABLE_P): Redefine.
10877 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10878 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10879 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10880 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10881 (msp430_modes_tieable_p): Make static.
10882 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10883 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10884 (TARGET_MODES_TIEABLE_P): Redefine.
10885 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10886 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10887 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10888 (TARGET_MODES_TIEABLE_P): Redefine.
10889 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10890 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10891 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10892 (TARGET_MODES_TIEABLE_P): Redefine.
10893 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10894 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10895 (pdp11_modes_tieable_p): New function.
10896 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10897 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10898 (rs6000_modes_tieable_p): New function.
10899 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10900 * config/powerpcspe/powerpcspe.md: Update comment.
10901 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10902 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10903 (TARGET_MODES_TIEABLE_P): Redefine.
10904 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10905 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10906 (rl78_modes_tieable_p): New function.
10907 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10908 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10909 (rs6000_modes_tieable_p): New function.
10910 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10911 * config/rs6000/rs6000.md: Update comment.
10912 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10913 * config/rx/rx.c (rx_modes_tieable_p): New function.
10914 (TARGET_MODES_TIEABLE_P): Redefine.
10915 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10916 * config/s390/s390.c (s390_modes_tieable_p): New function.
10917 (TARGET_MODES_TIEABLE_P): Redefine.
10918 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10919 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10920 (sh_modes_tieable_p): New function.
10921 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10922 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10923 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10924 (sparc_modes_tieable_p): Make static.
10925 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10926 * config/spu/spu.c (spu_modes_tieable_p): New function.
10927 (TARGET_MODES_TIEABLE_P): Redefine.
10928 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10929 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10930 (TARGET_MODES_TIEABLE_P): Redefine.
10931 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10932 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10933 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10934 * config/v850/v850.c (v850_modes_tieable_p): New function.
10935 (TARGET_MODES_TIEABLE_P): Redefine.
10936 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10937 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10938 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10939 (visium_modes_tieable_p): New function.
10940 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10941 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10942 (xtensa_modes_tieable_p): New function.
10943 * system.h (MODES_TIEABLE_P): Poison.
10945 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10946 Alan Hayward <alan.hayward@arm.com>
10947 David Sherwood <david.sherwood@arm.com>
10949 * target.def (hard_regno_mode_ok): New hook.
10950 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10951 (TARGET_HARD_REGNO_MODE_OK): ...this.
10952 * doc/tm.texi.in: Regenerate.
10953 * hooks.h (hook_bool_uint_mode_true): Declare.
10954 * hooks.c (hook_bool_uint_mode_true): New function.
10955 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10956 HARD_REGNO_MODE_OK.
10957 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10958 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10959 instead of HARD_REGNO_MODE_OK.
10960 * caller-save.c: Include target.h.
10961 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10962 HARD_REGNO_MODE_OK.
10963 * combine.c (can_combine_p): Likewise.
10964 (combinable_i3pat): Likewise.
10965 (can_change_dest_mode): Likewise.
10966 * expr.c (init_expr_target): Likewise.
10967 (convert_move): Likewise.
10968 (convert_modes): Likewise.
10969 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10970 (setup_prohibited_mode_move_regs): Likewise.
10971 * ira.h (target_ira): Likewise.
10972 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10973 * lra-constraints.c (process_alt_operands): Likewise.
10974 (split_reg): Likewise.
10975 * recog.c (peep2_find_free_register): Likewise.
10976 * ree.c (combine_reaching_defs): Likewise.
10977 * regcprop.c (maybe_mode_change): Likewise.
10978 * reginfo.c (init_reg_sets_1): Likewise.
10979 (choose_hard_reg_mode): Likewise.
10980 (simplifiable_subregs): Likewise.
10981 * regrename.c (check_new_reg_p): Likewise.
10982 * reload.c (find_valid_class): Likewise.
10983 (find_valid_class_1): Likewise.
10984 (reload_inner_reg_of_subreg): Likewise.
10985 (push_reload): Likewise.
10986 (combine_reloads): Likewise.
10987 (find_dummy_reload): Likewise.
10988 (find_reloads): Likewise.
10989 * reload1.c (find_reg): Likewise.
10990 (set_reload_reg): Likewise.
10991 (allocate_reload_reg): Likewise.
10992 (choose_reload_regs): Likewise.
10993 (reload_adjust_reg_for_temp): Likewise.
10994 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10995 (simplify_subreg_regno): Likewise.
10996 * sel-sched.c (init_regs_for_mode): Likewise.
10997 * varasm.c (make_decl_rtl): Likewise.
10998 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10999 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
11000 HARD_REGNO_MODE_OK.
11001 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
11002 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
11003 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11004 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11005 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11006 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11007 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11008 (arc_mode_class): Delete.
11009 (HARD_REGNO_MODE_OK): Delete.
11010 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11011 (arc_hard_regno_mode_ok): Rename old array to...
11012 (arc_hard_regno_mode_ok_modes): ...this.
11013 (arc_conditional_register_usage): Update accordingly.
11014 (arc_mode_class): Make static.
11015 (arc_hard_regno_mode_ok): New function.
11016 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11017 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11018 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11019 (arm_hard_regno_mode_ok): Make static.
11020 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11021 HARD_REGNO_MODE_OK.
11022 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11023 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11024 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11026 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11027 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11028 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11029 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11030 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11031 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11032 * config/bfin/predicates.md (valid_reg_operand): Use
11033 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11034 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11035 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11036 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11037 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11038 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11039 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11040 (cr16_hard_regno_mode_ok): Make static and return a bool.
11041 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11042 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11043 (cris_hard_regno_mode_ok): New function.
11044 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11045 (epiphany_mode_class): Delete.
11046 (HARD_REGNO_MODE_OK): Delete.
11047 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11048 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11049 (hard_regno_mode_ok): Rename to...
11050 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11051 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11052 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11053 HARD_REGNO_MODE_OK.
11054 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11055 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11056 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11057 (frv_hard_regno_mode_ok): Make static and return a bool.
11058 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11059 HARD_REGNO_MODE_OK.
11060 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11061 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11062 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11063 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11065 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11066 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11067 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11068 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11070 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11071 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11072 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11073 (ia64_hard_regno_mode_ok): New function.
11074 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11075 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11076 (iq2000_hard_regno_mode_ok): New function.
11077 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11078 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11079 (lm32_hard_regno_mode_ok): New function.
11080 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11081 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11082 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11083 instead of HARD_REGNO_MODE_OK.
11084 (m32c_hard_regno_ok): Rename to...
11085 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11086 (m32c_cannot_change_mode_class): Update accordingly.
11087 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11088 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11089 (m32r_mode_class): Delete.
11090 (HARD_REGNO_MODE_OK): Delete.
11091 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11092 (m32r_hard_regno_mode_ok): Rename to...
11093 (m32r_hard_regno_modes): ...this.
11094 (m32r_mode_class): Make static.
11095 (m32r_hard_regno_mode_ok): New function.
11096 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11097 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11098 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11099 (m68k_hard_regno_mode_ok): Make static.
11100 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11101 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11102 (mcore_hard_regno_mode_ok): New function.
11103 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11104 (HARD_REGNO_MODE_OK): Delete.
11105 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11107 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11108 (microblaze_hard_regno_mode_ok): New function.
11109 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11110 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11111 (mips_hard_regno_mode_ok): Delete.
11112 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11113 (mips_hard_regno_mode_ok_p): ...this and make static.
11114 (mips_hard_regno_mode_ok_p): Rename to...
11115 (mips_hard_regno_mode_ok_uncached): ...this.
11116 (mips_hard_regno_mode_ok): New function.
11117 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11118 of HARD_REGNO_MODE_OK.
11119 (mips_option_override): Update after above name changes.
11120 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11121 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11122 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11123 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11124 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11125 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11126 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11127 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11128 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11129 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11130 (msp430_hard_regno_mode_ok): Make static and return a bool.
11131 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11132 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11133 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11135 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11136 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11137 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11138 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11139 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11140 (PA_HARD_REGNO_MODE_OK): ...this
11141 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11142 (PA_HARD_REGNO_MODE_OK): ...this.
11143 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11144 (pa_hard_regno_mode_ok): New function.
11145 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11146 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11147 (pdp11_hard_regno_mode_ok): New function.
11148 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11149 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11151 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11153 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11154 (rs6000_hard_regno_mode_ok): Rename to...
11155 (rs6000_hard_regno_mode_ok_uncached): ...this.
11156 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11157 (rs6000_hard_regno_mode_ok): New function.
11158 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11159 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11160 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11161 (riscv_hard_regno_mode_ok): ...this and make static.
11162 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11163 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11164 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11165 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11166 (rl78_hard_regno_mode_ok): Make static and return bool.
11167 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11168 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11170 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11171 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11172 (rs6000_hard_regno_mode_ok): Rename to...
11173 (rs6000_hard_regno_mode_ok_uncached): ...this.
11174 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11175 (rs6000_hard_regno_mode_ok): New function.
11176 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11177 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11178 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11179 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11180 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11181 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11182 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11183 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11184 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11185 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11186 (sh_hard_regno_mode_ok): Make static.
11187 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11188 instead of HARD_REGNO_MODE_OK.
11189 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11190 (sparc_mode_class): Delete.
11191 (HARD_REGNO_MODE_OK): Delete.
11192 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11193 (hard_regno_mode_classes): Make static.
11194 (sparc_mode_class): Likewise.
11195 (sparc_hard_regno_mode_ok): New function.
11196 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11197 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11198 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11200 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11201 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11202 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11203 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11204 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11205 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11206 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11207 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11208 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11209 (visium_hard_regno_mode_ok): New function.
11210 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11211 instead of HARD_REGNO_MODE_OK.
11212 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11213 (HARD_REGNO_MODE_OK): Delete.
11214 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11215 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11216 (xtensa_option_override): Update accordingly.
11217 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11218 (xtensa_hard_regno_mode_ok): New function.
11219 * system.h (HARD_REGNO_MODE_OK): Poison.
11221 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11222 Alan Hayward <alan.hayward@arm.com>
11223 David Sherwood <david.sherwood@arm.com>
11225 * target.def (hard_regno_call_part_clobbered): New hook.
11226 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11227 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11228 * doc/tm.texi: Regenerate.
11229 * hooks.h (hook_bool_uint_mode_false): Declare.
11230 * hooks.c (hook_bool_uint_mode_false): New function.
11231 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11232 * cselib.c (cselib_process_insn): Use
11233 targetm.hard_regno_call_part_clobbered instead of
11234 HARD_REGNO_CALL_PART_CLOBBERED.
11235 * ira-conflicts.c (ira_build_conflicts): Likewise.
11236 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11237 * lra-constraints.c (need_for_call_save_p): Likewise.
11238 * lra-lives.c: Include target.h.
11239 (check_pseudos_live_through_calls): Use
11240 targetm.hard_regno_call_part_clobbered instead of
11241 HARD_REGNO_CALL_PART_CLOBBERED.
11242 * regcprop.c: Include target.h.
11243 (copyprop_hardreg_forward_1): Use
11244 targetm.hard_regno_call_part_clobbered instead of
11245 HARD_REGNO_CALL_PART_CLOBBERED.
11246 * reginfo.c (choose_hard_reg_mode): Likewise.
11247 * regrename.c (check_new_reg_p): Likewise.
11248 * reload.c (find_equiv_reg): Likewise.
11249 * reload1.c (emit_reload_insns): Likewise.
11250 * sched-deps.c (deps_analyze_insn): Likewise.
11251 * sel-sched.c (init_regs_for_mode): Likewise.
11252 (mark_unavailable_hard_regs): Likewise.
11253 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11254 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11255 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11257 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11258 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11259 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11261 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11263 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11264 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11265 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11267 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11268 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11269 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11271 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11272 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11274 * config/powerpcspe/powerpcspe.c
11275 (rs6000_hard_regno_call_part_clobbered): New function.
11276 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11277 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11278 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11280 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11281 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11282 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11284 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11285 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11286 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11288 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11289 Alan Hayward <alan.hayward@arm.com>
11290 David Sherwood <david.sherwood@arm.com>
11292 * rtl.h (subreg_memory_offset): Declare.
11293 * emit-rtl.c (subreg_memory_offset): New function.
11294 * expmed.c (store_bit_field_1): Use it.
11295 * expr.c (undefined_operand_subword_p): Likewise.
11296 * simplify-rtx.c (simplify_subreg): Likewise.
11298 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11300 PR rtl-optimization/57448
11303 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11304 using atomic_load pattern.
11305 (expand_atomic_store): Likewise.
11307 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11310 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11311 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11312 handling. Use replace_call_with_value with NULL instead of
11313 gsi_replace, unlink_stmt_vdef and release_defs.
11315 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11318 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11320 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11323 * rtl.h (emit_library_call_value_1): Declare.
11324 (emit_library_call): Replace declaration with a series of overloads.
11325 Remove the parameter count argument.
11326 (emit_library_call_value): Likewise.
11327 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11328 with an "rtx_mode_t *".
11329 (emit_library_call_value): Delete.
11330 (emit_library_call): Likewise.
11331 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11332 (asan_emit_allocas_unpoison): Likewise.
11333 * builtins.c (expand_builtin_powi): Likewise.
11334 (expand_asan_emit_allocas_unpoison): Likewise.
11335 * cfgexpand.c (expand_main_function): Likewise.
11336 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11337 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11338 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11339 * config/arm/arm.c (arm_trampoline_init): Likewise.
11340 (arm_call_tls_get_addr): Likewise.
11341 (arm_expand_divmod_libfunc): Likewise.
11342 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11343 (smulsi3_highpart): Likewise.
11344 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11345 (c6x_expand_compare): Likewise.
11346 (c6x_expand_movmem): Likewise.
11347 * config/frv/frv.c (frv_trampoline_init): Likewise.
11348 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11349 (ix86_expand_divmod_libfunc): Likewise.
11350 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11351 (ia64_expand_compare): Likewise.
11352 (ia64_profile_hook): Likewise.
11353 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11354 (nonlocal_goto): Likewise.
11355 (restore_stack_nonlocal): Likewise.
11356 * config/m32r/m32r.c (block_move_call): Likewise.
11357 (m32r_trampoline_init): Likewise.
11358 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11359 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11360 (m68k_call_m68k_read_tp): Likewise.
11361 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11362 (microblaze_expand_divide): Likewise.
11363 * config/mips/mips.h (mips_args): Likewise.
11364 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11365 (MIPS_ICACHE_SYNC): Likewise.
11366 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11367 (nios2_trampoline_init): Likewise.
11368 * config/pa/pa.c (hppa_tls_call): Likewise.
11369 (pa_trampoline_init): Likewise.
11370 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11371 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11372 (expand_strn_compare): Likewise.
11373 (rs6000_generate_compare): Likewise.
11374 (rs6000_expand_float128_convert): Likewise.
11375 (output_profile_hook): Likewise.
11376 (rs6000_trampoline_init): Likewise.
11377 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11378 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11379 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11380 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11381 (rs6000_generate_compare): Likewise.
11382 (rs6000_expand_float128_convert): Likewise.
11383 (output_profile_hook): Likewise.
11384 (rs6000_trampoline_init): Likewise.
11385 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11386 * config/sh/sh.c (sh_trampoline_init): Likewise.
11387 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11388 (sparc_emit_float_lib_cmp): Likewise.
11389 (sparc32_initialize_trampoline): Likewise.
11390 (sparc64_initialize_trampoline): Likewise.
11391 (sparc_profile_hook): Likewise.
11392 * config/spu/spu.c (ea_load_store): Likewise.
11393 * config/spu/spu.md (floatunssidf2): Likewise.
11394 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11395 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11396 * config/visium/visium.c (expand_block_move_4): Likewise.
11397 (expand_block_move_2): Likewise.
11398 (expand_block_move_1): Likewise.
11399 (expand_block_set_4): Likewise.
11400 (expand_block_set_2): Likewise.
11401 (expand_block_set_1): Likewise.
11402 (visium_trampoline_init): Likewise.
11403 (visium_profile_hook): Likewise.
11404 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11405 (xtensa_setup_frame_addresses): Likewise.
11406 (xtensa_trampoline_init): Likewise.
11407 * except.c (sjlj_emit_function_enter): Likewise.
11408 (sjlj_emit_function_exit): Likewise.
11409 * explow.c (allocate_dynamic_stack_space): Likewise.
11410 (probe_stack_range): Likewise.
11411 * expr.c (convert_mode_scalar): Likewise.
11412 * optabs.c (expand_binop): Likewise.
11413 (expand_twoval_binop_libfunc): Likewise.
11414 (expand_unop): Likewise.
11415 (prepare_cmp_insn): Likewise.
11416 (prepare_float_lib_cmp): Likewise.
11417 (expand_float): Likewise.
11418 (expand_fix): Likewise.
11419 (expand_fixed_convert): Likewise.
11420 (maybe_emit_sync_lock_test_and_set): Likewise.
11421 (expand_atomic_compare_and_swap): Likewise.
11422 (expand_mem_thread_fence): Likewise.
11423 (expand_atomic_fetch_op): Likewise.
11425 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11427 * doc/generic.texi (OpenACC): Adjust URL.
11428 * doc/invoke.texi (C Dialect Options): Ditto.
11430 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11432 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11433 predicate for operand 1. Add (m,<S>) constraint.
11434 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11435 Prevent memory operand 1 with register operand 2.
11437 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11439 PR rtl-optimization/82024
11440 * combine.c (try_combine): If the combination result is a PARALLEL,
11441 and we only need to retain the SET in there that would be placed
11442 at I2, check that we can place that at I3 instead, before doing so.
11444 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11447 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11449 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11450 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11453 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11454 Jeff Law <law@redhat.com>
11456 * varasm.c (bss_initializer_p): Do not put constants into .bss
11457 (categorize_decl_for_section): Handle bss_initializer_p returning
11458 false when DECL_INITIAL is NULL.
11460 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11463 * config/s390/s390.c (s390_can_inline_p): New function.
11465 2017-09-01 Jeff Law <law@redhat.com>
11467 PR tree-optimization/82052
11468 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11469 Always initialize the returned slot after a hash table miss
11470 when INSERT is true.
11472 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11474 * config/s390/s390.md (mem_signal_fence): Remove.
11475 * doc/md.texi (mem_signal_fence): Remove.
11476 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11478 * target-insns.def (mem_signal_fence): Remove.
11480 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11483 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11486 * asan.c (create_odr_indicator): Strip name encoding from assembler
11487 name before appending it after __odr_asan_.
11489 2017-09-01 Martin Liska <mliska@suse.cz>
11491 PR tree-optimization/82059
11492 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11493 frequency only when an edge is redirected.
11495 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11497 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11498 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11499 (arc_conditional_register_usage): Remove ARC600 lp_count
11501 (arc_file_start): Emit Tag_ARC_CPU_variation.
11502 (arc_can_use_doloop_p): New conditions to use ZOLs.
11503 (hwloop_fail): New function.
11504 (hwloop_optimize): Likewise.
11505 (hwloop_pattern_reg): Likewise.
11506 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11507 (arc_reorg_loops): New function, calls reorg_loops.
11508 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11509 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11511 (arc_loop_hazard): Remove function, functionality moved into
11513 (arc_hazard): Remove arc_loop_hazard call.
11514 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11515 into hwloop_optimize.
11516 (arc_label_align): Remove ZOL handling.
11517 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11518 * config/arc/arc.md (doloop_begin): Remove pattern.
11519 (doloop_begin_i): Likewise.
11520 (doloop_end_i): Likewise.
11521 (doloop_fallback): Likewise.
11522 (doloop_fallback_m): Likewise.
11523 (doloop_end): Reimplement expand.
11524 (arc_lp): New pattern for LP instruction.
11525 (loop_end): New pattern.
11526 (loop_fail): Likewise.
11527 (decrement_and_branch_until_zero): Likewise.
11528 * config/arc/arc.opt (mlpc-width): New option.
11529 * doc/invoke.texi (mlpc-width): Document option.
11531 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11533 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11534 (arc_ccfsm_advance): Fix checking for delay slots.
11535 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11537 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11539 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11541 (movhi_insn): Update store instruction constraint which are saving
11542 6-bit short immediates.
11543 (movsi_insn): Consider also short scaled load operations.
11544 (zero_extendhisi2_i): Use Usd constraint instead of T.
11545 (extendhisi2_i): Add q constraint.
11546 (arc_clzsi2): Add type and length attributes.
11547 (arc_ctzsi2): Likewise.
11548 * config/arc/constraints.md (Usc): Update constraint, the
11549 assembler can parse two relocations for a single instruction.
11551 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11553 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11554 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11556 2017-08-31 Olivier Hainque <hainque@adacore.com>
11558 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11559 match as powerpc-wrs-vxworks*.
11561 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11563 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11564 register constraint for by-element operand.
11565 (aarch64_mls_elt_merge<mode>): Likewise.
11567 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11569 * config/arc/arc.c (arc_can_follow_jump): Check for short
11572 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11574 * config.gcc: Use g.opt for arc.
11575 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11576 functionality moved to ...
11577 (legitimate_scaled_address_p): New function, ...here.
11578 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11579 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11580 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11582 (arc_override_options): Handle G option.
11583 (arc_output_pic_addr_const): Correct function definition.
11584 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11585 (arc_decl_anon_ns_mem_p): Delete.
11586 (arc_in_small_data_p): Overhaul this function to take into
11587 consideration the value given via G option.
11588 (arc_rewrite_small_data_1): Renamed and corrected old
11589 arc_rewrite_small_data function.
11590 (arc_rewrite_small_data): New function.
11591 (small_data_pattern): Don't use pic_offset_table_rtx.
11592 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11593 * config/arc/simdext.md (movmisalignv2hi): Use
11594 prepare_move_operands function.
11596 (movmisalign*): Likewise.
11597 * doc/invoke.texi (ARC options): Document -G option.
11599 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11601 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11603 * config/arc/arc.c (arc_print_operand): Output scalled address for
11604 sdata whenever is possible.
11605 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11606 load/stores are available.
11607 (compact_sda_memory_operand): Check for the alignment required by
11608 code density instructions.
11609 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11611 * config/arc/constraints.md (Usd): Update constraint.
11612 (Us0): New constraint.
11613 (Usc): Update constraint.
11615 2017-08-31 Richard Biener <rguenther@suse.de>
11617 PR middle-end/82054
11618 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11619 function only once.
11621 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11623 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11624 Resize type_signature.
11626 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11627 Alan Hayward <alan.hayward@arm.com>
11628 David Sherwood <david.sherwood@arm.com>
11630 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11631 subregs whose inner modes can be stored in GPRs.
11632 (aarch64_classify_index): Likewise.
11634 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11635 Alan Hayward <alan.hayward@arm.com>
11636 David Sherwood <david.sherwood@arm.com>
11638 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11639 (V_INT_EQUIV): ...this.
11640 (v_cmp_result): Rename to...
11641 (v_int_equiv): ...this.
11642 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11643 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11644 (copysign<mode>3): Likewise.
11645 (aarch64_simd_bsl<mode>_internal): Likewise.
11646 (aarch64_simd_bsl<mode>): Likewise.
11647 (vec_cmp<mode><mode>): Likewise.
11648 (vcond<mode><mode>): Likewise.
11649 (vcond<v_cmp_mixed><mode>): Likewise.
11650 (vcondu<mode><v_cmp_mixed>): Likewise.
11651 (aarch64_cm<optab><mode>): Likewise.
11652 (aarch64_cmtst<mode>): Likewise.
11653 (aarch64_fac<optab><mode>): Likewise.
11654 (vec_perm_const<mode>): Likewise.
11655 (vcond_mask_<mode><v_cmp_result>): Rename to...
11656 (vcond_mask_<mode><v_int_equiv>): ...this.
11657 (vec_cmp<mode><v_cmp_result>): Rename to...
11658 (vec_cmp<mode><v_int_equiv>): ...this.
11660 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11661 Alan Hayward <alan.hayward@arm.com>
11662 David Sherwood <david.sherwood@arm.com>
11664 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11666 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11667 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11668 (UNSPEC_LD4_DREG): New unspecs.
11669 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11670 (aarch64_ld2<mode>_dreg_be): Replace with...
11671 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11673 (aarch64_ld3<mode>_dreg_le)
11674 (aarch64_ld3<mode>_dreg_be): Replace with...
11675 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11677 (aarch64_ld4<mode>_dreg_le)
11678 (aarch64_ld4<mode>_dreg_be): Replace with...
11679 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11682 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11684 PR tree-optimization/81987
11685 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11686 insert an initializer in a location not dominated by the stride
11689 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11691 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11692 on the entire header of the finally block in the fallthru case.
11694 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11696 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11698 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11700 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11701 rs6000_emit_move_from_cr and call renamed function.
11702 (rs6000_emit_prologue): Call renamed functions.
11703 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11704 movesi_from_cr, remove volatile CRs.
11706 2017-08-30 Jon Beniston <jon@beniston.com>
11707 Richard Biener <rguenther@suse.de>
11709 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11710 of VECTOR_MODE_P check.
11711 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11712 element vector types.
11714 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11716 * df.h (df_read_modify_subreg_p): Remove in favor of...
11717 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11718 const_rtx instead of an rtx.
11719 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11720 * df-problems.c (df_word_lr_mark_ref): Likewise.
11721 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11722 (mark_pseudo_reg_dead): Likewise.
11723 (mark_ref_dead): Likewise.
11724 * reginfo.c (init_subregs_of_mode): Likewise.
11725 * sched-deps.c (sched_analyze_1): Likewise.
11726 * df-scan.c (df_def_record_1): Likewise.
11727 (df_uses_record): Likewise.
11728 (df_read_modify_subreg_p): Remove in favor of...
11729 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11730 const_rtx instead of an rtx.
11732 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11733 Alan Hayward <alan.hayward@arm.com>
11734 David Sherwood <david.sherwood@arm.com>
11736 * rtl.h (partial_subreg_p): New function.
11737 * caller-save.c (save_call_clobbered_regs): Use it.
11738 * calls.c (expand_call): Likewise.
11739 * combine.c (combinable_i3pat): Likewise.
11740 (simplify_set): Likewise.
11741 (make_extraction): Likewise.
11742 (make_compound_operation_int): Likewise.
11743 (gen_lowpart_or_truncate): Likewise.
11744 (force_to_mode): Likewise.
11745 (make_field_assignment): Likewise.
11746 (reg_truncated_to_mode): Likewise.
11747 (record_truncated_value): Likewise.
11748 (move_deaths): Likewise.
11749 * cse.c (record_jump_cond): Likewise.
11750 (cse_insn): Likewise.
11751 * cselib.c (cselib_lookup_1): Likewise.
11752 * expmed.c (extract_bit_field_using_extv): Likewise.
11753 * function.c (assign_parm_setup_reg): Likewise.
11754 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11755 * ira-build.c (create_insn_allocnos): Likewise.
11756 * lra-coalesce.c (merge_pseudos): Likewise.
11757 * lra-constraints.c (match_reload): Likewise.
11758 (simplify_operand_subreg): Likewise.
11759 (curr_insn_transform): Likewise.
11760 * lra-lives.c (process_bb_lives): Likewise.
11761 * lra.c (new_insn_reg): Likewise.
11762 (lra_substitute_pseudo): Likewise.
11763 * regcprop.c (mode_change_ok): Likewise.
11764 (maybe_mode_change): Likewise.
11765 (copyprop_hardreg_forward_1): Likewise.
11766 * reload.c (push_reload): Likewise.
11767 (find_reloads): Likewise.
11768 (find_reloads_subreg_address): Likewise.
11769 * reload1.c (alter_reg): Likewise.
11770 (eliminate_regs_1): Likewise.
11771 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11773 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11775 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11776 back to if statements, including unpack.
11778 2017-08-30 Martin Liska <mliska@suse.cz>
11780 PR inline-asm/82001
11781 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11783 (func_checker::compare_asm_inputs_outputs): ... this function.
11784 (func_checker::compare_gimple_asm): Use the function to compare
11785 also ASM constrains.
11786 * ipa-icf-gimple.h: Rename the function.
11788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11789 Alan Hayward <alan.hayward@arm.com>
11790 David Sherwood <david.sherwood@arm.com>
11792 * coretypes.h (complex_mode): New type.
11793 * gdbhooks.py (build_pretty_printer): Handle it.
11794 * machmode.h (complex_mode): New class.
11795 (complex_mode::includes_p): New function.
11796 (is_complex_int_mode): Likewise.
11797 (is_complex_float_mode): Likewise.
11798 * genmodes.c (get_mode_class): Handle complex mode classes.
11799 * function.c (expand_function_end): Use is_complex_int_mode.
11801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11802 Alan Hayward <alan.hayward@arm.com>
11803 David Sherwood <david.sherwood@arm.com>
11805 * coretypes.h (scalar_mode_pod): New typedef.
11806 * gdbhooks.py (build_pretty_printer): Handle it.
11807 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11808 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11809 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11810 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11811 as_a <scalar_mode>.
11813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11814 Alan Hayward <alan.hayward@arm.com>
11815 David Sherwood <david.sherwood@arm.com>
11817 * machmode.h (mode_for_vector): Take a scalar_mode instead
11819 * stor-layout.c (mode_for_vector): Likewise.
11820 * explow.c (promote_mode): Use as_a <scalar_mode>.
11821 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11824 Alan Hayward <alan.hayward@arm.com>
11825 David Sherwood <david.sherwood@arm.com>
11827 * target.def (preferred_simd_mode): Take a scalar_mode
11828 instead of a machine_mode.
11829 * targhooks.h (default_preferred_simd_mode): Likewise.
11830 * targhooks.c (default_preferred_simd_mode): Likewise.
11831 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11832 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11833 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11834 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11835 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11836 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11837 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11838 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11840 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11841 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11842 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11843 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11844 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11845 * doc/tm.texi: Regenerate.
11846 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11850 Alan Hayward <alan.hayward@arm.com>
11851 David Sherwood <david.sherwood@arm.com>
11853 * target.def (scalar_mode_supported_p): Take a scalar_mode
11854 instead of a machine_mode.
11855 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11856 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11857 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11858 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11859 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11860 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11861 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11862 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11863 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11864 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11865 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11866 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11867 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11868 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11870 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11871 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11872 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11873 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11874 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11876 * doc/tm.texi: Regenerate.
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 * coretypes.h (opt_scalar_mode): New typedef.
11883 * gdbhooks.py (build_pretty_printers): Handle it.
11884 * machmode.h (mode_iterator::get_2xwider): Add overload for
11886 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11888 * expr.c (convert_mode_scalar): Likewise.
11889 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11890 * optabs.c (expand_float): Likewise.
11891 (expand_fix): Likewise.
11892 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11894 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11895 Alan Hayward <alan.hayward@arm.com>
11896 David Sherwood <david.sherwood@arm.com>
11898 * optabs.c (expand_float): Explicitly check for scalars before
11899 using a branching expansion.
11900 (expand_fix): Likewise.
11902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11903 Alan Hayward <alan.hayward@arm.com>
11904 David Sherwood <david.sherwood@arm.com>
11906 * expr.c (convert_mode): Split scalar handling out into...
11907 (convert_mode_scalar): ...this new function. Treat the modes
11910 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11911 Alan Hayward <alan.hayward@arm.com>
11912 David Sherwood <david.sherwood@arm.com>
11914 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11916 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11919 Alan Hayward <alan.hayward@arm.com>
11920 David Sherwood <david.sherwood@arm.com>
11922 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11923 rather than a machine_mode.
11924 (fixed_from_string): Likewise.
11925 (fixed_convert): Likewise.
11926 (fixed_convert_from_int): Likewise.
11927 (fixed_convert_from_real): Likewise.
11928 (real_convert_from_fixed): Likewise.
11929 * fixed-value.c (fixed_from_double_int): Likewise.
11930 (fixed_from_string): Likewise.
11931 (fixed_convert): Likewise.
11932 (fixed_convert_from_int): Likewise.
11933 (fixed_convert_from_real): Likewise.
11934 (real_convert_from_fixed): Likewise.
11935 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11937 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11938 Alan Hayward <alan.hayward@arm.com>
11939 David Sherwood <david.sherwood@arm.com>
11941 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11942 of separate mode class checks. Do not allow vector modes here.
11943 (immed_wide_int_const): Use as_a <scalar_mode>.
11944 * explow.c (trunc_int_for_mode): Likewise.
11945 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11946 (wi::shwi): Likewise.
11947 (wi::min_value): Likewise.
11948 (wi::max_value): Likewise.
11949 * dwarf2out.c (loc_descriptor): Likewise.
11950 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11951 for CONST_WIDE_INT.
11953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11954 Alan Hayward <alan.hayward@arm.com>
11955 David Sherwood <david.sherwood@arm.com>
11957 * tree.h (SCALAR_TYPE_MODE): New macro.
11958 * expr.c (expand_expr_addr_expr_1): Use it.
11959 (expand_expr_real_2): Likewise.
11960 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11961 (fold_convert_const_fixed_from_int): Likewise.
11962 (fold_convert_const_fixed_from_real): Likewise.
11963 (native_encode_fixed): Likewise
11964 (native_encode_complex): Likewise
11965 (native_encode_vector): Likewise.
11966 (native_interpret_fixed): Likewise.
11967 (native_interpret_real): Likewise.
11968 (native_interpret_complex): Likewise.
11969 (native_interpret_vector): Likewise.
11970 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11971 (simd_clone_adjust_argument_types): Likewise.
11972 (simd_clone_init_simd_arrays): Likewise.
11973 (simd_clone_adjust): Likewise.
11974 * stor-layout.c (layout_type): Likewise.
11975 * tree.c (build_minus_one_cst): Likewise.
11976 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11977 * tree-inline.c (estimate_move_cost): Likewise.
11978 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11979 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11980 (vectorizable_reduction): Likewise.
11981 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11982 (vect_recog_mixed_size_cond_pattern): Likewise.
11983 (check_bool_pattern): Likewise.
11984 (adjust_bool_pattern): Likewise.
11985 (search_type_for_mask_1): Likewise.
11986 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11987 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11988 (vectorizable_load): Likewise.
11989 (vectorizable_store): Likewise.
11990 * ubsan.c (ubsan_encode_value): Likewise.
11991 * varasm.c (output_constant): Likewise.
11993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11994 Alan Hayward <alan.hayward@arm.com>
11995 David Sherwood <david.sherwood@arm.com>
11997 * coretypes.h (scalar_mode): New class.
11998 * machmode.h (scalar_mode): Likewise.
11999 (scalar_mode::includes_p): New function.
12000 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
12001 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
12002 * genmodes.c (get_mode_class): Handle remaining scalar modes.
12003 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12004 * expmed.c (store_bit_field_1): Likewise.
12005 (extract_bit_field_1): Likewise.
12006 * expr.c (write_complex_part): Likewise.
12007 (read_complex_part): Likewise.
12008 (emit_move_complex_push): Likewise.
12009 (expand_expr_real_2): Likewise.
12010 * function.c (assign_parm_setup_reg): Likewise.
12011 (assign_parms_unsplit_complex): Likewise.
12012 * optabs.c (expand_binop): Likewise.
12013 * rtlanal.c (subreg_get_info): Likewise.
12014 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12015 * varasm.c (output_constant_pool_2): Likewise.
12017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12018 Alan Hayward <alan.hayward@arm.com>
12019 David Sherwood <david.sherwood@arm.com>
12021 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12023 (expmed_mult_highpart_optab): Likewise.
12024 (expmed_mult_highpart): Likewise.
12026 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12027 Alan Hayward <alan.hayward@arm.com>
12028 David Sherwood <david.sherwood@arm.com>
12030 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12031 instead of a machine_mode.
12032 (builtin_memset_read_str): Likewise.
12033 * builtins.c (c_readstr): Likewise.
12034 (builtin_memcpy_read_str): Likewise.
12035 (builtin_strncpy_read_str): Likewise.
12036 (builtin_memset_read_str): Likewise.
12037 (builtin_memset_gen_str): Likewise.
12038 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12039 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12040 instead of a machine_mode.
12041 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12043 (make_extraction): Likewise.
12044 (try_widen_shift_mode): Take and return scalar_int_modes instead
12046 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12047 a scalar_int_mode instead of a machine_mode.
12048 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12049 (avr_addr_space_pointer_mode): Likewise.
12050 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12051 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12052 (msp430_unwind_word_mode): Likewise.
12053 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12054 (spu_addr_space_pointer_mode): Likewise.
12055 (spu_addr_space_address_mode): Likewise.
12056 (spu_libgcc_cmp_return_mode): Likewise.
12057 (spu_libgcc_shift_count_mode): Likewise.
12058 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12059 (rl78_addr_space_pointer_mode): Likewise.
12060 (fl78_unwind_word_mode): Likewise.
12061 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12063 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12064 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12065 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12066 (mips_valid_pointer_mode): Likewise.
12067 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12068 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12069 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12071 (ft32_addr_space_address_mode): Likewise.
12072 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12073 scalar_int_mode instead of a machine_mode.
12074 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12076 (m32c_addr_space_address_mode): Likewise.
12077 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12078 (rs6000_eh_return_filter_mode): Likewise.
12079 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12080 (rs6000_eh_return_filter_mode): Likewise.
12081 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12082 (s390_libgcc_shift_count_mode): Likewise.
12083 (s390_unwind_word_mode): Likewise.
12084 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12086 * target.def (mode_rep_extended): Likewise.
12087 (valid_pointer_mode): Likewise.
12088 (addr_space.valid_pointer_mode): Likewise.
12089 (eh_return_filter_mode): Return a scalar_int_mode rather than
12091 (libgcc_cmp_return_mode): Likewise.
12092 (libgcc_shift_count_mode): Likewise.
12093 (unwind_word_mode): Likewise.
12094 (addr_space.pointer_mode): Likewise.
12095 (addr_space.address_mode): Likewise.
12096 * doc/tm.texi: Regenerate.
12097 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12099 (do_jump): Use scalar_int_mode for local variables.
12100 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12101 rather than a machine_mode.
12102 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12103 (scompare_loc_descriptor_wide): Likewise.
12104 (scompare_loc_descriptor_narrow): Likewise.
12105 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12107 * except.c (sjlj_emit_dispatch_table): Likewise.
12108 (expand_builtin_eh_copy_values): Likewise.
12109 * explow.c (convert_memory_address_addr_space_1): Likewise.
12110 Take a scalar_int_mode rather than a machine_mode.
12111 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12112 than a machine_mode.
12113 (memory_address_addr_space): Use scalar_int_mode for local variables.
12114 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12115 rather than a machine_mode.
12116 * expmed.c (mask_rtx): Likewise.
12117 (init_expmed_one_conv): Likewise.
12118 (expand_mult_highpart_adjust): Likewise.
12119 (extract_high_half): Likewise.
12120 (expmed_mult_highpart_optab): Likewise.
12121 (expmed_mult_highpart): Likewise.
12122 (expand_smod_pow2): Likewise.
12123 (expand_sdiv_pow2): Likewise.
12124 (emit_store_flag_int): Likewise.
12125 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12127 (extract_low_bits): Likewise.
12128 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12130 * expr.c (pieces_addr::adjust): Likewise.
12131 (can_store_by_pieces): Likewise.
12132 (store_by_pieces): Likewise.
12133 (clear_by_pieces_1): Likewise.
12134 (expand_expr_addr_expr_1): Likewise.
12135 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12136 (expand_expr_real_1): Likewise.
12137 (try_casesi): Likewise.
12138 * final.c (shorten_branches): Likewise.
12139 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12140 type of "mode" to machine_mode.
12141 * internal-fn.c (expand_arith_overflow_result_store): Take a
12142 scalar_int_mode rather than a machine_mode.
12143 (expand_mul_overflow): Use scalar_int_mode for local variables.
12144 * loop-doloop.c (doloop_modify): Likewise.
12145 (doloop_optimize): Likewise.
12146 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12147 than a machine_mode.
12148 (expand_doubleword_shift_condmove): Likewise.
12149 (expand_doubleword_shift): Likewise.
12150 (expand_doubleword_clz): Likewise.
12151 (expand_doubleword_popcount): Likewise.
12152 (expand_doubleword_parity): Likewise.
12153 (expand_absneg_bit): Use scalar_int_mode for local variables.
12154 (prepare_float_lib_cmp): Likewise.
12155 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12156 rather than a machine_mode.
12157 (convert_memory_address_addr_space): Likewise.
12158 (get_mode_bounds): Likewise.
12159 (get_address_mode): Return a scalar_int_mode rather than a
12161 * rtlanal.c (get_address_mode): Likewise.
12162 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12163 than a machine_mode.
12164 * targhooks.c (default_mode_rep_extended): Likewise.
12165 (default_valid_pointer_mode): Likewise.
12166 (default_addr_space_valid_pointer_mode): Likewise.
12167 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12168 than a machine_mode.
12169 (default_libgcc_cmp_return_mode): Likewise.
12170 (default_libgcc_shift_count_mode): Likewise.
12171 (default_unwind_word_mode): Likewise.
12172 (default_addr_space_pointer_mode): Likewise.
12173 (default_addr_space_address_mode): Likewise.
12174 * targhooks.h (default_eh_return_filter_mode): Likewise.
12175 (default_libgcc_cmp_return_mode): Likewise.
12176 (default_libgcc_shift_count_mode): Likewise.
12177 (default_unwind_word_mode): Likewise.
12178 (default_addr_space_pointer_mode): Likewise.
12179 (default_addr_space_address_mode): Likewise.
12180 (default_mode_rep_extended): Take a scalar_int_mode rather than
12182 (default_valid_pointer_mode): Likewise.
12183 (default_addr_space_valid_pointer_mode): Likewise.
12184 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12186 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12187 rather than a machine_mode.
12188 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12189 for local variables.
12190 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12191 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12192 than a machine_mode.
12194 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12195 Alan Hayward <alan.hayward@arm.com>
12196 David Sherwood <david.sherwood@arm.com>
12198 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12199 the mode argument to scalar_int_mode.
12200 (do_jump_by_parts_zero_rtx): Likewise.
12201 (do_jump_by_parts_equality_rtx): Likewise.
12202 (do_jump_by_parts_greater): Take a mode argument.
12203 (do_jump_by_parts_equality): Likewise.
12204 (do_jump_1): Update calls accordingly.
12206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12207 Alan Hayward <alan.hayward@arm.com>
12208 David Sherwood <david.sherwood@arm.com>
12210 * is-a.h (safe_dyn_cast): New function.
12211 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12212 (jump_table_for_label): Likewise.
12213 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12214 instead of an rtx_insn *.
12215 (shorten_branches): Use dyn_cast instead of LABEL_P and
12216 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12217 rtx_jump_table_data::get_data_mode.
12218 (final_scan_insn): Likewise.
12220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12221 Alan Hayward <alan.hayward@arm.com>
12222 David Sherwood <david.sherwood@arm.com>
12224 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12225 trying to combine a full-register integer set with a subreg
12228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12229 Alan Hayward <alan.hayward@arm.com>
12230 David Sherwood <david.sherwood@arm.com>
12232 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12233 that is always either address_mode or pointer_mode.
12235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12236 Alan Hayward <alan.hayward@arm.com>
12237 David Sherwood <david.sherwood@arm.com>
12239 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12240 when the two are known to be equal.
12242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12243 Alan Hayward <alan.hayward@arm.com>
12244 David Sherwood <david.sherwood@arm.com>
12246 * simplify-rtx.c (simplify_const_unary_operation): Use
12247 is_a <scalar_int_mode> instead of checking for a nonzero
12248 precision. Forcibly convert op_mode to a scalar_int_mode
12249 in that case. More clearly differentiate the operand and
12250 result modes and use the former when deciding what the value
12251 of a count-bits operation should be. Use is_int_mode instead
12252 of checking for a MODE_INT. Remove redundant check for whether
12253 this mode has a zero precision.
12255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12256 Alan Hayward <alan.hayward@arm.com>
12257 David Sherwood <david.sherwood@arm.com>
12259 * optabs.c (widen_leading): Change the type of the mode argument
12260 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12261 (widen_bswap): Likewise.
12262 (expand_parity): Likewise.
12263 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12264 (expand_ffs): Likewise.
12265 (epand_unop): Check for scalar integer modes before calling the
12268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12269 Alan Hayward <alan.hayward@arm.com>
12270 David Sherwood <david.sherwood@arm.com>
12272 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12274 (expand_expr_real_1): Update call accordingly.
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 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12281 value_mode arguments. Use scalar_int_mode internally.
12282 (store_bit_field_1): Rename the new integer mode from imode
12283 to op0_mode and use it instead of GET_MODE (op0). Update calls
12284 to store_split_bit_field, store_bit_field_using_insv and
12285 store_fixed_bit_field.
12286 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12287 Use scalar_int_mode internally. Use a bit count rather than a mode
12288 when calculating the largest bit size for get_best_mode.
12289 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12290 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12291 Remove assertion that OP0 has a scalar integer mode.
12292 (store_split_bit_field): Add op0_mode and value_mode arguments.
12293 Update calls to extract_fixed_bit_field.
12294 (extract_bit_field_using_extv): Add an op0_mode argument.
12295 Use scalar_int_mode internally.
12296 (extract_bit_field_1): Rename the new integer mode from imode to
12297 op0_mode and use it instead of GET_MODE (op0). Update calls to
12298 extract_split_bit_field, extract_bit_field_using_extv and
12299 extract_fixed_bit_field.
12300 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12301 to extract_split_bit_field and extract_fixed_bit_field_1.
12302 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12303 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12304 on the target mode.
12305 (extract_split_bit_field): Add an op0_mode argument. Update call
12306 to extract_fixed_bit_field.
12308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12309 Alan Hayward <alan.hayward@arm.com>
12310 David Sherwood <david.sherwood@arm.com>
12312 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12313 * explow.c (hard_function_value): Likewise.
12314 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12315 convert_to_mode call outside the loop.
12316 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12317 for the mode iterator. Require the mode specified by max_pieces
12319 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12321 (copy_blkmode_to_reg): Likewise.
12322 (set_storage_via_setmem): Likewise.
12323 * optabs.c (prepare_cmp_insn): Likewise.
12324 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12325 * stor-layout.c (finish_bitfield_representative): Likewise.
12327 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12328 Alan Hayward <alan.hayward@arm.com>
12329 David Sherwood <david.sherwood@arm.com>
12331 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12332 * expr.c (convert_move): Use them.
12333 (convert_modes): Likewise.
12334 (store_expr_with_bounds): Likewise.
12336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12337 Alan Hayward <alan.hayward@arm.com>
12338 David Sherwood <david.sherwood@arm.com>
12340 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12341 parameter for the mode of "x". Remove the "known_x", "known_mode"
12342 and "known_ret" arguments. Change the type of the mode argument
12343 to scalar_int_mode.
12344 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12345 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12346 (reg_num_sign_bit_copies_for_combine): Likewise.
12347 * rtlanal.c (nonzero_bits1): Likewise.
12348 (num_sign_bit_copies1): Likewise.
12349 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12350 (reg_num_sign_bit_copies_general): Likewise.
12351 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12352 (reg_nonzero_bits_general): Likewise.
12354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12355 Alan Hayward <alan.hayward@arm.com>
12356 David Sherwood <david.sherwood@arm.com>
12358 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12359 than in subroutines. Return 1 for non-integer modes.
12360 (cached_num_sign_bit_copies): Change the type of the mode parameter
12361 to scalar_int_mode.
12362 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12363 classes. Handle CONST_INT_P first and then check whether X also
12364 has a scalar integer mode. Check the same thing for inner registers
12365 of a SUBREG and for values that are being extended or truncated.
12367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12368 Alan Hayward <alan.hayward@arm.com>
12369 David Sherwood <david.sherwood@arm.com>
12371 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12372 in subroutines. Return the mode mask for non-integer modes.
12373 (cached_nonzero_bits): Change the type of the mode parameter
12374 to scalar_int_mode.
12375 (nonzero_bits1): Likewise. Remove early exit for other mode
12376 classes. Handle CONST_INT_P first and then check whether X
12377 also has a scalar integer mode.
12379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12380 Alan Hayward <alan.hayward@arm.com>
12381 David Sherwood <david.sherwood@arm.com>
12383 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12384 Return a scalar_int_mode and assert that the size is greater than
12386 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12387 (op_by_pieces_d::op_by_pieces_d): Likewise.
12388 (op_by_pieces_d::run): Likewise.
12389 (can_store_by_pieces): Likewise.
12391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12392 Alan Hayward <alan.hayward@arm.com>
12393 David Sherwood <david.sherwood@arm.com>
12395 * combine.c (extract_left_shift): Add a mode argument and update
12397 (make_compound_operation_int): Change the type of the mode parameter
12398 to scalar_int_mode and update the call to extract_left_shift.
12400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12401 Alan Hayward <alan.hayward@arm.com>
12402 David Sherwood <david.sherwood@arm.com>
12404 * combine.c (simplify_and_const_int): Change the type of the mode
12405 parameter to scalar_int_mode.
12406 (simplify_and_const_int_1): Likewise. Update recursive call.
12408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12409 Alan Hayward <alan.hayward@arm.com>
12410 David Sherwood <david.sherwood@arm.com>
12412 * combine.c (simplify_compare_const): Check that the mode is a
12413 scalar_int_mode (rather than VOIDmode) before testing its
12415 (simplify_comparison): Move COMPARISON_P handling out of the
12416 loop and restrict the latter part of the loop to scalar_int_modes.
12417 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12418 and when considering SUBREG_REGs. Use is_int_mode instead of
12419 checking GET_MODE_CLASS against MODE_INT.
12421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12422 Alan Hayward <alan.hayward@arm.com>
12423 David Sherwood <david.sherwood@arm.com>
12425 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12426 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12427 shift_unit_mode and for modes involved in scalar shifts.
12429 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12430 Alan Hayward <alan.hayward@arm.com>
12431 David Sherwood <david.sherwood@arm.com>
12433 * combine.c (force_int_to_mode): New function, split out from...
12434 (force_to_mode): ...here. Keep xmode up-to-date and use it
12435 instead of GET_MODE (x).
12437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12438 Alan Hayward <alan.hayward@arm.com>
12439 David Sherwood <david.sherwood@arm.com>
12441 * optabs-query.h (extraction_insn::struct_mode): Change type to
12442 opt_scalar_int_mode and update comment.
12443 (extraction_insn::field_mode): Change type to scalar_int_mode.
12444 (extraction_insn::pos_mode): Likewise.
12445 * combine.c (make_extraction): Update accordingly.
12446 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12447 (get_optab_extraction_insn): Likewise.
12448 * recog.c (simplify_while_replacing): Likewise.
12449 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12450 parameter to opt_scalar_int_mode.
12452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12453 Alan Hayward <alan.hayward@arm.com>
12454 David Sherwood <david.sherwood@arm.com>
12456 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12457 to a scalar_int_mode instead of a machine_mode.
12458 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12459 (get_best_mode): Return a boolean and use a pointer argument to store
12460 the selected mode. Replace the limit mode parameter with a bit limit.
12461 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12462 for the values returned by bit_field_mode_iterator::next_mode.
12463 (store_bit_field): Update call to get_best_mode.
12464 (store_fixed_bit_field): Likewise.
12465 (extract_fixed_bit_field): Likewise.
12466 * expr.c (optimize_bitfield_assignment_op): Likewise.
12467 * fold-const.c (optimize_bit_field_compare): Likewise.
12468 (fold_truth_andor_1): Likewise.
12469 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12470 Update for new type of m_mode.
12471 (get_best_mode): As above.
12473 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12474 Alan Hayward <alan.hayward@arm.com>
12475 David Sherwood <david.sherwood@arm.com>
12477 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12478 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12479 (store_bit_field): Check is_a <scalar_int_mode> before calling
12480 strict_volatile_bitfield_p.
12481 (extract_bit_field): Likewise.
12483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12484 Alan Hayward <alan.hayward@arm.com>
12485 David Sherwood <david.sherwood@arm.com>
12487 * target.def (cstore_mode): Return a scalar_int_mode.
12488 * doc/tm.texi: Regenerate.
12489 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12490 * targhooks.h (default_cstore_mode): Likewise.
12491 * targhooks.c (default_cstore_mode): Likewise, using a forced
12493 * expmed.c (emit_cstore): Expect the target of the cstore to be
12496 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12497 Alan Hayward <alan.hayward@arm.com>
12498 David Sherwood <david.sherwood@arm.com>
12500 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12502 (niter_desc): Likewise mode.
12503 (iv_analyze): Add a mode parameter.
12505 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12506 and change its type to scalar_int_mode.
12507 * loop-iv.c: Update commentary at head of file.
12508 (iv_constant): Pass the mode paraeter before the rtx it describes
12509 and change its type to scalar_int_mode. Remove VOIDmode handling.
12510 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12511 (iv_extend): Likewise.
12512 (shorten_into_mode): Likewise.
12513 (iv_add): Use scalar_int_mode.
12514 (iv_mult): Likewise.
12515 (iv_shift): Likewise.
12516 (canonicalize_iv_subregs): Likewise.
12517 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12518 it describes and change its mode to scalar_int_mode. Also change
12519 the type of the returned inner_mode to scalar_int_mode.
12520 (get_biv_step): Likewise, turning outer_mode from a pointer
12521 into a direct parameter. Update call to get_biv_step_1.
12522 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12523 iv_constant and get_biv_step.
12524 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12525 and change its type to scalar_int_mode. Don't initialise iv->mode
12526 to VOIDmode and remove later checks for its still being VOIDmode.
12527 Update calls to iv_analyze_op and iv_analyze_expr. Check
12528 is_a <scalar_int_mode> when changing the mode under consideration.
12529 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12530 Update call to iv_analyze_expr.
12531 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12532 inner register is not also a scalar_int_mode. Update call to
12534 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12535 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12536 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12537 separate mode class checks. Update calls to iv_analyze. Remove
12538 fix-up of VOIDmodes after iv_analyze_biv.
12539 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12540 don't have a scalar_int_mode. Update call to biv_p.
12542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12543 Alan Hayward <alan.hayward@arm.com>
12544 David Sherwood <david.sherwood@arm.com>
12546 * cfgexpand.c (convert_debug_memory_address): Use
12547 as_a <scalar_int_mode>.
12548 * combine.c (expand_compound_operation): Likewise.
12549 (make_extraction): Likewise.
12550 (change_zero_ext): Likewise.
12551 (simplify_comparison): Likewise.
12552 * cse.c (cse_insn): Likewise.
12553 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12554 (mem_loc_descriptor): Likewise.
12555 (loc_descriptor): Likewise.
12556 * expmed.c (init_expmed_one_mode): Likewise.
12557 (synth_mult): Likewise.
12558 (emit_store_flag_1): Likewise.
12559 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12560 of a comparison with size.
12561 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12562 (reduce_to_bit_field_precision): Likewise.
12563 * function.c (expand_function_end): Likewise.
12564 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12565 * loop-doloop.c (doloop_modify): Likewise.
12566 * optabs.c (expand_binop): Likewise.
12567 (expand_unop): Likewise.
12568 (expand_copysign_absneg): Likewise.
12569 (prepare_cmp_insn): Likewise.
12570 (maybe_legitimize_operand): Likewise.
12571 * recog.c (const_scalar_int_operand): Likewise.
12572 * rtlanal.c (get_address_mode): Likewise.
12573 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12574 (simplify_cond_clz_ctz): Likewise.
12575 * tree-nested.c (get_nl_goto_field): Likewise.
12576 * tree.c (build_vector_type_for_mode): Likewise.
12577 * var-tracking.c (use_narrower_mode): Likewise.
12579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12580 Alan Hayward <alan.hayward@arm.com>
12581 David Sherwood <david.sherwood@arm.com>
12583 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12584 * builtins.c (expand_builtin_signbit): Use it.
12585 * cfgexpand.c (expand_debug_expr): Likewise.
12586 * dojump.c (do_jump): Likewise.
12587 (do_compare_and_jump): Likewise.
12588 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12589 * expmed.c (make_tree): Likewise.
12590 * expr.c (expand_expr_real_2): Likewise.
12591 (expand_expr_real_1): Likewise.
12592 (try_casesi): Likewise.
12593 * fold-const-call.c (fold_const_call_ss): Likewise.
12594 * fold-const.c (unextend): Likewise.
12595 (extract_muldiv_1): Likewise.
12596 (fold_single_bit_test): Likewise.
12597 (native_encode_int): Likewise.
12598 (native_encode_string): Likewise.
12599 (native_interpret_int): Likewise.
12600 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12601 * internal-fn.c (expand_addsub_overflow): Likewise.
12602 (expand_neg_overflow): Likewise.
12603 (expand_mul_overflow): Likewise.
12604 (expand_arith_overflow): Likewise.
12605 * match.pd: Likewise.
12606 * stor-layout.c (layout_type): Likewise.
12607 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12608 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12609 * tree-ssanames.c (get_range_info): Likewise.
12610 * tree-switch-conversion.c (array_value_type) Likewise.
12611 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12612 (vect_recog_divmod_pattern): Likewise.
12613 (vect_recog_mixed_size_cond_pattern): Likewise.
12614 * tree-vrp.c (extract_range_basic): Likewise.
12615 (simplify_float_conversion_using_ranges): Likewise.
12616 * tree.c (int_fits_type_p): Likewise.
12617 * ubsan.c (instrument_bool_enum_load): Likewise.
12618 * varasm.c (mergeable_string_section): Likewise.
12619 (narrowing_initializer_constant_valid_p): Likewise.
12620 (output_constant): 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 * machmode.h (NARROWEST_INT_MODE): New macro.
12627 * expr.c (alignment_for_piecewise_move): Use it instead of
12628 GET_CLASS_NARROWEST_MODE (MODE_INT).
12629 (push_block): Likewise.
12630 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12632 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
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 * postreload.c (move2add_valid_value_p): Change the type of the
12639 mode parameter to scalar_int_mode.
12640 (move2add_use_add2_insn): Add a mode parameter and use it instead
12642 (move2add_use_add3_insn): Likewise.
12643 (reload_cse_move2add): Update accordingly.
12645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12646 Alan Hayward <alan.hayward@arm.com>
12647 David Sherwood <david.sherwood@arm.com>
12649 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12651 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12652 * optabs.c (expand_unop): Likewise.
12654 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12655 Alan Hayward <alan.hayward@arm.com>
12656 David Sherwood <david.sherwood@arm.com>
12658 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12659 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12660 (popcount_loc_descriptor): Likewise.
12661 (bswap_loc_descriptor): Likewise.
12662 (rotate_loc_descriptor): Likewise.
12663 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12664 calling the functions above.
12666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12667 Alan Hayward <alan.hayward@arm.com>
12668 David Sherwood <david.sherwood@arm.com>
12670 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12672 (try_combine): Likewise.
12673 (simplify_if_then_else): Likewise.
12674 * cse.c (cse_insn): Likewise.
12675 * dwarf2out.c (mem_loc_descriptor): Likewise.
12676 * emit-rtl.c (gen_lowpart_common): Likewise.
12677 * simplify-rtx.c (simplify_truncation): Likewise.
12678 (simplify_binary_operation_1): Likewise.
12679 (simplify_const_relational_operation): Likewise.
12680 (simplify_ternary_operation): Likewise.
12681 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12684 Alan Hayward <alan.hayward@arm.com>
12685 David Sherwood <david.sherwood@arm.com>
12687 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12688 * reload.c (push_reload): Likewise.
12689 (find_reloads): Likewise.
12691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12692 Alan Hayward <alan.hayward@arm.com>
12693 David Sherwood <david.sherwood@arm.com>
12695 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12696 (make_compound_operation_int): Likewise.
12697 (change_zero_ext): Likewise.
12698 * expr.c (convert_move): Likewise.
12699 (convert_modes): Likewise.
12700 * fwprop.c (forward_propagate_subreg): Likewise.
12701 * loop-iv.c (get_biv_step_1): Likewise.
12702 * optabs.c (widen_operand): Likewise.
12703 * postreload.c (move2add_valid_value_p): Likewise.
12704 * recog.c (simplify_while_replacing): Likewise.
12705 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12706 (simplify_binary_operation_1): Likewise. Remove redundant
12707 mode equality check.
12709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12710 Alan Hayward <alan.hayward@arm.com>
12711 David Sherwood <david.sherwood@arm.com>
12713 * combine.c (combine_simplify_rtx): Add checks for
12714 is_a <scalar_int_mode>.
12715 (simplify_if_then_else): Likewise.
12716 (make_field_assignment): Likewise.
12717 (simplify_comparison): Likewise.
12718 * ifcvt.c (noce_try_bitop): Likewise.
12719 * loop-invariant.c (canonicalize_address_mult): Likewise.
12720 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12723 Alan Hayward <alan.hayward@arm.com>
12724 David Sherwood <david.sherwood@arm.com>
12726 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12727 is_a <scalar_int_mode> instead of != BLKmode.
12729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12730 Alan Hayward <alan.hayward@arm.com>
12731 David Sherwood <david.sherwood@arm.com>
12733 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12734 instead of != VOIDmode.
12735 * combine.c (if_then_else_cond): Likewise.
12736 (change_zero_ext): Likewise.
12737 * dwarf2out.c (mem_loc_descriptor): Likewise.
12738 (loc_descriptor): Likewise.
12739 * rtlanal.c (canonicalize_condition): Likewise.
12740 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12743 Alan Hayward <alan.hayward@arm.com>
12744 David Sherwood <david.sherwood@arm.com>
12746 * simplify-rtx.c (simplify_binary_operation_1): Use
12747 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12750 Alan Hayward <alan.hayward@arm.com>
12751 David Sherwood <david.sherwood@arm.com>
12753 * wide-int.h (int_traits<unsigned char>) New class.
12754 (int_traits<unsigned short>) Likewise.
12755 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12756 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12758 * combine.c (set_nonzero_bits_and_sign_copies): Use
12759 is_a <scalar_int_mode>.
12760 (find_split_point): Likewise.
12761 (combine_simplify_rtx): Likewise.
12762 (simplify_logical): Likewise.
12763 (expand_compound_operation): Likewise.
12764 (expand_field_assignment): Likewise.
12765 (make_compound_operation): Likewise.
12766 (extended_count): Likewise.
12767 (change_zero_ext): Likewise.
12768 (simplify_comparison): Likewise.
12769 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12770 (ucompare_loc_descriptor): Likewise.
12771 (minmax_loc_descriptor): Likewise.
12772 (mem_loc_descriptor): Likewise.
12773 (loc_descriptor): Likewise.
12774 * expmed.c (init_expmed_one_mode): Likewise.
12775 * lra-constraints.c (lra_constraint_offset): Likewise.
12776 * optabs.c (prepare_libcall_arg): Likewise.
12777 * postreload.c (move2add_note_store): Likewise.
12778 * reload.c (operands_match_p): Likewise.
12779 * rtl.h (load_extend_op): Likewise.
12780 * rtlhooks.c (gen_lowpart_general): Likewise.
12781 * simplify-rtx.c (simplify_truncation): Likewise.
12782 (simplify_unary_operation_1): Likewise.
12783 (simplify_binary_operation_1): Likewise.
12784 (simplify_const_binary_operation): Likewise.
12785 (simplify_const_relational_operation): Likewise.
12786 (simplify_subreg): Likewise.
12787 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12788 * var-tracking.c (adjust_mems): Likewise.
12789 (prepare_call_arguments): Likewise.
12791 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12792 Alan Hayward <alan.hayward@arm.com>
12793 David Sherwood <david.sherwood@arm.com>
12795 * machmode.h (is_int_mode): New fuction.
12796 * combine.c (find_split_point): Use it.
12797 (combine_simplify_rtx): Likewise.
12798 (simplify_if_then_else): Likewise.
12799 (simplify_set): Likewise.
12800 (simplify_shift_const_1): Likewise.
12801 (simplify_comparison): Likewise.
12802 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12803 * cse.c (notreg_cost): Likewise.
12804 (cse_insn): Likewise.
12805 * cselib.c (cselib_lookup_1): Likewise.
12806 * dojump.c (do_jump_1): Likewise.
12807 (do_compare_rtx_and_jump): Likewise.
12808 * dse.c (get_call_args): Likewise.
12809 * dwarf2out.c (rtl_for_decl_init): Likewise.
12810 (native_encode_initializer): Likewise.
12811 * expmed.c (emit_store_flag_1): Likewise.
12812 (emit_store_flag): Likewise.
12813 * expr.c (convert_modes): Likewise.
12814 (store_field): Likewise.
12815 (expand_expr_real_1): Likewise.
12816 * fold-const.c (fold_read_from_constant_string): Likewise.
12817 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12818 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12819 * optabs.c (expand_binop): Likewise.
12820 (expand_unop): Likewise.
12821 (expand_abs_nojump): Likewise.
12822 (expand_one_cmpl_abs_nojump): Likewise.
12823 * simplify-rtx.c (mode_signbit_p): Likewise.
12824 (val_signbit_p): Likewise.
12825 (val_signbit_known_set_p): Likewise.
12826 (val_signbit_known_clear_p): Likewise.
12827 (simplify_relational_operation_1): Likewise.
12828 * tree.c (vector_type_mode): Likewise.
12830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12831 Alan Hayward <alan.hayward@arm.com>
12832 David Sherwood <david.sherwood@arm.com>
12834 * machmode.h (smallest_mode_for_size): Fix formatting.
12835 (smallest_int_mode_for_size): New function.
12836 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12837 instead of smallest_mode_for_size.
12838 * combine.c (make_extraction): Likewise.
12839 * config/arc/arc.c (arc_expand_movmem): Likewise.
12840 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12841 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12842 * config/s390/s390.c (s390_expand_insv): Likewise.
12843 * config/sparc/sparc.c (assign_int_registers): Likewise.
12844 * config/spu/spu.c (spu_function_value): Likewise.
12845 (spu_function_arg): Likewise.
12846 * coverage.c (get_gcov_type): Likewise.
12847 (get_gcov_unsigned_t): Likewise.
12848 * dse.c (find_shift_sequence): Likewise.
12849 * expmed.c (store_bit_field_1): Likewise.
12850 * expr.c (convert_move): Likewise.
12851 (store_field): Likewise.
12852 * internal-fn.c (expand_arith_overflow): Likewise.
12853 * optabs-query.c (get_best_extraction_insn): Likewise.
12854 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12855 * stor-layout.c (layout_type): Likewise.
12856 (initialize_sizetypes): Likewise.
12857 * targhooks.c (default_get_mask_mode): Likewise.
12858 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12861 Alan Hayward <alan.hayward@arm.com>
12862 David Sherwood <david.sherwood@arm.com>
12864 * machmode.h (opt_mode::else_blk): New function.
12865 (int_mode_for_mode): Declare.
12866 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12867 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12869 * cfgexpand.c (expand_debug_expr): Likewise.
12870 * combine.c (gen_lowpart_or_truncate): Likewise.
12871 (gen_lowpart_for_combine): Likewise.
12872 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12873 * config/avr/avr.c (avr_to_int_mode): Likewise.
12874 (avr_out_plus_1): Likewise.
12875 (avr_out_plus): Likewise.
12876 (avr_out_round): Likewise.
12877 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12878 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12879 (s390_expand_vcond): Likewise.
12880 * config/spu/spu.c (spu_split_immediate): Likewise.
12881 (spu_expand_mov): Likewise.
12882 * dse.c (get_stored_val): Likewise.
12883 * expmed.c (store_bit_field_1): Likewise.
12884 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12886 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12887 (extract_low_bits): Likewise.
12888 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12889 handling rather than repeating the check.
12890 (emit_group_store): Likewise.
12891 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12892 * optabs.c (expand_absneg_bit): Likewise.
12893 (expand_copysign_absneg): Likewise.
12894 (expand_copysign_bit): Likewise.
12895 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12896 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12897 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12898 * var-tracking.c (prepare_call_arguments): Likewise.
12899 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12900 int_mode_for_mode instead of mode_for_size.
12901 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): 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 * machmode.h (int_mode_for_size): New function.
12908 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12909 instead of mode_for_size.
12910 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12912 * combine.c (expand_field_assignment): Use int_mode_for_size
12913 instead of mode_for_size.
12914 (make_extraction): Likewise.
12915 (simplify_shift_const_1): Likewise.
12916 (simplify_comparison): Likewise.
12917 * dojump.c (do_jump): Likewise.
12918 * dwarf2out.c (mem_loc_descriptor): Likewise.
12919 * emit-rtl.c (init_derived_machine_modes): Likewise.
12920 * expmed.c (flip_storage_order): Likewise.
12921 (convert_extracted_bit_field): Likewise.
12922 * expr.c (copy_blkmode_from_reg): Likewise.
12923 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12924 * internal-fn.c (expand_mul_overflow): Likewise.
12925 * lower-subreg.c (simple_move): Likewise.
12926 * optabs-libfuncs.c (init_optabs): Likewise.
12927 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12928 * tree.c (vector_type_mode): Likewise.
12929 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12930 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12931 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12932 * tree-vect-stmts.c (vectorizable_load): Likewise.
12933 (vectorizable_store): Likewise.
12935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12936 Alan Hayward <alan.hayward@arm.com>
12937 David Sherwood <david.sherwood@arm.com>
12939 * coretypes.h (pod_mode): New type.
12940 (scalar_int_mode_pod): New typedef.
12941 * machmode.h (pod_mode): New class.
12942 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12943 * genmodes.c (emit_mode_int_n): Update accordingly.
12944 * lower-subreg.h (target_lower_subreg): Change type to
12945 scalar_int_mode_pod.
12946 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12947 scalar_int_mode_pod.
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 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12954 machine_mode to scalar_int_mode.
12955 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12956 (rs6000_option_override_internal): Remove cast to int.
12957 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12958 machine_mode to scalar_int_mode.
12959 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12960 (rs6000_option_override_internal): Remove cast to int.
12961 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12962 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12964 * config/s390/s390.c (s390_expand_builtin): Likewise.
12965 * coretypes.h (scalar_int_mode): New type.
12966 (opt_scalar_int_mode): New typedef.
12967 * machmode.h (scalar_int_mode): New class.
12968 (scalar_int_mode::includes_p): New function.
12969 (byte_mode): Change type to scalar_int_mode.
12970 (word_mode): Likewise.
12971 (ptr_mode): Likewise.
12972 * emit-rtl.c (byte_mode): Likewise.
12973 (word_mode): Likewise.
12974 (ptr_mode): Likewise.
12975 (init_derived_machine_modes): Update accordingly.
12976 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12977 and MODE_PARTIAL_INT.
12978 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12979 opt_scalar_int_mode.
12981 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12982 Alan Hayward <alan.hayward@arm.com>
12983 David Sherwood <david.sherwood@arm.com>
12985 * target.def (libgcc_floating_mode_supported_p): Take a
12987 * doc/tm.texi: Regenerate.
12988 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12990 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12991 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12995 Alan Hayward <alan.hayward@arm.com>
12996 David Sherwood <david.sherwood@arm.com>
12998 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12999 * doc/tm.texi: Regenerate.
13000 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
13001 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
13002 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
13003 * targhooks.h (default_floatn_mode): Likewise.
13004 * targhooks.c (default_floatn_mode): Likewise.
13005 * tree.c (build_common_tree_nodes): Update accordingly.
13007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13008 Alan Hayward <alan.hayward@arm.com>
13009 David Sherwood <david.sherwood@arm.com>
13011 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13012 (mode_iterator::iterate_p): Likewise.
13013 (mode_iterator::get_wider): Likewise.
13014 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13017 Alan Hayward <alan.hayward@arm.com>
13018 David Sherwood <david.sherwood@arm.com>
13020 * coretypes.h (opt_scalar_float_mode): New typedef.
13021 * machmode.h (float_mode_for_size): New function.
13022 * emit-rtl.c (double_mode): Delete.
13023 (init_emit_once): Use float_mode_for_size.
13024 * stor-layout.c (layout_type): Likewise.
13025 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13028 Alan Hayward <alan.hayward@arm.com>
13029 David Sherwood <david.sherwood@arm.com>
13031 * output.h (assemble_real): Take a scalar_float_mode.
13032 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13033 * config/arm/arm.md (consttable_4): Likewise.
13034 (consttable_8): Likewise.
13035 (consttable_16): Likewise.
13036 * config/mips/mips.md (consttable_float): Likewise.
13037 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13038 * varasm.c (assemble_real): Take a scalar_float_mode.
13039 (output_constant_pool_2): Update accordingly.
13040 (output_constant): Likewise.
13042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13043 Alan Hayward <alan.hayward@arm.com>
13044 David Sherwood <david.sherwood@arm.com>
13046 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13047 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13048 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13049 (native_encode_real): Likewise.
13050 (native_interpret_real): Likewise.
13051 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13052 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13055 Alan Hayward <alan.hayward@arm.com>
13056 David Sherwood <david.sherwood@arm.com>
13058 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13059 <scalar_float_mode>. Simplify.
13060 (gen_extend_conv_libfunc): Likewise.
13062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13063 Alan Hayward <alan.hayward@arm.com>
13064 David Sherwood <david.sherwood@arm.com>
13066 * coretypes.h (scalar_float_mode): New type.
13067 * machmode.h (mode_traits::from_int): Use machine_mode if
13068 USE_ENUM_MODES is defined.
13069 (is_a): New function.
13071 (dyn_cast): Likewise.
13072 (scalar_float_mode): New class.
13073 (scalar_float_mode::includes_p): New function.
13074 (is_float_mode): Likewise.
13075 * gdbhooks.py (MachineModePrinter): New class.
13076 (build_pretty_printer): Use it for scalar_float_mode.
13077 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13078 (format_helper::format_helper): Turn into a template.
13079 * genmodes.c (get_mode_class): New function.
13080 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13081 or machine_mode if none.
13082 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13083 as_a <scalar_float_mode>.
13084 * dwarf2out.c (mem_loc_descriptor): Likewise.
13085 (insert_float): Likewise.
13086 (add_const_value_attribute): Likewise.
13087 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13088 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13089 (expand_unop): Update accordingly.
13090 (expand_abs_nojump): Likewise.
13091 (expand_copysign_absneg): Take a scalar_float_mode.
13092 (expand_copysign_bit): Likewise.
13093 (expand_copysign): Update accordingly.
13095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13096 Alan Hayward <alan.hayward@arm.com>
13097 David Sherwood <david.sherwood@arm.com>
13099 * coretypes.h (opt_mode): New class.
13100 * machmode.h (opt_mode): Likewise.
13101 (opt_mode::else_void): New function.
13102 (opt_mode::require): Likewise.
13103 (opt_mode::exists): Likewise.
13104 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13105 (GET_MODE_2XWIDER_MODE): Likewise.
13106 (mode_iterator::get_wider): Update accordingly.
13107 (mode_iterator::get_2xwider): Likewise.
13108 (mode_iterator::get_known_wider): Likewise, turning into a template.
13109 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13110 forcing a wider mode to exist.
13111 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13112 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13113 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13114 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13115 * lower-subreg.c (init_lower_subreg): Likewise.
13116 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13117 on the final iteration.
13118 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13119 a wider mode exists before asking for a move pattern.
13120 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13121 forcing a wider mode to exist.
13122 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13123 returning false if no such mode exists.
13124 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13125 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13126 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13127 Avoid checking for a MODE_INT if we already know the mode is not a
13129 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13130 forcing a wider mode to exist.
13131 (expmed_mult_highpart_optab): Likewise.
13132 (expmed_mult_highpart): Likewise.
13133 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13135 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13136 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13137 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13138 * internal-fn.c (expand_mul_overflow): Update use of
13139 GET_MODE_2XWIDER_MODE.
13140 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13141 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13142 GET_MODE_WIDER_MODE.
13143 (convert_plusminus_to_widen): Likewise.
13144 * tree-switch-conversion.c (array_value_type): Likewise.
13145 * var-tracking.c (emit_note_insn_var_location): Likewise.
13146 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13147 Return false inside rather than outside the loop if no wider mode
13149 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13150 and GET_MODE_2XWIDER_MODE
13151 (can_compare_p): Use else_void.
13152 * gdbhooks.py (OptMachineModePrinter): New class.
13153 (build_pretty_printer): Use it for opt_mode.
13155 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13156 Alan Hayward <alan.hayward@arm.com>
13157 David Sherwood <david.sherwood@arm.com>
13159 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13160 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13163 Alan Hayward <alan.hayward@arm.com>
13164 David Sherwood <david.sherwood@arm.com>
13166 * machmode.h (mode_traits): New structure.
13167 (get_narrowest_mode): New function.
13168 (mode_iterator::start): Likewise.
13169 (mode_iterator::iterate_p): Likewise.
13170 (mode_iterator::get_wider): Likewise.
13171 (mode_iterator::get_known_wider): Likewise.
13172 (mode_iterator::get_2xwider): Likewise.
13173 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13174 (FOR_EACH_MODE): Likewise.
13175 (FOR_EACH_MODE_FROM): Likewise.
13176 (FOR_EACH_MODE_UNTIL): Likewise.
13177 (FOR_EACH_WIDER_MODE): Likewise.
13178 (FOR_EACH_2XWIDER_MODE): Likewise.
13179 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13180 * combine.c (simplify_comparison): Likewise
13181 * config/i386/i386.c (type_natural_mode): Likewise.
13182 * cse.c (cse_insn): Likewise.
13183 * dse.c (find_shift_sequence): Likewise.
13184 * emit-rtl.c (init_derived_machine_modes): Likewise.
13185 (init_emit_once): Likewise.
13186 * explow.c (hard_function_value): Likewise.
13187 * expmed.c (extract_fixed_bit_field_1): Likewise.
13188 (extract_bit_field_1): Likewise.
13189 (expand_divmod): Likewise.
13190 (emit_store_flag_1): Likewise.
13191 * expr.c (init_expr_target): Likewise.
13192 (convert_move): Likewise.
13193 (alignment_for_piecewise_move): Likewise.
13194 (widest_int_mode_for_size): Likewise.
13195 (emit_block_move_via_movmem): Likewise.
13196 (copy_blkmode_to_reg): Likewise.
13197 (set_storage_via_setmem): Likewise.
13198 (compress_float_constant): Likewise.
13199 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13200 * optabs-query.c (get_best_extraction_insn): Likewise.
13201 * optabs.c (expand_binop): Likewise.
13202 (expand_twoval_unop): Likewise.
13203 (expand_twoval_binop): Likewise.
13204 (widen_leading): Likewise.
13205 (widen_bswap): Likewise.
13206 (expand_parity): Likewise.
13207 (expand_unop): Likewise.
13208 (prepare_cmp_insn): Likewise.
13209 (prepare_float_lib_cmp): Likewise.
13210 (expand_float): Likewise.
13211 (expand_fix): Likewise.
13212 (expand_sfix_optab): Likewise.
13213 * postreload.c (move2add_use_add2_insn): Likewise.
13214 * reg-stack.c (reg_to_stack): Likewise.
13215 * reginfo.c (choose_hard_reg_mode): Likewise.
13216 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13217 * stor-layout.c (mode_for_size): Likewise.
13218 (smallest_mode_for_size): Likewise.
13219 (mode_for_vector): Likewise.
13220 (finish_bitfield_representative): Likewise.
13221 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13222 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13223 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13224 * var-tracking.c (prepare_call_arguments): Likewise.
13226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13227 Alan Hayward <alan.hayward@arm.com>
13228 David Sherwood <david.sherwood@arm.com>
13230 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13231 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13232 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13233 * machmode.h (mode_size): Move earlier in file.
13234 (mode_precision): Likewise.
13235 (mode_inner): Likewise.
13236 (mode_nunits): Likewise.
13237 (mode_unit_size): Likewise.
13238 (unit_unit_precision): Likewise.
13239 (mode_wider): Likewise.
13240 (mode_2xwider): Likewise.
13241 (machine_mode): New class.
13242 (mode_to_bytes): New function.
13243 (mode_to_bits): Likewise.
13244 (mode_to_precision): Likewise.
13245 (mode_to_inner): Likewise.
13246 (mode_to_unit_size): Likewise.
13247 (mode_to_unit_precision): Likewise.
13248 (mode_to_nunits): Likewise.
13249 (GET_MODE_SIZE): Use mode_to_bytes.
13250 (GET_MODE_BITSIZE): Use mode_to_bits.
13251 (GET_MODE_PRECISION): Use mode_to_precision.
13252 (GET_MODE_INNER): Use mode_to_inner.
13253 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13254 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13255 (GET_MODE_NUNITS): Use mode_to_nunits.
13256 * system.h (ALWAYS_INLINE): New macro.
13257 * config/powerpcspe/powerpcspe-c.c
13258 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13259 int for arg1_mode and arg2_mode.
13261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13262 Alan Hayward <alan.hayward@arm.com>
13263 David Sherwood <david.sherwood@arm.com>
13265 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13266 Prefix mode names with E_ in case statements.
13267 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13268 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13269 (aarch64_split_simd_move): Likewise.
13270 (aarch64_gen_storewb_pair): Likewise.
13271 (aarch64_gen_loadwb_pair): Likewise.
13272 (aarch64_gen_store_pair): Likewise.
13273 (aarch64_gen_load_pair): Likewise.
13274 (aarch64_get_condition_code_1): Likewise.
13275 (aarch64_constant_pool_reload_icode): Likewise.
13276 (get_rsqrte_type): Likewise.
13277 (get_rsqrts_type): Likewise.
13278 (get_recpe_type): Likewise.
13279 (get_recps_type): Likewise.
13280 (aarch64_gimplify_va_arg_expr): Likewise.
13281 (aarch64_simd_container_mode): Likewise.
13282 (aarch64_emit_load_exclusive): Likewise.
13283 (aarch64_emit_store_exclusive): Likewise.
13284 (aarch64_expand_compare_and_swap): Likewise.
13285 (aarch64_gen_atomic_cas): Likewise.
13286 (aarch64_emit_bic): Likewise.
13287 (aarch64_emit_atomic_swap): Likewise.
13288 (aarch64_emit_atomic_load_op): Likewise.
13289 (aarch64_evpc_trn): Likewise.
13290 (aarch64_evpc_uzp): Likewise.
13291 (aarch64_evpc_zip): Likewise.
13292 (aarch64_evpc_ext): Likewise.
13293 (aarch64_evpc_rev): Likewise.
13294 (aarch64_evpc_dup): Likewise.
13295 (aarch64_gen_ccmp_first): Likewise.
13296 (aarch64_gen_ccmp_next): Likewise.
13297 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13298 (alpha_emit_xfloating_libcall): Likewise.
13299 (emit_insxl): Likewise.
13300 (alpha_arg_type): Likewise.
13301 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13302 (arc_preferred_simd_mode): Likewise.
13303 (arc_secondary_reload): Likewise.
13304 (get_arc_condition_code): Likewise.
13305 (arc_print_operand): Likewise.
13306 (arc_legitimate_constant_p): Likewise.
13307 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13308 * config/arc/arc.md (casesi_load): Likewise.
13309 (casesi_compact_jump): Likewise.
13310 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13311 (cc_use_register): Likewise.
13312 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13313 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13314 (arm_init_iwmmxt_builtins): Likewise.
13315 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13316 (neon_expand_vector_init): Likewise.
13317 (arm_attr_length_move_neon): Likewise.
13318 (maybe_get_arm_condition_code): Likewise.
13319 (arm_emit_vector_const): Likewise.
13320 (arm_preferred_simd_mode): Likewise.
13321 (arm_output_iwmmxt_tinsr): Likewise.
13322 (thumb1_output_casesi): Likewise.
13323 (thumb2_output_casesi): Likewise.
13324 (arm_emit_load_exclusive): Likewise.
13325 (arm_emit_store_exclusive): Likewise.
13326 (arm_expand_compare_and_swap): Likewise.
13327 (arm_evpc_neon_vuzp): Likewise.
13328 (arm_evpc_neon_vzip): Likewise.
13329 (arm_evpc_neon_vrev): Likewise.
13330 (arm_evpc_neon_vtrn): Likewise.
13331 (arm_evpc_neon_vext): Likewise.
13332 (arm_validize_comparison): Likewise.
13333 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13334 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13335 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13336 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13337 (c6x_preferred_simd_mode): Likewise.
13338 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13339 (epiphany_rtx_costs): Likewise.
13340 * config/epiphany/predicates.md (proper_comparison_operator):
13342 * config/frv/frv.c (condexec_memory_operand): Likewise.
13343 (frv_emit_move): Likewise.
13344 (output_move_single): Likewise.
13345 (output_condmove_single): Likewise.
13346 (frv_hard_regno_mode_ok): Likewise.
13347 (frv_matching_accg_mode): Likewise.
13348 * config/h8300/h8300.c (split_adds_subs): Likewise.
13349 (h8300_rtx_costs): Likewise.
13350 (h8300_print_operand): Likewise.
13351 (compute_mov_length): Likewise.
13352 (output_logical_op): Likewise.
13353 (compute_logical_op_length): Likewise.
13354 (compute_logical_op_cc): Likewise.
13355 (h8300_shift_needs_scratch_p): Likewise.
13356 (output_a_shift): Likewise.
13357 (compute_a_shift_length): Likewise.
13358 (compute_a_shift_cc): Likewise.
13359 (expand_a_rotate): Likewise.
13360 (output_a_rotate): Likewise.
13361 * config/i386/i386.c (classify_argument): Likewise.
13362 (function_arg_advance_32): Likewise.
13363 (function_arg_32): Likewise.
13364 (function_arg_64): Likewise.
13365 (function_value_64): Likewise.
13366 (ix86_gimplify_va_arg): Likewise.
13367 (ix86_legitimate_constant_p): Likewise.
13368 (put_condition_code): Likewise.
13369 (split_double_mode): Likewise.
13370 (ix86_avx256_split_vector_move_misalign): Likewise.
13371 (ix86_expand_vector_logical_operator): Likewise.
13372 (ix86_split_idivmod): Likewise.
13373 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13374 (ix86_build_const_vector): Likewise.
13375 (ix86_build_signbit_mask): Likewise.
13376 (ix86_match_ccmode): Likewise.
13377 (ix86_cc_modes_compatible): Likewise.
13378 (ix86_expand_branch): Likewise.
13379 (ix86_expand_sse_cmp): Likewise.
13380 (ix86_expand_sse_movcc): Likewise.
13381 (ix86_expand_int_sse_cmp): Likewise.
13382 (ix86_expand_vec_perm_vpermi2): Likewise.
13383 (ix86_expand_vec_perm): Likewise.
13384 (ix86_expand_sse_unpack): Likewise.
13385 (ix86_expand_int_addcc): Likewise.
13386 (ix86_split_to_parts): Likewise.
13387 (ix86_vectorize_builtin_gather): Likewise.
13388 (ix86_vectorize_builtin_scatter): Likewise.
13389 (avx_vpermilp_parallel): Likewise.
13390 (inline_memory_move_cost): Likewise.
13391 (ix86_tieable_integer_mode_p): Likewise.
13392 (x86_maybe_negate_const_int): Likewise.
13393 (ix86_expand_vector_init_duplicate): Likewise.
13394 (ix86_expand_vector_init_one_nonzero): Likewise.
13395 (ix86_expand_vector_init_one_var): Likewise.
13396 (ix86_expand_vector_init_concat): Likewise.
13397 (ix86_expand_vector_init_interleave): Likewise.
13398 (ix86_expand_vector_init_general): Likewise.
13399 (ix86_expand_vector_set): Likewise.
13400 (ix86_expand_vector_extract): Likewise.
13401 (emit_reduc_half): Likewise.
13402 (ix86_emit_i387_round): Likewise.
13403 (ix86_mangle_type): Likewise.
13404 (ix86_expand_round_sse4): Likewise.
13405 (expand_vec_perm_blend): Likewise.
13406 (canonicalize_vector_int_perm): Likewise.
13407 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13408 (expand_vec_perm_1): Likewise.
13409 (expand_vec_perm_interleave3): Likewise.
13410 (expand_vec_perm_even_odd_pack): Likewise.
13411 (expand_vec_perm_even_odd_1): Likewise.
13412 (expand_vec_perm_broadcast_1): Likewise.
13413 (ix86_vectorize_vec_perm_const_ok): Likewise.
13414 (ix86_expand_vecop_qihi): Likewise.
13415 (ix86_expand_mul_widen_hilo): Likewise.
13416 (ix86_expand_sse2_abs): Likewise.
13417 (ix86_expand_pextr): Likewise.
13418 (ix86_expand_pinsr): Likewise.
13419 (ix86_preferred_simd_mode): Likewise.
13420 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13421 * config/i386/sse.md (*andnot<mode>3): Likewise.
13422 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13423 (*<code><mode>3): Likewise.
13424 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13425 (ia64_expand_atomic_op): Likewise.
13426 (ia64_arg_type): Likewise.
13427 (ia64_mode_to_int): Likewise.
13428 (ia64_scalar_mode_supported_p): Likewise.
13429 (ia64_vector_mode_supported_p): Likewise.
13430 (expand_vec_perm_broadcast): Likewise.
13431 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13432 (iq2000_function_arg_advance): Likewise.
13433 (iq2000_function_arg): Likewise.
13434 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13435 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13436 (m68k_libcall_value): Likewise.
13437 (m68k_function_value): Likewise.
13438 (sched_attr_op_type): Likewise.
13439 * config/mcore/mcore.c (mcore_output_move): Likewise.
13440 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13442 (microblaze_function_arg): Likewise.
13443 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13444 (mips_print_operand): Likewise.
13445 (mips_mode_ok_for_mov_fmt_p): Likewise.
13446 (mips_vector_mode_supported_p): Likewise.
13447 (mips_preferred_simd_mode): Likewise.
13448 (mips_expand_vpc_loongson_even_odd): Likewise.
13449 (mips_expand_vec_unpack): Likewise.
13450 (mips_expand_vi_broadcast): Likewise.
13451 (mips_expand_vector_init): Likewise.
13452 (mips_expand_vec_reduc): Likewise.
13453 (mips_expand_msa_cmp): Likewise.
13454 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13455 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13456 (cc_flags_for_mode): Likewise.
13457 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13458 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13459 (nds32_output_casesi_pc_relative): Likewise.
13460 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13461 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13462 (nvptx_gen_unpack): Likewise.
13463 (nvptx_gen_pack): Likewise.
13464 (nvptx_gen_shuffle): Likewise.
13465 (nvptx_gen_wcast): Likewise.
13466 (nvptx_preferred_simd_mode): Likewise.
13467 * config/pa/pa.c (pa_secondary_reload): Likewise.
13468 * config/pa/predicates.md (base14_operand): Likewise.
13469 * config/powerpcspe/powerpcspe-c.c
13470 (altivec_resolve_overloaded_builtin): Likewise.
13471 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13473 (rs6000_preferred_simd_mode): Likewise.
13474 (output_vec_const_move): Likewise.
13475 (rs6000_expand_vector_extract): Likewise.
13476 (rs6000_split_vec_extract_var): Likewise.
13477 (reg_offset_addressing_ok_p): Likewise.
13478 (rs6000_legitimate_offset_address_p): Likewise.
13479 (rs6000_legitimize_address): Likewise.
13480 (rs6000_emit_set_const): Likewise.
13481 (rs6000_const_vec): Likewise.
13482 (rs6000_emit_move): Likewise.
13483 (spe_build_register_parallel): Likewise.
13484 (rs6000_darwin64_record_arg_recurse): Likewise.
13485 (swap_selector_for_mode): Likewise.
13486 (spe_init_builtins): Likewise.
13487 (paired_init_builtins): Likewise.
13488 (altivec_init_builtins): Likewise.
13489 (do_load_for_compare): Likewise.
13490 (rs6000_generate_compare): Likewise.
13491 (rs6000_expand_float128_convert): Likewise.
13492 (emit_load_locked): Likewise.
13493 (emit_store_conditional): Likewise.
13494 (rs6000_output_function_epilogue): Likewise.
13495 (rs6000_handle_altivec_attribute): Likewise.
13496 (rs6000_function_value): Likewise.
13497 (emit_fusion_gpr_load): Likewise.
13498 (emit_fusion_p9_load): Likewise.
13499 (emit_fusion_p9_store): Likewise.
13500 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13501 (fusion_gpr_mem_load): Likewise.
13502 (fusion_addis_mem_combo_load): Likewise.
13503 (fusion_addis_mem_combo_store): Likewise.
13504 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13505 (fusion_gpr_mem_load): Likewise.
13506 (fusion_addis_mem_combo_load): Likewise.
13507 (fusion_addis_mem_combo_store): Likewise.
13508 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13510 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13511 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13512 (rs6000_preferred_simd_mode): Likewise.
13513 (output_vec_const_move): Likewise.
13514 (rs6000_expand_vector_extract): Likewise.
13515 (rs6000_split_vec_extract_var): Likewise.
13516 (reg_offset_addressing_ok_p): Likewise.
13517 (rs6000_legitimate_offset_address_p): Likewise.
13518 (rs6000_legitimize_address): Likewise.
13519 (rs6000_emit_set_const): Likewise.
13520 (rs6000_const_vec): Likewise.
13521 (rs6000_emit_move): Likewise.
13522 (rs6000_darwin64_record_arg_recurse): Likewise.
13523 (swap_selector_for_mode): Likewise.
13524 (paired_init_builtins): Likewise.
13525 (altivec_init_builtins): Likewise.
13526 (rs6000_expand_float128_convert): Likewise.
13527 (emit_load_locked): Likewise.
13528 (emit_store_conditional): Likewise.
13529 (rs6000_output_function_epilogue): Likewise.
13530 (rs6000_handle_altivec_attribute): Likewise.
13531 (rs6000_function_value): Likewise.
13532 (emit_fusion_gpr_load): Likewise.
13533 (emit_fusion_p9_load): Likewise.
13534 (emit_fusion_p9_store): Likewise.
13535 * config/rx/rx.c (rx_gen_move_template): Likewise.
13536 (flags_from_mode): Likewise.
13537 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13538 (s390_slb_comparison): Likewise.
13539 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13540 (s390_vector_mode_supported_p): Likewise.
13541 (s390_cc_modes_compatible): Likewise.
13542 (s390_match_ccmode_set): Likewise.
13543 (s390_canonicalize_comparison): Likewise.
13544 (s390_emit_compare_and_swap): Likewise.
13545 (s390_branch_condition_mask): Likewise.
13546 (s390_rtx_costs): Likewise.
13547 (s390_secondary_reload): Likewise.
13548 (__SECONDARY_RELOAD_CASE): Likewise.
13549 (s390_expand_cs): Likewise.
13550 (s390_preferred_simd_mode): Likewise.
13551 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13552 * config/sh/sh.c (sh_print_operand): Likewise.
13553 (dump_table): Likewise.
13554 (sh_secondary_reload): Likewise.
13555 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13556 * config/sh/sh.md (casesi_worker_1): Likewise.
13557 (casesi_worker_2): Likewise.
13558 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13559 (fcc_comparison_operator): Likewise.
13560 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13561 (emit_soft_tfmode_cvt): Likewise.
13562 (sparc_preferred_simd_mode): Likewise.
13563 (output_cbranch): Likewise.
13564 (sparc_print_operand): Likewise.
13565 (sparc_expand_vec_perm_bmask): Likewise.
13566 (vector_init_bshuffle): Likewise.
13567 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13568 (spu_vector_mode_supported_p): Likewise.
13569 (spu_expand_insv): Likewise.
13570 (spu_emit_branch_or_set): Likewise.
13571 (spu_handle_vector_attribute): Likewise.
13572 (spu_builtin_splats): Likewise.
13573 (spu_builtin_extract): Likewise.
13574 (spu_builtin_promote): Likewise.
13575 (spu_expand_sign_extend): Likewise.
13576 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13577 (tilegx_simd_int): Likewise.
13578 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13579 (tilepro_simd_int): Likewise.
13580 * config/v850/v850.c (const_double_split): Likewise.
13581 (v850_print_operand): Likewise.
13582 (ep_memory_offset): Likewise.
13583 * config/vax/vax.c (vax_rtx_costs): Likewise.
13584 (vax_output_int_move): Likewise.
13585 (vax_output_int_add): Likewise.
13586 (vax_output_int_subtract): Likewise.
13587 * config/visium/predicates.md (visium_branch_operator): Likewise.
13588 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13589 (visium_print_operand_address): Likewise.
13590 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13591 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13592 (xtensa_expand_conditional_branch): Likewise.
13593 (xtensa_copy_incoming_a7): Likewise.
13594 (xtensa_output_literal): Likewise.
13595 * dfp.c (decimal_real_maxval): Likewise.
13596 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13599 Alan Hayward <alan.hayward@arm.com>
13600 David Sherwood <david.sherwood@arm.com>
13602 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13603 (mode_nunits_inline): Likewise.
13604 (mode_inner_inline): Likewise.
13605 (mode_unit_size_inline): Likewise.
13606 (mode_unit_precision_inline): Likewise.
13607 (emit_insn_modes_h): Likewise. Also emit a #define of the
13609 (emit_mode_wider): Add an E_ prefix to mode names.
13610 (emit_mode_complex): Likewise.
13611 (emit_mode_inner): Likewise.
13612 (emit_mode_adjustments): Likewise.
13613 (emit_mode_int_n): Likewise.
13614 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13615 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13616 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13617 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13618 (CRC32_BUILTIN, ENTRY): Likewise.
13619 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13620 (aarch64_pop_regs): Likewise.
13621 (aarch64_process_components): Likewise.
13622 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13623 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13624 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13625 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13626 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13627 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13629 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13630 (pcmpistr): Likewise.
13631 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13632 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13633 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13635 * config/rl78/rl78.c (mduc_regs): Likewise.
13636 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13637 (htm_expand_builtin): Likewise.
13638 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13639 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13640 * config/xtensa/xtensa.c (print_operand): Likewise.
13641 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13642 (NUM_MODE_VECTOR_INT): Likewise.
13643 * genoutput.c (null_operand): Likewise.
13644 (output_operand_data): Likewise.
13645 * genrecog.c (print_parameter_value): Likewise.
13646 * lra.c (debug_operand_data): Likewise.
13648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13650 * dwarf2out.c (scompare_loc_descriptor_wide)
13651 (scompare_loc_descriptor_narrow): New functions, split out from...
13652 (scompare_loc_descriptor): ...here.
13653 * expmed.c (emit_store_flag_int): New function, split out from...
13654 (emit_store_flag): ...here.
13656 2017-08-30 Richard Biener <rguenther@suse.de>
13658 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13659 (dwarf2out_early_finish): Move setting of AT_pubnames from
13660 early debug output to early finish.
13662 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13664 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13665 and -mdata-region to the assembler.
13667 2017-08-30 Richard Biener <rguenther@suse.de>
13669 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13671 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13672 (dwarf2out_early_global_decl): Only generate a DIE for the
13673 abstract origin if it doesn't already exist or is a declaration DIE.
13674 (resolve_addr): Do not add the linkage name twice when
13675 generating a stub DIE for the DW_TAG_GNU_call_site target.
13677 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13679 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13680 Use machine_mode rather than int for arg1_mode.
13682 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13685 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13686 that the second argument of the built-in functions to unpack
13687 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13688 switch statement instead a lot of if statements.
13689 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13690 Allow 64-bit values to be in Altivec registers as well as
13691 traditional floating point registers.
13692 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13694 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13696 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13697 MAX_REGS_PER_ADDRESS == 1.
13699 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13701 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13702 * config/i386/i386.c (ix86_option_override_internal): Simplify
13703 setting of opts->x_flag_entry.
13705 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13706 Jakub Jelinek <jakub@redhat.com>
13707 Richard Biener <rguenther@suse.de>
13709 PR tree-optimization/81503
13710 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13711 folded constant fits in the target type; reorder tests for clarity.
13713 2017-08-29 Martin Liska <mliska@suse.cz>
13715 * passes.def: Include pass_lower_switch.
13716 * stmt.c (dump_case_nodes): Remove and move to
13717 tree-switch-conversion.
13718 (case_values_threshold): Likewise.
13719 (expand_switch_as_decision_tree_p): Likewise.
13720 (emit_case_decision_tree): Likewise.
13721 (expand_case): Likewise.
13722 (balance_case_nodes): Likewise.
13723 (node_has_low_bound): Likewise.
13724 (node_has_high_bound): Likewise.
13725 (node_is_bounded): Likewise.
13726 (emit_case_nodes): Likewise.
13727 (struct simple_case_node): New struct.
13728 (add_case_node): Remove.
13729 (emit_case_dispatch_table): Use vector instead of case_list.
13730 (reset_out_edges_aux): Remove.
13731 (compute_cases_per_edge): Likewise.
13732 (expand_case): Build list of simple_case_node.
13733 (expand_sjlj_dispatch_table): Use it.
13734 * tree-switch-conversion.c (struct case_node): Moved from
13735 stmt.c and adjusted.
13736 (emit_case_nodes): Likewise.
13737 (node_has_low_bound): Likewise.
13738 (node_has_high_bound): Likewise.
13739 (node_is_bounded): Likewise.
13740 (case_values_threshold): Likewise.
13741 (reset_out_edges_aux): Likewise.
13742 (compute_cases_per_edge): Likewise.
13743 (add_case_node): Likewise.
13744 (dump_case_nodes): Likewise.
13745 (balance_case_nodes): Likewise.
13746 (expand_switch_as_decision_tree_p): Likewise.
13747 (emit_jump): Likewise.
13748 (emit_case_decision_tree): Likewise.
13749 (try_switch_expansion): Likewise.
13750 (do_jump_if_equal): Likewise.
13751 (emit_cmp_and_jump_insns): Likewise.
13752 (fix_phi_operands_for_edge): New function.
13753 (record_phi_operand_mapping): Likewise.
13754 (class pass_lower_switch): New pass.
13755 (pass_lower_switch::execute): New function.
13756 (make_pass_lower_switch): Likewise.
13757 (conditional_probability):
13758 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13759 * tree-pass.h: Add make_pass_lower_switch.
13761 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13764 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13767 2017-08-29 Richard Biener <rguenther@suse.de>
13769 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13770 we do not add a DW_AT_inline attribute twice.
13771 (gen_subprogram_die): Remove code setting DW_AT_inline on
13772 DECL_ABSTRACT_P nodes.
13774 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13776 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13777 calls to internal functions.
13778 (gimplify_modify_expr): Likewise.
13779 * tree-call-cdce.c (use_internal_fn): Likewise.
13780 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13781 (convert_to_divmod): Set the nothrow flag.
13782 * tree-if-conv.c (predicate_mem_writes): Likewise.
13783 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13784 (vectorizable_call): Likewise.
13785 (vectorizable_store): Likewise.
13786 (vectorizable_load): Likewise.
13787 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13788 (vect_recog_mask_conversion_pattern): Likewise.
13790 2017-08-29 Martin Liska <mliska@suse.cz>
13793 * gcc.c (driver_handle_option): Add new argument.
13794 * opts-common.c (handle_option): Pass
13795 target_option_override_hook.
13796 * opts-global.c (lang_handle_option): Add new option.
13797 (set_default_handlers): Add new argument.
13798 (decode_options): Likewise.
13799 * opts.c (target_handle_option): Likewise.
13800 (common_handle_option): Call target_option_override_hook.
13801 * opts.h (struct cl_option_handler_func): Add hook for
13802 target option override.
13803 (struct cl_option_handlers): Likewise.
13804 (set_default_handlers): Add new argument.
13805 (decode_options): Likewise.
13806 (common_handle_option): Likewise.
13807 (target_handle_option): Likewise.
13808 * toplev.c (toplev::main): Pass targetm.target_option.override
13811 2017-08-29 Richard Biener <rguenther@suse.de>
13812 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13814 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13815 life to the active subtree.
13817 2017-08-28 Jeff Law <law@redhat.com>
13819 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13820 derive_equivalences.
13821 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13822 Code moved into....
13823 (edge_info::derive_equivalences): New private member function
13825 * tree-ssa-dom.c (class edge_info): Changed from a struct
13826 to a class. Add ctor/dtor, methods and data members.
13827 (edge_info::edge_info): Renamed from allocate_edge_info.
13828 Initialize additional members.
13829 (edge_info::~edge_info): New.
13830 (free_dom_edge_info): Delete the edge info.
13831 (record_edge_info): Use new class & associated member functions.
13832 Tighten forms for testing for edge equivalences.
13833 (record_temporary_equivalences): Iterate over the simple
13834 equivalences rather than assuming there's only one per edge.
13835 (cprop_into_successor_phis): Iterate over the simple
13836 equivalences rather than assuming there's only one per edge.
13837 (optimize_stmt): Use operand_equal_p rather than pointer
13838 equality for mini-DSE code.
13840 2017-08-28 Nathan Sidwell <nathan@acm.org>
13842 * gcc.c (execute): Fold SIGPIPE handling into switch
13843 statement. Adjust internal error message.
13845 2017-08-28 Richard Biener <rguenther@suse.de>
13848 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13849 Do nothing for removed DIEs.
13851 2017-08-28 Richard Biener <rguenther@suse.de>
13853 PR tree-optimization/81977
13854 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13857 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13860 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13861 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13862 using targetm.gen_mem_thread_fence.
13864 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13867 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13868 predicate to register_operand. Reorder operands.
13869 (*btr<mode>): Ditto.
13870 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13871 (*btr<mode>_mask): Ditto.
13873 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13875 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13876 * config/rs6000/xmmintrin.h: New file.
13877 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13879 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13882 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13883 parameter and_insn and return it.
13884 (recombine_lvx_pattern): Insert a copy to ensure availability of
13885 the base register of the copied masking operation at the point of
13886 the instruction replacement.
13887 (recombine_stvx_pattern): Likewise.
13889 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13891 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13892 undocumented switches.
13893 (-mpower9-dform-vector): Likewise.
13894 (-mpower9-dform): Likewise.
13895 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13896 comments to delete references to -mpower9-dform* switches.
13897 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13898 Delete reference to -mpower9-dform* switches, test for
13899 -mpower9-vector instead.
13900 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13901 (OTHER_P9_VECTOR_MASKS): Likewise.
13902 (POWERPC_MASKS): Likewise.
13903 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13904 tests against -mpower9-dform* to -mpower9-vector. Delete code
13905 that checked for -mpower9-dform* consistancy with other options.
13906 Add test for -mpower9-misc to enable other power9 switches.
13907 (rs6000_init_hard_regno_mode_ok): Likewise.
13908 (rs6000_option_override_internal): Likewise.
13909 (rs6000_emit_prologue): Likewise.
13910 (rs6000_emit_epilogue): Likewise.
13911 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13912 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13913 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13915 (emit_fusion_p9_store): Likewise.
13916 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13917 resetting these macros if the assembler does not support ISA 3.0
13919 (TARGET_P9_DFORM_VECTOR): Likewise.
13920 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13921 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13923 2017-08-25 Alan Modra <amodra@gmail.com>
13925 PR rtl-optimization/81747
13926 * cse.c (cse_extended_basic_block): Don't attempt to record
13927 equivalences for degenerate conditional branches that branch
13928 to their fall-through.
13930 2017-08-24 Martin Sebor <msebor@redhat.com>
13932 PR middle-end/81908
13933 * gimple-fold.c (size_must_be_zero_p): New function.
13934 (gimple_fold_builtin_memory_op): Call it.
13936 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13938 * config/rs6000/mm_malloc.h: New file.
13940 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13942 PR tree-optimization/81913
13943 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13944 analysis when either IVs in condition can wrap.
13946 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13948 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13949 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13951 2017-08-24 Richard Biener <rguenther@suse.de>
13954 * targhooks.c (default_target_can_inline_p): Properly
13955 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13956 is present and always compare.
13957 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13958 infer -mfpmath=sse from TARGET_SSE_P.
13959 (ix86_can_inline_p): Properly use target_option_default_node when
13960 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13962 2017-08-24 Richard Biener <rguenther@suse.de>
13965 * dwarf2out.c (output_die): Handle flag_generate_offload like
13967 (output_comp_unit): Likewise.
13968 (gen_array_type_die): Likewise.
13969 (dwarf2out_early_finish): Likewise.
13970 (note_variable_value_in_expr): Likewise.
13971 (dwarf2out_finish): Likewise. Adjust assert.
13972 * cgraphunit.c (symbol_table::compile): Move setting of
13973 flag_generate_offload earlier ...
13974 (symbol_table::finalize_compilation_unit): ... here, before
13975 early debug finalization.
13977 2017-08-24 Richard Biener <rguenther@suse.de>
13979 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13980 and ipa-fnsummary.h.
13981 (ix86_can_inline_p): When ix86_fpmath flags do not match
13982 check whether the callee uses FP math at all.
13984 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13986 PR middle-end/81931
13987 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13988 instead of TYPE_PRECISION.
13990 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13991 Alan Hayward <alan.hayward@arm.com>
13992 David Sherwood <david.sherwood@arm.com>
13994 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13996 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13998 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
14000 (ix86_frame::stack_realign_allocate): New field.
14001 (struct machine_frame_state): Modify comments.
14002 (machine_frame_state::sp_realigned_fp_end): New field.
14003 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
14004 layout calculation.
14005 (sp_valid_at): Add assertion to assure no attempt to access invalid
14006 offset of a realigned stack.
14007 (fp_valid_at): Likewise.
14008 (choose_baseaddr): Modify comments.
14009 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14010 ix86_expand_prologue.
14011 (ix86_expand_prologue): Modify stack realignment and allocation.
14012 (ix86_expand_epilogue): Modify comments.
14013 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14014 avx2_runtime, avx512f, and avx512f_runtime.
14016 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14018 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14019 (mstackrealign): Do not init to -1.
14020 * config/i386/i386.c (ix86_option_override_internal):
14021 Check opts_set, not opts when setting default value of
14022 opts->x_ix86_force_align_arg_pointer.
14024 2017-08-23 Richard Biener <rguenther@suse.de>
14026 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14027 lang_hooks.decl_printable_name.
14028 * print-rtl-function.c (print_rtx_function): Likewise.
14029 * tree-pretty-print.c (dump_function_header): Likewise.
14031 2017-08-23 Richard Biener <rguenther@suse.de>
14034 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14035 -g0 at compile-time.
14037 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14039 PR middle-end/19706
14040 * doc/sourcebuild.texi (Other hardware attributes):
14043 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14045 PR middle-end/19706
14046 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14047 Fix single-use check.
14049 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14051 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14053 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14055 * doc/install.texi: Modify to add more details on running selected
14058 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14060 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14061 is combined with -mabi=ms.
14062 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14065 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14067 PR tree-optimization/81488
14068 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14069 and cached_basis fields.
14070 (MAX_SPREAD): New constant.
14071 (alloc_cand_and_find_basis): Initialize new fields.
14072 (clear_visited): New function.
14073 (create_phi_basis_1): Rename from create_phi_basis, set visited
14074 and cached_basis fields.
14075 (create_phi_basis): New wrapper function.
14076 (phi_add_costs_1): Rename from phi_add_costs, add spread
14077 parameter, set visited field, short-circuit when limits reached.
14078 (phi_add_costs): New wrapper function.
14079 (record_phi_increments_1): Rename from record_phi_increments, set
14081 (record_phi_increments): New wrapper function.
14082 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14083 (phi_incr_cost): New wrapper function.
14084 (all_phi_incrs_profitable_1): Rename from
14085 all_phi_incrs_profitable, set visited field.
14086 (all_phi_incrs_profitable): New wrapper function.
14088 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14089 Alan Hayward <alan.hayward@arm.com>
14090 David Sherwood <david.sherwood@arm.com>
14092 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14093 that takes the outer and inner modes.
14094 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14095 comparison as the canonical test for a paradoxical subreg.
14096 * combine.c (simplify_set): Use paradoxical_subreg_p.
14097 (make_extraction): Likewise.
14098 (force_to_mode): Likewise.
14099 (rtx_equal_for_field_assignment_p): Likewise.
14100 (gen_lowpart_for_combine): Likewise.
14101 (simplify_comparison): Likewise.
14102 * cse.c (equiv_constant): Likewise.
14103 * expmed.c (store_bit_field_1): Likewise.
14104 * final.c (alter_subreg): Likewise.
14105 * fwprop.c (propagate_rtx): Likewise.
14106 (forward_propagate_subreg): Likewise.
14107 * ira-conflicts.c (ira_build_conflicts): Likewise.
14108 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14109 * lra-constraints.c (curr_insn_transform): Likewise.
14110 (split_reg): Likewise.
14111 * lra-eliminations.c (move_plus_up): Likewise.
14112 (lra_eliminate_regs_1): Likewise.
14113 * recog.c (general_operand): Likewise.
14114 * ree.c (combine_reaching_defs): Likewise.
14115 * reload.c (push_reload): Likewise.
14116 (find_reloads): Likewise.
14117 * reload1.c (elimination_effects): Likewise.
14118 (compute_reload_subreg_offset): Likewise.
14119 (choose_reload_regs): Likewise.
14120 * rtlanal.c (subreg_lsb_1): Likewise.
14121 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14122 (simplify_subreg): Likewise.
14123 * var-tracking.c (track_loc_p): Likewise.
14124 * emit-rtl.c (byte_lowpart_offset): Likewise.
14125 (paradoxical_subreg_p): Delete out-of-line definition.
14127 2017-08-22 Jeff Law <law@redhat.com>
14129 PR tree-optimization/81741
14130 PR tree-optimization/71947
14131 * tree-ssa-dom.c: Include tree-inline.h.
14132 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14133 equivalences if one is more expensive to compute than the other.
14134 * tree-ssa-scopedtables.h (class const_or_copies): Make
14135 record_const_or_copy_raw method private.
14136 (class avail_exprs_stack): New method simplify_binary_operation.
14137 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14138 avail_exprs_stack::simplify_binary_operation as needed.
14139 (avail_exprs_stack::simplify_binary_operation): New function.
14141 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14143 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14144 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14145 (DOT_SYMBOLS): Likewise.
14146 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14147 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14148 (RS6000_ABI_NAME): Likewise.
14149 (TARGET_CMODEL): Likewise.
14150 (TOC_SECTION_ASM_OP): Likewise.
14151 (SET_CMODEL): New macro.
14152 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14154 2017-08-22 Richard Biener <rguenther@suse.de>
14156 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14157 to allow for free-lang-data replacements similar to verify_type_variant.
14159 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14160 Alan Hayward <alan.hayward@arm.com>
14161 David Sherwood <david.sherwood@arm.com>
14163 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14164 VOIDmode for the LABEL_REF.
14166 2017-08-22 Richard Biener <rguenther@suse.de>
14168 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14170 2017-08-22 Richard Biener <rguenther@suse.de>
14172 * common.opt (feliminate-dwarf2-dups): Ignore.
14173 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14174 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14175 same_die_p_wrap, compute_section_prefix,
14176 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14177 (comdat_symbol_id, comdat_symbol_number): Likewise.
14178 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14180 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14181 (output_die): Mark unreachable path unreachable.
14182 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14183 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14184 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14185 (dwarf2out_early_finish): Likewise.
14187 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14189 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14191 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14194 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14195 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14196 (avr_attribute_table) <io, io_low, address>: Initialize
14197 .decl_required with true.
14199 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14201 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14202 undocumented debugging options.
14203 (-mvsx-scalar-double): Likewise.
14204 (-mallow-df-permute): Likewise.
14205 (-mvectorize-builtins): Likewise.
14206 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14207 (rs6000_builtin_vectorized_function): Likewise.
14208 (rs6000_builtin_md_vectorized_function): Likewise.
14209 (rs6000_opt_vars): Likewise.
14211 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14214 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14215 (*btrq_imm): Rename from *btrq.
14216 (*btcq_imm): Rename from *btcq.
14217 (btsc): New code attribute.
14218 (*<btsc><mode>): New insn pattern.
14219 (*btr<mode>): Ditto.
14220 (*<btsc><mode>_mask): New insn_and_split pattern.
14221 (*btr<mode>_mask): Ditto.
14223 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14224 Alan Hayward <alan.hayward@arm.com>
14225 David Sherwood <david.sherwood@arm.com>
14227 * function.c (pad_below): Simplify padding calculation.
14229 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14230 Alan Hayward <alan.hayward@arm.com>
14231 David Sherwood <david.sherwood@arm.com>
14233 * target.def (function_prologue): Remove frame size argument.
14234 (function_epilogue): Likewise.
14235 * doc/tm.texi: Regenerate.
14236 * final.c (final_start_function): Update call to function_prologue.
14237 (final_end_function): Update call to function_epilogue.
14238 (default_function_pro_epilogue): Remove frame size argument.
14239 * output.h (default_function_pro_epilogue): Likewise.
14240 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14241 (arm_output_function_prologue): Likewise.
14242 * config/frv/frv.c (frv_function_prologue): Likewise.
14243 (frv_function_epilogue): Likewise.
14244 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14245 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14246 (ia64_output_function_epilogue): Likewise.
14247 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14248 (m32r_output_function_epilogue): Likewise.
14249 * config/microblaze/microblaze.c (microblaze_function_prologue)
14250 (microblaze_function_epilogue): Likewise.
14251 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14252 (mips_output_function_epilogue): Likewise.
14253 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14254 (mmix_target_asm_function_epilogue): Likewise.
14255 * config/msp430/msp430.c (msp430_start_function): Likewise.
14256 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14257 (nds32_asm_function_epilogue): Likewise.
14258 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14259 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14260 (pa_output_function_epilogue): Likewise.
14261 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14262 (rs6000_output_function_epilogue): Likewise.
14263 * config/rl78/rl78.c (rl78_start_function): Likewise.
14264 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14265 (rs6000_output_function_epilogue): Likewise.
14266 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14267 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14268 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14269 (sparc_asm_function_epilogue): Likewise.
14271 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14273 * tree.h (type_has_mode_precision_p): New function.
14274 * convert.c (convert_to_integer_1): Use it.
14275 * expr.c (expand_expr_real_2): Likewise.
14276 (expand_expr_real_1): Likewise.
14277 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14278 * match.pd: Likewise.
14279 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14280 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14281 * tree-tailcall.c (process_assignment): Likewise.
14282 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14283 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14284 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14285 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14286 (vectorizable_assignment): Likewise.
14287 (vectorizable_shift): Likewise.
14288 (vectorizable_operation): Likewise.
14289 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14291 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14293 * match.pd: Add pow (C, x) simplification.
14295 2017-08-21 Richard Biener <rguenther@suse.de>
14297 PR tree-optimization/81900
14298 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14299 for blocks with abnormal predecessors.
14300 (compute_antic): Do not set visited flag prematurely.
14302 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14305 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14307 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14309 * stor-layout.h (vector_type_mode): Move to...
14310 * tree.h (vector_type_mode): ...here.
14311 * stor-layout.c (vector_type_mode): Move to...
14312 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14314 2017-08-21 Richard Biener <rguenther@suse.de>
14316 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14317 register_external_die hooks.
14318 (debug_false_tree_charstarstar_uhwistar): Declare.
14319 (debug_nothing_tree_charstar_uhwi): Likewise.
14320 * debug.c (do_nothing_debug_hooks): Adjust.
14321 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14322 (debug_nothing_tree_charstar_uhwi): Likewise.
14323 * dbxout.c (dbx_debug_hooks): Adjust.
14324 (xcoff_debug_hooks): Likewise.
14325 * sdbout.c (sdb_debug_hooks): Likewise.
14326 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14327 * dwarf2out.c (macinfo_label_base): New global.
14328 (dwarf2out_register_external_die): New function for the
14329 register_external_die hook.
14330 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14331 (dwarf2_debug_hooks): Use them.
14332 (dwarf2_lineno_debug_hooks): Adjust.
14333 (struct die_struct): Add with_offset flag.
14334 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14335 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14336 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14337 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14338 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14339 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14340 defining section names for the early LTO debug variants.
14341 (reset_indirect_string): New helper.
14342 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14343 (print_dw_val): Add support for offsetted symbol references.
14344 (get_ultimate_context): Split out from is_cxx.
14345 (is_cxx): Use get_ultimate_context.
14346 (is_fortran): Add decl overload.
14347 (compute_comp_unit_symbol): Split out worker from
14348 compute_section_prefix.
14349 (compute_section_prefix): Call compute_comp_unit_symbol and
14350 set comdat_type_p here.
14351 (output_die): Skip DIE symbol output for the LTO added one.
14352 Handle DIE symbol references with offset.
14353 (output_comp_unit): Guard section name mangling properly.
14354 For LTO debug sections emit a symbol at the section beginning
14355 which we use to refer to its DIEs.
14356 (add_abstract_origin_attribute): For DIEs registered via
14357 dwarf2out_register_external_die directly refer to the early
14358 DIE rather than indirectly through the shadow one we created.
14359 Remove obsolete call to dwarf2out_abstract_function for
14360 non-function/block origins.
14361 (gen_array_type_die): When generating early LTO debug do
14362 not emit DW_AT_string_length.
14363 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14364 late when in LTO. As suggested place a gcc_unreachable for
14365 the DECL_ABSTRACT_P case.
14366 (gen_subprogram_die): Avoid another specification DIE
14367 for early built declarations/definitions for the late LTO case.
14368 (gen_variable_die): Add type references for late duplicated VLA dies
14370 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14371 we have the abstract instance already.
14372 (process_scope_var): Adjust decl DIE contexts in LTO which
14373 first puts them in limbo.
14374 (gen_decl_die): Do not generate type DIEs late apart from
14375 types for VLAs or for decls we do not yet have a DIE. Do not
14376 call dwarf2out_abstract_function late.
14377 (dwarf2out_early_global_decl): Make sure to create DIEs
14378 for abstract instances of a decl first.
14379 (dwarf2out_late_global_decl): Adjust comment.
14380 (output_macinfo_op): With multiple macro sections use
14381 macinfo_label_base to distinguish labels.
14382 (output_macinfo): Likewise. Update macinfo_label_base.
14383 Pass in the line info label.
14384 (note_variable_value_in_expr): When generating LTO resolve
14385 all variable values here by generating DIEs as needed.
14386 (init_sections_and_labels): Add early LTO debug flag parameter
14387 and generate different sections and names if set. Add generation
14388 counter for the labels so we can have multiple of them.
14389 (reset_dies): Helper to allow DIEs to be output multiple times.
14390 (dwarf2out_finish): When outputting DIEs to the fat part of an
14391 LTO object first reset DIEs.
14392 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14393 (modified_type_die): Check for decl_ultimate_origin being self
14395 (gen_type_die_with_usage): Likewise.
14396 (gen_typedef_die): Allow decl_ultimate_origin being self.
14397 (set_decl_abstract_flags): Remove.
14398 (set_block_abstract_flags): Likewise.
14399 (dwarf2out_abstract_function): Treat the early generated DIEs
14400 as the abstract copy and only add DW_AT_inline and
14401 DW_AT_artificial here and call set_decl_origin_self.
14402 If the DIE has an abstract origin don't do anything.
14403 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14404 if we have none yet (Go fails to build one, PR78628).
14405 (variably_modified_type_p): Prevent endless recursion for Ada
14406 cyclic pointer types.
14407 * lto-streamer-in.c: Include debug.h.
14408 (dref_queue): New global.
14409 (lto_read_tree_1): Stream in DIE references.
14410 (lto_input_tree): Register DIE references.
14411 (input_function): Stream DECL_DEBUG_ARGS.
14412 * lto-streamer-out.c: Include debug.h.
14413 (lto_write_tree_1): Output DIE references.
14414 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14415 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14416 (output_function): Stream DECL_DEBUG_ARGS.
14417 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14418 Stream DECL_ABSTRACT_ORIGIN.
14419 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14420 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14421 DECL_CONTEXT for file-scope decls.
14422 * lto-streamer.h (struct dref_entry): Declare.
14423 (dref_queue): Likewise.
14424 * cfgexpand.c (pass_expand::execute): Do not call the
14425 outlining_inline_function hook here.
14426 * lto-wrapper.c (debug_obj): New global.
14427 (tool_cleanup): Unlink it if required.
14428 (debug_objcopy): New function.
14429 (run_gcc): Handle early debug sections in the IL files by
14430 extracting them to separate files, partially linkin them and
14431 feeding the result back as result to the linker.
14432 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14433 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14434 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14435 sections into a separate segment.
14436 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14438 (darwin_asm_dwarf_section): Likewise.
14439 (darwin_asm_output_dwarf_offset): Likewise.
14440 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14442 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14443 Alan Hayward <alan.hayward@arm.com>
14444 David Sherwood <david.sherwood@arm.com>
14446 * read-md.h (md_reader::record_potential_iterator_use): Replace
14447 pointer argument with an rtx and an index.
14448 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14449 (apply_mode_iterator): Likewise.
14450 (apply_code_iterator): Likewise.
14451 (apply_int_iterator): Likewise.
14452 (apply_subst_iterator): Likewise.
14453 (record_iterator_use): Likewise.
14454 (record_attribute_use): Likewise.
14455 (md_reader::record_potential_iterator_use): Likewise. Update calls
14456 to record_iterator_use and apply_iterator.
14457 (iterator_use): Replace ptr with x and index.
14458 (attribute_use): Likewise.
14459 (apply_attribute_uses): Update calls to apply_iterator.
14460 (apply_iterators): Likewise. Update initialization of iterator_use.
14461 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14462 and record_potential_iterator_use.
14463 (rtx_reader::read_rtx_operand): Likewise.
14465 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14466 Alan Hayward <alan.hayward@arm.com>
14467 David Sherwood <david.sherwood@arm.com>
14469 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14472 2017-08-21 Richard Biener <rguenther@suse.de>
14474 PR middle-end/81884
14475 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14476 at struct end conservatively when comparing common bases.
14478 2017-08-21 Richard Biener <rguenther@suse.de>
14480 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14481 (mem_ref_in_stmt): Remove.
14482 (determine_max_movement): Use ref index to get at the reference.
14483 (invariantness_dom_walker::before_dom_children): Deal with
14484 lim data already initialized.
14485 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14487 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14489 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14490 (TARGET_ISA_ROUND): Ditto.
14491 (TARGET_ROUND): Ditto.
14492 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14493 * config/i386/i386.md: Ditto.
14494 * config/i386/sse.md: Ditto.
14495 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14496 with OPTION_MASK_ISA_SSE4_1.
14498 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14501 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14502 __builtin_ia32_lzcnt_u16.
14504 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14507 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14508 (rs6000_set_current_function): Rewrite function to use it.
14510 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14513 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14514 and TYPE_WARN_IF_NOT_ALIGN.
14515 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14516 (handle_warn_if_not_align): New.
14517 (place_union_field): Call handle_warn_if_not_align.
14518 (place_field): Call handle_warn_if_not_align.
14519 Copy TYPE_WARN_IF_NOT_ALIGN.
14520 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14521 (layout_type): Likewise.
14522 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14524 (tree_decl_common): Add warn_if_not_align.
14525 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14526 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14527 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14528 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14529 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14530 * doc/extend.texi: Document warn_if_not_aligned attribute.
14531 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14533 2017-08-17 Martin Liska <mliska@suse.cz>
14536 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14537 (get_data_dependence): Use it as pointer type.
14538 (distribute_loop): Likewise.
14540 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14542 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14543 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14544 (p8_vmrgow_<mode>_direct): New define_insn.
14545 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14546 handle endianness for vmrgew and vmrgow permute patterns.
14548 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14550 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14551 * config/rs6000/rs6000-cpus.def: Remove comment.
14552 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14553 (POWERPC_MASKS): Likewise.
14554 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14555 use of TARGET_VSX_TIMODE.
14556 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14557 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14558 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14559 (rs6000_option_override_internal): Remove dead code.
14560 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14561 (rs6000_legitimize_reload_address): Likewise.
14562 (rs6000_legitimate_address_p): Likewise.
14563 (rs6000_opt_masks): Delete "vsx-timode".
14564 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14565 from function comment.
14566 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14567 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14568 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14570 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14571 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14572 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14573 (VSX_TI): Likewise.
14575 (define_peephole2): Likewise.
14577 2017-08-17 Martin Sebor <msebor@redhat.com>
14580 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14581 past the end of an array.
14582 (test_pp_format): Add test cases.
14584 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14586 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14587 missing ECF_NOTHROW flags.
14589 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14592 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14593 operands residing in integer registers.
14594 (*vsx_le_perm_load_<mode>): Likewise.
14595 (*vsx_le_perm_store_<mode>): Likewise.
14596 (define_peephole2): Add peepholes to optimize the above.
14598 2017-08-17 Marek Polacek <polacek@redhat.com>
14600 PR middle-end/81814
14601 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14602 to mimic what shorten_compare did. Change the return type to bool.
14603 (fold_cond_expr_with_comparison): Update call to
14604 operand_equal_for_comparison_p.
14605 (fold_ternary_loc): Likewise.
14607 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14609 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14611 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14613 2017-08-17 Richard Biener <rguenther@suse.de>
14615 * tree-ssa-structalias.c (solve_graph): When propagating
14616 to successors update the graphs succ edges and avoid duplicate work.
14618 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14621 * config/i386/i386.c (ix86_option_override_internal): Save target
14622 specific options after ix86_stack_protector_guard_reg was changed.
14624 2017-08-17 Richard Biener <rguenther@suse.de>
14626 PR tree-optimization/81827
14627 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14628 (new_var_info): Initialize it conservatively.
14629 (get_call_vi): Mark register vars.
14630 (new_scalar_tmp_constraint_exp): Likewise.
14631 (handle_rhs_call): Likewise.
14632 (handle_const_call): Likewise.
14633 (create_function_info_for): Likewise.
14634 (solve_constraints): Sort varinfos to separate register from
14635 non-register vars to pack points-to solution bitmaps during
14638 2017-08-17 Marek Polacek <polacek@redhat.com>
14640 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14642 2017-08-17 Richard Biener <rguenther@suse.de>
14644 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14645 to true when overflow is undefined and we saturated the result.
14647 2017-08-17 Alan Modra <amodra@gmail.com>
14650 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14651 Don't use store multiple if only one reg needs saving.
14652 (interesting_frame_related_regno): New function.
14653 (rs6000_frame_related): Don't emit frame info for regs that
14655 (rs6000_emit_epilogue): Likewise.
14657 2017-08-16 Nathan Sidwell <nathan@acm.org>
14659 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14660 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14661 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14662 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14663 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14664 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14665 (verify_type): Adjust for TYPE_BINFO move.
14666 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14667 process TYPE_BINFO directly.
14668 (hash_tree): Likewise.
14669 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14671 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14674 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14676 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14678 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14681 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14682 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14683 Add ix86_binary_operator_ok to insn constraint.
14684 (*iordi_1_bts): Ditto.
14685 (*xordi_1_btc): Ditto.
14686 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14687 Update corresponding peephole2 pattern.
14691 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14693 PR tree-optimization/81832
14694 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14695 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14697 2017-08-16 Marek Polacek <polacek@redhat.com>
14700 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14701 regarding size_zero_node.
14703 2017-08-16 Martin Liska <mliska@suse.cz>
14706 * config.gcc: Respect previously set extra_objs in case
14709 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14711 PR tree-optimization/81835
14712 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14713 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14714 not depend on the phi.
14716 2017-08-16 Alan Modra <amodra@gmail.com>
14718 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14721 2017-08-16 Alan Modra <amodra@gmail.com>
14723 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14724 (save_reg_p): ..into this. Update all callers.
14725 (first_reg_to_save): Simplify.
14727 2017-08-16 Alan Modra <amodra@gmail.com>
14729 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14732 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14736 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14737 constant count if that count is less than 32.
14739 2017-08-15 Nathan Sidwell <nathan@acm.org>
14741 * gcc.c (execute): Emit friendlier message if inferior is killed
14742 by an external cause.
14744 2017-08-15 Richard Biener <rguenther@suse.de>
14746 PR tree-optimization/81790
14747 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14748 CONSTRUCTORs from simplifying and VN.
14750 2017-08-14 Martin Sebor <msebor@redhat.com>
14752 * builtin-attrs.def: Add comments.
14754 2017-08-14 Martin Sebor <msebor@redhat.com>
14757 * doc/extend.texi (attribute nonstring): Document new attribute.
14759 2017-08-14 Martin Sebor <msebor@redhat.com>
14762 * tree-diagnostic.c (default_tree_printer): Handle %G.
14763 * gimple-pretty-print.h (percent_G_format): Declare new function.
14764 * gimple-pretty-print.c (percent_G_format): Define.
14765 * tree-pretty-print.c (percent_K_format): Add argument.
14767 2017-08-14 Martin Sebor <msebor@redhat.com>
14769 PR translation/79998
14770 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14771 Remove a stray space.
14773 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14776 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14777 (*iordi_1_bts): Ditto.
14778 (*xordi_1_btc): Ditto.
14780 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14783 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14784 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14786 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14787 quoted strings, and make more translator-friendly.
14788 (darwin_rs6000_override_options): Likewise.
14789 (rs6000_option_override_internal): Likewise.
14790 (rs6000_return_in_memory): Fix overlong line.
14791 (init_cmulative_args): Use quoted strings, and make more
14792 translator-friendly.
14793 (rs6000_pass_by_reference): Fix overlong line.
14794 (def_builtin): Use quoted strings.
14795 (altivec_expand_predicate_builtin): Use quoted strings, and make
14796 more translator-friendly.
14797 (htm_expand_builtin): Use quoted strings.
14798 (cpu_expand_builtin): Use quoted strings, and make more
14799 translator-friendly.
14800 (altivec_expand_builtin): Likewise.
14801 (paired_expand_predicate_builtin): Likewise.
14802 (rs6000_invalid_builtin): Likewise.
14803 (builtin_function_type): Use quoted strings.
14804 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14805 more translator-friendly.
14806 (rs6000_trampoline_init): Likewise.
14807 (rs6000_handle_altivec_attribute): Likewise.
14808 (rs6000_inner_target_options): Use quoted strings.
14809 (rs6000_disable_incompatible_switches): Likewise.
14810 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14811 strings, and make more translator-friendly.
14812 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14814 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14816 PR tree-optimization/81799
14817 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14818 cond_expr to simple gimple operand.
14820 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14822 PR middle-end/46932
14823 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14825 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14829 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14830 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14831 TARGET_GASISR_PROLOGUES.
14832 * config/avr/avr.c (avr_option_override): Same.
14833 (avr_pass_pre_proep::execute): Same.
14835 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14838 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14839 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14841 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14843 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14844 define_insn_and_split. Split to a memory load from 0 in
14845 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14846 using PTR mode iterator.
14847 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14848 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14849 (*add_tp_<mode>): Redefine as define_insn_and_split.
14850 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14851 address space. Merge with *add_tp_x32 using PTR mode iterator.
14852 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14853 Split to an add with a memory load from 0 in
14854 DEFAULT_TLS_SEG_REG address space.
14856 2017-08-12 Andrew Pinski <apinski@cavium.com>
14858 * config/aarch64/aarch64-option-extensions.def (rdma):
14859 Fix feature string to what Linux prints out in /proc/cpuinfo.
14861 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14864 * dwarf2out.c (modified_type_die): For C typedef types that have
14865 an ultimate origin, process the ultimate origin instead of the
14867 (gen_typedef_die): Assert that input DECLs have no ultimate
14869 (gen_type_die_with_usage): For typedef variants that have an
14870 ultimate origin, just call gen_decl_die on the original DECL.
14871 (process_scope_var): Avoid creating DIEs for local typedefs and
14872 concrete static variables.
14874 2017-08-12 Alan Modra <amodra@gmail.com>
14878 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14879 match gnu-user.h startfile.
14880 (ENDFILE_LINUX_SPEC): Similarly.
14882 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14885 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14887 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14889 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14890 * config/aarch64/aarch64.md (mov<mode>): Change.
14891 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14892 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14893 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14895 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14897 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14898 for storage order barriers.
14900 2017-08-11 Martin Liska <mliska@suse.cz>
14903 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14904 variables of void type.
14906 2017-08-11 Martin Liska <mliska@suse.cz>
14908 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14909 TARGET_SUPPORTS_ALIASES.
14910 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14911 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14912 (optimize_weakref): Likewise.
14913 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14914 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14915 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14917 2017-08-11 Martin Liska <mliska@suse.cz>
14920 * config/i386/i386.c (make_resolver_func): Do complete
14921 refactoring of the function.
14923 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14926 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14927 * config/i386/i386.c (ix86_stack_protect_guard): Use
14928 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14929 * doc/invoke.texi (x86 Options): Document
14930 -mstack-protector-guard-symbol= option.
14932 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14934 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14935 * config/i386/i386.c (ix86_split_stack_guard): New function.
14936 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14937 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14938 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14939 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14940 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14941 (split_stack_space_check): Call ix86_split_stack_guard.
14943 2017-08-10 Martin Sebor <msebor@redhat.com>
14945 * print-tree.c (print_node): Print location using the established
14947 Replace spaces with colons.
14948 (debug_raw, debug): Ditto.
14950 2017-08-10 Martin Sebor <msebor@redhat.com>
14953 * pretty-print.c (pp_format): Correct the handling of %s precision.
14955 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14958 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14960 (ix86_finalize_stack_frame_flags): This. Also clear
14961 frame_pointer_needed if -fno-omit-frame-pointer is used without
14963 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14964 with ix86_finalize_stack_frame_flags.
14965 (ix86_expand_epilogue): Likewise.
14966 (ix86_expand_split_stack_prologue): Likewise.
14967 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14969 2017-08-10 Martin Liska <mliska@suse.cz>
14972 * c-attribs.c (handle_target_attribute):
14973 Report warning for an empty string argument of target attribute.
14975 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14978 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14980 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14981 or DECL_NONLOCAL labels.
14982 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14983 or DECL_NONLOCAL labels here.
14985 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14987 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14988 to indicate when early gimple folding has been disabled.
14989 (rs6000_gimple_fold_builtin): Add debug content.
14990 (rs6000_invalid_builtin): Fix whitespace.
14991 (rs6000_expand_builtin): Fix whitespace.
14992 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14994 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14997 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14998 SAVE_MULTIPLE if not all the registers that saves, should be saved.
15000 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
15002 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
15003 (qdf24xx): Likewise.
15004 * config/aarch64/aarch64.md: Include falkor.md.
15005 * config/aarch64/falkor.md: New.
15007 2017-08-09 Marek Polacek <polacek@redhat.com>
15010 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15011 * diagnostic.c (emit_diagnostic): Add a comment.
15012 (emit_diagnostic_valist): New function.
15014 2017-08-09 Marek Polacek <polacek@redhat.com>
15017 * input.c (make_location): New overload.
15018 * input.h (make_location): Declare.
15020 2017-08-08 Alan Modra <amodra@gmail.com>
15021 H.J. Lu <hongjiu.lu@intel.com>
15024 * gcc.c (NO_PIE_SPEC): Delete.
15025 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15027 (LINK_PIE_SPEC): ..to here.
15028 (LINK_COMMAND_SPEC): Support -no-pie.
15029 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15030 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15032 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15033 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15034 (ENDFILE_CRTEND_SPEC): Similarly.
15036 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15039 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15040 (mstack-protector-guard-offset=): Ditto.
15041 * config/i386/i386.c (ix86_option_override): Handle
15042 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15044 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15045 ix86_stack_protect_guard_offset variables.
15046 (TARGET_STACK_PROTECT_GUARD): Always define.
15047 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15048 and -mstack-protector-guard-offset= options.
15050 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15052 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15053 boundary case for the last candidate.
15055 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15057 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15058 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15060 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15062 PR middle-end/19706
15063 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15064 * config/aarch64/aarch64-builtins.c
15065 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15066 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15067 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15069 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15070 Andrew Pinski <pinskia@gmail.com>
15072 PR middle-end/19706
15073 * internal-fn.def (XORSIGN): New.
15074 * optabs.def (xorsign_optab): New.
15075 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15076 (convert_expand_mult_copysign): New.
15077 (pass_optimize_widening_mul::execute): Call
15078 convert_expand_mult_copysign.
15080 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15082 PR tree-optimization/81354
15083 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15084 Insert on edges rather than explicitly creating landing pads.
15085 (analyze_candidates_and_replace): Commit edge inserts.
15087 2017-08-08 Richard Biener <rguenther@suse.de>
15089 PR middle-end/81719
15090 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15091 (expand_simple_operations): Also look through ADDR_EXPRs with
15092 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15094 2017-08-08 Richard Biener <rguenther@suse.de>
15096 PR tree-optimization/81723
15097 * tree-vect-slp.c (struct bst_traits): New hash traits.
15098 (bst_fail): New global.
15099 (vect_build_slp_tree_2): New worker, split out from ...
15100 (vect_build_slp_tree): ... this now wrapping it with using
15101 bst_fail set to cache SLP tree build fails. Properly handle
15103 (vect_analyze_slp_instance): Allocate and free bst_fail.
15105 2017-08-08 Martin Liska <mliska@suse.cz>
15108 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15109 LABEL_DECLs that can be from a different function.
15111 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15113 PR tree-optimization/81744
15114 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15115 loop's number of iterations.
15117 2017-08-08 Martin Liska <mliska@suse.cz>
15119 * asan.c: Include header files.
15120 * attribs.c (build_decl_attribute_variant): New function moved
15122 (build_type_attribute_qual_variant): Likewise.
15123 (cmp_attrib_identifiers): Likewise.
15124 (simple_cst_list_equal): Likewise.
15125 (omp_declare_simd_clauses_equal): Likewise.
15126 (attribute_value_equal): Likewise.
15127 (comp_type_attributes): Likewise.
15128 (build_type_attribute_variant): Likewise.
15129 (lookup_ident_attribute): Likewise.
15130 (remove_attribute): Likewise.
15131 (merge_attributes): Likewise.
15132 (merge_type_attributes): Likewise.
15133 (merge_decl_attributes): Likewise.
15134 (merge_dllimport_decl_attributes): Likewise.
15135 (handle_dll_attribute): Likewise.
15136 (attribute_list_equal): Likewise.
15137 (attribute_list_contained): Likewise.
15138 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15139 (lookup_attribute_by_prefix): Likewise.
15140 * bb-reorder.c: Include header files.
15141 * builtins.c: Likewise.
15142 * calls.c: Likewise.
15143 * cfgexpand.c: Likewise.
15144 * cgraph.c: Likewise.
15145 * cgraphunit.c: Likewise.
15146 * convert.c: Likewise.
15147 * dwarf2out.c: Likewise.
15148 * final.c: Likewise.
15149 * fold-const.c: Likewise.
15150 * function.c: Likewise.
15151 * gimple-expr.c: Likewise.
15152 * gimple-fold.c: Likewise.
15153 * gimple-pretty-print.c: Likewise.
15154 * gimple.c: Likewise.
15155 * gimplify.c: Likewise.
15156 * hsa-common.c: Likewise.
15157 * hsa-gen.c: Likewise.
15158 * internal-fn.c: Likewise.
15159 * ipa-chkp.c: Likewise.
15160 * ipa-cp.c: Likewise.
15161 * ipa-devirt.c: Likewise.
15162 * ipa-fnsummary.c: Likewise.
15163 * ipa-inline.c: Likewise.
15164 * ipa-visibility.c: Likewise.
15166 * lto-cgraph.c: Likewise.
15167 * omp-expand.c: Likewise.
15168 * omp-general.c: Likewise.
15169 * omp-low.c: Likewise.
15170 * omp-offload.c: Likewise.
15171 * omp-simd-clone.c: Likewise.
15172 * opts-global.c: Likewise.
15173 * passes.c: Likewise.
15174 * predict.c: Likewise.
15175 * sancov.c: Likewise.
15176 * sanopt.c: Likewise.
15177 * symtab.c: Likewise.
15178 * toplev.c: Likewise.
15179 * trans-mem.c: Likewise.
15180 * tree-chkp.c: Likewise.
15181 * tree-eh.c: Likewise.
15182 * tree-into-ssa.c: Likewise.
15183 * tree-object-size.c: Likewise.
15184 * tree-parloops.c: Likewise.
15185 * tree-profile.c: Likewise.
15186 * tree-ssa-ccp.c: Likewise.
15187 * tree-ssa-live.c: Likewise.
15188 * tree-ssa-loop.c: Likewise.
15189 * tree-ssa-sccvn.c: Likewise.
15190 * tree-ssa-structalias.c: Likewise.
15191 * tree-ssa.c: Likewise.
15192 * tree-streamer-in.c: Likewise.
15193 * tree-vectorizer.c: Likewise.
15194 * tree-vrp.c: Likewise.
15195 * tsan.c: Likewise.
15196 * ubsan.c: Likewise.
15197 * varasm.c: Likewise.
15198 * varpool.c: Likewise.
15199 * tree.c: Remove functions moved to attribs.[ch].
15200 * tree.h: Likewise.
15201 * config/aarch64/aarch64.c: Add attrs.h header file.
15202 * config/alpha/alpha.c: Likewise.
15203 * config/arc/arc.c: Likewise.
15204 * config/arm/arm.c: Likewise.
15205 * config/avr/avr.c: Likewise.
15206 * config/bfin/bfin.c: Likewise.
15207 * config/c6x/c6x.c: Likewise.
15208 * config/cr16/cr16.c: Likewise.
15209 * config/cris/cris.c: Likewise.
15210 * config/darwin.c: Likewise.
15211 * config/epiphany/epiphany.c: Likewise.
15212 * config/fr30/fr30.c: Likewise.
15213 * config/frv/frv.c: Likewise.
15214 * config/ft32/ft32.c: Likewise.
15215 * config/h8300/h8300.c: Likewise.
15216 * config/i386/winnt.c: Likewise.
15217 * config/ia64/ia64.c: Likewise.
15218 * config/iq2000/iq2000.c: Likewise.
15219 * config/lm32/lm32.c: Likewise.
15220 * config/m32c/m32c.c: Likewise.
15221 * config/m32r/m32r.c: Likewise.
15222 * config/m68k/m68k.c: Likewise.
15223 * config/mcore/mcore.c: Likewise.
15224 * config/microblaze/microblaze.c: Likewise.
15225 * config/mips/mips.c: Likewise.
15226 * config/mmix/mmix.c: Likewise.
15227 * config/mn10300/mn10300.c: Likewise.
15228 * config/moxie/moxie.c: Likewise.
15229 * config/msp430/msp430.c: Likewise.
15230 * config/nds32/nds32-isr.c: Likewise.
15231 * config/nds32/nds32.c: Likewise.
15232 * config/nios2/nios2.c: Likewise.
15233 * config/nvptx/nvptx.c: Likewise.
15234 * config/pa/pa.c: Likewise.
15235 * config/pdp11/pdp11.c: Likewise.
15236 * config/powerpcspe/powerpcspe.c: Likewise.
15237 * config/riscv/riscv.c: Likewise.
15238 * config/rl78/rl78.c: Likewise.
15239 * config/rx/rx.c: Likewise.
15240 * config/s390/s390.c: Likewise.
15241 * config/sh/sh.c: Likewise.
15242 * config/sol2.c: Likewise.
15243 * config/sparc/sparc.c: Likewise.
15244 * config/spu/spu.c: Likewise.
15245 * config/stormy16/stormy16.c: Likewise.
15246 * config/tilegx/tilegx.c: Likewise.
15247 * config/tilepro/tilepro.c: Likewise.
15248 * config/v850/v850.c: Likewise.
15249 * config/vax/vax.c: Likewise.
15250 * config/visium/visium.c: Likewise.
15251 * config/xtensa/xtensa.c: Likewise.
15253 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15256 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15257 constraints since the -mupper-regs-* switches have been
15259 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15260 into a vector from a double word element that was extracted from
15261 another vector, and eliminate extra XXPERMDI instructions.
15262 (vsx_concat_<mode>_2): Likewise.
15263 (vsx_concat_<mode>_3): Likewise.
15264 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15265 concat to allow optimizing inserts from previous extracts.
15267 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15269 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15270 memory reference to a SSP offset in TLS address space.
15271 (ix86_print_operand) <case '@'>: Remove.
15272 (ix86_print_operand_punct_valid_p): Remove '@' code.
15273 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15274 UNSPEC_SP_TLS_TEST.
15275 (stack_tls_protect_set_<mode>): Remove.
15276 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15277 (stack_tls_protect_test_<mode>): Remove.
15278 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15280 2017-08-07 Olivier Hainque <hainque@adacore.com>
15283 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15285 2017-08-07 Douglas Rupp <rupp@adacore.com>
15287 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15288 variable was referenced as multidir in command.
15290 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15293 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15296 2017-08-07 Martin Liska <mliska@suse.cz>
15298 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15299 * config/rl78/rl78.c: Add include of attribs.h.
15300 * config/sh/sh.c: Likewise.
15301 * config/v850/v850.c: Likewise.
15303 2017-08-07 Tom de Vries <tom@codesourcery.com>
15305 PR middle-end/78266
15306 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15308 2017-08-07 Martin Liska <mliska@suse.cz>
15310 * config/mips/mips.c: Include attribs.h.
15312 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15315 * doc/invoke.texi: Document change in behvaior for -Ofast for
15318 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15320 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15322 (aarch64_pop_regs): Likewise.
15323 (aarch64_gen_load_pair): Likewise.
15324 (aarch64_save_callee_saves): Likewise.
15325 (aarch64_restore_callee_saves): Likewise.
15327 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15329 * config/i386/i386.c: Revert the last change.
15331 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15334 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15336 (ix86_finalize_stack_frame_flags): This. Also clear
15337 frame_pointer_needed if -fno-omit-frame-pointer is used without
15339 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15340 with ix86_finalize_stack_frame_flags.
15341 (ix86_expand_epilogue): Likewise.
15342 (ix86_expand_split_stack_prologue): Likewise.
15344 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15347 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15348 to P_AES for Westmere.
15350 2017-08-07 Jonathan Yong <10walls@gmail.com>
15352 * config/i386/mingw.opt (fset-stack-executable): Removed.
15353 * config/i386/cygming.opt (fset-stack-executable): Moved
15355 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15357 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15359 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15361 2017-08-07 Marek Polacek <polacek@redhat.com>
15363 PR middle-end/81737
15364 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15366 2017-08-07 Martin Liska <mliska@suse.cz>
15368 * attribs.h (canonicalize_attr_name): New function.
15369 (cmp_attribs): Move from c-format.c and adjusted.
15370 (is_attribute_p): Moved from tree.h.
15371 * tree-inline.c: Add new includes.
15372 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15373 (private_is_attribute_p): Remove.
15374 (private_lookup_attribute): Likewise.
15375 (private_lookup_attribute_by_prefix): Simplify.
15376 (remove_attribute): Use is_attribute_p.
15377 * tree.h: Remove removed declarations.
15379 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15381 PR middle-end/81698
15382 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15383 instead of computing it in the function. Formatting fix.
15384 (expand_case): Don't rely on default_edge being the first edge,
15385 clear it if removing it, pass default_edge to
15386 emit_case_dispatch_table.
15387 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15390 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15392 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15393 insn in the function, emit NOP after the insn.
15395 2017-08-06 Tom de Vries <tom@codesourcery.com>
15397 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15400 2017-08-06 Tom de Vries <tom@codesourcery.com>
15402 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15405 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15407 PR tree-optimization/57371
15408 * match.pd: New pattern.
15410 2017-08-04 Marek Polacek <polacek@redhat.com>
15412 PR middle-end/81695
15413 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15414 perform the computation in offset_int.
15416 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15418 PR tree-optimization/81136
15419 * tree-vectorizer.h: Include tree-hash-traits.h.
15420 (vec_base_alignments): New typedef.
15421 (vec_info): Add a base_alignments field.
15422 (vect_record_base_alignments): Declare.
15423 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15425 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15426 (create_data_ref): Add an is_conditional_in_stmt argument.
15427 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15428 the is_conditional_in_stmt field.
15429 (data_ref_loc): Add an is_conditional_in_stmt field.
15430 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15431 (find_data_references_in_stmt): Update call to create_data_ref.
15432 (graphite_find_data_references_in_stmt): Likewise.
15433 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15434 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15435 (vect_record_base_alignment): New function.
15436 (vect_record_base_alignments): Likewise.
15437 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15438 for nested statements even if we fail to compute a misalignment.
15439 Use pooled base alignments for unconditional references.
15440 (vect_find_same_alignment_drs): Compare base addresses instead
15442 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15443 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15445 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15447 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15448 Add an explicit name for the enum. Use auto_vec for slp_instances
15449 and grouped_stores.
15450 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15452 (_bb_vec_info): Add a constructor and destructor.
15453 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15454 (destroy_loop_vec_info): Delete.
15455 (vect_destroy_datarefs): Likewise.
15456 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15457 (vec_info::vec_info): New function.
15458 (vec_info::~vec_info): Likewise.
15459 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15460 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15461 destroy_loop_vec_info.
15462 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15463 (_loop_vec_info::_loop_vec_info): ...this.
15464 (destroy_loop_vec_info): Replace with...
15465 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15466 the stmt_vec_infos. Leave handling of vec_info information to its
15467 destructor. Remove explicit vector releases.
15468 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15469 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15470 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15471 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15472 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15473 (destroy_bb_vec_info): Replace with...
15474 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15475 information to its destructor.
15476 (vect_slp_analyze_bb_1): Use new and delete instead of
15477 new_bb_vec_info and destroy_bb_vec_info.
15478 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15481 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15483 * tree-data-ref.h (subscript): Add access_fn field.
15484 (data_dependence_relation): Add could_be_independent_p.
15485 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15486 (same_access_functions): Move to tree-data-ref.c.
15487 * tree-data-ref.c (ref_contains_union_access_p): New function.
15488 (access_fn_component_p): Likewise.
15489 (access_fn_components_comparable_p): Likewise.
15490 (dr_analyze_indices): Add a reference to access_fn_component_p.
15491 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15493 (constant_access_functions): Likewise.
15494 (add_other_self_distances): Likewise.
15495 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15496 (initialize_data_dependence_relation): Use XCNEW and remove
15497 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15498 of access functions that have the same type. Allow the
15499 subsequence to end with different bases in some circumstances.
15500 Record the chosen access functions in SUB_ACCESS_FN.
15501 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15502 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15503 (subscript_dependence_tester_1): Likewise dra and drb.
15504 (build_classic_dist_vector): Update calls accordingly.
15505 (subscript_dependence_tester): Likewise.
15506 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15507 DDR_COULD_BE_INDEPENDENT_P.
15508 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15509 comp_alias_ddrs instead of may_alias_ddrs.
15510 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15512 (vect_analyze_data_ref_dependence): Use it if
15513 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15514 distance vectors if that fails.
15515 (dependence_distance_ge_vf): New function.
15516 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15517 LOOP_VINFO_MAY_ALIAS_DDRS.
15519 2017-08-04 Richard Biener <rguenther@suse.de>
15521 PR middle-end/81705
15522 * fold-const.c (fold_binary_loc): Properly restrict
15523 minus_var0 && minus_var1 case when associating undefined overflow
15526 2017-08-04 Tom de Vries <tom@codesourcery.com>
15528 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15530 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15532 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15533 Don't start diagnostic messages with a capital letter.
15534 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15536 (rs6000_invalid_builtin): Likewise.
15537 (rs6000_trampoline_init): Likewise.
15539 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15542 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15543 after setting changeable df flags.
15545 2017-08-03 Richard Biener <rguenther@suse.de>
15547 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15548 up if the use is in USE - X.
15550 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15552 * toplev.c (dumpfile.h): New include.
15553 (internal_error_reentered): New static function. Use it...
15554 (internal_error_function): ...here to handle reentered internal_error.
15556 2017-08-03 Richard Biener <rguenther@suse.de>
15558 PR middle-end/81148
15559 * fold-const.c (split_tree): Add minus_var and minus_con
15560 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15561 here but always use minus_*.
15562 (associate_trees): Assert we never associate with MINUS_EXPR
15563 and NULL first operand. Do not recurse for PLUS_EXPR operands
15564 when associating as MINUS_EXPR either.
15565 (fold_binary_loc): Track minus_var and minus_con.
15567 2017-08-03 Tom de Vries <tom@codesourcery.com>
15570 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15571 ACCEL_COMPILER, apply finish_options on
15572 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15574 2017-08-03 Tom de Vries <tom@codesourcery.com>
15577 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15578 function_entry_patch_area_size > 0.
15580 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15583 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15584 instead of 10??LU, perform unit multiplication in wide_int,
15585 don't change alloc_object_size_limit if the limit is larger
15588 PR tree-optimization/81655
15589 PR tree-optimization/81588
15590 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15591 the case when ranges[i].low and high are 1 for unsigned type with
15594 PR middle-end/81052
15595 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15596 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15598 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15600 * tree-vrp.h: Add include guard.
15602 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15605 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15606 (ud2): New insn pattern.
15607 * config/i386/i386.c (ix86_expand_epilogue):
15608 For naked functions, generate ud2 instead of trap insn.
15610 2017-08-02 Marek Polacek <polacek@redhat.com>
15613 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15615 2017-08-02 Tom de Vries <tom@codesourcery.com>
15616 Cesar Philippidis <cesar@codesourcery.com>
15618 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15619 Add missing edge probabilities.
15621 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15623 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15624 Correct endianness.
15626 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15628 PR middle-end/79499
15629 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15630 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15631 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15632 prologue_seq sequences - if any.
15634 2017-08-02 Richard Biener <rguenther@suse.de>
15636 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15637 via vectors if supported, integer extracts via punning if supported
15638 or otherwise vector extracts.
15640 2017-08-02 Richard Biener <rguenther@suse.de>
15642 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15644 (bitmap_insert_into_set): ... this.
15646 2017-08-02 Richard Biener <rguenther@suse.de>
15648 PR tree-optimization/81633
15650 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15652 PR tree-optimization/71752
15653 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15655 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15657 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15658 (machine_function::call_ms2sysv_pad_out): Remove field.
15659 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15660 (ix86_compute_frame_layout): Likewise.
15662 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15665 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15666 attribute with interrupt attribute.
15668 2017-08-01 Andrew Pinski <apinski@cavium.com>
15670 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15671 BIT_INSERT_EXPR's operand 1
15672 to see if the types precision matches.
15674 2017-08-01 Martin Liska <mliska@suse.cz>
15676 PR middle-end/70140
15677 * builtins.c (expand_builtin_memcpy_args): Remove.
15678 (expand_builtin_memcpy): Call newly added function
15679 expand_builtin_memory_copy_args.
15680 (expand_builtin_memcpy_with_bounds): Likewise.
15681 (expand_builtin_mempcpy): Remove last argument.
15682 (expand_builtin_mempcpy_with_bounds): Likewise.
15683 (expand_builtin_memory_copy_args): New function created from
15684 expand_builtin_mempcpy_args with small modifications.
15685 (expand_builtin_mempcpy_args): Remove.
15686 (expand_builtin_stpcpy): Remove unused argument.
15687 (expand_builtin): Likewise.
15688 (expand_builtin_with_bounds): Likewise.
15690 2017-08-01 Martin Liska <mliska@suse.cz>
15693 Make mempcpy more optimal (PR middle-end/70140).
15695 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15698 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15699 __builtin_vec_cmpne verify both arguments are compatible vectors
15700 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15701 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15702 move computation of aligned to after checking the argument types.
15706 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15707 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15710 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15711 Doug Rupp <rupp@adacore.com>
15712 Olivier Hainque <hainque@adacore.com>
15714 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15715 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15717 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15718 for TARGET_OS_CPP_BUILTIN.
15719 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15720 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15722 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15723 passing required abi options to the assembler for EABI configurations.
15724 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15725 of .text.hot and .text.unlikely sections for kernel modules when
15726 using ARM style exceptions.
15727 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15728 options. Add EXTRA_CC1_SPEC.
15729 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15731 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15733 (ARM_TARGET2_DWARF_FORMAT): Define.
15734 * config/arm/t-vxworks: Adjust multilib control to removal of the
15735 Diab command line options.
15737 2017-08-01 Martin Liska <mliska@suse.cz>
15739 PR gcov-profile/81561
15740 * gcov.c (unblock): Make unblocking safe as we need to preserve
15741 index correspondence of blocks and block_lists.
15743 2017-08-01 Richard Biener <rguenther@suse.de>
15745 PR tree-optimization/81181
15746 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15747 (compute_antic): ... end of iteration here.
15749 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15751 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15752 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15753 (ftree-slp-vectorize): Likewise.
15754 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15755 can no longer be set independent of flag_tree_loop_vectorize.
15756 * omp-general.c (emp_max_vf): Likewise.
15757 * opts.c (enable_fdo_optimizations): Remove references to
15758 flag_tree_vectorize, these are now implicit.
15759 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15760 and leave it for the options machinery.
15762 2017-08-01 Martin Liska <mliska@suse.cz>
15764 PR middle-end/70140
15765 * builtins.c (expand_builtin_memcpy_args): Remove.
15766 (expand_builtin_memcpy): Call newly added function
15767 expand_builtin_memory_copy_args.
15768 (expand_builtin_memcpy_with_bounds): Likewise.
15769 (expand_builtin_mempcpy): Remove last argument.
15770 (expand_builtin_mempcpy_with_bounds): Likewise.
15771 (expand_builtin_memory_copy_args): New function created from
15772 expand_builtin_mempcpy_args with small modifications.
15773 (expand_builtin_mempcpy_args): Remove.
15774 (expand_builtin_stpcpy): Remove unused argument.
15775 (expand_builtin): Likewise.
15776 (expand_builtin_with_bounds): Likewise.
15778 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15781 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15782 print "ds:" only for immediates in generic address space.
15784 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15787 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15788 (ix86_function_ok_for_sibcall): Return false for naked functions.
15790 2017-08-01 Richard Biener <rguenther@suse.de>
15792 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15793 (compute_antic): Seed worklist with exit block predecessors.
15794 * cfganal.c (dfs_find_deadend): For a cycle return the source
15795 of the edge closing it.
15797 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15799 * config/aarch64/aarch64.c
15800 (aarch64_can_const_movi_rtx_p): Move 0 check.
15802 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15804 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15805 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15807 * match.pd: Ditto in address comparison pattern.
15809 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15811 PR tree-optimization/81627
15812 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15813 closed ssa form for store-store chain.
15815 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15817 PR tree-optimization/81620
15818 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15819 for store-store chain.
15821 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15823 PR tree-optimization/81588
15824 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15825 ranges[i].in_p, invert comparison code ccode. For >/>=,
15826 swap rhs1 and rhs2 and comparison code unconditionally,
15827 for </<= don't do that. Don't swap rhs1/rhs2 again if
15828 ranges[i].in_p, instead invert comparison code ccode if
15829 opcode or oe->rank is BIT_IOR_EXPR.
15832 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15833 a direct optab to conversion optab.
15834 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15835 with GET_MODE_INNER as last argument instead of optab_handler.
15836 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15837 vector extraction if possible and optab is available.
15838 * expr.c (store_constructor): Use convert_optab_handler instead
15839 of optab_handler. Use vector initialization from smaller
15840 vectors if possible and optab is available.
15841 * tree-vect-stmts.c (vectorizable_load): Likewise.
15842 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15843 now have two modes.
15844 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15845 of vec_init from half-sized vectors with the same element mode.
15846 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15847 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15848 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15849 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15850 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15851 after mode in gen_vec_extract* calls.
15852 (vec_extract<mode>): Renamed to ...
15853 (vec_extract<mode><ssescalarmodelower>): ... this.
15854 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15855 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15856 element mode after mode in gen_vec_init* calls.
15857 (VEC_INIT_HALF_MODE): New mode iterator.
15858 (vec_init<mode>): Renamed to ...
15859 (vec_init<mode><ssescalarmodelower>): ... this.
15860 (vec_init<mode><ssehalfvecmodelower>): New expander.
15861 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15862 (vec_extractv2sfsf): ... this.
15863 (vec_initv2sf): Renamed to ...
15864 (vec_initv2sfsf): ... this.
15865 (vec_extractv2si): Renamed to ...
15866 (vec_extractv2sisi): ... this.
15867 (vec_initv2si): Renamed to ...
15868 (vec_initv2sisi): ... this.
15869 (vec_extractv4hi): Renamed to ...
15870 (vec_extractv4hihi): ... this.
15871 (vec_initv4hi): Renamed to ...
15872 (vec_initv4hihi): ... this.
15873 (vec_extractv8qi): Renamed to ...
15874 (vec_extractv8qiqi): ... this.
15875 (vec_initv8qi): Renamed to ...
15876 (vec_initv8qiqi): ... this.
15877 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15878 (vec_init<mode>): Renamed to ...
15879 (vec_init<mode><VEC_base_l>): ... this.
15880 (vec_extract<mode>): Renamed to ...
15881 (vec_extract<mode><VEC_base_l>): ... this.
15882 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15883 (vec_initv2sfsf): ... this.
15884 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15885 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15886 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15887 element mode after mode in gen_vec_init* calls.
15888 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15889 (vec_init<mode><Vel>): ... this.
15890 (vec_extract<mode>): Renamed to ...
15891 (vec_extract<mode><Vel>): ... this.
15892 * config/aarch64/iterators.md (Vel): New mode attribute.
15893 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15894 Add element mode after mode in gen_vec_extract* calls.
15895 * config/s390/vector.md (non_vec_l): New mode attribute.
15896 (vec_extract<mode>): Renamed to ...
15897 (vec_extract<mode><non_vec_l>): ... this.
15898 (vec_init<mode>): Renamed to ...
15899 (vec_init<mode><non_vec_l>): ... this.
15900 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15901 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15903 * config/arm/iterators.md (V_elem_l): New mode attribute.
15904 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15905 (vec_extract<mode><V_elem_l>): ... this.
15906 (vec_extractv2di): Renamed to ...
15907 (vec_extractv2didi): ... this.
15908 (vec_init<mode>): Renamed to ...
15909 (vec_init<mode><V_elem_l>): ... this.
15910 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15911 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15912 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15913 Add element mode after gen_vec_extract* calls.
15914 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15915 (vec_init<mode><unitmode>): ... this.
15916 (vec_extract<mode>): Renamed to ...
15917 (vec_extract<mode><unitmode>): ... this.
15918 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15919 (vec_init<mode><unitmode>): ... this.
15920 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15921 (vec_initv2sfsf): ... this.
15922 (vec_extractv2sf): Renamed to ...
15923 (vec_extractv2sfsf): ... this.
15924 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15925 Add element mode after gen_vec_extract* calls.
15926 * config/mips/mips.md (unitmode): New mode iterator.
15927 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15928 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15929 * config/spu/spu.md (inner_l): New mode attribute.
15930 (vec_init<mode>): Renamed to ...
15931 (vec_init<mode><inner_l>): ... this.
15932 (vec_extract<mode>): Renamed to ...
15933 (vec_extract<mode><inner_l>): ... this.
15934 * config/sparc/sparc.md (veltmode): New mode iterator.
15935 (vec_init<VMALL:mode>): Renamed to ...
15936 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15937 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15938 (vec_initv2sisi): ... this.
15939 (vec_initv2sf): Renamed to ...
15940 (vec_initv2sfsf): ... this.
15941 (vec_extractv2sf): Renamed to ...
15942 (vec_extractv2sfsf): ... this.
15943 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15944 (vec_init<mode>): Renamed to ...
15945 (vec_init<mode><VEC_base_l>): ... this.
15946 (vec_extract<mode>): Renamed to ...
15947 (vec_extract<mode><VEC_base_l>): ... this.
15948 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15949 (vec_initv2sfsf): ... this.
15950 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15951 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15952 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15953 gen_vec_init* calls.
15955 2017-08-01 Richard Biener <rguenther@suse.de>
15957 PR tree-optimization/81297
15958 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15959 TREE_OVERFLOW from INTEGER_CSTs.
15961 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15963 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15965 2017-07-31 Carl Love <cel@us.ibm.com>
15967 * config/rs6000/rs6000-c: Add support for built-in functions
15968 vector signed char vec_xl_be (signed long long, signed char *);
15969 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15970 vector signed int vec_xl_be (signed long long, signed int *);
15971 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15972 vector signed long long vec_xl_be (signed long long, signed long long *);
15973 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15974 vector signed short vec_xl_be (signed long long, signed short *);
15975 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15976 vector double vec_xl_be (signed long long, double *);
15977 vector float vec_xl_be (signed long long, float *);
15978 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15979 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15980 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15982 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15983 (altivec_expand_builtin): Add switch statement to call
15984 altivec_expand_xl_be for each builtin.
15985 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15986 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15987 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15988 __builtin_vsx_le_be_v16qi.
15989 * doc/extend.texi: Update the built-in documentation file for the
15990 new built-in functions.
15992 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15995 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15997 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15999 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16001 * config.gcc: Add z14.
16002 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
16003 CPU model numbers for z13s and z14.
16004 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16006 * config/s390/s390-opts.h (enum processor_type): Rename
16007 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16008 * config/s390/s390.c (processor_table): Add field for CPU name to
16009 be passed to Binutils.
16010 (s390_asm_output_machine_for_arch): Use the new field in
16011 processor_table for Binutils.
16012 (s390_expand_builtin): Replace arch12 with z14.
16013 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16014 (s390_get_sched_attrmask): Likewise.
16015 (s390_get_unit_mask): Likewise.
16016 * config/s390/s390.opt: Add z14 to processor_type enum.
16018 2017-07-31 Martin Jambor <mjambor@suse.cz>
16021 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16022 regardless of optimization level.
16024 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16025 Martin Liska <mliska@suse.cz>
16027 * predict.def: Remove old comment and adjust probability.
16028 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16029 PREDICT statements.
16031 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16034 * config/i386/i386.c (ix86_function_naked): New function.
16035 (ix86_can_use_return_insn_p): Return false for naked functions.
16036 (ix86_expand_prologue): Skip prologue for naked functions.
16037 (ix86_expand_epilogue): Skip epilogue for naked functions
16038 and emit trap instruction.
16039 (ix86_warn_func_return): New function.
16040 (ix86_attribute_table): Add "naked" attribute specification.
16041 (TARGET_WARN_FUNC_RETURN): Define.
16042 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16044 2017-07-31 Martin Liska <mliska@suse.cz>
16046 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16047 (dump_gimple_bb_header): Always dump BB info.
16048 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16050 2017-07-31 Martin Liska <mliska@suse.cz>
16053 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16054 also with current_function_decl non-null equality.
16056 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16059 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16060 change type to the element type, instead add eltype variable and
16061 use it where we are interested in the element type.
16063 PR tree-optimization/81603
16064 * ipa-polymorphic-call.c
16065 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16066 offset arithmetic in offset_int, bail out if the resulting bit offset
16067 doesn't fit into shwi.
16069 2017-07-31 Martin Liska <mliska@suse.cz>
16071 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16072 (gimplify_save_expr): Fix comment.
16074 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16077 * config/i386/i386.c (ix86_function_arg): Update arguments for
16079 (ix86_compute_frame_layout): Set the initial stack offset to
16080 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16081 INCOMING_FRAME_SP_OFFSET.
16082 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16083 stack before exception handler returns.
16084 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16085 the 'ERROR_CODE' for exception handler.
16087 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16089 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16090 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16091 (ASM_OUTPUT_REG_POP): Ditto.
16092 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16093 instead of asm_fprintf to output pure string.
16095 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16097 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16098 to imported_module_or_decl hook.
16099 (debug_nothing_tree_tree_tree_bool): Remove.
16100 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16101 * debug.c (do_nothing_debug_hooks): Use
16102 debug_nothing_tree_tree_tree_bool_bool instead of
16103 debug_nothing_tree_tree_tree_bool.
16104 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16105 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16106 * sdbout.c (sdb_debug_hooks): Likewise.
16107 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16108 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16110 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16111 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16112 attribute, don't add anything.
16114 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16116 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16117 (fold_build2_stat_loc): Likewise.
16118 (fold_build3_stat_loc): Likewise.
16119 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16120 (fold_build1_loc): Remove macro.
16121 (fold_build2_loc): Likewise.
16122 (fold_build3_loc): Likewise.
16124 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16126 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16127 (gimple_build_debug_bind_source_stat): Likewise.
16128 * gimple.h (gimple_build_debug_bind): Remove macro.
16129 (gimple_build_debug_bind_source): Likewise.
16131 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16133 * bitmap.c (bitmap_alloc): Adjust.
16134 (bitmap_gc_alloc): Likewise.
16135 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16137 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16139 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16140 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16141 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16142 (bitmap_gc_alloc_stat): Likewise.
16143 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16145 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16147 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16148 * rtl.h (shallow_copy_rtx): Remove macro.
16150 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16152 * emit-rtl.c (gen_raw_REG): Adjust.
16153 * gengenrtl.c (gendef): Likewise.
16154 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16155 * rtl.h (rtx_alloc): Remove macro.
16157 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16159 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16160 (build_tree_list_stat): Likewise.
16161 * tree.h (build_tree_list): Remove macro.
16162 (build_tree_list_vec): Likewise.
16164 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16166 * tree.c (make_vector_stat): Remove _stat from name.
16167 (build_vector_stat): Likewise.
16168 * tree.h (make_vector_stat): Remove macro.
16169 (build_vector_stat): Likewise.
16171 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16173 * tree.h (build_var_debug_value): Remove prototype.
16175 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16177 * tree.c (tree_cons_stat): Remove _stat from name.
16178 * tree.h (tree_cons): Remove macro.
16180 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16182 * tree.c (build_vl_exp_stat): Remove _stat from name.
16183 * tree.h (build_vl_exp): Remove macro.
16185 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16187 * tree.c (build_decl_stat): Remove _stat from name.
16188 * tree.h (build_decl): Remove macro.
16190 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16192 * gimple.c (gimple_build_with_ops_stat): Adjust.
16193 (gimple_alloc_stat): Remove _stat from name.
16194 * gimple.h (gimple_alloc): Remove macro.
16196 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16198 * tree.c (make_tree_vec_stat): Remove _stat from name.
16199 (grow_tree_vec_stat): Likewise.
16200 * tree.h (make_tree_vec_stat): Adjust prototype.
16201 (grow_tree_vec_stat): Likewise.
16202 (make_tree_vec): Remove macro.
16203 (grow_tree_vec): Likewise.
16205 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16207 * fold-const.c (fold_build1_stat_loc): Adjust.
16208 (fold_build2_stat_loc): Likewise.
16209 (fold_build3_stat_loc): Likewise.
16210 * tree.c (build0_stat): Remove _stat from name.
16211 (build1_stat): Likewise.
16212 (build2_stat): Likewise.
16213 (build3_stat): Likewise.
16214 (build4_stat): Likewise.
16215 (build5_stat): Likewise.
16216 * tree.h (build1_loc): Remove macro, and rename _stat function
16218 (build2_loc): Likewise.
16219 (build3_loc): Likewise.
16220 (build4_loc): Likewise.
16221 (build5_loc): Likewise.
16223 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16225 * tree.c (make_int_cst_stat): Remove _stat from name.
16226 * tree.h (make_int_cst_stat): Adjust prototype.
16227 (make_int_cst): Remove macro.
16229 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16231 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16232 * tree.h (make_tree_binfo_stat): Adjust prototype.
16233 (make_tree_binfo): Remove.
16235 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16237 * tree.c (copy_node_stat): Rename to copy_node.
16238 (build_distinct_type_copy): Adjust.
16239 * tree.h (copy_node_stat): Adjust prototype.
16240 (copy_node): Remove macro.
16242 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16244 * tree.c (make_node_stat): rename to make_node.
16245 (build_tree_list_stat): Adjust.
16246 (build0_stat): Likewise.
16247 (build2_stat): Likewise.
16248 (build3_stat): Likewise.
16249 (build4_stat): Likewise.
16250 (build5_stat): Likewise.
16251 (build_decl_stat): Likewise.
16252 * tree.h (make_node_stat): Adjust prototype.
16253 (make_node): remove macro.
16255 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16257 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16258 (PPC_FEATURE2_SCV): Likewise.
16259 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16261 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16263 * config/aarch64/aarch64.c
16264 (aarch64_internal_mov_immediate): Add new special pattern.
16265 * config/aarch64/aarch64.md (*movdi_aarch64):
16266 Add reg/32bit const mov case.
16268 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16269 Richard Sandiford <richard.sandiford@linaro.org>
16271 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16272 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16273 Add integer and movi cases.
16274 (movi-split-hf-df-sf split, fp16): New.
16275 (enabled): Added TARGET_FP_F16INST.
16276 * config/aarch64/iterators.md (GPF_HF): New.
16277 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16279 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16281 * config/aarch64/aarch64.c
16282 (aarch64_simd_container_mode): Add prototype.
16283 (aarch64_expand_mov_immediate): Add HI support.
16284 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16285 (aarch64_can_const_movi_rtx_p): New.
16286 (aarch64_preferred_reload_class):
16287 Remove restrictions of using FP registers for certain SIMD operations.
16288 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16289 (aarch64_valid_floating_const): Add integer move validation.
16290 (aarch64_simd_imm_scalar_p): Remove.
16291 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16292 (aarch64_legitimate_constant_p): Expand list of supported cases.
16293 * config/aarch64/aarch64-protos.h
16294 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16295 (aarch64_reinterpret_float_as_int): New.
16296 (aarch64_simd_imm_scalar_p): Remove.
16297 * config/aarch64/constraints.md (Uvi): New.
16298 (Dd): Split into Ds and new Dd.
16299 * config/aarch64/aarch64.md (*movsi_aarch64):
16301 (*movdi_aarch64): Add SIMD mov case.
16303 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16305 * tree-predcom.c: (struct chain): Handle store-store chain in which
16306 stores for elimination only store loop invariant values.
16307 (execute_pred_commoning_chain): Ditto.
16308 (prepare_initializers_chain_store_elim): Ditto.
16309 (prepare_finalizers): Ditto.
16310 (is_inv_store_elimination_chain): New function.
16311 (initialize_root_vars_store_elim_1): New function.
16313 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16315 * tree-predcom.c: Revise general description of the pass.
16316 (enum chain_type): New enum type for store elimination.
16317 (struct chain): New field supporting store elimination.
16318 (struct component): Ditto.
16319 (dump_chain): Dump store-stores chain.
16320 (release_chain): Release resources.
16321 (split_data_refs_to_components): Compute and create component
16322 contains only stores for elimination.
16323 (get_chain_last_ref_at): New function.
16324 (make_invariant_chain): Initialization.
16325 (make_rooted_chain): Specify chain type in parameter and record it.
16326 (add_looparound_copies): Skip for store-stores chain.
16327 (determine_roots_comp): Compute type of chain and pass it to
16329 (initialize_root_vars_store_elim_2): New function.
16330 (finalize_eliminated_stores): New function.
16331 (remove_stmt): Handle store for elimination.
16332 (execute_pred_commoning_chain): Execute predictive commoning on
16333 store-store chains.
16334 (determine_unroll_factor): Skip unroll for store-stores chain.
16335 (prepare_initializers_chain_store_elim): New function.
16336 (prepare_initializers_chain): Hanlde store-store chain.
16337 (prepare_finalizers_chain, prepare_finalizers): New function.
16338 (tree_predictive_commoning_loop): Return integer value indicating
16339 if loop is unrolled or lcssa form is corrupted.
16340 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16342 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16344 * tree-predcom.c (initialize_root): Delete.
16345 (execute_pred_commoning_chain): Initialize root vars and replace
16346 reference of non-combined chain directly, rather than call above
16349 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16351 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16352 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16354 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16356 * tree-predcom.c (struct chain): New field init_seq.
16357 (release_chain): Release init_seq.
16358 (prepare_initializers_chain): Record intialization stmts in above
16360 (insert_init_seqs): New function.
16361 (tree_predictive_commoning_loop): Call insert_init_seqs.
16363 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16365 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16367 2017-07-28 Richard Biener <rguenther@suse.de>
16369 * match.pd: Remove superfluous :c.
16370 * genmatch.c (simplify::id): Add member.
16371 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16373 (current_id): New global.
16374 (dt_node::parent): Move from ...
16375 (dt_operand::parent): ... here. Add for_id member.
16376 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16377 (decision_tree::find_node): Relax order requirement when
16378 merging DT_TRUE nodes to ones inbetween the current simplify
16379 and the one we try to merge with. Add diagnostic whenever
16380 we need to enforce pattern order by not merging.
16381 (decision_tree::insert): Set current_id.
16382 (decision_tree::print_node): Dump parent node and for_id.
16383 (parser::last_id): Add member.
16384 (parser::push_simplify): Assign unique id.
16385 (parser::parser): Initialize last_id.
16387 2017-07-28 Martin Liska <mliska@suse.cz>
16390 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16391 gimple_build_debug_bind.
16393 2017-07-28 Richard Biener <rguenther@suse.de>
16395 PR tree-optimization/81502
16396 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16398 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16400 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16401 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16402 for BIT_FIELD_REF args.
16403 * fold-const.c (make_bit_field_ref): Likewise.
16404 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16406 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16409 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16410 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16411 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16412 Or it into SANITIZER_UNDEFINED.
16413 * ubsan.c: Include gimple-fold.h and varasm.h.
16414 (ubsan_expand_ptr_ifn): New function.
16415 (instrument_pointer_overflow): New function.
16416 (maybe_instrument_pointer_overflow): New function.
16417 (instrument_object_size): Formatting fix.
16418 (pass_ubsan::execute): Call instrument_pointer_overflow
16419 and maybe_instrument_pointer_overflow.
16420 * internal-fn.c (expand_UBSAN_PTR): New function.
16421 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16422 * sanitizer.def (__ubsan_handle_pointer_overflow,
16423 __ubsan_handle_pointer_overflow_abort): New builtins.
16424 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16425 * internal-fn.def (UBSAN_PTR): New internal function.
16426 * opts.c (sanitizer_opts): Add pointer-overflow.
16427 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16428 * fold-const.c (build_range_check): Compute pointer range check in
16429 integral type if pointer arithmetics would be needed. Formatting
16432 2017-07-28 Martin Liska <mliska@suse.cz>
16435 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16436 parameters that are of a variable-length.
16438 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16440 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16442 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16443 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16444 (CRT_CALL_STATIC_FUNCTION): Likewise.
16445 (ASM_DEFAULT_SPEC): New define.
16446 (ASM_SPEC32): Likewise.
16447 (ASM_SPEC64): Likewise.
16448 (ASM_SPEC_COMMON): Likewise.
16449 (ASM_SPEC): Likewise.
16450 (INVALID_64BIT): Likewise.
16451 (LINK_OS_DEFAULT_SPEC): Likewise.
16452 (LINK_OS_SPEC32): Likewise.
16453 (LINK_OS_SPEC64): Likewise.
16454 (POWERPC_LINUX): Likewise.
16455 (PTRDIFF_TYPE): Likewise.
16456 (RESTORE_FP_PREFIX): Likewise.
16457 (RESTORE_FP_SUFFIX): Likewise.
16458 (SAVE_FP_PREFIX): Likewise.
16459 (SAVE_FP_SUFFIX): Likewise.
16460 (SIZE_TYPE): Likewise.
16461 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16462 (TARGET_64BIT): Likewise.
16463 (TARGET_64BIT): Likewise.
16464 (TARGET_AIX): Likewise.
16465 (WCHAR_TYPE_SIZE): Likewise.
16466 (WCHAR_TYPE): Undefine.
16467 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16468 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16469 (CPP_OS_RTEMS_SPEC): Delete.
16470 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16471 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16473 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16475 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16477 PR tree-optimization/81578
16478 * tree-parloops.c (build_new_reduction): Bail out if
16479 reduction_code isn't one of the standard OpenMP reductions.
16480 Move the details printing after that decision.
16482 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16484 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16485 related to reload_in_progress.
16486 (splat_input_operand): Likewise.
16487 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16489 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16491 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16492 (TARGET_INSTANTIATE_DECLS): Likewise.
16493 (legitimate_indexed_address_p): Delete reload_in_progress code.
16494 (rs6000_debug_legitimate_address_p): Likewise.
16495 (rs6000_eliminate_indexed_memrefs): Likewise.
16496 (rs6000_emit_le_vsx_store): Likewise.
16497 (rs6000_emit_move_si_sf_subreg): Likewise.
16498 (rs6000_emit_move): Likewise.
16499 (register_to_reg_type): Likewise.
16500 (rs6000_pre_atomic_barrier): Likewise.
16501 (rs6000_machopic_legitimize_pic_address): Likewise.
16502 (rs6000_allocate_stack_temp): Likewise.
16503 (rs6000_address_for_fpconvert): Likewise.
16504 (rs6000_address_for_altivec): Likewise.
16505 (rs6000_secondary_memory_needed_rtx): Delete function.
16506 (rs6000_check_sdmode): Likewise.
16507 (rs6000_alloc_sdmode_stack_slot): Likewise.
16508 (rs6000_instantiate_decls): Likewise.
16509 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16510 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16511 Delete reload_in_progress.
16512 (*vec_reload_and_plus_<mptrsize>): Likewise.
16513 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16514 (vsx_div_v2di): Likewise.
16515 (vsx_udiv_v2di): Likewise.
16517 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16519 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16520 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16521 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16522 (rs6000_debug_reg_global): Delete print of LRA status.
16523 (rs6000_option_override_internal): Delete dead LRA related code.
16524 (rs6000_lra_p): Delete function.
16525 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16527 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16529 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16530 * config/riscv/rtems.h: New file.
16532 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16533 Sudakshina Das <sudi.das@arm.com>
16535 * config/aarch64/aarch64.md
16536 (define_split for and<mode>3nr_compare): Move
16537 non aarch64_logical_operand to a register.
16538 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16539 register immediate operand to a register.
16540 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16542 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16544 PR middle-end/81564
16545 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16547 2017-07-27 Richard Biener <rguenther@suse.de>
16549 PR tree-optimization/81573
16550 PR tree-optimization/81494
16551 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16552 multi defuse cycle case.
16554 2017-07-27 Richard Biener <rguenther@suse.de>
16556 PR tree-optimization/81571
16557 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16560 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16562 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16563 earlier and only if MASK_FPU is set. Adjust formatting.
16565 2017-07-27 Martin Liska <mliska@suse.cz>
16567 * opt-functions.awk: Add validation of value of Init.
16568 * optc-gen.awk: Pass new argument.
16570 2017-07-27 Martin Liska <mliska@suse.cz>
16572 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16573 Fix wrong condition.
16575 2017-07-27 Martin Liska <mliska@suse.cz>
16577 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16578 BBs and edges seen by autoFDO.
16580 2017-07-27 Richard Biener <rguenther@suse.de>
16582 PR tree-optimization/81502
16583 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16584 with incompatible but same sized type.
16585 (execute_update_addresses_taken): Likewise.
16587 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16589 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16590 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16592 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16595 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16596 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16597 Change s_operand to memory_operand.
16599 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16601 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16602 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16603 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16604 Emit instructions rather than returning an expression. Handle TFmode
16605 and KFmode by casting to TImode.
16606 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16607 (rs6000_emit_le_vsx_store): Likewise.
16608 * config/rs6000/vsx.md (VSX_TI): New iterator.
16609 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16610 (*vsx_le_undo_permute_<mode>): Likewise.
16611 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16612 emit the split sequence.
16613 (*vsx_le_perm_store_<mode>): Likewise.
16615 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16617 PR tree-optimization/81555
16618 PR tree-optimization/81556
16619 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16620 if true, force CHANGED for the recursive invocation.
16621 (reassociate_bb): Remember original length of ops array, pass
16622 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16624 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16625 attributes for noipa attribute. For naked attribute use
16626 lookup_attribute first before lookup_attribute_spec.
16627 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16629 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16630 for functions with noipa attribute.
16631 (cgraph_externally_visible_p): Return true for functions with noipa
16633 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16634 for functions with noipa attribute.
16635 * doc/extend.texi: Document noipa function attribute.
16636 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16637 also for functions with noipa attribute.
16638 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16640 2017-07-26 Andrew Pinski <apinski@cavium.com>
16642 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16643 vec_unalign_load_cost and vec_unalign_store_cost.
16645 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16647 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16648 -mvsx-small-integer option.
16649 (ISA_3_0_MASKS_IEEE): Likewise.
16650 (OTHER_VSX_VECTOR_MASKS): Likewise.
16651 (POWERPC_MASKS): Likewise.
16652 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16653 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16654 code, only testing for DImode being allowed in non-VSX floating
16656 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16657 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16659 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16660 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16661 TARGET_P8_VECTOR test.
16662 (rs6000_secondary_reload_simple_move): Likewise.
16663 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16664 since TARGET_P9_VECTOR was already tested.
16665 (rs6000_opt_masks): Remove -mvsx-small-integer.
16666 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16667 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16669 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16670 test for TARGET_VEXTRACTUB was used, and that uses
16672 (p9 extract splitter): Likewise.
16673 (vsx_extract_<mode>_di_p9): Likewise.
16674 (vsx_extract_<mode>_store_p9): Likewise.
16675 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16676 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16677 the elimination of TARGET_VSX_SMALL_INTEGER.
16678 (vsx_extract_<mode>_p8): Likewise.
16679 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16680 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16681 (vsx_set_<mode>_p9): Likewise.
16682 (vsx_set_v4sf_p9): Likewise.
16683 (vsx_set_v4sf_p9_zero): Likewise.
16684 (vsx_insert_extract_v4sf_p9): Likewise.
16685 (vsx_insert_extract_v4sf_p9_2): Likewise.
16686 * config/rs6000/rs6000.md (sign extend splitter): Change
16687 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16688 (floatsi<mode>2_lfiwax_mem): Likewise.
16689 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16690 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16691 since a test for TARGET_P9_VECTOR was used.
16692 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16693 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16694 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16695 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16696 TARGET_P8_VECTOR test.
16697 (fix_trunc<mode>si2_stfiwx): Likewise.
16698 (fix_trunc<mode>si2_internal): Likewise.
16699 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16700 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16702 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16703 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16704 TARGET_P8_VECTOR test.
16705 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16706 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16707 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16709 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16710 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16711 since a test for TARGET_P9_VECTOR was used.
16712 (splitter for loading small constants): Likewise.
16714 2017-07-26 Andrew Pinski <apinski@cavium.com>
16716 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16719 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16722 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16723 (fp_valid_at): Likewise.
16725 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16727 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16728 (qdf24xx_addrcost_table): Likewise.
16729 (cortexa57_tunings): Update to use generic_branch_cost.
16730 (cortexa72_tunings): Likewise.
16731 (cortexa73_tunings): Likewise.
16732 (qdf24xx_tunings): Likewise.
16734 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16736 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16737 (thunderx2t99_branch_cost): Likewise.
16738 (cortexa35_tunings): Update to use generic_branch_cost.
16739 (cortexa53_tunings): Likewise.
16740 (cortexa57_tunings): Likewise.
16741 (cortexa72_tunings): Likewise.
16742 (cortexa73_tunings): Likewise.
16743 (thunderx2t99_tunings): Likewise.
16745 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16747 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16748 (sparc_option_override): Honour MASK_FSMULD.
16749 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16750 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16751 * config/sparc/sparc.opt (mfsmuld): New option.
16752 * doc/invoke.texi (mfsmuld): Document option.
16754 2017-07-26 Marek Polacek <polacek@redhat.com>
16756 PR middle-end/70992
16757 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16759 2017-07-26 Richard Biener <rguenther@suse.de>
16761 * gimple-match-head.c (do_valueize): Return OP if valueize
16763 (get_def): New helper to get at the def stmt of a SSA name
16764 if valueize allows.
16765 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16766 do_valueize to get at the def stmt.
16767 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16769 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16771 PR middle-end/46932
16772 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16774 2017-07-26 Martin Liska <mliska@suse.cz>
16777 * function.c (expand_function_start): Make expansion of
16778 nonlocal_goto_save_area after parm_birth_insn.
16780 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16782 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16783 from all CPU target flags enable members.
16785 2017-07-26 Richard Biener <rguenther@suse.de>
16787 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16788 (decision_tree::gen): Make 'type' const.
16789 (write_predicate): Likewise.
16791 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16793 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16794 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16795 (rs6000_option_override_internal): Likewise.
16796 (rs6000_expand_vector_set): Likewise.
16797 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16798 (TARGET_UPPER_REGS_SF): Likewise.
16799 (TARGET_UPPER_REGS_DI): Likewise.
16800 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16801 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16802 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16803 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16804 (Splitters for DI constants in Altivec registers): Likewise.
16805 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16806 (vsx_set_v4sf_p9): Likewise.
16807 (vsx_set_v4sf_p9_zero): Likewise.
16808 (vsx_insert_extract_v4sf_p9): Likewise.
16809 (vsx_insert_extract_v4sf_p9_2): Likewise.
16811 2017-07-25 Carl Love <cel@us.ibm.com>
16813 * doc/extend.texi: Update the built-in documentation file for the
16814 existing built-in functions
16815 vector signed char vec_cnttz (vector signed char);
16816 vector unsigned char vec_cnttz (vector unsigned char);
16817 vector signed short vec_cnttz (vector signed short);
16818 vector unsigned short vec_cnttz (vector unsigned short);
16819 vector signed int vec_cnttz (vector signed int);
16820 vector unsigned int vec_cnttz (vector unsigned int);
16821 vector signed long long vec_cnttz (vector signed long long);
16822 vector unsigned long long vec_cnttz (vector unsigned long long);
16824 2017-07-25 Andrew Pinski <apinski@cavium.com>
16826 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16827 accesses where the use is for the first operand of a BIT_INSERT.
16829 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16832 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16833 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16835 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16837 * config/i386/gstabs.h: Delete.
16838 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16840 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16842 * config/i386/i386.c (ix86_decompose_address): Do not check for
16843 register RTX when looking at index_reg or base_reg.
16844 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16846 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16848 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16849 to update EH info here.
16851 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16853 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16855 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16857 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16859 2017-07-25 Torsten Duwe <duwe@suse.de>
16861 * common.opt: Introduce -fpatchable-function-entry
16862 command line option, and its variables function_entry_patch_area_size
16863 and function_entry_patch_area_start.
16864 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16865 including a two-value parser.
16866 * target.def (print_patchable_function_entry): New target hook.
16867 * targhooks.h (default_print_patchable_function_entry): New function.
16868 * targhooks.c (default_print_patchable_function_entry): Likewise.
16869 * toplev.c (process_options): Switch off IPA-RA if
16870 patchable function entries are being generated.
16871 * varasm.c (assemble_start_function): Look at the
16872 patchable-function-entry command line switch and current
16873 function attributes and maybe generate NOP instructions by
16874 calling the print_patchable_function_entry hook.
16875 * doc/extend.texi: Document patchable_function_entry attribute.
16876 * doc/invoke.texi: Document -fpatchable_function_entry
16877 command line option.
16878 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16880 * doc/tm.texi: Re-generate.
16882 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16885 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16886 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16888 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16890 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16892 2017-07-25 Richard Biener <rguenther@suse.de>
16894 PR tree-optimization/81455
16895 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16896 not walk in cycles when looking for guards.
16898 2017-07-25 Richard Biener <rguenther@suse.de>
16900 PR tree-optimization/81529
16901 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16902 when optimizing backedge uses.
16904 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16906 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16908 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16909 to dl_section_ref. On AIX, append an expression to subtract
16910 the size of the section length to dl_section_ref.
16912 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16914 * configure.ac: If any of the config.* scripts fail, exit 1.
16915 * configure: Regenerate.
16917 2017-07-25 Richard Biener <rguenther@suse.de>
16919 PR middle-end/81546
16920 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16921 of immediate uses, be more verbose on errors.
16923 2017-07-25 Richard Biener <rguenther@suse.de>
16925 PR tree-optimization/81510
16926 * tree-vect-loop.c (vect_is_simple_reduction): When the
16927 reduction stmt is not inside the loop bail out.
16929 2017-07-25 Richard Biener <rguenther@suse.de>
16931 PR tree-optimization/81303
16932 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16933 profitability check against LOOP_VINFO_NITERSM1.
16935 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16937 * domwalk.c (cmp_bb_postorder): Simplify.
16938 (sort_bbs_postorder): New function. Use it...
16939 (dom_walker::walk): ...here to optimize common cases.
16941 2017-07-25 Martin Liska <mliska@suse.cz>
16944 * ipa-visibility.c (function_and_variable_visibility): Make the
16945 redirection just on target that supports aliasing.
16946 Fix GNU coding style.
16948 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16951 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16953 * config/arm/rtems.h: Likewise.
16954 * config/bfin/rtems.h: Likewise.
16955 * config/i386/rtemself.h: Likewise.
16956 * config/lm32/rtems.h: Likewise.
16957 * config/m32c/rtems.h: Likewise.
16958 * config/m68k/rtemself.h: Likewise.
16959 * config/microblaze/rtems.h: Likewise.
16960 * config/mips/rtems.h: Likewise.
16961 * config/moxie/rtems.h: Likewise.
16962 * config/nios2/rtems.h: Likewise.
16963 * config/powerpcspe/rtems.h: Likewise.
16964 * config/rs6000/rtems.h: Likewise.
16965 * config/rtems.h: Likewise.
16966 * config/sh/rtems.h: Likewise.
16967 * config/sh/rtemself.h: Likewise.
16968 * config/sparc/rtemself.h: Likewise.
16970 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16973 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16974 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16975 * tree-ssa-structalias.c (alias_get_name): Same.
16977 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16980 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16981 instructions if no du chain is found.
16983 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16985 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16987 2017-07-25 Richard Biener <rguenther@suse.de>
16989 PR middle-end/81505
16990 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16993 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16995 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16996 upper-regs options.
16997 (ISA_2_7_MASKS_SERVER): Likewise.
16998 (ISA_3_0_MASKS_IEEE): Likewise.
16999 (OTHER_P8_VECTOR_MASKS): Likewise.
17000 (OTHER_VSX_VECTOR_MASKS): Likewise.
17001 (POWERPC_MASKS): Likewise.
17002 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
17003 duplicate list of options.
17004 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17005 explicit -mupper-regs options.
17006 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17007 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17008 alias for -mupper-regs-df.
17009 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17010 (rs6000_init_hard_regno_mode_ok): Likewise.
17011 (rs6000_option_override_internal): Likewise.
17012 (rs6000_opt_masks): Likewise.
17013 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17014 options in terms of whether -mvsx or -mpower8-vector was used.
17015 (TARGET_UPPER_REGS_DI): Likewise.
17016 (TARGET_UPPER_REGS_SF): Likewise.
17017 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17018 -mupper-regs-* options.
17020 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17022 * passes.c (emergency_dump_function): Print some empty lines and a
17023 header before the RTL dump.
17025 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17027 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17029 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17032 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17033 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17035 2017-07-24 Carl Love <cel@us.ibm.com>
17037 * config/rs6000/rs6000-c.c: Add support for built-in functions
17038 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17039 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17040 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17041 vec_extract_fp_from_shortl): Add defines for the two builtins.
17042 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17043 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17045 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17046 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17047 * doc/extend.texi: Update the built-in documentation file for the
17048 new built-in function.
17050 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17053 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17055 * doc/generic.texi: Likewise.
17056 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17057 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17059 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17061 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17062 (aarch64_mls_elt_merge<mode>): Likewise.
17064 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17066 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17067 having __cxa_atexit.
17069 2017-07-23 Michael Collison <michael.collison@arm.com>
17071 * config/arm/arm.c (arm_option_override): Deprecate
17072 use of -mstructure-size-boundary.
17073 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17074 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17076 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17079 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17080 Reduce cost estimate for direct moves.
17082 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17085 * config/i386/i386.c (ix86_option_override_internal): Disable
17086 BMI, BMI2 and TBM instructions for -m16.
17088 2017-07-21 Carl Love <cel@us.ibm.com>
17090 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17091 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17092 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17093 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17094 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17095 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17096 VMULOSW): New enum "unspec" values.
17097 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17098 altivec_vmulosw): New patterns.
17099 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17100 VMULOSW): Add definitions.
17102 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17104 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17105 (qdf24xx): Likewise.
17106 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17107 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17108 (AARCH64_FL_V8_1): Renumber.
17109 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17110 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17111 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17112 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17113 rdma to feature modifiers list.
17115 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17117 PR middle-end/56727
17118 * ipa-visibility (function_and_variable_visibility): Convert
17119 recursive PLT call to direct call if appropriate.
17121 2017-07-21 Andrew Pinski <apinski@cavium.com>
17123 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17124 operand 1 to see if the types precision matches.
17125 * fold-const.c (operand_equal_p): Likewise.
17127 2017-07-21 Richard Biener <rguenther@suse.de>
17129 PR tree-optimization/81303
17130 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17131 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17132 (vect_peeling_hash_get_lowest_cost): Adjust.
17133 (vect_enhance_data_refs_alignment): Likewise. Use
17134 vect_get_peeling_costs_all_drs to compute the penalty for no
17135 peeling to match up costs.
17137 2017-07-21 Richard Biener <rguenther@suse.de>
17139 PR tree-optimization/81500
17140 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17141 we didn't identify a reduction path.
17143 2017-07-21 Tom de Vries <tom@codesourcery.com>
17144 Cesar Philippidis <cesar@codesourcery.com>
17146 PR gcov-profile/81442
17147 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17150 2017-07-21 Tom de Vries <tom@codesourcery.com>
17153 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17155 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17156 nvptx_override_options_after_change.
17158 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17160 * dwarf2out.c (output_file_names): Avoid double testing for
17161 dwarf_version >= 5.
17163 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17165 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17167 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17169 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17171 (try_crossjump_to_edge): Do not punt on partitioned functions.
17173 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17175 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17176 Put all BBs reachable only via paths crossing cold region to cold
17178 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17180 2016-07-21 Richard Biener <rguenther@suse.de>
17182 PR tree-optimization/81303
17183 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17184 into account prologue and epilogue iterations when raising
17185 min_profitable_iters to sth at least covering one vector iteration.
17187 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17189 * config/arm/arm.c (arm_test_cpu_arch_dat):
17192 2017-07-20 Nathan Sidwell <nathan@acm.org>
17194 Remove TYPE_METHODS.
17195 * tree.h (TYPE_METHODS): Delete.
17196 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17197 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17198 (dbxout_type_methods): Scan TYPE_FIELDS.
17199 (dbxout_type): Don't check TYPE_METHODS here.
17200 * function.c (use_register_for_decl): Always ignore register for
17201 class types when not optimizing.
17202 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17203 * tree.c (free_lang_data_in_type): Stitch out member functions and
17204 templates from TYPE_FIELDS.
17205 (build_distinct_type_copy, verify_type_variant,
17206 verify_type): Member fns are on TYPE_FIELDS.
17207 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17208 * tree-pretty-print.c (dump_generic_node): Likewise.
17210 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17213 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17214 V2TImode and V4TImode.
17215 (ix86_expand_vector_extract): Likewise.
17216 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17217 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17218 (ssescalarmode): Handle V4TImode and V2TImode.
17219 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17220 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17221 (VEXTRACTI128_MODE): New mode iterator.
17222 (splitter for *vec_extractv?ti first element): New.
17223 (VEC_INIT_MODE): New mode iterator.
17224 (vec_init<mode>): Consolidate 3 expanders into one using
17225 VEC_INIT_MODE mode iterator.
17227 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17229 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17230 non_spilled_static_chain_regno_p.
17232 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17234 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17236 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17238 * bb-reorder.c (connect_traces): Allow copying of blocks within
17241 2017-07-20 Richard Biener <rguenther@suse.de>
17243 * gimple.h (gimple_phi_result): Add gphi * overload.
17244 (gimple_phi_result_ptr): Likewise.
17245 (gimple_phi_arg): Likewise. Adjust index assert to only
17246 allow actual argument accesses rather than all slots available
17248 (gimple_phi_arg_def): Add gphi * overload.
17249 * tree-phinodes.c (make_phi_node): Initialize only actual
17251 (resize_phi_node): Clear memory not covered by old node,
17252 do not initialize excess argument slots.
17253 (reserve_phi_args_for_new_edge): Initialize new argument slot
17256 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17258 PR tree-optimization/81388
17260 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17262 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17263 by removing computation of may_be_zero.
17265 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17266 Tom de Vries <tom@codesourcery.com>
17268 PR middle-end/81030
17269 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17270 when gimple level profile disagrees with what RTL expander did.
17272 2017-07-20 Richard Biener <rguenther@suse.de>
17274 PR tree-optimization/61171
17275 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17276 (vect_analyze_stmt): Add slp instance parameter.
17277 (vectorizable_reduction): Likewise.
17278 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17279 (vect_is_simple_reduction): Deal with chains not detected
17280 as SLP reduction chain, specifically not properly associated
17281 chains containing a mix of plus/minus.
17282 (get_reduction_op): Remove.
17283 (get_initial_defs_for_reduction): Simplify, pass in whether
17284 this is a reduction chain, pass in the SLP node for the PHIs.
17285 (vect_create_epilog_for_reduction): Get the SLP instance as
17287 (vectorizable_reduction): Get the SLP instance as arg.
17288 During analysis remember the SLP node with the PHIs in the
17289 instance. Simplify getting at the vectorized reduction PHIs.
17290 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17291 through SLP instance.
17292 (vect_slp_analyze_operations): Likewise.
17293 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17294 (vect_transform_stmt): Likewise.
17296 2017-07-20 Tom de Vries <tom@codesourcery.com>
17298 PR tree-optimization/81489
17299 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17300 read of phi arg location to before loop that modifies phi.
17302 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17304 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17307 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17309 PR middle-end/81331
17310 * except.c (execute): Fix ordering issue.
17312 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17314 PR rtl-optimization/81423
17315 * combine.c (make_compound_operation_int): Don't try to optimize
17316 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17318 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17320 PR rtl-optimization/81423
17321 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17322 with a constant that is -1 in the truncated to mode.
17324 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17326 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17327 (determine_unlikely_bbs): ... here.
17328 * predict.h (propagate_unlikely_bbs_forward): Declare.
17329 * cfgexpand.c (pass_expand::execute): Use it.
17330 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17332 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17333 propagate_unlikely_bbs_forward.
17335 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17337 PR middle-end/81331
17338 * except.c (maybe_add_nop_after_section_switch): New function.
17341 2017-07-19 Tom de Vries <tom@codesourcery.com>
17343 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17345 2017-07-19 Tom de Vries <tom@codesourcery.com>
17347 * gimple.h (gimple_phi_arg): Make assert more strict.
17349 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17351 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17352 * config/rs6000/mmintrin.h: New file.
17353 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17355 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17357 PR tree-optimization/81346
17358 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17360 2017-07-19 Tom de Vries <tom@codesourcery.com>
17362 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17364 2017-07-19 Tom de Vries <tom@codesourcery.com>
17366 * config/nvptx/nvptx-modes.def: Add V2DImode.
17367 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17368 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17369 (nvptx_output_mov_insn): Handle lack of mov.b128.
17370 (nvptx_print_operand): Handle 'H' and 'L' codes.
17371 (nvptx_vector_mode_supported): Allow V2DImode.
17372 (nvptx_preferred_simd_mode): New function.
17373 (nvptx_data_alignment): New function.
17374 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17375 nvptx_preferred_simd_mode.
17376 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17378 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17380 2017-07-19 Tom de Vries <tom@codesourcery.com>
17382 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17383 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17384 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17385 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17386 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17387 (mov<VECIM>_insn): New define_insn.
17388 (define_expand "mov<VECIM>): New define_expand.
17390 2017-07-19 Tom de Vries <tom@codesourcery.com>
17392 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17394 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17396 PR tree-optimization/81346
17397 * fold-const.h (fold_div_compare, range_check_type): Declare.
17398 * fold-const.c (range_check_type): New function.
17399 (build_range_check): Use range_check_type.
17400 (fold_div_compare): No longer static, rewritten into
17401 a match.pd helper function.
17402 (fold_comparison): Don't call fold_div_compare here.
17403 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17404 as helper function.
17406 2017-07-19 Nathan Sidwell <nathan@acm.org>
17408 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17409 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17410 * tree.c (find_decls_types_r, verify_type): Use
17411 TYPE_{MIN,MAX}_VALUE_RAW.
17412 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17413 (hash_tree): Likewise.
17414 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17416 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17419 2017-07-18 Tom de Vries <tom@codesourcery.com>
17421 PR middle-end/81464
17422 * omp-expand.c (expand_omp_for_static_chunk): Handle
17423 equal-argument loop exit phi.
17425 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17428 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17429 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17430 operand 2 predicate.
17431 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17432 operand 2 predicate.
17433 (ror,rol -> rorx splitters): Use const_int_operand as
17434 operand 2 predicate.
17436 2017-06-18 Richard Biener <rguenther@suse.de>
17438 PR tree-optimization/81410
17439 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17440 the gap in the ! slp_perm SLP case after each group.
17442 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17444 PR middle-end/81463
17445 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17448 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17450 PR middle-end/81462
17451 * predict.c (set_even_probabilities): Cleanup; do not affect
17452 probabilities that are already known.
17453 (combine_predictions_for_bb): Call even when count is set.
17455 2017-07-18 Nathan Sidwell <nathan@acm.org>
17457 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17460 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17463 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17464 optimization for loop niter analysis.
17466 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17469 * config/avr/avr.c (avr_optimize_casesi): Don't use
17470 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17472 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17474 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17475 body_cost_vec from _vect_peel_extended_info.
17476 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17477 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17480 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17482 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17484 2017-07-18 Richard Biener <rguenther@suse.de>
17486 PR tree-optimization/80620
17487 PR tree-optimization/81403
17488 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17489 info when re-using a VN table entry.
17491 2017-07-18 Richard Biener <rguenther@suse.de>
17493 PR tree-optimization/81418
17494 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17495 vectype_in. Verify that with lane-reducing reduction operations
17496 we have a single def-use cycle.
17498 2017-07-17 Carl Love <cel@us.ibm.com>
17500 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17502 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17503 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17504 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17505 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17506 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17507 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17508 VMULOSW): New enum "unspec" values.
17509 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17510 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17511 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17512 altivec_vmulosw): New patterns.
17513 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17514 VMULOSW): Add definitions.
17515 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17517 * config/alpha/alpha.c: Include predict.h.
17519 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17521 * tree-vrp.c (compare_assert_loc): Fix comparison function
17522 to return predictable results.
17524 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17526 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17528 (subdi3): Likewise.
17529 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17530 * doc/invoke.texi (mexpand-adddi): Update text.
17532 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17534 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17535 that also clobbers the CC register. The old expand code is moved
17537 (*arc_clzsi2): ... here.
17538 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17539 the CC register. The old expand code is moved to ...
17540 (arc_ctzsi2): ... here.
17542 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17544 * config/arc/arc.opt (mindexed-loads): Use initial value
17545 TARGET_INDEXED_LOADS_DEFAULT.
17546 (mauto-modify-reg): Use initial value
17547 TARGET_AUTO_MODIFY_REG_DEFAULT.
17548 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17549 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17550 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17551 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17553 2017-07-17 Martin Liska <mliska@suse.cz>
17556 * opts.c (finish_options): Do not allow -fgnu-tm
17557 w/ -fsanitize={kernel-,}address. Say sorry.
17559 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17562 * tree-loop-distribution.c (classify_partition): Only assert on
17563 numer of iterations.
17564 (merge_dep_scc_partitions): Delete prameter. Update function call.
17565 (distribute_loop): Remove code handling loop with unknown niters.
17566 (pass_loop_distribution::execute): Skip loop with unknown niters.
17568 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17571 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17572 function sort_partitions_by_post_order.
17574 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17576 PR tree-optimization/81374
17577 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17578 the max index of basic blocks, rather than number of basic blocks.
17580 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17582 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17584 (arc_legitimate_pic_operand_p): Likewise.
17585 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17587 (arc_needs_pcl_p): Likewise.
17588 (arc_legitimate_pc_offset_p): Likewise.
17589 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17590 function is also used in constrains.md.
17591 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17592 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17593 PLUS. Only return true/false in known cases, otherwise assert.
17594 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17595 is already called in arc_legitimate_constant_p.
17596 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17598 (LEGITIMATE_PIC_OPERAND_P): Use
17599 arc_raw_symbolic_reference_mentioned_p function.
17600 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17605 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17606 Andrew Burgess <andrew.burgess@embecosm.com>
17608 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17609 (arc_return_address_register): New function.
17610 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17611 (arc_handle_fndecl_attribute): Add naked attribute.
17612 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17613 (TARGET_WARN_FUNC_RETURN): Likewise.
17614 (arc_allocate_stack_slots_for_args): New function.
17615 (arc_warn_func_return): Likewise.
17616 (machine_function): Change type fn_type.
17617 (arc_compute_function_type): Consider new naked function type,
17618 change function return type.
17619 (arc_must_save_register): Adapt to handle new
17620 arc_compute_function_type's return type.
17621 (arc_expand_prologue): Likewise.
17622 (arc_expand_epilogue): Likewise.
17623 (arc_return_address_regs): Delete.
17624 (arc_return_address_register): New function.
17625 (arc_epilogue_uses): Use above function.
17626 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17627 (arc_function_type): Change encoding, add naked type.
17628 (ARC_INTERRUPT_P): Change to handle the new encoding.
17629 (ARC_FAST_INTERRUPT_P): Likewise.
17630 (ARC_NORMAL_P): Define.
17631 (ARC_NAKED_P): Likewise.
17632 (arc_compute_function_type): Delete prototype.
17633 * config/arc/arc.md (in_ret_delay_slot): Use
17634 arc_return_address_register function.
17635 (simple_return): Likewise.
17636 (p_return_i): Likewise.
17638 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17640 PR tree-optimization/81428
17641 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17642 can't be built for those types.
17644 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17646 Remove stuff dead since r239246.
17648 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17649 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17650 (avr_inform_devices): Remove dead stuff.
17652 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17654 * config/arm/arm_neon.h: Fix softp typo.
17656 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17658 PR tree-optimization/81365
17659 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17660 aggregate moves onto bb predecessor edges, make sure there are no
17661 loads that could alias the lhs in between the start of bb and the
17664 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17667 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17668 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17669 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17671 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17673 PR tree-optimization/81396
17674 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17675 (init_symbolic_number): Initialize it to 1.
17676 (perform_symbolic_merge): Add n_ops from both operands into the new
17678 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17679 without base_addr as useless if they need more than one operation.
17680 (bswap_replace): Handle !bswap case for NULL base_addr.
17682 2017-07-17 Tom de Vries <tom@codesourcery.com>
17685 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17688 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17690 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17691 conditional builtin define __FIX_LEON3FT_B2BST.
17693 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17695 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17696 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17699 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17701 PR rtl-optimization/81424
17702 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17703 to remove potential trapping from operands if -fnon-call-exceptions.
17705 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17707 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17708 profile_proability for scalling.
17709 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17711 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17713 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17715 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17717 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17718 fixpoint arithmetics.
17720 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17722 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17723 fixpoint arithmetics.
17725 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17727 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17728 fixpoint arithmetics.
17730 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17732 * profile-count.h (profile_probability::from_reg_br_prob_note,
17733 profile_probability::to_reg_br_prob_note): New functions.
17734 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17735 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17736 * predict.c (probability_reliable_p): Update.
17737 (edge_probability_reliable_p): Update.
17738 (br_prob_note_reliable_p): Update.
17739 (invert_br_probabilities): Update.
17740 (add_reg_br_prob_note): New function.
17741 (combine_predictions_for_insn): Update.
17742 * asan.c (asan_clear_shadow): Update.
17743 * cfgbuild.c (compute_outgoing_frequencies): Update.
17744 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17745 (update_br_prob_note): Update.
17746 (rtl_verify_edges): Update.
17747 (purge_dead_edges): Update.
17748 (fixup_reorder_chain): Update.
17749 * emit-rtl.c (try_split): Update.
17750 * ifcvt.c (cond_exec_process_insns): Update.
17751 (cond_exec_process_if_block): Update.
17752 (dead_or_predicable): Update.
17753 * internal-fn.c (expand_addsub_overflow): Update.
17754 (expand_neg_overflow): Update.
17755 (expand_mul_overflow): Update.
17756 * loop-doloop.c (doloop_modify): Update.
17757 * loop-unroll.c (compare_and_jump_seq): Update.
17758 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17759 * predict.h: Update.
17760 * reorg.c (mostly_true_jump): Update.
17762 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17763 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17764 * config/arc/arc.c: (emit_unlikely_jump): Update.
17765 * config/arm/arm.c: (emit_unlikely_jump): Update.
17766 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17767 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17768 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17769 (ix86_print_operand): Update.
17770 (ix86_split_fp_branch): Update.
17771 (predict_jump): Update.
17772 * config/ia64/ia64.c (ia64_print_operand): Update.
17773 * config/mmix/mmix.c (mmix_print_operand): Update.
17774 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17775 (rs6000_expand_split_stack_prologue): Update.
17776 * config/rs6000/rs6000.c: Update.
17777 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17778 (s390_expand_vec_movstr): Update.
17779 (s390_expand_cs_tdsi): Update.
17780 (s390_expand_split_stack_prologue): Update.
17781 * config/sh/sh.c (sh_print_operand): Update.
17782 (expand_cbranchsi4): Update.
17783 (expand_cbranchdi4): Update.
17784 * config/sparc/sparc.c (output_v9branch): Update.
17785 * config/spu/spu.c (get_branch_target): Update.
17786 (ea_load_store_inline): Update.
17787 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17788 * config/tilepro/tilepro.c: Update.
17790 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17792 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17793 (gimplify_save_expr): Likewise.
17795 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17797 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17799 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17801 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17803 * timevar.def (TV_IPA_FNSUMMARY): Define.
17805 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17807 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17808 to back store errata sensitive sequence from being generated.
17809 (sqrtdf2_fix): Likewise.
17811 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17813 * tree-ssa-threadupdate.c (compute_path_counts,
17814 update_joiner_offpath_counts): Use profile_probability.
17816 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17819 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17821 * config/arm/arm-c.c (arm_cpu_builtins): Define
17822 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17824 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17826 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17827 array entries to represent __ieee128 versions of the
17828 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17829 scalar_extract_sig, and scalar_insert_exp built-in functions.
17830 (altivec_resolve_overloaded_builtin): Add special case handling
17831 for the __builtin_scalar_insert_exp function, as represented by
17832 the P9V_BUILTIN_VEC_VSIEDP constant.
17833 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17834 exponent support for __ieee128 argument.
17835 (VSESQP): Add scalar extract signature support for __ieee128
17837 (VSTDCNQP): Add scalar test negative support for __ieee128
17839 (VSIEQP): Add scalar insert exponent support for __int128 argument
17840 with __ieee128 result.
17841 (VSIEQPF): Add scalar insert exponent support for __ieee128
17842 argument with __ieee128 result.
17843 (VSTDCQP): Add scalar test data class support for __ieee128
17845 (VSTDCNQP): Add overload support for scalar test negative with
17846 __ieee128 argument.
17847 (VSTDCQP): Add overload support for scalar test data class
17848 __ieee128 argument.
17849 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17850 UNSPEC_VSX_SXSIGDP.
17851 (UNSPEC_VSX_SIEXPQP): New constant.
17852 (xsxexpqp): New insn for VSX scalar extract exponent quad
17854 (xsxsigqp): New insn for VSX scalar extract significand quad
17856 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17857 precision with floating point argument.
17858 (xststdcqp): New expand for VSX scalar test data class quad
17860 (xststdcnegqp): New expand for VSX scalar test negative quad
17862 (xststdcqp): New insn to match expansions for VSX scalar test data
17863 class quad precision and VSX scalar test negative quad precision.
17864 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17865 special case operand checking to enforce that second operand of
17866 VSX scalar test data class with quad precision argument is a 7-bit
17868 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17869 prototypes and descriptions of __ieee128 versions of
17870 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17871 scalar_test_data_class, and scalar_test_neg built-in functions.
17873 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17875 PR tree-optimization/81162
17876 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17877 replace a negate with an add.
17879 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17881 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17883 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17885 * config/arm/arm-c.c (arm_cpu_builtins): Define
17886 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17888 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17890 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17891 (armv8-r): Set ARM Cortex-R52 as default CPU.
17892 * config/arm/arm-tables.opt: Regenerate.
17893 * config/arm/arm-tune.md: Regenerate.
17894 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17896 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17897 extension for -mcpu=cortex-r52.
17899 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17901 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17902 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17903 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17904 (fp-armv8): Define it as FP_ARMv8 only.
17905 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17906 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17908 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17909 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17910 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17911 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17912 than TARGET_FPU_ARMV8.
17913 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17914 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17915 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17917 * config/arm/neon.md (neon_vrint): Likewise.
17918 (neon_vcvt): Likewise.
17919 (neon_<fmaxmin_op><mode>): Likewise.
17920 (<fmaxmin><mode>3): Likewise.
17921 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17922 * config/arm/predicates.md (arm_cond_move_operator): Check against
17923 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17925 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17927 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17928 to top of function.
17930 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17932 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17933 loop in comment with memset.
17935 2017-07-14 Martin Liska <mliska@suse.cz>
17937 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17938 * dwarf2out.c (is_java): Remove the function.
17939 (output_pubname): Remove usage of the function.
17940 (lower_bound_default): Remove usage of DW_LANG_Java.
17941 (gen_compile_unit_die): Likewise.
17942 * gcc.c: Remove compiler defaults for .java and .zip files.
17943 * gimple-expr.c (remove_suffix): Change as there's no longer
17944 extension than 4-letter one.
17945 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17946 (gimplify_save_expr): Likewise.
17947 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17948 as it's possible even for other languages than Java.
17949 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17950 * lto-opts.c (lto_write_options): Remove reference to Java.
17951 * opts.c (strip_off_ending): Update file extension handling.
17952 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17953 * tree-eh.c (lower_resx): Likewise.
17954 * tree.c (free_lang_data_in_type): Remove dead code.
17955 (find_decls_types_r): Likewise.
17956 (build_common_builtin_nodes): Remove Java from a comment.
17957 (verify_type): Remove dead code.
17958 * varasm.c (assemble_external): Remove Java from a comment.
17960 2017-07-14 Martin Liska <mliska@suse.cz>
17962 * opts.c (finish_options): Add quotes.
17963 (common_handle_option): Likewise.
17965 2017-07-14 Martin Liska <mliska@suse.cz>
17967 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17968 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17969 Remove N_SO_PASCAL.
17970 * dwarf2out.c (lower_bound_default): Do not handle
17972 (gen_compile_unit_die): Likewise.
17973 * gcc.c: Remove default extension binding for GNU Pascal.
17974 * stmt.c: Remove Pascal language from a comment.
17975 * xcoffout.c: Likewise.
17977 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17980 * diagnostic-show-locus.c (fixit_cmp): New function.
17981 (layout::layout): Sort m_fixit_hints.
17982 (column_range::column_range): Assert that the values are valid.
17983 (struct char_span): New struct.
17984 (correction::overwrite): New method.
17985 (struct source_line): New struct.
17986 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17987 calls in terms of classes source_line and char_span, and
17988 correction::overwrite.
17989 (selftest::test_overlapped_fixit_printing_2): New function.
17990 (selftest::diagnostic_show_locus_c_tests): Call it.
17992 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17994 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17995 early if there is no lhs.
17997 2017-07-13 Martin Liska <mliska@suse.cz>
17999 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
18000 (gen_reference_type_die): Likewise.
18001 * stor-layout.c: Remove Pascal-related comment.
18003 2017-07-13 Martin Liska <mliska@suse.cz>
18005 * opts.c (finish_options): Add quotes to error messages.
18006 (parse_sanitizer_options): Likewise.
18008 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18010 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18012 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18014 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18016 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18018 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18020 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18022 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18025 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18026 provides the hardware capability bits, define the macro
18027 __BUILTIN_CPU_SUPPORTS__.
18028 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18029 if GLIBC does not provide the hardware capability bits. Add a
18030 gcc_unreachable call if the built-in cpu function is neither
18031 __builtin_cpu_is nor __builtin_cpu_supports.
18032 (rs6000_get_function_versions_dispatcher): Change the warning
18033 that an old GLIBC is used which does not export the capability
18034 bits to be an error.
18035 * doc/extend.texi (target_clones attribute): Document the
18036 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18037 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18038 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18039 the macros defined by GCC if the newer GLIBC is available.
18041 2017-07-12 Jeff Law <law@redhat.com>
18043 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18044 remaining includes slightly.
18045 * config/riscv/riscv-builtins.c: Include profile-count.h.
18047 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18050 * config/avr/avr.c (avr_set_current_function): In diagnostic
18051 messages: Quote keywords and (parts of) identifiers.
18052 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18055 2017-07-12 Carl Love <cel@us.ibm.com>
18057 * config/rs6000/rs6000-c.c: Add support for built-in functions
18058 vector bool char vec_revb (vector bool char);
18059 vector bool short vec_revb (vector short char);
18060 vector bool int vec_revb (vector bool int);
18061 vector bool long long vec_revb (vector bool long long);
18062 * doc/extend.texi: Update the built-in documentation file for the
18063 new built-in functions.
18065 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18067 * config/s390/s390.md: Remove movcc splitter.
18069 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18071 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18072 load/store on condition.
18074 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18077 * config/avr/avr.c (avr_encode_section_info)
18078 [progmem && !TREE_READONLY]: Error if progmem object needs
18081 2017-07-11 Michael Collison <michael.collison@arm.com>
18083 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18086 2017-07-11 Carl Love <cel@us.ibm.com>
18088 * config/rs6000/rs6000-c.c: Add support for builtins
18089 vector unsigned int vec_parity_lsbb (vector signed int);
18090 vector unsigned int vec_parity_lsbb (vector unsigned int);
18091 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18092 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18093 vector unsigned long long vec_parity_lsbb (vector signed long long);
18094 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18095 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18096 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18097 * doc/extend.texi: Update the built-in documentation file for the
18098 new built-in functions.
18100 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18102 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18103 (layout::m_primary_loc): New field.
18104 (layout::layout): Initialize new field. Move location filtering
18105 logic from here to...
18106 (layout::maybe_add_location_range): ...this new method. Add
18107 support for filtering to just the lines already specified by other
18109 (layout::will_show_line_p): New method.
18110 (gcc_rich_location::add_location_if_nearby): New method.
18111 (selftest::test_add_location_if_nearby): New test function.
18112 (selftest::diagnostic_show_locus_c_tests): Call it.
18113 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18116 2017-07-11 Tom de Vries <tom@codesourcery.com>
18118 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18119 (bb_first_real_insn): New function.
18120 (nvptx_single): Add extra initialization of broadcasted condition
18123 2017-07-11 Nathan Sidwell <nathan@acm.org>
18125 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18127 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18129 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18130 Binutils doc as TEXI will mess them up.
18131 * doc/invoke.texi (AVR Options): Same here.
18133 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18135 * config/sparc/sparc.opt (mfix-ut700): New option.
18136 (mfix-gr712rc): Likewise.
18137 (sparc_fix_b2bst): New variable.
18138 * doc/invoke.texi (SPARC options): Document them.
18139 (ARM options): Fix warnings.
18140 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18141 instructions to prevent sequences that can trigger the store-store
18142 errata for certain LEON3FT processors.
18143 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18144 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18145 * config/sparc/sparc.md (fix_b2bst): New attribute.
18146 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18148 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18151 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18153 (*rsqrtsf2_sse): Ditto.
18155 (div<mode>3): Macroize insn from divdf3 and divsf3
18156 using MODEF mode iterator.
18158 2017-07-10 Martin Sebor <msebor@redhat.com>
18160 PR tree-optimization/80397
18161 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18162 instead of testing for equality to INTEGER_TYPE.
18164 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18166 * config.gcc: Remove uclibc from arc target spec.
18168 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18170 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18172 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18175 * lto-wrapper.c (remove_option): New function.
18176 (merge_and_complain): Merge PIC/PIE options more realistically.
18178 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18180 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18184 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18185 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18186 * config.in: Regenerate.
18187 * configure: Regenerate.
18188 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18189 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18190 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18191 (TARGET_GASISR_PROLOGUES): ...target mask.
18192 * common/config/avr/avr-common.c
18193 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18194 Set -mgas-isr-prologues.
18195 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18196 INSERT_PASS_BEFORE for it.
18197 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18198 * config/avr/avr.c (avr_option_override)
18199 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18200 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18201 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18202 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18203 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18205 (make_avr_pass_pre_proep): New function.
18206 (emit_push_sfr) <treg>: Add argument to function and use it
18207 instead of TMP_REG.
18208 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18209 and set machine->gasisr.yes.
18210 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18211 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18212 __gcc_isr.n_pushed to .L__stack_usage.
18213 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18214 (avr_asm_final_postscan_insn): ...this new static function.
18215 * config/avr/avr.h (machine_function)
18216 <is_no_gccisr, use_L__stack_usage>: New fields.
18217 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18218 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18219 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18220 (gasisr, *gasisr): New expander and insn.
18221 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18222 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18223 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18225 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18227 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18230 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18232 Move jump-tables out of .text again.
18235 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18236 (ASM_OUTPUT_ADDR_VEC): New function.
18237 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18238 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18239 INSN_ADDRESSes as asm comment.
18240 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18241 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18242 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18243 * config/avr/avr.md (*tablejump): Adjust comment.
18244 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18245 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18247 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18248 (avr_output_addr_vec): New proto.
18249 (avr_log_t) <insn_addresses>: New field.
18251 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18254 * config/i386/i386.c (ix86_function_arg_advance): Set
18255 outgoing_args_on_stack to true if there are outgoing arguments
18257 (ix86_function_arg): Likewise.
18258 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18259 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18260 * config/i386/i386.h (machine_function): Add
18261 outgoing_args_on_stack.
18263 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18265 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18266 supporting pthreds.
18267 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18269 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18271 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18272 (REAL_H): Remove $(MACHMODE_H).
18273 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18275 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18276 $(MACHMODE_H) and double-int.h.
18277 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18279 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18282 2017-07-07 Andrew Pinski <apinski@cavium.com>
18284 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18285 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18287 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18289 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18290 Add warning if GCC was not configured to link against a GLIBC that
18291 exports the hardware capability bits.
18292 (make_resolver_func): Make resolver function private and not a
18293 COMDAT function. Create the name with clone_function_name instead
18294 of make_unique_name.
18297 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18298 correct operand in doing the split.
18300 2017-07-07 Carl Love <cel@us.ibm.com>
18302 * config/rs6000/rs6000-c: Add support for built-in function
18303 vector unsigned short vec_pack_to_short_fp32 (vector float,
18305 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18306 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18307 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18308 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18309 (convert_4f32_8i16): Add define_expand.
18310 * doc/extend.texi: Update the built-in documentation file for the
18311 new built-in function.
18313 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18315 * config/sparc/m8.md: New file.
18316 * config/sparc/sparc.md: Include m8.md.
18318 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18320 * config/sparc/sparc.opt: New option -mvis4b.
18321 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18322 (sparc_option_override): Handle VIS4B.
18323 (enum sparc_builtins): Define
18324 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18325 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18326 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18327 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18328 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18329 (check_constant_argument): New function.
18330 (sparc_vis_init_builtins): Define builtins
18331 __builtin_vis_dictunpack{8,16,32},
18332 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18333 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18334 __builtin_vis_fpcmpde{8,16,32}shl and
18335 __builtin_vis_fpcmpur{8,16,32}shl.
18336 (sparc_expand_builtin): Check that the constant operands to
18337 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18338 constant and in range.
18339 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18341 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18342 (SPARC_IMM5_P): Likewise.
18343 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18344 (enabled): Handle vis4b.
18345 (UNSPEC_DICTUNPACK): New unspec.
18346 (UNSPEC_FPCMPSHL): Likewise.
18347 (UNSPEC_FPUCMPSHL): Likewise.
18348 (UNSPEC_FPCMPDESHL): Likewise.
18349 (UNSPEC_FPCMPURSHL): Likewise.
18350 (cpu_feature): New CPU feature `vis4b'.
18351 (dictunpack{8,16,32}): New insns.
18352 (FPCSMODE): New mode iterator.
18353 (fpcscond): New code iterator.
18354 (fpcsucond): Likewise.
18355 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18356 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18357 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18358 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18359 * config/sparc/constraints.md: Define constraints `q' for unsigned
18360 2-bit integer constants and `t' for unsigned 5-bit integer
18362 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18364 (imm5_operand_dictunpack16): Likewise.
18365 (imm5_operand_dictunpack32): Likewise.
18366 (imm2_operand): Likewise.
18367 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18368 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18369 ditunpack* and fpcmp*shl builtins.
18371 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18373 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18374 * config.in: Add HAVE_AS_SPARC6 define.
18375 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18377 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18379 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18380 (CPP_CPU_SPEC): Handle m8.
18381 (ASM_CPU_SPEC): Likewise.
18382 * config/sparc/sparc-opts.h (enum processor_type): Add
18384 * config/sparc/sparc.c (m8_costs): New struct.
18385 (sparc_option_override): Handle TARGET_CPU_m8.
18386 (sparc32_initialize_trampoline): Likewise.
18387 (sparc64_initialize_trampoline): Likewise.
18388 (sparc_issue_rate): Likewise.
18389 (sparc_register_move_cost): Likewise.
18390 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18391 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18392 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18393 (CPP_CPU_SPEC): Handle M8.
18394 (ASM_CPU_SPEC): Likewise.
18395 (AS_M8_FLAG): Define.
18396 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18397 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18398 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18400 * configure: Regenerate.
18401 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18404 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18406 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18408 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18409 ("*movdi_insn_sp32"): Do not set v3pipe.
18410 ("*movsi_insn"): Likewise.
18411 ("*movdi_insn_sp64"): Likewise.
18412 ("*movsf_insn"): Likewise.
18413 ("*movdf_insn_sp32"): Likewise.
18414 ("*movdf_insn_sp64"): Likewise.
18415 ("*zero_extendsidi2_insn_sp64"): Likewise.
18416 ("*sign_extendsidi2_insn"): Likewise.
18417 ("*mov<VM32:mode>_insn"): Likewise.
18418 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18419 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18420 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18421 ("<vlop:code><VL:mode>3"): Likewise.
18422 ("*not_<vlop:code><VL:mode>3"): Likewise.
18423 ("*nand<VL:mode>_vis"): Likewise.
18424 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18425 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18426 ("one_cmpl<VL:mode>2"): Likewise.
18427 ("faligndata<VM64:mode>_vis"): Likewise.
18428 ("alignaddrsi_vis"): Likewise.
18429 ("alignaddrdi_vis"): Likweise.
18430 ("alignaddrlsi_vis"): Likewise.
18431 ("alignaddrldi_vis"): Likewise.
18432 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18433 ("bmaskdi_vis"): Likewise.
18434 ("bmasksi_vis"): Likewise.
18435 ("bshuffle<VM64:mode>_vis"): Likewise.
18436 ("cmask8<P:mode>_vis"): Likewise.
18437 ("cmask16<P:mode>_vis"): Likewise.
18438 ("cmask32<P:mode>_vis"): Likewise.
18439 ("pdistn<P:mode>_vis"): Likewise.
18440 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18442 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18444 * config/sparc/sparc.md ("subtype"): New insn attribute.
18445 ("*wrgsr_sp64"): Set insn subtype.
18446 ("*rdgsr_sp64"): Likewise.
18447 ("alignaddrsi_vis"): Likewise.
18448 ("alignaddrdi_vis"): Likewise.
18449 ("alignaddrlsi_vis"): Likewise.
18450 ("alignaddrldi_vis"): Likewise.
18451 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18452 ("fexpand_vis"): Likewise.
18453 ("fpmerge_vis"): Likewise.
18454 ("faligndata<VM64:mode>_vis"): Likewise.
18455 ("bshuffle<VM64:mode>_vis"): Likewise.
18456 ("cmask8<P:mode>_vis"): Likewise.
18457 ("cmask16<P:mode>_vis"): Likewise.
18458 ("cmask32<P:mode>_vis"): Likewise.
18459 ("fchksm16_vis"): Likewise.
18460 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18461 ("fmean16_vis"): Likewise.
18462 ("fp<plusminus_insn>64_vis"): Likewise.
18463 ("<plusminus_insn>v8qi3"): Likewise.
18464 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18465 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18466 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18467 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18468 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18469 ("*movqi_insn"): Likewise.
18470 ("*movhi_insn"): Likewise.
18471 ("*movsi_insn"): Likewise.
18472 ("movsi_pic_gotdata_op"): Likewise.
18473 ("*movdi_insn_sp32"): Likewise.
18474 ("*movdi_insn_sp64"): Likewise.
18475 ("movdi_pic_gotdata_op"): Likewise.
18476 ("*movsf_insn"): Likewise.
18477 ("*movdf_insn_sp32"): Likewise.
18478 ("*movdf_insn_sp64"): Likewise.
18479 ("*zero_extendhisi2_insn"): Likewise.
18480 ("*zero_extendqihi2_insn"): Likewise.
18481 ("*zero_extendqisi2_insn"): Likewise.
18482 ("*zero_extendqidi2_insn"): Likewise.
18483 ("*zero_extendhidi2_insn"): Likewise.
18484 ("*zero_extendsidi2_insn_sp64"): Likewise.
18485 ("ldfsr"): Likewise.
18486 ("prefetch_64"): Likewise.
18487 ("prefetch_32"): Likewise.
18488 ("tie_ld32"): Likewise.
18489 ("tie_ld64"): Likewise.
18490 ("*tldo_ldub_sp32"): Likewise.
18491 ("*tldo_ldub1_sp32"): Likewise.
18492 ("*tldo_ldub2_sp32"): Likewise.
18493 ("*tldo_ldub_sp64"): Likewise.
18494 ("*tldo_ldub1_sp64"): Likewise.
18495 ("*tldo_ldub2_sp64"): Likewise.
18496 ("*tldo_ldub3_sp64"): Likewise.
18497 ("*tldo_lduh_sp32"): Likewise.
18498 ("*tldo_lduh1_sp32"): Likewise.
18499 ("*tldo_lduh_sp64"): Likewise.
18500 ("*tldo_lduh1_sp64"): Likewise.
18501 ("*tldo_lduh2_sp64"): Likewise.
18502 ("*tldo_lduw_sp32"): Likewise.
18503 ("*tldo_lduw_sp64"): Likewise.
18504 ("*tldo_lduw1_sp64"): Likewise.
18505 ("*tldo_ldx_sp64"): Likewise.
18506 ("*mov<VM32:mode>_insn"): Likewise.
18507 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18508 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18510 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18512 * config/sparc/sparc.md ("type"): New insn type viscmp.
18513 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18515 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18516 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18517 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18518 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18520 ("n7_vis_logical_11cycle"): Likewise.
18521 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18522 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18523 * config/sparc/niagara.md ("niag_vis"): Likewise.
18524 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18525 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18527 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18529 * config/sparc/sparc.md: New instruction type `bmask'.
18530 (bmaskdi_vis): Use the `bmask' type.
18531 (bmasksi_vis): Likewise.
18532 * config/sparc/ultra3.md (us3_array): Likewise.
18533 * config/sparc/niagara7.md (n7_array): Likewise.
18534 * config/sparc/niagara4.md (n4_array): Likewise.
18535 * config/sparc/niagara2.md (niag2_vis): Likewise.
18536 (niag3_vis): Likewise.
18537 * config/sparc/niagara.md (niag_vis): Likewise.
18539 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18541 * ipa-comdats.c: Remove optimize check from gate.
18542 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18543 for functions not optimized.
18544 (ipa_fn_summary_read): Skip optimize check.
18545 (ipa_fn_summary_write): Likewise.
18546 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18548 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18550 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18551 (check_callers): Give up on caller which is not optimized.
18552 (inline_small_functions): Likewise.
18553 (ipa_inline): Do not give up when not optimizing.
18554 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18555 away unoptimizes cdtors.
18556 (whole_program_function_and_variable_visibility): Do
18557 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18558 * ipa.c (process_references): Do not check optimize.
18559 (symbol_table::remove_unreachable_nodes): Update optimize check.
18560 (set_writeonly_bit): Update optimize check.
18561 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18562 (pass_ipa_single_use::gate): Remove.
18564 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18566 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18567 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18568 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18569 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18570 permute_load, permute_store, adjust_extract, adjust_splat,
18571 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18572 replace_swap_with_copy, dump_swap_insn_table,
18573 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18574 recombine_lvx_pattern, recombine_stvx_pattern,
18575 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18576 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18577 to file rs6000-p8swap.c.
18578 * config/rs6000/rs6000-p8swap.c: New file.
18579 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18580 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18581 and rs6000*-*-* targets.
18583 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18585 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18587 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18589 * lto-wrapper.c (merge_and_complain): Do not merge
18590 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18591 fsigned_zeros, ftrapping_math, fwrapv.
18592 (append_compiler_options): Do not track these options.
18593 (append_linker_options): Likewie
18595 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18597 * cgraphunit.c (cgraph_node::finalize_function): When
18598 !flag_toplevel_reorde set no_reorder flag.
18599 (varpool_node::finalize_decl): Likewise.
18600 (symbol_table::compile): Drop no toplevel reorder path.
18602 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18604 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18605 edges; zero probability is not better than uninitialized.
18607 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18609 * asan.h (asan_sanitize_allocas_p): Declare.
18610 * asan.c (asan_sanitize_allocas_p): New function.
18611 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18612 (handle_builtin_alloca): Likewise.
18613 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18614 if !asan_sanitize_allocas_p.
18615 * params.def (asan-instrument-allocas): Add new option.
18616 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18617 * opts.c (common_handle_option): Disable allocas sanitization for
18620 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18622 * asan.c: Include gimple-fold.h.
18623 (get_last_alloca_addr): New function.
18624 (handle_builtin_stackrestore): Likewise.
18625 (handle_builtin_alloca): Likewise.
18626 (asan_emit_allocas_unpoison): Likewise.
18627 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18628 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18629 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18630 (instrument_builtin_call): Pass gimple iterator to
18631 get_mem_refs_of_builtin_call.
18632 (last_alloca_addr): New global.
18633 * asan.h (asan_emit_allocas_unpoison): Declare.
18634 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18635 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18636 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18637 if function calls alloca.
18638 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18639 * gimple-fold.h (replace_call_with_value): Declare.
18640 * internal-fn.c: Include asan.h.
18641 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18642 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18644 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18646 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18647 (C_SELFTEST_FLAGS): New.
18648 (CPP_SELFTEST_FLAGS): New.
18649 (SELFTEST_DEPS): New, from deps of s-selftest.
18650 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18651 (CPP_SELFTEST_DEPS): New.
18652 (selftest): Add dependency on s-selftest-c++.
18653 (s-selftest): Rename to...
18654 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18655 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18656 than SELFTEST_FLAGS.
18657 (selftest-gdb): Rename to...
18658 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18660 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18661 (selftest-valgrind): Rename to...
18662 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18664 (selftest-valgrind): Reintroduce as an alias for
18665 selftest-c-valgrind.
18666 (s-selftest-c++): New.
18667 (selftest-c++-gdb): New.
18668 (selftest-c++-valgrind): New.
18670 2017-07-06 Olivier Hainque <hainque@adacore.com>
18672 * gcc.c (process_command): When deciding if undefined variables
18673 should be ignored when processing specs, accept "gcc -v" as well.
18675 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18677 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18678 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18680 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18682 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18683 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18684 * config/arm/arm-tables.opt: Regenerate.
18685 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18687 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18689 2017-07-06 Carl Love <cel@us.ibm.com>
18691 * ChangeLog: Clean up from mid air collision
18693 2017-07-06 Carl Love <cel@us.ibm.com>
18695 * config/rs6000/rs6000-c.c: Add support for built-in functions
18696 vector signed int vec_subc (vector signed int, vector signed int);
18697 vector signed __int128 vec_subc (vector signed __int128,
18698 vector signed __int128);
18699 vector unsigned __int128 vec_subc (vector unsigned __int128,
18700 vector unsigned __int128);
18701 vector signed int vec_sube (vector signed int, vector signed int,
18702 vector signed int);
18703 vector unsigned int vec_sube (vector unsigned int,
18704 vector unsigned int,
18705 vector unsigned int);
18706 vector signed __int128 vec_sube (vector signed __int128,
18707 vector signed __int128,
18708 vector signed__int128);
18709 vector unsigned __int128 vec_sube (vector unsigned __int128,
18710 vector unsigned __int128,
18711 vector unsigned __int128);
18712 vector signed int vec_subec (vector signed int, vector signed int,
18713 vector signed int);
18714 vector unsigned int vec_subec (vector unsigned int,
18715 vector unsigned int,
18716 vector unsigned int);
18717 vector signed __int128 vec_subec (vector signed __int128,
18718 vector signed __int128,
18719 vector signed__int128);
18720 vector unsigned __int128 vec_subec (vector unsigned __int128,
18721 vector unsigned __int128,
18722 vector unsigned __int128);
18723 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18724 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18725 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18726 BU_ALTIVEC_OVERLOAD_X definitions.
18727 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18728 * doc/extend.texi: Update the built-in documentation file for the new
18729 built-in functions.
18731 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18734 * diagnostic-show-locus.c (layout::layout): Use start and finish
18735 spelling location for the start and finish of each range.
18736 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18737 Add unused aspect param.
18738 * input.c (expand_location_1): Add "aspect" param, and use it
18739 to access the correct part of the location.
18740 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18742 (expand_location_to_spelling_point): Likewise.
18743 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18744 param, and pass it to expand_location_1.
18746 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18748 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18749 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18750 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18751 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18752 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18753 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18754 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18755 _mm_maskz_getmant_ss): New intrinsics.
18756 (__builtin_ia32_getexpss128_mask): Changed to ...
18757 __builtin_ia32_getexpss128_round ... this.
18758 (__builtin_ia32_getexpsd128_mask): Changed to ...
18759 __builtin_ia32_getexpsd128_round ... this.
18760 * config/i386/i386-builtin-types.def
18761 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18762 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18763 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18764 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18765 __builtin_ia32_getmantss_mask_round): New builtins.
18766 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18767 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18768 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18769 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18770 * config/i386/sse.md
18771 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18772 avx512f_sgetexp<mode><mask_scalar_name>
18773 <round_saeonly_scalar_name> ... this.
18774 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18775 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18776 vgetexp<ssescalarmodesuffix>
18777 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18778 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18779 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18780 avx512f_vgetmant<mode><mask_scalar_name>
18781 <round_saeonly_scalar_name> ... this.
18782 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18783 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18784 vgetmant<ssescalarmodesuffix>
18785 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18786 %0<mask_scalar_operand4>, %1, %2
18787 <round_saeonly_scalar_mask_op4>, %3} ... this.
18788 * config/i386/subst.md (mask_scalar_operand4,
18789 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18790 round_saeonly_scalar_nimm_predicate): New subst attributes.
18792 2017-07-06 Julia Koval <julia.koval@intel.com>
18794 * config/i386/i386.c (ix86_erase_embedded_rounding):
18795 Remove code for old rounding pattern.
18797 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18799 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18801 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18803 * doc/sourcebuild.texi (Test Directives, Variants of
18804 dg-require-support): Add documentation for dg-require-stack-check.
18806 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18808 * config/i386/subst.md (mask_scalar, round_scalar,
18809 round_saeonly_scalar): New meta-templates.
18810 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18811 round_scalar_mask_operand3, round_scalar_mask_op3,
18812 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18813 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18814 round_saeonly_scalar_constraint,
18815 round_saeonly_scalar_prefix): New subst attribute.
18816 * config/i386/sse.md
18817 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18818 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18819 <round_scalar_name> ... this.
18820 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18821 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18822 <round_scalar_name> ... this.
18823 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18824 <sse>_vm<code><mode>3<mask_scalar_name>
18825 <round_saeonly_scalar_name> ... this.
18826 (v<plusminus_mnemonic><ssescalarmodesuffix>
18827 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18828 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18829 v<plusminus_mnemonic><ssescalarmodesuffix>
18830 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18831 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18832 (v<multdiv_mnemonic><ssescalarmodesuffix>
18833 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18834 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18835 v<multdiv_mnemonic><ssescalarmodesuffix>
18836 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18837 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18838 (v<maxmin_float><ssescalarmodesuffix>
18839 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18840 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18841 v<maxmin_float><ssescalarmodesuffix>
18842 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18843 %0<mask_scalar_operand3>, %1, %<iptr>2
18844 <round_saeonly_scalar_mask_op3>} ... this.
18846 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18848 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18849 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18851 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18852 Alan Hayward <alan.hayward@arm.com>
18853 David Sherwood <david.sherwood@arm.com>
18855 * combine.c (simplify_if_then_else): Remove "enum" before
18857 * compare-elim.c (can_eliminate_compare): Likewise.
18858 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18860 (aarch64_lookup_simd_builtin_type): Likewise.
18861 (aarch64_simd_builtin_type): Likewise.
18862 (aarch64_init_simd_builtin_types): Likewise.
18863 (aarch64_simd_expand_args): Likewise.
18864 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18866 (aarch64_reverse_mask): Likewise.
18867 (aarch64_simd_emit_reg_reg_move): Likewise.
18868 (aarch64_gen_adjusted_ldpstp): Likewise.
18869 (aarch64_ccmp_mode_to_code): Likewise.
18870 (aarch64_operands_ok_for_ldpstp): Likewise.
18871 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18872 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18874 (aarch64_min_divisions_for_recip_mul): Likewise.
18875 (aarch64_reassociation_width): Likewise.
18876 (aarch64_get_condition_code_1): Likewise.
18877 (aarch64_simd_emit_reg_reg_move): Likewise.
18878 (aarch64_simd_attr_length_rglist): Likewise.
18879 (aarch64_reverse_mask): Likewise.
18880 (aarch64_operands_ok_for_ldpstp): Likewise.
18881 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18882 (aarch64_gen_adjusted_ldpstp): Likewise.
18883 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18885 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18886 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18887 (arm_lookup_simd_builtin_type): Likewise.
18888 (arm_simd_builtin_type): Likewise.
18889 (arm_init_simd_builtin_types): Likewise.
18890 (arm_expand_builtin_args): Likewise.
18891 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18892 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18893 (ft32_setup_incoming_varargs): Likewise.
18894 (ft32_function_arg): Likewise.
18895 (ft32_function_arg_advance): Likewise.
18896 (ft32_pass_by_reference): Likewise.
18897 (ft32_arg_partial_bytes): Likewise.
18898 (ft32_valid_pointer_mode): Likewise.
18899 (ft32_addr_space_pointer_mode): Likewise.
18900 (ft32_addr_space_legitimate_address_p): Likewise.
18901 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18903 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18904 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18905 (iamcu_alignment): Likewise.
18906 (canonicalize_vector_int_perm): Likewise.
18907 (ix86_noce_conversion_profitable_p): Likewise.
18908 (ix86_mpx_bound_mode): Likewise.
18909 (ix86_operands_ok_for_move_multiple): Likewise.
18910 * config/microblaze/microblaze-protos.h
18911 (microblaze_expand_conditional_branch_reg): Likewise.
18912 * config/microblaze/microblaze.c
18913 (microblaze_expand_conditional_branch_reg): Likewise.
18914 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18916 (rs6000_reassociation_width): Likewise.
18917 (rs6000_invalid_binary_op): Likewise.
18918 (fusion_p9_p): Likewise.
18919 (emit_fusion_p9_load): Likewise.
18920 (emit_fusion_p9_store): Likewise.
18921 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18923 (riscv_hard_regno_mode_ok_p): Likewise.
18924 (riscv_address_insns): Likewise.
18925 (riscv_split_symbol): Likewise.
18926 (riscv_legitimize_move): Likewise.
18927 (riscv_function_value): Likewise.
18928 (riscv_hard_regno_nregs): Likewise.
18929 (riscv_expand_builtin): Likewise.
18930 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18931 (riscv_build_integer): Likewise.
18932 (riscv_split_integer): Likewise.
18933 (riscv_legitimate_constant_p): Likewise.
18934 (riscv_cannot_force_const_mem): Likewise.
18935 (riscv_regno_mode_ok_for_base_p): Likewise.
18936 (riscv_valid_base_register_p): Likewise.
18937 (riscv_valid_offset_p): Likewise.
18938 (riscv_valid_lo_sum_p): Likewise.
18939 (riscv_classify_address): Likewise.
18940 (riscv_legitimate_address_p): Likewise.
18941 (riscv_address_insns): Likewise.
18942 (riscv_load_store_insns): Likewise.
18943 (riscv_force_binary): Likewise.
18944 (riscv_split_symbol): Likewise.
18945 (riscv_force_address): Likewise.
18946 (riscv_legitimize_address): Likewise.
18947 (riscv_move_integer): Likewise.
18948 (riscv_legitimize_const_move): Likewise.
18949 (riscv_legitimize_move): Likewise.
18950 (riscv_address_cost): Likewise.
18951 (riscv_subword): Likewise.
18952 (riscv_output_move): Likewise.
18953 (riscv_canonicalize_int_order_test): Likewise.
18954 (riscv_emit_int_order_test): Likewise.
18955 (riscv_function_arg_boundary): Likewise.
18956 (riscv_pass_mode_in_fpr_p): Likewise.
18957 (riscv_pass_fpr_single): Likewise.
18958 (riscv_pass_fpr_pair): Likewise.
18959 (riscv_get_arg_info): Likewise.
18960 (riscv_function_arg): Likewise.
18961 (riscv_function_arg_advance): Likewise.
18962 (riscv_arg_partial_bytes): Likewise.
18963 (riscv_function_value): Likewise.
18964 (riscv_pass_by_reference): Likewise.
18965 (riscv_setup_incoming_varargs): Likewise.
18966 (riscv_print_operand): Likewise.
18967 (riscv_elf_select_rtx_section): Likewise.
18968 (riscv_save_restore_reg): Likewise.
18969 (riscv_for_each_saved_reg): Likewise.
18970 (riscv_register_move_cost): Likewise.
18971 (riscv_hard_regno_mode_ok_p): Likewise.
18972 (riscv_hard_regno_nregs): Likewise.
18973 (riscv_class_max_nregs): Likewise.
18974 (riscv_memory_move_cost): Likewise.
18975 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18976 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18977 (rl78_addr_space_address_mode): Likewise.
18978 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18980 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18981 (rs6000_reassociation_width): Likewise.
18982 (rs6000_invalid_binary_op): Likewise.
18983 (fusion_p9_p): Likewise.
18984 (emit_fusion_p9_load): Likewise.
18985 (emit_fusion_p9_store): Likewise.
18986 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18987 (ok_for_simple_move_operands): Likewise.
18988 (ok_for_simple_move_strict_operands): Likewise.
18989 (ok_for_simple_arith_logic_operands): Likewise.
18990 (visium_legitimize_reload_address): Likewise.
18991 (visium_select_cc_mode): Likewise.
18992 (output_cbranch): Likewise.
18993 (visium_split_double_move): Likewise.
18994 (visium_expand_copysign): Likewise.
18995 (visium_expand_int_cstore): Likewise.
18996 (visium_expand_fp_cstore): Likewise.
18997 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18998 (visium_function_arg): Likewise.
18999 (visium_function_arg_advance): Likewise.
19000 (visium_libcall_value): Likewise.
19001 (visium_setup_incoming_varargs): Likewise.
19002 (visium_legitimate_constant_p): Likewise.
19003 (visium_legitimate_address_p): Likewise.
19004 (visium_legitimize_address): Likewise.
19005 (visium_secondary_reload): Likewise.
19006 (visium_register_move_cost): Likewise.
19007 (visium_memory_move_cost): Likewise.
19008 (prepare_move_operands): Likewise.
19009 (ok_for_simple_move_operands): Likewise.
19010 (ok_for_simple_move_strict_operands): Likewise.
19011 (ok_for_simple_arith_logic_operands): Likewise.
19012 (visium_function_value_1): Likewise.
19013 (rtx_ok_for_offset_p): Likewise.
19014 (visium_legitimize_reload_address): Likewise.
19015 (visium_split_double_move): Likewise.
19016 (visium_expand_copysign): Likewise.
19017 (visium_expand_int_cstore): Likewise.
19018 (visium_expand_fp_cstore): Likewise.
19019 (visium_split_cstore): Likewise.
19020 (visium_select_cc_mode): Likewise.
19021 (visium_split_cbranch): Likewise.
19022 (output_cbranch): Likewise.
19023 (visium_print_operand_address): Likewise.
19024 * expmed.c (flip_storage_order): Likewise.
19025 * expmed.h (emit_cstore): Likewise.
19026 (flip_storage_order): Likewise.
19027 * genrecog.c (validate_pattern): Likewise.
19028 * hsa-gen.c (gen_hsa_addr): Likewise.
19029 * internal-fn.c (expand_arith_overflow): Likewise.
19030 * ira-color.c (allocno_copy_cost_saving): Likewise.
19031 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19032 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19033 (process_invariant_for_inheritance): Likewise.
19034 * lra-eliminations.c (move_plus_up): Likewise.
19035 * omp-low.c (lower_oacc_reductions): Likewise.
19036 * simplify-rtx.c (simplify_subreg): Likewise.
19037 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19038 (TARGET_CHKP_BOUND_MODE): Likewise..
19039 * targhooks.c (default_chkp_bound_mode): Likewise.
19040 (default_setup_incoming_vararg_bounds): Likewise.
19041 * targhooks.h (default_chkp_bound_mode): Likewise.
19042 (default_setup_incoming_vararg_bounds): Likewise.
19043 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19044 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19045 (have_whole_vector_shift): Likewise.
19046 * tree-vect-stmts.c (vectorizable_load): Likewise.
19047 * doc/tm.texi: Regenerate.
19049 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19051 Graceful degrade if Binutils PR21472 is not available.
19054 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19055 .rodata in flash test fails.
19056 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19057 * confgure: Regenerate.
19058 * config.in: Regenerate.
19059 * config/avr/avr.c (avr_asm_named_section)
19060 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19061 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19062 (avr_asm_init_sections): Same.
19064 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19066 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19067 (fma<VH:mode>4_intrinsic): Likewise.
19068 (*fmsub<VCVTF:mode>4): Likewise.
19069 (*fmsub<VH:mode>4_intrinsic): Likewise.
19071 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19074 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19075 Don't depend on "optimize > 0".
19076 (out_movhi_r_mr, out_movqi_mr_r): Same.
19077 (out_movhi_mr_r, out_movqi_r_mr): Same.
19078 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19079 io_address_operand on "optimize > 0".
19081 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19083 * tree-loop-distribution.c: Add general explanantion on the pass.
19084 (generate_loops_for_partition): Mark distributed loop.
19085 (pg_add_dependence_edges): New parameter. Handle alias data
19086 dependence specially and record it in the parameter if asked.
19087 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19088 (init_partition_graph_vertices, add_partition_graph_edge): New.
19089 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19090 (free_partition_graph_vdata, build_partition_graph): New.
19091 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19092 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19093 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19094 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19095 (version_for_distribution_p, finalize_partitions): New.
19096 (distribute_loop): Handle alias data dependence specially. Factor
19097 out loop fusion code as functions and call these functions.
19099 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19101 * tree-loop-distribution.c (classify_partition): New parameter and
19102 better handle reduction statement.
19103 (rdg_build_partitions): Revise comment.
19104 (distribute_loop): Compute statements in all partitions and pass it
19105 to classify_partition.
19107 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19109 * tree-loop-distribution.c (enum partition_type): New.
19110 (struct partition): New field type.
19111 (partition_merge_into): Add parameter. Update partition type.
19112 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19113 (build_rdg_partition_for_vertex): Compute partition type.
19114 (rdg_build_partitions): Dump partition type.
19115 (distribute_loop): Update calls to partition_merge_into.
19117 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19119 * tree-loop-distribution.c (struct ddr_hasher): New.
19120 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19122 (classify_partition): Call get_data_dependence.
19123 (pg_add_dependence_edges): Ditto.
19124 (distribute_loop): Release data dependence hash table.
19126 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19128 * tree-loop-distribution.c (ref_base_address): Delete.
19129 (similar_memory_accesses): Rename ...
19130 (share_memory_accesses): ... to this. Check if partitions access
19131 the same memory reference.
19132 (distribute_loop): Call share_memory_accesses.
19134 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19136 * tree-loop-distribution.c (struct partition): New field recording
19137 its data reference.
19138 (partition_alloc, partition_free): Init and release data refs.
19139 (partition_merge_into): Merge data refs.
19140 (build_rdg_partition_for_vertex): Collect data refs for partition.
19141 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19143 (distribute_loop): Remve data refs from vertice data of partition
19146 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19148 * tree-loop-distribution.c (params.h): Include header file.
19149 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19150 (datarefs_vec): New global var.
19151 (create_rdg_vertices): Use datarefs_vec directly.
19152 (free_rdg): Don't free data references.
19153 (build_rdg): Update use. Don't free data references.
19154 (distribute_loop): Compute global variable for data references.
19155 Bail out if there are too many data references.
19157 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19159 * tree-loop-distribution.c (loop_nest): New global var.
19160 (build_rdg): Use loop directly, rather than loop nest.
19161 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19163 (distribute_loop): Compute global variable loop nest. Update use.
19165 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19167 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19168 (partition_merge_into): New parameter. Dump reason for fusion.
19169 (distribute_loop): Update use of partition_merge_into.
19171 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19173 * tree-loop-distribution.c (bb_top_order_index): New.
19174 (bb_top_order_index_size, bb_top_order_cmp): New.
19175 (stmts_from_loop): Use topological order.
19176 (pass_loop_distribution::execute): Compute and release topological
19177 order for basic blocks.
19179 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19181 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19184 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19186 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19187 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19188 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19189 * internal-fn.def (LOOP_DIST_ALIAS): New.
19190 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19191 (fold_loop_internal_call): ... this.
19192 (vect_loop_dist_alias_call): New function.
19193 (set_uid_loop_bbs): Call fold_loop_internal_call.
19194 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19197 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19200 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19201 Require dead FLAGS_REG at the beginning of a peephole.
19203 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19206 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19207 arguments in the call to __builtin_ia32_sbb_u32.
19208 (_subborrow_u64): Swap _X and _Y arguments in the call to
19209 __builtin_ia32_sbb_u64.
19211 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19214 * tree-vrp.c (compare_assert_loc): Turn into a function template
19215 with stable template parameter. Only test if a->e is NULL,
19216 !a->e == !b->e has been verified already. Use e == NULL or
19217 e != NULL instead of e or ! e tests. If stable is true, don't use
19218 iterative_hash_expr, on the other side allow a or b or both NULL
19219 and sort the NULLs last.
19220 (process_assert_insertions): Sort using compare_assert_loc<false>
19221 instead of compare_assert_loc, later sort using
19222 compare_assert_loc<true> before calling process_assert_insertions_for
19223 in a loop. Use break instead of continue once seen NULL pointer.
19225 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19227 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19228 Cortex-R7 and Cortex-R8 processors.
19230 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19232 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19233 uninitialized while src is not.
19235 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19237 * common/config/arm/arm-common.c: Adjust include path for
19239 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19240 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19241 (arm-cpu-data.h): Likewise.
19242 (arm-cpu-cdata.h): Likewise.
19243 * config/arm/arm-cpu.h: Delete.
19244 * config/arm/arm-cpu-cdata.h: Delete.
19245 * config/arm/arm-cpu-data.h: Delete.
19247 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19249 * config/arm/arm-cpus.in (cortex-a55): New.
19250 (cortex-a75): Likewise.
19251 (cortex-a75.cortex-a55): Likewise.
19252 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19254 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19255 * config/arm/arm-cpu-cdata.h: Regenerate.
19256 * config/arm/arm-cpu-data.h: Regenerate.
19257 * config/arm/arm-cpu.h: Regenerate.
19258 * config/arm/arm-tables.opt: Regenerate.
19259 * config/arm/arm-tune.md: Regenerate.
19261 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19263 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19265 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19267 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19270 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19272 PR tree-optimization/81292
19273 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19274 full_string_p, also call adjust_related_strinfos if the adjustment
19275 is simple, otherwise invalidate related strinfos.
19277 2017-07-04 Martin Liska <mliska@suse.cz>
19280 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19281 newly created variable as DECL_IGNORED_P.
19283 2017-07-04 Martin Liska <mliska@suse.cz>
19286 * ipa-inline.c (inline_small_functions):
19287 Use xstrdup_for_dump.
19289 2017-07-04 Tom de Vries <tom@codesourcery.com>
19291 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19293 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19296 * config/darwin.c (darwin_function_switched_text_sections):
19299 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19301 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19303 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19305 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19307 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19309 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19310 min_profitable_iters, and th as inclusive lower bounds.
19311 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19312 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19313 for min_profitable_iters and min_profitable_estimate.
19314 (vect_transform_loop): Treat th as an inclusive lower bound.
19315 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19317 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19320 * config/darwin.c (darwin_function_switched_text_sections):
19321 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19322 in two pieces, and suppress the use of buf.
19324 2017-07-03 Nathan Sidwell <nathan@acm.org>
19326 * hash-table.h (hash_table_mod1): Fix indentation.
19328 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19330 PR middle-end/81290
19331 * predict.c (force_edge_cold): Be more careful about propagation
19333 * profile-count.h (profile_probability::guessed,
19334 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19336 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19338 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19340 * doc/invoke.texi (rcpc architecture extension): Document it.
19342 2017-07-03 Richard Biener <rguenther@suse.de>
19344 PR tree-optimization/60510
19345 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19346 the scalar reduction PHI and use it.
19347 (vectorizable_reduction): Properly guard the single_defuse_cycle
19348 path for non-SLP reduction chains where we cannot use it.
19349 Rework reduc_def/index and vector type deduction. Rework
19350 vector operand gathering during reduction op code-gen.
19351 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19352 chains dissolve the chain and leave it to non-SLP reduction
19355 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19357 * tree-data-ref.h (dr_alignment): Declare.
19358 * tree-data-ref.c (dr_alignment): New function.
19359 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19360 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19362 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19364 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19366 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19367 and base_misalignment fields.
19368 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19369 * tree-data-ref.c: Include builtins.h.
19370 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19371 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19372 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19373 * tree-vect-data-refs.c: Include tree-cfg.h.
19374 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19375 fields instead of calculating an alignment here.
19376 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19377 innermost_loop_behavior fields.
19379 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19381 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19383 (DR_STEP_ALIGNMENT): New macro.
19384 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19385 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19386 (create_data_ref): Print it.
19387 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19388 to tell whether the step preserves vector (mis)alignment.
19389 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19390 Move the check for an integer step and generalise to all INTEGER_CST.
19391 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19392 Print the outer step alignment.
19394 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19396 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19397 with offset_alignment.
19398 (DR_ALIGNED_TO): Delete.
19399 (DR_OFFSET_ALIGNMENT): New macro.
19400 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19401 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19402 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19403 (create_data_ref): Likewise.
19404 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19405 (vect_analyze_data_refs): Likewise.
19406 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19407 creating dummy innermost behavior.
19409 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19411 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19412 with a "innermost_loop_behavior *" and refeence tree.
19413 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19414 (create_data_ref): Update call accordingly.
19415 * tree-predcom.c (find_looparound_phi): Likewise.
19417 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19419 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19420 fields with dr_wrt_vec_loop.
19421 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19422 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19423 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19424 (vect_dr_behavior): New function.
19425 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19426 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19427 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19428 track whether the step preserves the misalignment.
19429 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19430 Use vect_dr_behavior.
19431 (vect_setup_realignment): Update call accordingly.
19432 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19433 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19434 call to vect_create_addr_base_for_vector_ref.
19435 (vect_create_cond_for_align_checks): Likewise.
19436 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19437 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19438 (vect_recog_mask_conversion_pattern): Likewise.
19439 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19440 (new_stmt_vec_info): Remove redundant zeroing.
19442 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19444 * common/config/arm/arm-common.c (arm_be8_option): New function.
19445 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19446 (ISA_ARMv6): Add isa_bit_be8.
19447 * config/arm/arm.h (arm_be8_option): Add prototype.
19448 (BE8_SPEC_FUNCTION): New define.
19449 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19450 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19451 (mlittle-endian): Similarly.
19452 (mbe8, mbe32): New options.
19453 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19454 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19456 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19458 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19460 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19462 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19463 (cleanup_tree_cfg_bb): Use it.
19464 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19466 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19468 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19471 * loop-doloop.c (add_test): Update profile.
19473 2017-07-03 Martin Liska <mliska@suse.cz>
19476 * sanopt.c (rewrite_usage_of_param): New function.
19477 (sanitize_rewrite_addressable_params): Likewise.
19478 (pass_sanopt::execute): Call rewrite_usage_of_param.
19480 2017-07-03 Richard Biener <rguenther@suse.de>
19482 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19483 back to using VIEW_CONVERT_EXPR.
19485 2017-07-03 Martin Liska <mliska@suse.cz>
19488 * doc/extend.texi: Document when a resolver function is
19489 generated for target_clones.
19491 2017-07-03 Martin Liska <mliska@suse.cz>
19493 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19494 and shadow memory of auto variables which are subject of
19495 use-after-scope sanitization.
19496 (asan_expand_mark_ifn): Add do set only when is_poison.
19498 2016-07-03 Richard Biener <rguenther@suse.de>
19500 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19502 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19503 (vectorizable_reduction): Perform reduction PHI creation when
19504 visiting a reduction PHI and adjust and simplify code generation
19505 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19506 (vect_transform_loop): Visit reduction PHIs.
19507 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19508 defs into the SLP tree.
19509 (vect_build_slp_tree): Reduction defs terminate the recursion.
19510 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19512 (vect_get_vec_defs_for_stmt_copy): Export.
19513 (vect_get_vec_defs): Likewise.
19514 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19516 (vect_get_vec_defs_for_stmt_copy): Declare.
19517 (vect_get_vec_defs): Likewise.
19519 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19521 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19522 parameter with a "loop" parameter and use it instead of the
19523 loop containing DR_STMT. Don't check simple_iv when doing
19524 BB analysis. Describe the two analysis modes in the comment.
19526 2017-07-03 Tom de Vries <tom@codesourcery.com>
19528 PR tree-optimization/69468
19529 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19530 (find_same_succ_bb): Handle ignore_edge_flags.
19532 2017-07-03 Tom de Vries <tom@codesourcery.com>
19534 PR tree-optimization/81192
19535 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19537 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19539 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19541 2017-07-03 Tom de Vries <tom@codesourcery.com>
19543 PR tree-optimization/81192
19544 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19545 BB_SAME_SUCC (bb) == NULL.
19547 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19549 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19552 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19554 * dumpfile.c: Include profile-count.h
19555 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19557 (insert_cond_bb): Update profile.
19558 * tree-cfg.h (insert_cond_bb): Update prototype.
19559 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19560 * tree-dump.c: Do not include tree-cfg.
19562 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19564 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19566 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19568 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19571 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19573 * tree-complex.c (expand_complex_div_wide): update profile.
19575 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19576 Alan Hayward <alan.hayward@arm.com>
19577 David Sherwood <david.sherwood@arm.com>
19579 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19580 (CORETYPES_H): New define.
19581 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19582 (insn-modes-inline.h, s-modes-inline-h): New rules.
19583 (generated_files): Add insn-modes-inline.h.
19584 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19585 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19586 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19587 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19588 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19589 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19590 (build/gencodes.o, build/genconditions.o): Likewise.
19591 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19592 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19593 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19594 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19595 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19596 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19597 * coretypes.h: Include everything up to real.h for generators.
19598 Include insn-modes.h first. Include wide-int-print.h after
19599 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19600 * machmode.h: Don't include insn-modes.h here.
19601 * function-tests.c: Remove includes of signop.h, machmode.h,
19602 double-int.h and wide-int.h.
19604 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19606 * optc-save-gen.awk: Likewise.
19607 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19608 * godump.c: Remove include of wide-int-print.h.
19609 * pretty-print.h: Likewise.
19610 * wide-int-print.cc: Likewise.
19611 * wide-int.cc: Likewise.
19612 * hash-map-tests.c: Remove include of signop.h.
19613 * hash-set-tests.c: Likewise.
19614 * rtl-tests.c: Likewise.
19615 * mkconfig.sh: Remove include of machmode.h.
19616 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19618 (emit_insn_modes_inline_h): ...this new function. Emit the code
19619 into an insn-modes-inline.h header file, adding appropriate
19620 include guards and end comments.
19621 (emit_insn_modes_c_header): Remove include of machmode.h.
19622 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19624 (main): Handle -i flag and call emit_insn_modes_inline_h when
19627 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19629 * tree-ssa-strlen.c (strinfo): Rename the length field to
19630 nonzero_chars. Add a full_string_p field.
19631 (compare_nonzero_chars, zero_length_string_p): New functions.
19632 (get_addr_stridx): Add an offset_out parameter.
19633 Use compare_nonzero_chars.
19634 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19635 (new_strinfo): Update after above changes to strinfo.
19636 (set_endptr_and_length): Set full_string_p.
19637 (get_string_length): Update after above changes to strinfo.
19638 (unshare_strinfo): Update call to new_strinfo.
19639 (maybe_invalidate): Likewise.
19640 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19641 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19642 as a uhwi instead of an shwi. Update after above changes to
19643 strinfo and new_strinfo.
19644 (zero_length_string): Assert that chainsi contains full strings.
19645 Use zero_length_string_p. Update call to new_strinfo.
19646 (adjust_related_strinfos): Update after above changes to strinfo.
19647 Copy full_string_p from origsi.
19648 (adjust_last_stmt): Use zero_length_string_p.
19649 (handle_builtin_strlen): Update after above changes to strinfo and
19650 new_strinfo. Install the lhs as the string length if the previous
19651 entry didn't describe a full string.
19652 (handle_builtin_strchr): Update after above changes to strinfo
19654 (handle_builtin_strcpy): Likewise.
19655 (handle_builtin_strcat): Likewise.
19656 (handle_builtin_malloc): Likewise.
19657 (handle_pointer_plus): Likewise.
19658 (handle_builtin_memcpy): Likewise. Track nonzero characters
19659 that aren't necessarily followed by a nul terminator.
19660 (handle_char_store): Likewise.
19662 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19664 PR tree-optimization/80769
19665 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19666 for malloc and calloc. Document the new invariant that all related
19667 strinfos have delayed lengths or none do.
19668 (verify_related_strinfos): Move earlier in file.
19669 (set_endptr_and_length): New function, split out from...
19670 (get_string_length): ...here. Also set the lengths of related
19672 (zero_length_string): Assert that chainsi has known (rather than
19674 (adjust_related_strinfos): Likewise.
19676 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19678 PR tree-optimization/81136
19679 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19680 assert that two references with the same misalignment have the same
19681 compile-time misalignment if those compile-time misalignments
19684 2017-07-01 Andi Kleen <ak@linux.intel.com>
19686 * print-tree.c (print_node): Print all attributes.
19688 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19690 * cfg.c (scale_bbs_frequencies): New function.
19691 * cfg.h (scale_bbs_frequencies): Declare it.
19692 * cfgloopanal.c (single_likely_exit): Cleanup.
19693 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19695 (scale_loop_profile): Likewise.
19696 (loop_version): Likewise.
19697 (create_empty_loop_on_edge): Update.
19698 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19699 scale_loop_frequencies, scale_loop_profile, loopify,
19700 loop_version): Update prototypes.
19701 * modulo-sched.c (sms_schedule): Update.
19702 * predict.c (unlikely_executed_edge_p): Also check probability.
19703 (probably_never_executed_edge_p): Fix typo.
19704 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19705 * tree-parloops.c (gen_parallel_loop): Update.
19706 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19707 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19708 * tree-ssa-loop-split.c (split_loop): Update.
19709 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19710 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19711 (vect_loop_versioning): Update.
19712 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19714 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19716 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19718 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19720 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19721 to keep profile consistent.
19723 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19725 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19726 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19727 * profile-count.h (max_safe_multiplier): Make unsigned.
19728 (profile_count::guessed_zero): New.
19730 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19732 * bb-reorder.c (fix_up_crossing_landing_pad,
19733 fix_crossing_conditional_branches): Use make_single_succ_edge
19734 to keep profile consistent.
19736 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19738 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19741 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19744 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19745 the right scopes, make sure cond_jump isn't preserved between multiple
19746 iterations. Search for fallthru edge whenever there are 3+ edges and
19747 use find_fallthru_edge for it.
19749 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19751 Patch by Alexander Monakov <amonakov@ispras.ru>
19752 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19753 probabilities consistently.
19755 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19757 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19758 emit_cmp_and_jump_insns.
19760 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19763 * tree-inline.c (expand_call_inline): Combine profile statuses.
19765 2017-06-30 Andrew Pinski <apinski@cavium.com>
19767 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19768 fold_stmt returned true.
19770 2017-06-30 Nathan Sidwell <nathan@acm.org>
19772 * ggc.h (empty_string): Delete.
19773 * cfgexpand.c (expand_asm_stmt): Use plain "".
19774 * optabs.c (expand_asm_memory_barrier): Likewise.
19775 * stringpool.c (empty_string): Delete.
19776 (digit_vector, digit_string): Delete.
19777 (ggc_alloc_string): Use plain "", don't optimize single digit
19778 strings. Use ggc_alloc_atomic.
19780 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19782 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19783 comparison set and one other set, use the cost of the non-comparison
19786 2017-06-30 Nathan Sidwell <nathan@acm.org>
19788 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19791 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19793 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19794 loops. Remove now unneeded calls to gimple_switch_set_label() that
19795 just set removed labels to NULL_TREE.
19797 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19799 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19800 (set_range_info): ...here. Only call set_range_info_raw if domain
19802 (set_nonzero_bits): Call set_range_info_raw.
19803 * tree-ssanames.h (set_range_info_raw): New.
19805 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19808 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19809 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19810 of nonimmediate_operand and <store_mask_constraint> instead of m
19811 for the input operand. For V8FI iterator, always split if input
19812 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19813 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19814 <store_mask_predicate> instead of register_operand and
19815 <store_mask_constraint> instead of v for the input operand. Make
19816 sure both operands aren't MEMs for if not <mask_applied>.
19818 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19820 * lto-wrapper.c (copy_file) Close both file descriptors before
19823 2017-06-30 Martin Liska <mliska@suse.cz>
19826 * multiple_target.c (create_dispatcher_calls): Make ifunc
19827 also for function that don't have calls or are not referenced.
19829 2017-06-30 Richard Biener <rguenther@suse.de>
19831 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19832 analyze the first scalar stmt. Move vector type computation
19833 for the BB case here from ...
19834 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19835 live operation processing in the SLP case properly.
19837 2017-06-30 Richard Biener <rguenther@suse.de>
19839 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19841 2017-06-30 Martin Liska <mliska@suse.cz>
19844 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19845 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19847 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19849 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19852 2017-06-30 Richard Biener <rguenther@suse.de>
19854 PR tree-optimization/81249
19855 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19856 condition reduction result to original scalar type.
19858 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19860 * profile-count.h (enum profile_quality): Fix typos and whitespace
19863 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19865 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19866 type for branch probabilities.
19868 2017-06-29 Julian Brown <julian@codesourcery.com>
19869 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19871 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19872 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19873 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19874 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19876 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19878 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19879 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19880 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19881 CC usage from generic code to here.
19882 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19883 CC usage into the target macros.
19885 2017-06-29 Maya Rashish <coypu@sdf.org>
19887 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19890 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19892 * arm/arm-builtins.c: Include profile-count.h
19893 * except.c (sjlj_emit_function_enter): Use
19894 profile_probability::unlikely.
19896 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19898 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19899 and tocrel_offset be pointer args rather than implicitly using
19901 (legitimate_constant_pool_address_p, rs6000_emit_move,
19902 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19903 tocrel_offset and use in toc_relative_expr_p call.
19904 (print_operand, print_operand_address): Use static tocrel_base_oac
19905 and tocrel_offset_oac.
19906 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19909 2017-06-29 Maya Rashish <coypu@sdf.org>
19911 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19913 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19915 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19916 objects, take into account only the alignment of 'op0' and 'mode1' if
19919 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19921 * ccmp.c (ccmp_tree_comparison_p): New function.
19922 (ccmp_candidate_p): Update to use above function.
19923 (get_compare_parts): New function.
19924 (expand_ccmp_next): Update to use new functions.
19925 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19927 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19928 take mode as argument.
19929 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19930 * expr.c (expand_expr_real_1): Pass mode as argument.
19932 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19934 * combine.c (combine_instructions): Print insns to dump_file, together
19937 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19939 * asan.c (asan_emit_stack_protection): Update.
19940 (create_cond_insert_point): Update.
19941 * auto-profile.c (afdo_propagate_circuit): Update.
19942 * basic-block.h (struct edge_def): Turn probability to
19943 profile_probability.
19944 (EDGE_FREQUENCY): Update.
19945 * bb-reorder.c (find_traces_1_round): Update.
19946 (better_edge_p): Update.
19947 (sanitize_hot_paths): Update.
19948 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19949 (make_single_succ_edge): Update.
19950 (check_bb_profile): Update.
19951 (dump_edge_info): Update.
19952 (update_bb_profile_for_threading): Update.
19953 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19954 probabilitycount to 0.
19955 * cfgbuild.c (compute_outgoing_frequencies): Update.
19956 * cfgcleanup.c (try_forward_edges): Update.
19957 (outgoing_edges_match): Update.
19958 (try_crossjump_to_edge): Update.
19959 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19960 (expand_gimple_tailcall): Update.
19961 (construct_init_block): Use make_single_succ_edge.
19962 (construct_exit_block): Use make_single_succ_edge.
19963 * cfghooks.c (verify_flow_info): Update.
19964 (redirect_edge_succ_nodup): Update.
19965 (split_edge): Update.
19966 (account_profile_record): Update.
19967 * cfgloopanal.c (single_likely_exit): Update.
19968 * cfgloopmanip.c (scale_loop_profile): Update.
19969 (set_zero_probability): Remove.
19970 (duplicate_loop_to_header_edge): Update.
19971 * cfgloopmanip.h (loop_version): Update prototype.
19972 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19973 (force_nonfallthru_and_redirect): Update.
19974 (update_br_prob_note): Update.
19975 (rtl_verify_edges): Update.
19976 (purge_dead_edges): Update.
19977 (rtl_lv_add_condition_to_bb): Update.
19978 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19979 * cgraphunit.c (init_lowered_empty_function): Update.
19980 (cgraph_node::expand_thunk): Update.
19981 * cilk-common.c: Include profile-count.h
19982 * dojump.c (inv): Remove.
19983 (jumpifnot): Update.
19984 (jumpifnot_1): Update.
19985 (do_jump_1): Update.
19987 (do_jump_by_parts_greater_rtx): Update.
19988 (do_compare_rtx_and_jump): Update.
19989 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19990 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19991 * dwarf2cfi.c: Include profile-count.h
19992 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19993 (sjlj_emit_dispatch_table): Likewise.
19994 * explow.c: Include profile-count.h
19995 * expmed.c (emit_store_flag_force): Update.
19996 (do_cmp_and_jump): Update.
19997 * expr.c (compare_by_pieces_d::generate): Update.
19998 (compare_by_pieces_d::finish_mode): Update.
19999 (emit_block_move_via_loop): Update.
20000 (store_expr_with_bounds): Update.
20001 (store_constructor): Update.
20002 (expand_expr_real_2): Update.
20003 (expand_expr_real_1): Update.
20004 * expr.h (try_casesi, try_tablejump): Update prototypes.
20005 * gimple-pretty-print.c (dump_probability): Update.
20006 (dump_profile): New.
20007 (dump_gimple_label): Update.
20008 (dump_gimple_bb_header): Update.
20009 * graph.c (draw_cfg_node_succ_edges): Update.
20010 * hsa-gen.c (convert_switch_statements): Update.
20011 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20012 (find_if_case_1): Update.
20013 (find_if_case_2): Update.
20014 * internal-fn.c (expand_arith_overflow_result_store): Update.
20015 (expand_addsub_overflow): Update.
20016 (expand_neg_overflow): Update.
20017 (expand_mul_overflow): Update.
20018 (expand_vector_ubsan_overflow): Update.
20019 * ipa-cp.c (good_cloning_opportunity_p): Update.
20020 * ipa-split.c (split_function): Use make_single_succ_edge.
20021 * ipa-utils.c (ipa_merge_profiles): Update.
20022 * loop-doloop.c (add_test): Update.
20023 (doloop_modify): Update.
20024 * loop-unroll.c (compare_and_jump_seq): Update.
20025 (unroll_loop_runtime_iterations): Update.
20026 * lra-constraints.c (lra_inheritance): Update.
20027 * lto-streamer-in.c (input_cfg): Update.
20028 * lto-streamer-out.c (output_cfg): Update.
20029 * mcf.c (adjust_cfg_counts): Update.
20030 * modulo-sched.c (sms_schedule): Update.
20031 * omp-expand.c (expand_omp_for_init_counts): Update.
20032 (extract_omp_for_update_vars): Update.
20033 (expand_omp_ordered_sink): Update.
20034 (expand_omp_for_ordered_loops): Update.
20035 (expand_omp_for_generic): Update.
20036 (expand_omp_for_static_nochunk): Update.
20037 (expand_omp_for_static_chunk): Update.
20038 (expand_cilk_for): Update.
20039 (expand_omp_simd): Update.
20040 (expand_omp_taskloop_for_outer): Update.
20041 (expand_omp_taskloop_for_inner): Update.
20042 * omp-simd-clone.c (simd_clone_adjust): Update.
20043 * optabs.c (expand_doubleword_shift): Update.
20044 (expand_abs): Update.
20045 (emit_cmp_and_jump_insn_1): Update.
20046 (expand_compare_and_swap_loop): Update.
20047 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20048 * predict.c (predictable_edge_p): Update.
20049 (edge_probability_reliable_p): Update.
20050 (set_even_probabilities): Update.
20051 (combine_predictions_for_insn): Update.
20052 (combine_predictions_for_bb): Update.
20053 (propagate_freq): Update.
20054 (estimate_bb_frequencies): Update.
20055 (force_edge_cold): Update.
20056 * profile-count.c (profile_count::dump): Add missing space into dump.
20057 (profile_count::debug): Add newline.
20058 (profile_count::differs_from_p): Explicitly convert to unsigned.
20059 (profile_count::stream_in): Update.
20060 (profile_probability::dump): New member function.
20061 (profile_probability::debug): New member function.
20062 (profile_probability::differs_from_p): New member function.
20063 (profile_probability::differs_lot_from_p): New member function.
20064 (profile_probability::stream_in): New member function.
20065 (profile_probability::stream_out): New member function.
20066 * profile-count.h (profile_count_quality): Rename to ...
20067 (profile_quality): ... this one.
20068 (profile_probability): New.
20069 (profile_count): Update.
20070 * profile.c (compute_branch_probabilities): Update.
20071 * recog.c (peep2_attempt): Update.
20072 * sched-ebb.c (schedule_ebbs): Update.
20073 * sched-rgn.c (find_single_block_region): Update.
20074 (compute_dom_prob_ps): Update.
20075 (schedule_region): Update.
20076 * sel-sched-ir.c (compute_succs_info): Update.
20077 * stmt.c (struct case_node): Update.
20078 (do_jump_if_equal): Update.
20079 (get_outgoing_edge_probs): Update.
20080 (conditional_probability): Update.
20081 (emit_case_dispatch_table): Update.
20082 (expand_case): Update.
20083 (expand_sjlj_dispatch_table): Update.
20084 (emit_case_nodes): Update.
20085 * targhooks.c: Update.
20086 * tracer.c (better_p): Update.
20087 (find_best_successor): Update.
20088 * trans-mem.c (expand_transaction): Update.
20089 * tree-call-cdce.c: Update.
20090 * tree-cfg.c (gimple_split_edge): Upate.
20091 (move_sese_region_to_fn): Upate.
20092 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20093 * tree-eh.c (lower_resx): Upate.
20094 (cleanup_empty_eh_move_lp): Upate.
20095 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20096 * tree-inline.c (copy_edges_for_bb): Update.
20097 (copy_cfg_body): Update.
20098 * tree-parloops.c (gen_parallel_loop): Update.
20099 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20100 (gimple_gen_time_profiler): Update.
20101 * tree-ssa-dce.c (remove_dead_stmt): Update.
20102 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20103 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20104 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20105 (unloop_loops): Update.
20106 (try_peel_loop): Update.
20107 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20108 * tree-ssa-loop-split.c (connect_loops): Update.
20109 (split_loop): Update.
20110 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20111 (hoist_guard): Update.
20112 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20113 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20114 (value_replacement): Update.
20115 * tree-ssa-reassoc.c (branch_fixup): Update.
20116 * tree-ssa-tail-merge.c (replace_block_by): Update.
20117 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20118 (create_edge_and_update_destination_phis): Update.
20119 (compute_path_counts): Update.
20120 (recompute_probabilities): Update.
20121 (update_joiner_offpath_counts): Update.
20122 (freqs_to_counts_path): Update.
20123 (duplicate_thread_path): Update.
20124 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20125 (struct switch_conv_info): Update.
20126 (gen_inbound_check): Update.
20127 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20128 (vect_do_peeling): Update.
20129 (vect_loop_versioning): Update.
20130 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20131 (optimize_mask_stores): Update.
20132 * ubsan.c (ubsan_expand_null_ifn): Update.
20133 * value-prof.c (gimple_divmod_fixed_value): Update.
20134 (gimple_divmod_fixed_value_transform): Update.
20135 (gimple_mod_pow2): Update.
20136 (gimple_mod_pow2_value_transform): Update.
20137 (gimple_mod_subtract): Update.
20138 (gimple_mod_subtract_transform): Update.
20139 (gimple_ic): Update.
20140 (gimple_stringop_fixed_value): Update.
20141 (gimple_stringops_transform): Update.
20142 * value-prof.h: Update.
20144 2017-06-29 Carl Love <cel@us.ibm.com>
20146 * config/rs6000/rs6000-c.c: Add support for built-in functions
20147 vector signed int vec_signed (vector float);
20148 vector signed long long vec_signed (vector double);
20149 vector signed int vec_signed2 (vector double, vector double);
20150 vector signed int vec_signede (vector double);
20151 vector signed int vec_signedo (vector double);
20152 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20153 instruction generator.
20154 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20155 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20156 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20158 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20159 vunsignede_v2df): Add define_expands.
20160 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20161 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20162 VEC_UNSIGNEDO): Add definitions.
20163 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20164 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20165 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20166 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20167 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20168 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20169 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20170 vec_unsignede, vec_unsignedo): Add builtin defines.
20171 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20173 * doc/extend.texi: Update the built-in documentation file for the
20174 new built-in functions.
20176 2017-06-29 Richard Biener <rguenther@suse.de>
20178 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20179 reduction chains to LOOP_VINFO_REDUCTIONS.
20180 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20181 SLP reductions after processing reduction chains.
20183 2017-06-29 Nathan Sidwell <nathan@acm.org>
20185 * builtins.c (fold_builtin_FUNCTION): Use
20186 lang_hooks.decl_printable_name.
20188 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20190 PR middle-end/81194
20191 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20192 with only one label.
20193 * stmt.c (expand_case): Assert NCASES is greater than one.
20195 2017-06-29 Richard Biener <rguenther@suse.de>
20197 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20199 (group_case_labels): Likewise.
20200 (find_taken_edge): Push sanity checking on val to workers...
20201 (find_taken_edge_cond_expr): ... here
20202 (find_taken_edge_switch_expr): ... and here, handle cases
20203 with just a default label.
20204 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20205 (group_case_labels): Likewise.
20206 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20207 group_case_labels does anything cleanup the CFG again.
20209 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20211 PR tree-optimization/81196
20212 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20213 exit condition comparing two IVs.
20215 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20217 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20218 profile to the dummy entry at the end of the list of architectures.
20219 * config/arm/arm-cpu-cdata.h: Regenerated.
20221 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20222 Michael Collison <michael.collison@arm.com>
20225 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20227 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20228 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20229 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20230 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20231 of shift when the shift amount is masked with constant equal to
20232 the size of the mode.
20233 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20236 2017-06-29 Martin Liska <mliska@suse.cz>
20238 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20240 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20243 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20246 (edge_growth_cache_entry::edge_growth_cache_entry): New
20248 (reset_edge_growth_cache): Update to use constructor.
20250 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20252 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20253 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20254 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20256 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20258 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20259 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20261 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20263 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20264 (*-linux-uclibc*): Add t-uclibc tmake_file.
20265 * config/t-musl: New.
20266 * config/t-uclibc: New.
20268 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20270 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20272 (gen_comm_data): Emit architectural setting of arch_prof.
20273 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20275 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20276 (armv8-m.base, armv8-m.main): Likewise.
20277 * arm-protos.h (arm_build_target): Add profile field.
20278 (arch_option): Likewise.
20279 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20281 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20282 arm_active_target.profile.
20284 2017-06-28 Richard Biener <rguenther@suse.de>
20286 PR middle-end/81227
20287 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20288 TYPE_OVERFLOW_WRAPS.
20289 * match.pd (negate_expr_p): Likewise.
20290 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20291 fold_build2, not fold_binary.
20293 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20295 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20296 Convert memory address to Pmode.
20297 (aarch64_print_operand): Assert MEM operands are always Pmode.
20299 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20302 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20303 Remove redundant if.
20304 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20305 * config/arm/aarch-common-protos.h
20306 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20307 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20309 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20312 * multiple_target.c (create_dispatcher_calls): Set the default
20313 clone to be static, not public.
20315 2017-06-28 Richard Biener <rguenther@suse.de>
20317 * tree-vect-loop.c (vectorizable_reduction): Move special
20318 cond reduction IV var creation ...
20319 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20320 parameter. Use STMT_VINFO_VECTYPE.
20321 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20324 2017-06-28 Martin Liska <mliska@suse.cz>
20327 * ipa-visibility.c (non_local_p): Handle visibility.
20329 2017-06-28 Martin Liska <mliska@suse.cz>
20332 * common.opt: Add IntegerRange to various options.
20333 * opt-functions.awk (integer_range_info): New function.
20334 * optc-gen.awk: Add integer_range_info to cl_options struct.
20335 * opts-common.c (decode_cmdline_option): Handle
20336 CL_ERR_INT_RANGE_ARG.
20337 (cmdline_handle_error): Likewise.
20338 * opts.c (print_filtered_help): Show valid interval in
20339 when --help is provided.
20340 * opts.h (struct cl_option): Add range_min and range_max fields.
20341 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20343 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20345 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20346 (x * C EQ/NE y * C): New transformation.
20348 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20350 * genmultilib (combination_space): Accept '+' in option names.
20352 2017-06-28 Martin Liska <mliska@suse.cz>
20355 * asan.c (instrument_derefs): Bail out inner references
20356 that are hard register variables.
20358 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20361 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20362 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20364 2017-06-28 Richard Biener <rguenther@suse.de>
20366 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20367 (vect_get_slp_defs): Adjust.
20368 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20370 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20372 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20373 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20374 (vectorizable_reduction): Adjust.
20375 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20377 (vect_get_slp_defs): Likewise.
20378 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20379 (vectorizable_bswap): Adjust.
20380 (vectorizable_call): Likewise.
20381 (vectorizable_conversion): Likewise.
20382 (vectorizable_assignment): Likewise.
20383 (vectorizable_shift): Likewise.
20384 (vectorizable_operation): Likewise.
20385 (vectorizable_store): Likewise.
20386 (vectorizable_condition): Likewise.
20387 (vectorizable_comparison): Likewise.
20389 2017-06-28 Michael Collison <michael.collison@arm.com>
20392 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20394 (arm_gen_movmemqi): Removed unused variable 'i'.
20395 Convert 'for' loop into 'while' loop.
20396 (arm_expand_prologue): Remove last unnecessary set of insn.
20397 (thumb_pop): Remove unused variable 'pushed_words'.
20398 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20400 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20402 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20403 * config/s390/s390-protos.h: Add prototype of
20404 s390_rel_address_ok_p.
20405 * config/s390/s390.c (s390_got_symbol): New function.
20406 (s390_rel_address_ok_p): New function.
20407 (legitimize_pic_address): Use s390_rel_address_ok_p.
20408 (s390_load_got): Use s390_got_symbol.
20409 (s390_option_override): Issue error if
20410 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20411 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20413 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20415 2017-06-27 Andrew Pinski <apinski@cavium.com>
20417 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20418 (X * copysign (1.0, X)): New pattern.
20419 (X * copysign (1.0, -X)): New pattern.
20420 (copysign (-1.0, CST)): New pattern.
20422 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20424 * genmultilib (combination_space): Remove variable.
20425 Validate reuse rules against regular expression for any sequence
20426 of multilib options in any order.
20428 2017-06-27 Michael Collison <michael.collison@arm.com>
20430 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20431 call aarch64_split_simd_combine.
20432 * (aarch64_combine_internal<mode>): Delete pattern.
20433 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20434 Allow register and subreg operands.
20436 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20438 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20439 specific need, just fallback on defaults.
20440 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20442 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20443 Olivier Hainque <hainque@adacore.com>
20445 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20447 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20448 targets. Pick a default if no particular attempt applied.
20449 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20452 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20454 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20455 (x86_64-wrs-vxworks7): Likewise.
20457 2017-06-27 Marek Polacek <polacek@redhat.com>
20460 * ubsan.c (instrument_null): Check get_base_address's result for null.
20462 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20464 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20466 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20468 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20469 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20470 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20471 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20472 New function types.
20473 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20474 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20475 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20476 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20477 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20478 BUILT_IN_FEUPDATEENV): New builtins.
20479 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20480 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20481 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20482 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20484 (builtin_structptr_types): Adjust size.
20485 * tree.c (builtin_structptr_types): Add four entries.
20487 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20488 Olivier Hainque <hainque@adacore.com>
20490 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20491 (TLS_SYM): New local macro, forcing reference to __tls__ on
20492 link command lines for VxWorks 7 RTPs, triggering initialization
20494 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20495 OS features TLS support, true for RTPs on VxWorks 7.
20496 * config/vxworks.c (vxworks_override_options): Setup emutls
20499 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20501 * predict.c (test_prediction_value_range): Use -1U instead of -1
20502 to avoid narrowing conversion warning.
20503 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20504 to avoid narrowing conversion warning.
20505 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20507 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20509 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20511 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20512 64bit configurations.
20513 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20514 (SIZE_TYPE): Likewise.
20515 * config/vxworks.c (vxworks_emutls_var_fields): Use
20516 long_unsigned_type_node instead of unsigned_type_node as the offset
20517 field type, which is "pointer" mode in emutls.c.
20519 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20522 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20524 PR middle-end/81207
20525 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20526 gimple_vuse copying separately from gimple_vdef copying.
20528 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20530 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20533 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20534 Olivier Hainque <hainque@adacore.com>
20536 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20537 for all vxworks7 targets.
20538 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20539 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20540 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20541 variations for VX6/VX7 and 32/64bits later on in ...
20542 (VXWORKS_LIB_SPEC): Leverage new macros.
20543 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20544 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20546 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20548 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20549 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20551 2017-06-26 Carl Love <cel@us.ibm.com>
20553 * config/rs6000/rs6000-c.c: Add support for built-in functions
20554 vector bool char vec_reve (vector bool char);
20555 vector signed char vec_reve (vector signed char);
20556 vector unsigned char vec_reve (vector unsigned char);
20557 vector bool int vec_reve (vector bool int);
20558 vector signed int vec_reve (vector signed int);
20559 vector unsigned int vec_reve (vector unsigned int);
20560 vector bool long long vec_reve (vector bool long long);
20561 vector signed long long vec_reve (vector signed long long);
20562 vector unsigned long long vec_reve (vector unsigned long long);
20563 vector bool short vec_reve (vector bool short);
20564 vector signed short vec_reve (vector signed short);
20565 vector double vec_reve (vector double);
20566 vector float vec_reve (vector float);
20567 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20568 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20569 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20570 (altivec_vreve): New pattern.
20571 * config/rs6000/altivec.h (vec_reve): New define.
20572 * doc/extend.texi (vec_rev): Update the built-in documentation file
20573 for the new built-in functions.
20575 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20577 PR tree-optimization/71815
20578 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20580 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20582 (slsr_process_phi): Likewise.
20583 (replace_uncond_cands_and_profitable_phis): Don't replace a
20584 multiply candidate with a stride of 1 (copy or cast).
20585 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20587 (lowest_cost_path): Likewise.
20588 (total_savings): Likewise.
20590 2017-06-26 Richard Biener <rguenther@suse.de>
20593 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20594 Use def_builtin_pure for all gather builtins.
20596 2017-06-26 Richard Biener <rguenther@suse.de>
20598 PR tree-optimization/81203
20599 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20600 non-dominating BBs.
20602 2017-06-26 Marek Polacek <polacek@redhat.com>
20605 * doc/invoke.texi: Document -Wmultistatement-macros.
20607 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20609 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20610 arm_neon_ok_no_float_abi effective target.
20612 2017-06-26 Richard Biener <rguenther@suse.de>
20614 PR tree-optimization/80928
20615 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20616 (copy_bbs): Set BB_DUPLICATED flag early.
20617 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20619 (execute_on_shrinking_pred): Likewise.
20620 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20621 BB_DUPLICATED blocks.
20622 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20623 iterate over all PHIs considering removal of *gsi.
20625 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20627 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20630 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20632 * config/rs6000/rs6000-string.c: (expand_block_clear,
20633 do_load_for_compare, select_block_compare_mode,
20634 compute_current_alignment, expand_block_compare,
20635 expand_strncmp_align_check, expand_strn_compare,
20636 expand_block_move, rs6000_output_load_multiple)
20637 Move functions related to string/block move/compare
20638 to a separate file.
20639 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20640 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20641 for this function which is now used in two files.
20642 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20643 * config.gcc: Add rs6000-string.o to extra_objs for
20644 targets powerpc*-*-* and rs6000*-*-*.
20646 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20649 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20650 32-bit, since indexed is not valid for DImode.
20651 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20652 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20653 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20654 (define_peephole2 for Altivec d-form store): Likewise.
20657 * multiple_target.c (create_dispatcher_calls): Only create the
20658 dispatcher call if the function is the default clone of a
20659 versioned function.
20661 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20663 PR middle-end/80902
20664 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20665 a call, force the call to not be a tail call.
20667 2017-06-23 Jeff Law <law@redhat.com>
20669 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20672 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20674 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20675 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20676 handling for early expansion of vector shifts (sl,sr,sra,rl).
20677 (builtin_function_type): Add vector shift right instructions
20678 to the unsigned argument list.
20680 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20682 rtl-optimizatoin/79286
20683 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20684 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20685 trap. PIC register plus a const unspec without offset can never trap.
20687 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20689 * tree.h (builtin_structptr_type): New type.
20690 (builtin_structptr_types): Declare new array.
20691 * tree.c (builtin_structptr_types): New array.
20692 (free_lang_data, build_common_tree_nodes): Use it.
20694 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20697 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20700 2017-06-22 Matt Turner <mattst88@gmail.com>
20702 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20703 Lake models to skylake case. Assume skylake for unknown
20704 models with clflushopt.
20706 2017-06-22 Jeff Law <law@redhat.com>
20708 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20709 frame sizes that do not satisfy aarch64_uimm12_shift.
20711 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20713 * profile-count.h (apply_probability,
20714 apply_scale, probability_in): Fix checks for zero.
20716 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20718 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20719 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20721 2016-06-22 Richard Biener <rguenther@suse.de>
20723 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20724 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20725 REDUC_MAX_EXPR support.
20726 (vectorizable_reduction): Likewise.
20727 (vect_create_epilog_for_reduction): Likewise.
20729 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20731 * match.pd (A / (1 << B) -> A >> B): New.
20732 * generic-match-head.c: Include optabs-tree.h.
20733 * gimple-match-head.c: Likewise.
20734 * optabs-tree.h (target_supports_op_p): New.
20735 * optabs-tree.c (target_supports_op_p): New.
20737 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20739 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20740 $gcc_cv_ld --help output.
20741 (gcc_cv_ld_demangle): Likewise.
20742 (gcc_cv_ld_eh_frame_hdr): Likewise.
20743 (gcc_cv_ld_pie): Likewise.
20744 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20745 (gcc_cv_ld_buildid): Likewise.
20746 (gcc_cv_ld_sysroot): Likewise.
20747 (ld_bndplt_support): Likewise.
20748 (ld_pushpopstate_support): Likewise.
20749 * configure: Regenerate.
20750 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20752 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20755 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20756 operands indexes to avoid gap between operands and match_dups.
20758 2017-06-21 Andrew Pinski <apinski@cavium.com>
20760 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20761 Increment Arith_shift and Arith_shift_reg by 1.
20762 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20764 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20765 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20766 (aarch64_strip_extend): Add new argument and test for it.
20767 (aarch64_cheap_mult_shift_p): New function.
20768 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20769 add a cost if it is true.
20770 Update calls to aarch64_strip_extend.
20771 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20773 2017-06-21 Andrew Pinski <apinski@cavium.com>
20775 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20777 (thunderxt88): Likewise.
20778 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20779 (thunderx_prefetch_tune): New variable.
20780 (thunderx2t99_prefetch_tune): Update for the correct values.
20781 (thunderxt88_tunings): New variable.
20782 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20783 generic_prefetch_tune.
20784 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20786 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20788 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20789 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20790 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20791 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20792 (aarch64_atomic_cas<mode>, GPI): Likewise.
20794 2017-06-21 Martin Liska <mliska@suse.cz>
20796 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20797 statements on cold and hot labels.
20798 * predict.c (tree_estimate_probability_bb): Remove the
20799 prediction from this place.
20801 2017-06-21 Martin Liska <mliska@suse.cz>
20803 PR tree-optimization/79489
20804 * gimplify.c (maybe_add_early_return_predict_stmt): New
20806 (gimplify_return_expr): Call the function.
20807 * predict.c (tree_estimate_probability_bb): Remove handling
20809 * predict.def: Update comment about early return predictor.
20810 * gimple-predict.h (is_gimple_predict): New function.
20811 * predict.def: Change default value of early return to 66.
20812 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20814 * passes.def: Put pass_strip_predict_hints to the beginning of
20817 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20819 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20820 FUNCTION_DECL declarations.
20821 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20823 (dwaf2out_decl): Likewise.
20824 * godump.c (go_early_global_decl): Skip call to the real debug hook
20825 for FUNCTION_DECL declarations.
20826 * passes.c (rest_of_decl_compilation): Skip call to the
20827 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20828 -fdump-go-spec is passed.
20830 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20832 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20833 Reorder for compactness.
20834 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20835 (def_builtin_pure, def_builtin_pure2): New functions.
20836 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20838 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20840 * match.pd (nop_convert): New predicate.
20841 ((A +- CST1) +- CST2): Allow some NOP conversions.
20843 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20846 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20847 with ctors/dtors if GOVD_SHARED is set.
20849 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20851 * config/aarch64/aarch64.md (movti_aarch64):
20852 Emit mov rather than orr.
20853 (movtf_aarch64): Likewise.
20854 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20855 Emit mov rather than orr.
20857 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20859 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20860 Swap alternatives, make integer dup more expensive.
20862 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20864 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20865 Return true for non-tls symbols.
20867 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20869 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20870 (cortex-a75): Likewise.
20871 (cortex-a75.cortex-a55): Likewise.
20872 * config/aarch64/aarch64-tune.md: Regenerate.
20873 * doc/invoke.texi (-mtune): Document new values for -mtune.
20875 2017-06-21 Tom de Vries <tom@codesourcery.com>
20877 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20878 stack_size feature.
20879 (Effective-Target Keywords, Other attributes): Suggest using
20880 dg-add-options stack_size feature to get stack limit in stack_size
20881 effective target documentation.
20883 2017-06-21 Julian Brown <julian@codesourcery.com>
20884 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20886 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20887 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20888 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20890 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20891 attribute type list for neon_multiply.
20892 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20893 attribute type list for neon_multiply.
20894 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20895 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20896 attribute type list for neon_multiply.
20897 * config/arm/types.md (crypto_pmull): Add.
20898 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20899 attribute type list.
20901 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20903 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20906 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20908 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20909 to make sure not to dereference a NULL cost_classes_ptr pointer.
20911 2017-06-20 Carl Love <cel@us.ibm.com>
20913 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20914 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20915 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20916 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20917 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20918 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20919 VMULOSW): New enum "unspec" values.
20920 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20921 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20922 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20923 altivec_vmulosw): New patterns.
20924 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20925 VMULOSW): Add definitions.
20927 2017-06-20 Julia Koval <julia.koval@intel.com>
20929 * config/i386/i386.c: Fix rounding expand for new pattern.
20930 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20932 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20934 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20935 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20937 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20939 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20942 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20944 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20945 architecture, then by implementer ID.
20946 * config/aarch64/aarch64-tune.md: Regenerate.
20948 2017-06-20 Richard Biener <rguenther@suse.de>
20950 PR middle-end/81097
20951 * fold-const.c (split_tree): Fold to type before negating.
20953 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20955 * diagnostic-show-locus.c
20956 (selftest::test_fixit_deletion_affecting_newline): New function.
20957 (selftest::diagnostic_show_locus_c_tests): Call it.
20959 2017-06-20 Andreas Schwab <schwab@suse.de>
20962 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20965 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20967 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20968 __ARM_FEATURE_COPROC according to support.
20970 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20972 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20973 Rewritten to avoid overflow for > 32-bit pointers.
20976 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20977 by removing enum keyword.
20978 (ubsan_type_descriptor): Likewise. Formatting fix.
20981 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20982 splitter): Require TARGET_SSE2 in the condition.
20984 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20987 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20988 for doing vector set of SFmode on ISA 3.0.
20989 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20990 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20992 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20993 SFmode value into a V4SF variable that was extracted from another
20994 V4SF variable without converting the element to double precision
20995 and back to single precision vector format.
20996 (vsx_insert_extract_v4sf_p9_2): Likewise.
20998 2017-06-19 Jakub Jelinek <jakub@redhat.com>
21000 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
21001 in UWHI to avoid undefined overflow.
21004 * ubsan.h (enum ubsan_encode_value_phase): New.
21005 (ubsan_encode_value): Change second argument to
21006 enum ubsan_encode_value_phase with default value of
21007 UBSAN_ENCODE_VALUE_GENERIC.
21008 * ubsan.c (ubsan_encode_value): Change second argument to
21009 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21010 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21011 create_tmp_var_raw instead of create_tmp_var and use a
21013 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21014 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21015 ubsan_encode_value callers.
21018 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21019 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21020 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21022 2017-06-19 Richard Biener <rguenther@suse.de>
21024 PR middle-end/81118
21025 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21026 estimates if we changed anything.
21028 2017-06-19 Richard Biener <rguenther@suse.de>
21030 PR tree-optimization/80887
21031 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21032 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21033 simplified lookups, then reset mprts_hook.
21034 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21036 (try_to_simplify): Likewise.
21038 2017-06-19 Martin Liska <mliska@suse.cz>
21041 * gimplify.c (gimplify_switch_expr):
21042 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21044 2017-06-19 Martin Liska <mliska@suse.cz>
21046 * doc/install.texi: Document that PGO runs in 4 stages.
21048 2017-06-19 Martin Liska <mliska@suse.cz>
21051 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21052 to dispatcher function name.
21053 * multiple_target.c (replace_function_decl): New function.
21054 (create_dispatcher_calls): Redirect both edges and references.
21056 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21058 * profile-count.c (profile_count::dump): Dump quality.
21059 (profile_count::differs_from_p): Update for unsigned val.
21060 * profile-count.h (profile_count_quality): New enum.
21061 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21063 2017-06-19 Richard Biener <rguenther@suse.de>
21065 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21066 struct function as arg.
21067 (estimate_numbers_of_iterations): Export overload with loop arg.
21068 (free_numbers_of_iterations_estimates_loop): Use an overload of
21069 free_numbers_of_iterations_estimates instead.
21070 * tree-cfg.c (remove_bb): Adjust.
21071 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21072 * tree-parloops.c (gen_parallel_loop): Likewise.
21073 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21075 (tree_unroll_loops_completely): Likewise.
21076 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21077 Use an overload instead and export.
21078 (estimated_loop_iterations): Adjust.
21079 (max_loop_iterations): Likewise.
21080 (likely_max_loop_iterations): Likewise.
21081 (estimate_numbers_of_iterations): Take struct function as arg
21083 (loop_exits_before_overflow): Adjust.
21084 (free_numbers_of_iterations_estimates_loop): Use an overload.
21085 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21086 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21088 2017-06-19 Richard Biener <rguenther@suse.de>
21091 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21092 RANGE_EXPR conservatively.
21094 2017-06-16 Carl Love <cel@us.ibm.com>
21096 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21097 definitions for vec_float, vec_float2, vec_floato,
21098 vec_floate built-ins.
21099 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21100 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21102 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21103 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21104 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21105 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21106 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21107 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21108 vec_floato): Add builtin defines.
21109 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21110 Update the built-in documentation file for the new built-in
21113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21115 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21116 (mthumb): Mark as the negative of -marm.
21118 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21120 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21122 (ARM Options, -mtune): Document that this accepts the same
21123 extension options as -mcpu.
21124 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21128 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21129 permitted extensions.
21131 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21133 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21134 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21135 (armv8-m.main): Add option +nodsp.
21136 * config/arm/arm-cpu-cdata.h: Regenerated.
21138 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21140 * config/arm/t-fuchsia: New file.
21141 * config.gcc (arm*-*-fuchsia*): Use it.
21143 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21145 * config/arm/t-symbian: Rewrite for new option infrastructure.
21147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21149 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21150 (MULTILIB_REQUIRED): Likewise.
21152 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21154 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21155 (MULTILIB_RESUE): Likewise.
21156 (MULTILIB_MATCHES): Likewise.
21157 (MULTLIB_REQUIRED): Likewise.
21159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21161 * config/arm/t-rtems: Rewrite for new option framework.
21163 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21165 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21166 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21167 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21168 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21169 * config/arm/t-multilib: ... here.
21170 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21171 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21172 armv7-a and armv8*-a when A-profile libraries have not been built.
21173 * config/arm/t-rmprofile: Rewrite.
21175 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21177 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21178 with a backslash. Remove the backslash after substituting unescaped
21180 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21182 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21184 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21185 the driver to pass through the default setting of -mfloat-abi.
21186 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21188 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21189 (all_feat_combs): New rule.
21190 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21192 * config/arm/t-aprofile: Rewrite.
21194 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21196 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21197 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21198 fpu is not specified by the user/command-line.
21199 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21200 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21201 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21202 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21203 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21204 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21206 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21208 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21209 * config/arm/t-arm-elf: Rewritten.
21211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21213 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21214 have some floating-point instructions.
21215 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21216 (TARGET_MAYBE_HARD_FLOAT): New macro.
21217 * config/arm/arm-builtins.c (arm_init_builtins): Use
21218 TARGET_MAYBE_HARD_FLOAT.
21219 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21221 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21223 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21224 (configargs.h): Include it.
21225 (arm_print_hint_for_fpu_option): New function.
21226 (arm_parse_fpu_option): New function.
21227 (candidate_extension): New class.
21228 (arm_canon_for_multilib): New function.
21229 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21230 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21231 (ARCH_CANONICAL_SPECS): New macro.
21232 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21234 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21236 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21237 are set after handling multilib fragments. Set target_cpu_default2
21240 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21242 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21244 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21245 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21247 (check_arch): Likewise.
21248 * config/arm/arm.c (arm_configure_build_target): Handle
21249 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21250 options in the default.
21252 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21254 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21255 when an option is an alias of another.
21256 * config/arm/parsecpu.awk (optalias): New parser token.
21257 (gen_comm_data): Mark non-alias options as such. Emit entries
21258 for extension aliases.
21259 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21260 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21261 (armv6kz, armv6zk, armv6t2): Likewise.
21262 (armv7): Make vfpv3-d16 an alias.
21263 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21265 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21266 Sort in canonical order.
21267 (armv8-a): Sort in canonical order.
21268 (armv8.1-a, armv8.2-a): Likewise.
21269 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21271 (cortex-a9): Sort in canonical order.
21272 * config/arm/arm.c (selftests.h): Include it.
21273 (arm_test_cpu_arch_data): New function.
21274 (arm_run_self_tests): New function.
21275 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21276 (targetm): Move declaration to the end of the file.
21277 * arm-cpu-cdata.h: Regenerated.
21279 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21281 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21282 call to target_mode_check describing the type of option passed.
21283 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21284 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21285 arm_parse_cpu_option_name to match parameters against list of
21287 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21288 arm_arch_core_flags data structure.
21289 * config/arm/arm-cpu_cdata.h: Regenerated.
21291 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21293 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21295 (arm_print_hint_for_cpu_option): Likewise.
21296 (arm_print_hint_for_arch_option): Likewise.
21297 (arm_parse_cpu_option_name): Likewise.
21298 (arm_parse_arch_option_name): Likewise.
21299 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21300 of entries in the all_fpus list.
21301 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21302 (arm_parse_cpu_option_name): Declare.
21303 (arm_parse_arch_option_name): Declare.
21304 (arm_parse_option_features): Declare.
21305 (arm_intialize_isa): Declare.
21306 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21308 (gen_comm_data): ... here. Make definitions non-static.
21309 * config/arm/arm-cpu-data.h: Regenerated.
21310 * config/arm/arm-cpu-cdata.h: Regenerated.
21312 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21314 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21315 (cpu_arch_extension): New structure.
21316 (cpu_arch_option, arch_option, cpu_option): New structures.
21317 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21318 architecture types.
21319 (gen_data): Generate new format data tables.
21320 * config/arm/arm.c (cpu_tune): New structure.
21321 (cpu_option, processors): Delete.
21322 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21323 (arm_print_hint_for_cpu_option): ... this and ...
21324 (arm_print_hint_for_arch_option): ... this.
21325 (arm_parse_arch_cpu_name): Delete. Replace with ...
21326 (arm_parse_cpu_option_name): ... this and ...
21327 (arm_parse_arch_option_name): ... this.
21328 (arm_unrecognized_feature): Change type of target parameter to
21330 (arm_parse_arch_cpu_features): Delete. Replace with ...
21331 (arm_parse_option_features): ... this.
21332 (arm_configure_build_target): Rework to use new configuration data
21334 (arm_print_tune_info): Rework for new configuration data tables.
21335 * config/arm/arm-cpu-data.h: Regenerated.
21336 * config/arm/arm-cpu.h: Regenerated.
21338 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21340 * Makefile.in (OBJS): Move sbitmap.o from here ...
21341 (OBJS-libcommon): ... to here.
21343 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21345 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21346 (ISA_ALL_CRYPTO): New macro.
21347 (ISA_ALL_SIMD): New macro
21348 (ISA_ALL_FP): New macro.
21349 * config/arm/arm.c (fpu_bitlist): Update initializer.
21350 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21352 (arm9e): Add fpu. Add option for nofp
21353 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21354 (arm926ej-s, arm1026ej-s): Likewise.
21355 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21356 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21357 neon-fp16, neon-vfpv4, nofp and nosimd.
21358 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21359 (cortex-a8): Add fpu. Add option for nofp.
21360 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21361 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21362 (cortex-r4f): Add fpu.
21363 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21364 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21366 (cortex-r8): Likewise.
21367 (cortex-m4): Add fpu. Add option for nofp.
21368 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21369 (cortex-a17.cortex-a7): Likewise.
21370 (cortex-a32): Add fpu. Add options for crypto and nofp.
21371 (cortex-a35, cortex-a53): Likewise.
21372 (cortex-a57): Add fpu. Add option for crypto.
21373 (cortex-a72, cortex-a73): Likewise.
21374 (exynos-m1): Likewise.
21375 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21376 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21377 (cortex-m33): Add fpu. Add option for nofp.
21378 * config/arm/arm-cpu-cdata.h: Regenerated
21379 * config/arm/arm-cpu-data.h: Regenerated.
21381 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21383 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21384 (armv5te, armv5tej): Likewise.
21385 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21386 (armv7): Add options fp and vfpv3-d16.
21387 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21388 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21390 (armv7ve): Likewise.
21391 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21392 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21393 (armv8-a): Add nocrypto option.
21394 (armv8.1-a, armv8.2-a): Likewise.
21395 (armv8-m.main): add options fp, fp.dp and nofp.
21397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21399 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21401 (armv8-a+crc): Delete.
21402 (armv8.1-a): Add options simd, crypto and nofp.
21403 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21404 (armv8.2-a+fp16): Delete.
21405 (armv8-m.main): Add option dsp.
21406 (armv8-m.main+dsp): Delete.
21407 (cortex-a8): Add fpu. Add nofp option.
21408 (cortex-a9): Add fpu. Add nofp and nosimd options.
21409 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21410 link to main cpu and architecture data structures.
21411 (gen_comm_data): Only put isa attributes from the main architecture
21413 (option): New statement for architecture and CPU entries.
21414 * arm.c (struct cpu_option): New structure.
21415 (struct processors): Add entry for options.
21416 (arm_unrecognized_feature): New function.
21417 (arm_parse_arch_cpu_name): Ignore any characters after the first
21419 (arm_parse_arch_cpu_feature): New function.
21420 (arm_configure_build_target): Separate out any CPU and architecture
21421 features and parse separately. Don't error out if -mfpu=auto is
21422 used with only an architecture string.
21423 (arm_print_asm_arch_directives): New function.
21424 (arm_file_start): Call it.
21425 * config/arm/arm-cpu-cdata.h: Regenerated.
21426 * config/arm/arm-cpu-data.h: Likewise.
21427 * config/arm/arm-tables.opt: Likewise.
21429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21431 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21432 assembler when it is not -mfpu=auto.
21434 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21436 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21437 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21438 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21439 (ASM_CPU_SPEC): Rewrite.
21440 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21441 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21442 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21443 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21444 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21445 copied string is NUL-terminated. Also strip any characters prefixed
21447 (arm_rewrite_selected_arch): New function.
21448 (arm_rewrite_march): New function.
21450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21452 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21453 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21454 (march, mcpu, mtune): Convert to string-based options.
21455 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21456 (arm_parse_arch_cpu_name): New function.
21457 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21458 identify selected architecture or CPU.
21459 (arm_option_save): New function.
21460 (TARGET_OPTION_SAVE): Redefine.
21461 (arm_option_restore): Restore string options.
21462 (arm_option_print): Print string options.
21464 2017-06-16 Martin Sebor <msebor@redhat.com>
21466 PR tree-optimization/80933
21467 PR tree-optimization/80934
21468 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21469 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21470 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21471 (gimple_fold_builtin): Call them.
21473 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21475 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21476 as unlikely; update profile.
21478 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21480 * predict.c (force_edge_cold): Handle declaring edges impossible
21483 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21485 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21487 (try_unroll_loop_completely): Fix reporting.
21489 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21491 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21493 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21496 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21497 if given a non-constant argument for an intrinsic which requires a
21500 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21502 * profile.c (compare_freqs): New function.
21503 (branch_prob): Sort edge list.
21504 (find_spanning_tree): Assume that the list is priority sorted.
21506 2017-06-16 Richard Biener <rguenther@suse.de>
21508 PR tree-optimization/81090
21509 * passes.def (pass_record_bounds): Remove.
21510 * tree-pass.h (make_pass_record_bounds): Likewise.
21511 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21512 make_pass_record_bounds): Likewise.
21513 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21514 not free niter estimates at the beginning but at the end.
21515 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21517 2017-06-16 Richard Biener <rguenther@suse.de>
21519 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21520 initializer to workaround ICE in host GCC 4.8.
21522 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21524 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21526 (clone_inlined_nodes): Update.
21528 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21530 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21531 prefetch settings, and enable prefetching by default at -O3.
21533 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21535 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21536 Set flag_prefetch_loop_arrays according to tuning data.
21538 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21540 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21541 New tune structure.
21542 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21543 [Unrelated to main purpose of the patch] Place the pointer field last
21544 to enable type checking errors when tune structure are wrongly merged.
21545 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21546 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21547 (thunderx2t99_prefetch_tune): New tune constants.
21548 (tune_params *_tunings): Update all tunings (no functional change).
21549 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21550 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21551 from tunings structures.
21553 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21556 * ubsan.c (instrument_null): Add T argument, use it instead
21557 of computing it based on IS_LHS.
21558 (instrument_object_size): Likewise.
21559 (pass_ubsan::execute): Adjust instrument_null and
21560 instrument_object_size callers to pass gimple_get_lhs or
21561 gimple_assign_rhs1 result to it. Use instrument_null instead of
21562 calling get_base_address and instrument_mem_ref. Handle
21563 aggregate call arguments for object-size sanitization.
21565 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21567 PR tree-optimization/81089
21568 * tree-vrp.c (is_masked_range_test): Validate operands of
21571 2017-06-15 Martin Sebor <msebor@redhat.com>
21574 * dumpfile.c (dump_register): Avoid calling memset to initialize
21575 a class with a default ctor.
21576 * gcc.c (struct compiler): Remove const qualification.
21577 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21578 * hash-table.h: Ditto.
21579 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21581 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21582 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21584 * params.h (struct param_info): Make struct members non-const.
21585 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21586 with default initialization.
21587 * vec.h (vec_copy_construct, vec_default_construct): New helper
21589 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21590 with vec_copy_construct.
21591 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21592 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21593 * doc/invoke.texi (-Wclass-memaccess): Document.
21595 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21597 * emit-rtl.h (is_leaf): Update comment about local
21598 register allocator.
21600 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21603 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21604 for a variable to have a section before checking if the section has a
21606 Set section to.persistent if persistent attribute is set.
21607 Warn if .persistent attribute is used on an automatic variable.
21609 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21611 PR rtl-optimization/80474
21612 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21614 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21616 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21619 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21621 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21622 (casesi): Emit bounds checking as RTL.
21623 (casesi_internal_mips16_<mode>): Remove bounds checking.
21625 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21627 * config/xtensa/xtensa.c (xtensa_option_override): Append
21628 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21629 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21630 xtensa_doloop_hooks): Define unconditionally.
21631 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21633 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21634 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21635 for it in xtensa_option_override.
21636 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21637 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21639 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21641 * doc/cppopts.texi: Document '-' special value to -MF.
21643 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21645 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21646 (cortex_a53_fconst): Likewise.
21647 (cortex_a53_fpmul): Likewise.
21648 (cortex_a53_f_load_64): Likewise.
21649 (cortex_a53_f_load_many): Likewise.
21650 (cortex_a53_advsimd_alu): Likewise.
21651 (cortex_a53_advsimd_alu_q): Likewise.
21652 (cortex_a53_advsimd_mul): Likewise.
21653 (cortex_a53_advsimd_mul_q): Likewise.
21654 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21655 Add missing fmul, r2f_cvt and fconst cases.
21657 2017-06-14 Richard Biener <rguenther@suse.de>
21659 PR middle-end/81088
21660 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21662 (fold_binary_loc): When associating do not treat pre-existing
21663 TREE_OVERFLOW on literal constants as a reason to allow
21664 TREE_OVERFLOW on associated literal constants.
21666 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21668 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21669 (MASK_FEATURES): New macro.
21670 * config/sparc/sparc.c (sparc_option_override): Remove the special
21671 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21673 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21675 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21676 a division of 0 if non-call exceptions are enabled.
21678 2017-06-14 Andrew Pinski <apinski@cavium.com>
21679 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21682 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21683 Improve vector initialization code gen for only variable case.
21685 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21687 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21689 2017-06-14 Richard Biener <rguenther@suse.de>
21691 PR tree-optimization/81083
21692 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21695 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21697 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21698 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21699 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21700 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21701 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21702 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21704 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21706 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21707 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21709 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21711 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21713 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21715 * config/rs6000/t-rtems: Don't handle SPE.
21717 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21719 * config/rs6000/t-linux: Don't handle SPE.
21721 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21723 * config/rs6000/eabispe.h: Delete file.
21725 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21727 * config/rs6000/t-spe: Delete file.
21729 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21731 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21732 (rs6000_legitimate_offset_address_p): Return false for anything in
21733 V2SImode or V2SFmode.
21735 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21737 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21738 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21739 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21741 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21742 (rs6000_legitimate_offset_address_p): Ditto.
21743 (rs6000_emit_move): Ditto.
21744 (rs6000_init_builtins): Remove V4HI_type_node.
21746 2017-06-13 Martin Liska <mliska@suse.cz>
21749 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21750 (gate_asan): Likewise.
21751 * asan.h (asan_no_sanitize_address_p): Remove the function.
21752 (sanitize_flags_p): New function.
21753 * builtins.def: Fix coding style.
21754 * common.opt: Use renamed enum value.
21755 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21756 * doc/extend.texi: Document no_sanitize attribute.
21757 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21758 to SANITIZE_UNDEFINED_NONDEFAULT.
21759 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21760 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21761 Use sanitize_flags_p.
21762 * gimplify.c (gimplify_function_tree): Likewise.
21763 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21764 * opts.c (parse_no_sanitize_attribute): New function.
21765 (common_handle_option): Use renamed enum value.
21766 * opts.h (parse_no_sanitize_attribute): Declare.
21767 * tree.c (sanitize_flags_p): New function.
21768 * tree.h: Declared here.
21769 * tsan.c: Use sanitize_flags_p.
21770 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21771 (instrument_mem_ref): Likewise.
21772 (instrument_bool_enum_load): Likewise.
21773 (do_ubsan_in_current_function): Remove the function.
21774 (pass_ubsan::execute): Use sanitize_flags_p.
21775 * ubsan.h: Remove do_ubsan_in_current_function
21776 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21777 (dump_function_to_file): Use it here.
21779 2017-06-13 Martin Jambor <mjambor@suse.cz>
21781 PR tree-optimization/80803
21782 PR tree-optimization/81063
21783 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21784 (propagate_subaccesses_across_link): Enqueue subtree whenever
21785 necessary instead of relying on the caller.
21787 2017-06-13 Martin Jambor <mjambor@suse.cz>
21789 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21790 that have a first_link.
21791 (sort_and_splice_var_accesses): Do not check first_link before
21793 (subtree_mark_written_and_enqueue): Likewise.
21794 (propagate_all_subaccesses): Likewise and do not stop at first
21795 parent with a first_link.
21797 2017-06-13 Martin Jambor <mjambor@suse.cz>
21799 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21802 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21804 * match.pd: New pattern.
21806 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21808 * tree-vrp.c (is_masked_range_test): New function.
21809 (register_edge_assert_for): Determine ranges for
21812 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21814 PR tree-optimization/67328
21815 * fold-const.c (maskable_range_p): New function.
21816 (build_range_check): Generate bittests if possible.
21818 2017-06-13 Martin Liska <mliska@suse.cz>
21820 * gimple-pretty-print.c (dump_probability): Add new argument.
21821 (dump_edge_probability): Dump both probability and count.
21822 (dump_gimple_label): Likewise.
21823 (dump_gimple_bb_header): Likewise.
21825 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21828 * config/avr/avr-devices.c: Fix indentation.
21829 * config/avr/gen-avr-mmcu-specs.c: Dito.
21831 2017-06-13 Richard Biener <rguenther@suse.de>
21833 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21834 instead get vector type from stmt_info.
21835 (vectorizable_reduction): Adjust. Remove dead code.
21837 2017-06-13 Richard Biener <rguenther@suse.de>
21839 PR middle-end/81065
21840 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21841 case of C * (x * C2 + C3).
21842 (fold_addr_of_array_ref_difference): Properly fold index difference.
21844 2017-06-12 David S. Miller <davem@davemloft.net>
21847 * config/sparc/sparc.md (return expander): Emit frame blockage if
21848 function uses alloca.
21850 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21852 * combine.c (make_field_assignment): Check len rather than the mode
21853 precision when calling force_to_mode.
21855 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21857 Support multilibs and devices that see flash in RAM address range.
21860 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21861 (avr_mcu_t) <flash_pm_offset>: New field.
21862 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21863 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21864 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21865 (AVR_TINY_PM_OFFSET): Remove macro.
21866 * config/avr/avr.opt (-mshort-calls): New option.
21867 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21868 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21869 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21870 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21871 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21872 instead of avr_arch->have_jmp_call.
21873 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21874 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21875 avr_arch->flash_pm_offset to define.
21876 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21877 new field flash_pm_offset. Add entry for avrxmega3.
21878 (avr_texinfo): Add entry for avrxmega3.
21879 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21880 attiny212, attiny214,
21881 attiny412, attiny414, attiny416, attiny417,
21882 attiny814, attiny816, attiny817,
21883 attiny1614, attiny1616, attiny1617,
21884 attiny3214, attiny3216, attiny3217.
21885 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21886 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21887 (avr_print_operand_address) [AVR_TINY]: Same.
21888 (avr_asm_init_sections) <readonly_data_section>: Only patch
21889 callback if avr_arch->flash_pm_offset = 0.
21890 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21891 for rodata if avr_arch->flash_pm_offset != 0.
21892 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21893 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21894 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21895 (m_options): Append opt_rcall.
21896 (m_dirnames): Append dir_rcall.
21897 * config/avr/t-multilib: Regenerate.
21899 * configure.ac [target=avr]: Check whether avrxmega3 default
21900 linker description file works as needed.
21901 * configure: Regenerate.
21902 * doc/avr-mmcu.texi: Regenerate.
21903 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21904 <__AVR_ARCH__>: Document avrxmega3 and 103.
21905 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21906 <__AVR_SHORT_CALLS__>: Document it.
21907 <__AVR_PM_BASE_ADDRESS__>: Document it.
21908 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21909 (AVR Variable Attributes) <progmem>: Document this is
21910 not needed for avrxmega3.
21911 (AVR Named Address Spaces) <__flash>: Dito.
21913 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21915 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21917 2017-06-12 Doug Rupp <rupp@adacore.com>
21919 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21920 Append vxworks-stdint.h to the tm_file list.
21921 * config/vxworks-stdint.h: New file.
21923 2017-06-12 Martin Liska <mliska@suse.cz>
21925 PR tree-optimization/81041
21926 * tree-profile.c (gimple_gen_ic_func_profiler):
21927 Create an extra BB in profile-generate
21928 (gimple_gen_time_profiler): Likewise.
21930 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21932 PR tree-optimization/81003
21933 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21934 (update_range_test): Use it instead of force_gimple_operand_gsi.
21936 2017-06-12 Richard Biener <rguenther@suse.de>
21938 PR tree-optimization/81053
21939 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21940 with backedge value not defined in loop. Simplify def stmt
21943 2017-06-11 Tom de Vries <tom@codesourcery.com>
21946 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21948 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21949 nvptx_cannot_force_const_mem.
21951 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21953 * opts.c (finish_options): Move test for flag_split_stack after
21954 it has been initialized.
21956 2017-06-11 Jason Merrill <jason@redhat.com>
21958 * tree.h (id_equal): New.
21959 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21960 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21961 instead of strcmp of IDENTIFIER_POINTER.
21963 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21965 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21966 (mark_all_inlined_calls_cdtor): Fix formating.
21967 (inline_transform): Rescale profile before inlining.
21969 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21971 * cgraph.h (cgraph_edge::clone): Update prototype.
21972 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21973 (cgraph_node::create_clone): Update.
21974 (cgraph_node::create_version_clone): Update.
21975 * tree-inline.c (copy_bb): Update.
21976 (expand_call_inline): Update.
21978 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21980 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21981 factored out from ...
21982 (rs6000_emit_prologue): ... here.
21984 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21986 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21987 factored out from ...
21988 (rs6000_emit_prologue): ... here.
21990 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21992 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21994 (handle_missing_profiles): Fix computation of tp_first_run.
21995 (counts_to_freqs): Do not touch freqs when count is 0.
21997 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21999 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
22002 2017-06-10 Tom de Vries <tom@codesourcery.com>
22004 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22005 attributes): Document signal effective target.
22007 2017-06-10 Tom de Vries <tom@codesourcery.com>
22009 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22010 Document effective target stack_size.
22012 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22014 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22015 to the edit_context if they can be auto-applied.
22017 2017-06-9 Ian Lance Taylor <iant@golang.org>
22019 * opts.c (finish_options): If -fsplit-stack, disable implicit
22020 -forder-blocks-and-partition.
22021 * doc/invoke.texi (Optimize Options): Document that when using
22022 -fsplit-stack -forder-blocks-and-partition is not implicitly
22025 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22027 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22028 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22029 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22030 * builtins.def (abort, trap, unreachable): Declare cold.
22031 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22032 * tree-core.h (ECF_COLD): New.
22033 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22034 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22036 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22038 * predict.c (unlikely_executed_stmt_p): Cleanup.
22040 2017-06-09 Richard Biener <rguenther@suse.de>
22042 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22043 model if the ref is always written to.
22045 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22047 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22049 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22051 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22054 2017-06-09 Tom de Vries <tom@codesourcery.com>
22057 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22058 "target cannot support label values" when encountering LABEL_REF.
22060 2017-06-09 Martin Liska <mliska@suse.cz>
22062 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22063 (gimple_gen_ic_func_profiler): Emit direct comparison
22064 of __gcov_indirect_call_callee with NULL.
22065 (gimple_gen_time_profiler): Change probability from
22066 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22068 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22070 * profile.c (edge_gcov_counts): Turn to pointer.
22071 (compute_branch_probabilities, compute_branch_probabilities): Update.
22072 (branch_prob): Do not clear edge_gcov_count.
22073 * profile.h (edge_gcov_counts): Turn to pointer.
22074 (edge_gcov_count): Update.
22076 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22078 * gimple.h (gimple_check_failed): Mark cold.
22080 2017-06-09 Richard Biener <rguenther@suse.de>
22082 PR tree-optimization/66623
22083 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22084 refactor check_reduction into two parts, properly computing
22085 whether we have to check reduction validity for outer loop
22088 2017-06-09 Richard Biener <rguenther@suse.de>
22090 PR tree-optimization/79483
22091 * graphite-scop-detection.c (order): New global.
22092 (get_order): Compute bb to order mapping that satisfies code
22093 generation constraints.
22094 (cmp_pbbs): New helper.
22095 (build_scops): Start domwalk at entry block, sort generated
22098 2017-06-09 Richard Biener <rguenther@suse.de>
22100 PR middle-end/81007
22101 * ipa-polymorphic-call.c
22102 (ipa_polymorphic_call_context::restrict_to_inner_class):
22103 Skip FIELD_DECLs with error_mark_node type.
22104 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22107 2017-06-09 Martin Liska <mliska@suse.cz>
22109 * predict.c (struct branch_predictor): New struct.
22110 (test_prediction_value_range): New test.
22111 (predict_c_tests): New function.
22112 * selftest-run-tests.c (selftest::run_tests): Run the function.
22113 * selftest.h: Declare new tests.
22115 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22118 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22119 gen_add3_insn did not fail.
22120 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22121 r0, construct that number in a temporary reg and add that reg to r0.
22122 If asked to put the result in r0 as well, fail.
22124 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22126 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22127 for early expansion of vec_eqv.
22129 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22131 PR middle-end/81005
22132 * ubsan.c (instrument_null): Avoid pointless code temporary.
22133 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22135 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22139 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22142 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22145 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22147 * predict.c (unlikely_executed_edge_p): Move ahead.
22148 (probably_never_executed_edge_p): Use it.
22150 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22152 PR middle-end/79988
22153 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22154 gimple_call_builtin_p call.
22156 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22158 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22159 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22160 rtl_check_failed_type2, rtl_check_failed_code1,
22161 rtl_check_failed_code2, rtl_check_failed_code_mode,
22162 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22163 rtvec_check_failed_bounds, rtl_check_failed_flag,
22164 _fatal_insn_not_found, _fatal_insn): Likewise.
22165 * tree.h (tree_contains_struct_check_failed,
22166 tree_check_failed, tree_not_check_failed,
22167 tree_class_check_failed, tree_range_check_failed,
22168 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22169 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22170 tree_operand_check_failed, omp_clause_check_failed,
22171 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22174 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22176 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22177 flag_branch_probabilities.
22178 * ipa-inline.c (edge_badness): Likewise.
22179 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22180 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22181 * predict.c (maybe_hot_frequency_p): Likewise.
22182 (probably_never_executed): Likewise.
22183 * sched-ebb.c (schedule_ebbs): Likewise.
22184 * sched-rgn.c (find_single_block_region): Likewise.
22185 * tracer.c (tail_duplicate): Likewise.
22187 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22189 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22190 longer requires x_flag_profile_use.
22192 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22194 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22195 instead of flag_reorder_blocks_and_partition.
22196 * dbxout.c (dbxout_function_end): Likewise.
22197 * dwarf2out.c (gen_subprogram_die): Likewise.
22198 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22199 * hw-doloop.c (reorg_loops): Likewise.
22200 * varasm.c (assemble_start_function,
22201 assemble_end_function): Likewise.
22202 (decide_function_section): Do not check for
22203 flag_reorder_blocks_and_partition.
22205 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22207 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22209 (chkp_get_hard_register_fake_addr_expr): Ditto.
22210 (chkp_build_addr_expr): Add check for hard reg case.
22211 (chkp_parse_array_and_component_ref): Ditto.
22212 (chkp_find_bounds_1): Ditto.
22213 (chkp_process_stmt): Don't generate bounds store for
22216 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22218 * predict.c (maybe_hot_bb_p): Do not check profile status.
22219 (maybe_hot_edge_p): Likewise.
22220 (probably_never_executed): Check for zero counts even if profile
22222 (unlikely_executed_edge_p): New function.
22223 (unlikely_executed_stmt_p): New function.
22224 (unlikely_executed_bb_p): New function.
22225 (set_even_probabilities): Use unlikely predicates.
22226 (combine_predictions_for_bb): Likewise.
22227 (predict_paths_for_bb): Likewise.
22228 (predict_paths_leading_to_edge): Likewise.
22229 (determine_unlikely_bbs): New function.
22230 (estimate_bb_frequencies): Use it.
22231 (compute_function_frequency): Use zero counts even if profile is
22233 * profile-count.h: Fix typo.
22235 2017-08-08 Julia Koval <julia.koval@intel.com>
22237 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22238 _mm512_mask_cvtsepi16_storeu_epi8,
22239 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22240 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22241 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22242 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22243 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22244 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22245 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22246 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22247 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22248 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22249 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22250 __builtin_ia32_pmovuswb256mem_mask,
22251 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22252 __builtin_ia32_pmovwb512mem_mask): New builtins.
22254 2017-08-08 Julia Koval <julia.koval@intel.com>
22256 PR target/73350,80862
22257 * config/i386/subst.md (round): Fix round pattern.
22258 * config/i386/i386.c (ix86_erase_embedded_rounding):
22259 Fix erasing rounding for the fixed pattern.
22261 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22263 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22265 2017-06-08 Martin Liska <mliska@suse.cz>
22267 PR gcov-profile/80911
22268 * gcov.c (block_info::block_info): New constructor.
22270 2017-06-07 Carl Love <cel@us.ibm.com>
22272 * config/rs6000/rs6000-c: The return type of the following
22273 built-in functions was implemented as int not long long. Fix sign
22274 of return value for the unsigned version of vec_mulo and vec_mule.
22275 vector unsigned long long vec_bperm (vector unsigned long long,
22276 vector unsigned char)
22277 vector signed long long vec_mule (vector signed int,
22279 vector unsigned long long vec_mule (vector unsigned int,
22280 vector unsigned int)
22281 vector signed long long vec_mulo (vector signed int,
22283 vector unsigned long long vec_mulo (vector unsigned int,
22284 vector unsigned int)
22285 * doc/extend.texi: Fix the documentation for the built-in
22288 2017-06-07 Carl Love <cel@us.ibm.com>
22291 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22294 2017-06-07 Carl Love <cel@us.ibm.com>
22296 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22297 support, Generate doublehv for signed int/float for BE case only.
22299 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22301 * doc/invoke.texi (mcx16): Rewrite.
22303 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22305 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22306 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22307 *mov<mode>_softfloat, and an anonymous splitter): Use
22308 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22310 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22312 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22314 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22315 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22316 (rs6000_debug_reg_global): Adjust.
22317 (rs6000_init_hard_regno_mode_ok): Adjust.
22318 (rs6000_dbx_register_number): Adjust.
22319 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22320 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22321 Remove SPE_ACC and SPEFSCR.
22322 (REG_ALLOC_ORDER): Ditto.
22323 (FRAME_POINTER_REGNUM): Change to 111.
22324 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22325 (REG_CLASS_NAMES): Ditto.
22326 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22327 (REGISTER_NAMES): Ditto.
22328 (ADDITIONAL_REG_NAMES): Ditto.
22329 (rs6000_reg_names): Ditto.
22330 * config/rs6000/rs6000.md: Renumber some register number
22333 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22335 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22337 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22338 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22340 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22341 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22342 Delete the SPE high registers.
22343 (REG_ALLOC_ORDER): Ditto.
22344 (enum reg_class): Remove SPE_HIGH_REGS.
22345 (REG_CLASS_NAMES): Ditto.
22346 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22347 (REGISTER_NAMES): Ditto.
22348 (rs6000_reg_names): Ditto.
22349 * doc/tm.texi.in: Remove SPE as example.
22350 * doc/tm.texi: Regenerate.
22352 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22354 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22355 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22356 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22357 * config/rs6000/rs6000.md (type): Remove "brinc".
22359 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22361 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22362 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22363 * config/rs6000/linuxspe.h: Delete file.
22364 * config/rs6000/rs6000.md: Don't include spe.md.
22365 * config/rs6000/spe.h: Delete file.
22366 * config/rs6000/spe.md: Delete file.
22367 * config/rs6000/t-rs6000: Remove spe.md.
22369 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22371 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22372 (reg_or_none500mem_operand): Delete.
22373 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22374 instead of reg_or_none500mem_operand.
22376 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22378 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22379 handling of SPE flags.
22380 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22382 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22384 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22386 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22387 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22388 paired_divv2sf3): Similar.
22389 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22390 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22391 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22393 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22394 Rename the paired_* instruction patterns.
22395 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22397 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22398 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22399 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22400 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22401 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22402 PAIRED_VECTOR_MODE.
22403 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22404 (spe_func_has_64bit_regs_p): Delete.
22405 (spe_expand_predicate_builtin): Delete.
22406 (spe_expand_evsel_builtin): Delete.
22407 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22408 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22409 (invalid_e500_subreg): Delete.
22410 (rs6000_legitimize_address): Always force_reg op2 as well, for
22411 paired single memory accesses.
22412 (rs6000_member_type_forces_blk): Delete.
22413 (rs6000_spe_function_arg): Delete.
22414 (rs6000_expand_unop_builtin): Delete SPE handling.
22415 (rs6000_expand_binop_builtin): Ditto.
22416 (spe_expand_stv_builtin): Delete.
22417 (bdesc_2arg_spe): Delete.
22418 (spe_expand_builtin): Delete.
22419 (spe_expand_predicate_builtin): Delete.
22420 (spe_expand_evsel_builtin): Delete.
22421 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22422 (spe_init_builtins): Delete.
22423 (spe_func_has_64bit_regs_p): Delete.
22424 (savres_routine_name): Delete "info" parameter. Adjust callers.
22425 (rs6000_emit_stack_reset): Ditto.
22426 (rs6000_dwarf_register_span): Delete.
22427 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22428 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22429 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22431 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22433 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22434 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22435 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22436 mulv2sf3, divv2sf3): Delete expanders.
22438 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22440 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22442 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22444 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22445 * config/rs6000/rs6000.c: Ditto.
22447 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22449 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22450 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22451 comparison_operator.
22453 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22455 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22456 * config/rs6000/rs6000.opt: Ditto.
22457 * config/rs6000/t-rtems: Ditto.
22459 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22461 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22462 TARGET_E500_SINGLE by 0, simplify.
22463 * config/rs6000/rs6000.c: Ditto.
22464 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22465 (spe_build_register_parallel): Delete.
22466 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22467 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22468 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22469 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22470 (E500_CONVERT): Delete.
22471 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22473 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22475 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22476 * config/rs6000/dfp.md: Ditto.
22477 (negdd2, *negdd2_fpr): Merge.
22478 (absdd2, *absdd2_fpr): Merge.
22479 (negtd2, *negtd2_fpr): Merge.
22480 (abstd2, *abstd2_fpr): Merge.
22481 * config/rs6000/e500.h: Delete file.
22482 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22483 TARGET_FPRS by 1 and simplify.
22484 * config/rs6000/rs6000-c.c: Ditto.
22485 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22486 TARGET_DF_SPE by 0.
22487 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22489 * config/rs6000/rs6000.md: Ditto.
22490 (floatdidf2, *floatdidf2_fpr): Merge.
22491 (move_from_CR_gt_bit): Delete.
22492 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22493 (E500_CR_IOR_COMPARE): Delete.
22494 (All patterns that require !TARGET_FPRS): Delete.
22495 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22497 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22499 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22501 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22503 * graphds.c (add_edge): Intitialize edge's attached data.
22504 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22505 pointer parameter. Call pointed function on each edge during
22506 graph traversing. Skip traversing the edge when the function
22508 (graphds_dfs, graphds_scc): Ditto.
22509 (for_each_edge): New parameter. Pass the new parameter to callback
22511 * graphds.h (skip_edge_callback): New function pointer type.
22512 (graphds_dfs, graphds_scc): New function pointer parameter.
22513 (graphds_edge_callback, for_each_edge): New parameter.
22515 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22517 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22518 out code checking if runtime alias check is possible to below ...
22519 Call the new function.
22520 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22521 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22523 2017-06-07 Marek Polacek <polacek@redhat.com>
22526 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22527 TYPE_OVERFLOW_WRAPS check.
22529 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22531 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22532 if versioning is required.
22533 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22534 peeling with the check for versioning.
22536 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22538 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22539 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22540 Set true to new parameter if new ssa variable is defined.
22541 (vect_gen_vector_loop_niters): Refactor. Set range information
22542 for the new vector loop bound variable.
22543 (vect_do_peeling): Ditto.
22545 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22547 * tree-affine.c (ssa.h): Include header file.
22548 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22549 has wrapping overflow behavior.
22551 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22553 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22555 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22557 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22558 (tree_to_aff_combination): ... here.
22560 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22562 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22563 reg_pressure model function.
22564 (ivopts_global_cost_for_size): Delete.
22565 (determine_set_costs, iv_ca_recount_cost): Call new model function
22566 ivopts_estimate_reg_pressure.
22568 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22570 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22571 expensive than udiv. Remove floating point cases from mod.
22573 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22575 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22576 Increase idiv cost.
22578 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22580 * config/aarch64/aarch64.md
22581 (copysignsf3): Fix mask generation.
22583 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22585 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22587 (class dump_manager): Add register_dumps method.
22588 * dumpfile.c: Include langhooks.h.
22589 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22590 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22591 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22592 (dump_manager::dump_register): Start with 512 entries instead of 32.
22593 (dump_manager::register_dumps): New method.
22594 * toplev.c (general_init): Instead of invoking register_dumps
22595 langhook, invoke register_dumps method on the dump manager.
22596 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22599 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22601 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22602 Say that # requires an associated define_split to exist, and that
22603 the define_split must be suitable for use after register allocation.
22605 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22607 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22608 (compute_outgoing_frequencies): Also initialize zero counts.
22609 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22610 around loops; preserve more of profile when nothing changes.
22612 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22614 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22616 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22617 * config/arm/arm-cpu-cdata.h: Regenerate.
22618 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22619 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22620 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22621 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22622 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22624 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22626 * config/arm/t-rmprofile: Likewise.
22627 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22629 2017-06-06 David S. Miller <davem@davemloft.net>
22632 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22633 blockage if function uses alloca.
22635 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22637 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22638 New "uid" fields to hold pretty-print IDs of group and ref.
22639 Memory references are now identified as <group_id>:<ref_id>
22640 instead of using [random] addresses.
22641 (dump_mem_details): Simplify, no functional change.
22642 (dump_mem_ref): Simplify and make output more concise.
22643 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22644 (find_or_create_group): Initialize group uid.
22645 (record_ref): Initialize ref uid. Improve debug output.
22646 (prune_group_by_reuse, should_issue_prefetch_p,)
22647 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22648 (mark_nontemporal_store, determine_loop_nest_reuse):
22649 Improve debug output.
22651 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22653 * dbgcnt.def (prefetch): New debug counter.
22654 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22655 (schedule_prefetches): Stop issueing prefetches if debug counter
22658 2017-06-06 Tom de Vries <tom@codesourcery.com>
22660 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22661 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22663 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22665 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22666 Use aarch64_reg_or_zero predicate for operand 4.
22667 (aarch64_compare_and_swap<mode> define_insn_and_split):
22668 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22669 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22671 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22673 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22674 (arm_compute_save_core_reg_mask): This.
22675 (thumb1_compute_save_reg_mask): Rename into ...
22676 (thumb1_compute_save_core_reg_mask): This.
22677 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22678 (arm_compute_frame_layout): Likewise.
22680 2017-06-06 Richard Biener <rguenther@suse.de>
22682 PR tree-optimization/80974
22683 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22684 keep or clear leaders SSA info.
22686 2017-06-06 Tom de Vries <tom@codesourcery.com>
22688 * config/nvptx/nvptx.c (split_mode_p): New function.
22689 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22691 2017-06-06 Tom de Vries <tom@codesourcery.com>
22693 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22695 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22698 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22701 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22703 * shrink-wrap.c (handle_simple_exit): Update profile.
22704 (try_shrink_wrapping): Upate profile.
22706 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22708 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22709 (tree_guess_outgoing_edge_probabilities): New.
22710 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22711 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22713 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22715 * ipa-split.c (split_function): Initialize return bb profile.
22717 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22719 * profile.c (compute_branch_probabilities): Also initialize
22720 EXIT_BLOCK profile.
22722 2017-06-06 Richard Biener <rguenther@suse.de>
22724 PR tree-optimization/80928
22725 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22726 (vect_analyze_loop_operations): Properly guard analysis for
22728 (vect_transform_loop): Likewise.
22729 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22730 (vect_model_induction_cost): Do not cost for pure SLP.
22731 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22732 of induction in inner loop vectorization.
22733 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22734 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22735 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22737 (vect_analyze_slp_cost_1): Cost induction.
22738 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22739 (vect_get_slp_vect_defs): Likewise.
22740 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22741 (vect_transform_stmt): Handle SLP reductions.
22742 * tree-vectorizer.h (vectorizable_induction): Adjust.
22744 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22746 * config/rs6000/rs6000.c (make_resolver_func): Update
22747 init_lowered_empty_function call.
22749 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22751 * doc/invoke.texi: Document the -fprofile-abs-path option.
22752 * common.opt (fprofile-abs-path): New option.
22753 * gcov-io.h (gcov_write_filename): Declare.
22754 * gcov-io.c (gcov_write_filename): New function.
22755 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22756 * profile.c (output_location): Likewise.
22758 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22760 * shring-wrap.c: Revert accidental commit.
22762 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22764 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22766 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22768 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22770 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22771 profile in callgraph edge.
22772 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22773 (apply_scale): Likewise.
22774 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22775 Also scale profile when inlining function with zero profile.
22776 (initialize_cfun): Update exit block profile even when it is zero.
22777 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22778 when profile is read.
22780 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22782 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22783 (CLONE_*): New constants to define the processors we can generate
22784 code for with the target_clone attribute.
22785 (rs6000_clone_map): New array to identify which clone processors
22786 the current program is running on.
22787 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22788 target_clone attribute.
22789 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22790 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22791 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22792 (cpu_expand_builtin): Add support for target_clone attribute.
22793 (rs6000_valid_attribute_p): Allow "default" attribute.
22794 (get_decl_name): New debug function to simplify printing the
22795 current function name in debugging statements.
22796 (rs6000_clone_priority): New functions to support the target_clone
22797 attribute, and be able to generate code to switch between ISA 2.05
22798 through ISA 3.0 (power6 through power9).
22799 (rs6000_compare_version_priority): Likewise.
22800 (rs6000_get_function_versions_dispatcher): Likewise.
22801 (make_resolver_func): Likewise.
22802 (add_condition_to_bb): Likewise.
22803 (dispatch_function_versions): Likewise.
22804 (rs6000_generate_version_dispatcher_body): Likewise.
22805 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22806 (fusion_gpr_load_p): Fix a spacing issue.
22807 * doc/extend.texi (Common Function Attributes): Document that the
22808 PowerPC supports the target_clone attribute.
22810 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22812 * config/arm/arm.h: explain F symbol found in description of ARM
22813 register allocation in its legend.
22815 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22817 * config/mips/frame-header-opt.c: Include profile-count.h.
22818 * config/riscv/riscv.c: Include profile-count.h
22820 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22822 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22824 (sm_set_flag_if_changed): Add bbs field.
22825 (execute_sm_if_changed_flag_set): Pass BBS.
22826 (execute_sm): Update.
22828 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22830 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22833 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22835 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22836 (peephole2): New peephole2 to emit the above.
22837 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22839 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22841 * config/aarch64/aarch64.c (define_peephole2 above
22842 *sub_<shift>_<mode>): New peephole.
22844 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22846 * config/i386/i386.c (make_resolver_func): Update.
22847 * Makefile.in: Add profile-count.h and profile-count.o
22848 * auto-profile.c (afdo_indirect_call): Update to new API.
22849 (afdo_set_bb_count): Update.
22850 (afdo_propagate_edge): Update.
22851 (afdo_propagate_circuit): Update.
22852 (afdo_calculate_branch_prob): Update.
22853 (afdo_annotate_cfg): Update.
22854 * basic-block.h: Include profile-count.h
22855 (struct edge_def): Turn count to profile_count.
22856 (struct basic_block_def): Likewie.
22857 (REG_BR_PROB_BASE): Move to profile-count.h
22858 (RDIV): Move to profile-count.h
22859 * bb-reorder.c (max_entry_count): Turn to profile_count.
22860 (find_traces): Update.
22861 (rotate_loop):Update.
22862 (connect_traces):Update.
22863 (sanitize_hot_paths):Update.
22864 * bt-load.c (migrate_btr_defs): Update.
22865 * cfg.c (RDIV): Remove.
22866 (init_flow): Use alloc_block.
22867 (alloc_block): Uninitialize count.
22868 (unchecked_make_edge): Uninitialize count.
22869 (check_bb_profile): Update.
22870 (dump_edge_info): Update.
22871 (dump_bb_info): Update.
22872 (update_bb_profile_for_threading): Update.
22873 (scale_bbs_frequencies_int): Update.
22874 (scale_bbs_frequencies_gcov_type): Update.
22875 (scale_bbs_frequencies_profile_count): New.
22876 * cfg.h (update_bb_profile_for_threading): Update.
22877 (scale_bbs_frequencies_profile_count): Declare.
22878 * cfgbuild.c (compute_outgoing_frequencies): Update.
22879 (find_many_sub_basic_blocks): Update.
22880 * cfgcleanup.c (try_forward_edges): Update.
22881 (try_crossjump_to_edge): Update.
22882 * cfgexpand.c (expand_gimple_tailcall): Update.
22883 (construct_exit_block): Update.
22884 * cfghooks.c (verify_flow_info): Update.
22885 (dump_bb_for_graph): Update.
22886 (split_edge): Update.
22887 (make_forwarder_block): Update.
22888 (duplicate_block): Update.
22889 (account_profile_record): Update.
22890 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22891 (get_estimated_loop_iterations): Update.
22892 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22893 (single_likely_exit): Update.
22894 * cfgloopmanip.c (scale_loop_profile): Update.
22896 (set_zero_probability): Update.
22897 (lv_adjust_loop_entry_edge): Update.
22898 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22899 (purge_dead_edges): Update.
22900 (rtl_account_profile_record): Update.
22901 * cgraph.c (cgraph_node::create): Uninitialize count.
22902 (symbol_table::create_edge): Uninitialize count.
22903 (cgraph_update_edges_for_call_stmt_node): Update.
22904 (cgraph_edge::dump_edge_flags): Update.
22905 (cgraph_node::dump): Update.
22906 (cgraph_edge::maybe_hot_p): Update.
22907 * cgraph.h: Include profile-count.h
22908 (create_clone), create_edge, create_indirect_edge): Update.
22909 (cgraph_node): Turn count to profile_count.
22910 (cgraph_edge0: Likewise.
22911 (make_speculative, clone): Update.
22912 (create_edge): Update.
22913 (init_lowered_empty_function): Update.
22914 * cgraphclones.c (cgraph_edge::clone): Update.
22915 (duplicate_thunk_for_node): Update.
22916 (cgraph_node::create_clone): Update.
22917 * cgraphunit.c (cgraph_node::analyze): Update.
22918 (cgraph_node::expand_thunk): Update.
22919 * final.c (dump_basic_block_info): Update.
22920 * gimple-streamer-in.c (input_bb): Update.
22921 * gimple-streamer-out.c (output_bb): Update.
22922 * graphite.c (print_global_statistics): Update.
22923 (print_graphite_scop_statistics): Update.
22924 * hsa-brig.c: Include basic-block.h.
22925 * hsa-dump.c: Include basic-block.h.
22926 * hsa-gen.c (T sum_slice): Update.
22927 (convert_switch_statements):Update.
22928 * hsa-regalloc.c: Include basic-block.h.
22929 * ipa-chkp.c (chkp_produce_thunks): Update.
22930 * ipa-cp.c (struct caller_statistics): Update.
22931 (init_caller_stats): Update.
22932 (gather_caller_stats): Update.
22933 (ipcp_cloning_candidate_p): Update.
22934 (good_cloning_opportunity_p): Update.
22935 (get_info_about_necessary_edges): Update.
22936 (dump_profile_updates): Update.
22937 (update_profiling_info): Update.
22938 (update_specialized_profile): Update.
22939 (perhaps_add_new_callers): Update.
22940 (decide_about_value): Update.
22941 (ipa_cp_c_finalize): Update.
22942 * ipa-devirt.c (struct odr_type_warn_count): Update.
22943 (struct decl_warn_count): Update.
22944 (struct final_warning_record): Update.
22945 (possible_polymorphic_call_targets): Update.
22946 (ipa_devirt): Update.
22947 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22948 * ipa-icf.c (sem_function::merge): Update.
22949 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22950 * ipa-inline.c (compute_uninlined_call_time): Update.
22951 (compute_inlined_call_time): Update.
22952 (want_inline_small_function_p): Update.
22953 (want_inline_self_recursive_call_p): Update.
22954 (edge_badness): Update.
22955 (lookup_recursive_calls): Update.
22956 (recursive_inlining): Update.
22957 (inline_small_functions): Update.
22958 (dump_overall_stats): Update.
22959 (dump_inline_stats): Update.
22960 * ipa-profile.c (ipa_profile_generate_summary): Update.
22961 (ipa_propagate_frequency): Update.
22962 (ipa_profile): Update.
22963 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22964 * ipa-utils.c (ipa_merge_profiles): Update.
22965 * loop-doloop.c (doloop_modify): Update.
22966 * loop-unroll.c (report_unroll): Update.
22967 (unroll_loop_runtime_iterations): Update.
22968 * lto-cgraph.c (lto_output_edge): Update.
22969 (lto_output_node): Update.
22970 (input_node): Update.
22971 (input_edge): Update.
22972 (merge_profile_summaries): Update.
22973 * lto-streamer-in.c (input_cfg): Update.
22974 * lto-streamer-out.c (output_cfg): Update.
22975 * mcf.c (create_fixup_graph): Update.
22976 (adjust_cfg_counts): Update.
22977 (sum_edge_counts): Update.
22978 * modulo-sched.c (sms_schedule): Update.
22979 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22980 * predict.c (maybe_hot_count_p): Update.
22981 (probably_never_executed): Update.
22982 (dump_prediction): Update.
22983 (combine_predictions_for_bb): Update.
22984 (propagate_freq): Update.
22985 (handle_missing_profiles): Update.
22986 (counts_to_freqs): Update.
22987 (rebuild_frequencies): Update.
22988 (force_edge_cold): Update.
22989 * predict.h: Include profile-count.h
22990 (maybe_hot_count_p, counts_to_freqs): UPdate.
22991 * print-rtl-function.c: Do not include cfg.h
22992 * print-rtl.c: Include basic-block.h
22993 * profile-count.c: New file.
22994 * profile-count.h: New file.
22995 * profile.c (is_edge_inconsistent): Update.
22996 (correct_negative_edge_counts): Update.
22997 (is_inconsistent): Update.
22998 (set_bb_counts): Update.
22999 (read_profile_edge_counts): Update.
23000 (compute_frequency_overlap): Update.
23001 (compute_branch_probabilities): Update; Initialize and deinitialize
23003 (branch_prob): Update.
23004 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23005 (edge_gcov_count): New.
23006 (bb_gcov_count): New.
23007 * shrink-wrap.c (try_shrink_wrapping): Update.
23008 * tracer.c (better_p): Update.
23009 * trans-mem.c (expand_transaction): Update.
23010 (ipa_tm_insert_irr_call): Update.
23011 (ipa_tm_insert_gettmclone_call): Update.
23012 * tree-call-cdce.c: Update.
23013 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23014 (gimple_duplicate_sese_tail): Update.
23015 (gimple_account_profile_record): Update.
23016 (execute_fixup_cfg): Update.
23017 * tree-inline.c (copy_bb): Update.
23018 (copy_edges_for_bb): Update.
23019 (initialize_cfun): Update.
23020 (freqs_to_counts): Update.
23021 (copy_cfg_body): Update.
23022 (expand_call_inline): Update.
23023 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23024 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23025 (try_unroll_loop_completely): Update.
23026 (try_peel_loop): Update.
23027 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23028 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23029 * tree-ssa-loop-split.c (connect_loops): Update.
23030 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23031 * tree-ssa-reassoc.c (branch_fixup): Update.
23032 * tree-ssa-tail-merge.c (replace_block_by): Update.
23033 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23034 (compute_path_counts): Update.
23035 (update_profile): Update.
23036 (recompute_probabilities): Update.
23037 (update_joiner_offpath_counts): Update.
23038 (estimated_freqs_path): Update.
23039 (freqs_to_counts_path): Update.
23040 (clear_counts_path): Update.
23041 (ssa_fix_duplicate_block_edges): Update.
23042 (duplicate_thread_path): Update.
23043 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23044 (struct switch_conv_info): Update.
23045 * tree-tailcall.c (decrease_profile): Update.
23046 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23047 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23048 * value-prof.c (check_counter): Update.
23049 (gimple_divmod_fixed_value): Update.
23050 (gimple_mod_pow2): Update.
23051 (gimple_mod_subtract): Update.
23052 (gimple_ic_transform): Update.
23053 (gimple_stringop_fixed_value): Update.
23054 * value-prof.h (gimple_ic): Update.
23056 2017-06-02 Carl Love <cel@us.ibm.com>
23058 * config/rs6000/rs6000-c: Add support for built-in functions
23059 vector double vec_doublee (vector signed int);
23060 vector double vec_doublee (vector unsigned int);
23061 vector double vec_doublee (vector float);
23062 vector double vec_doubleh (vector signed int);
23063 vector double vec_doubleh (vector unsigned int);
23064 vector double vec_doubleh (vector float);
23065 vector double vec_doublel (vector signed int);
23066 vector double vec_doublel (vector unsigned int);
23067 vector double vec_doublel (vector float);
23068 vector double vec_doubleo (vector signed int);
23069 vector double vec_doubleo (vector unsigned int);
23070 vector double vec_doubleo (vector float);.
23071 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23072 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23074 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23075 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23076 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23078 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23079 vec_doublel, vec_doubleh.
23080 * doc/extend.texi: Update the built-in documentation file for the
23081 new built-in functions.
23083 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23086 * ipa-inline-analysis.c (free_growth_caches): Set
23087 edge_removal_hook_holder to NULL after removing it.
23089 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23091 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23092 comparision with zero.
23094 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23095 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23096 for early expansion of vec_min and vec_max builtins.
23097 (builtin_function_type): Add min/max unsigned variants to those
23098 identified as having unsigned arguments.
23100 2017-06-02 Olivier Hainque <hainque@adacore.com>
23102 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23104 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23106 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23107 Use VALL_F16 iterator rather than VALL.
23109 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23111 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23112 Emit CBNZ inside loop when doing a strong exchange and comparing
23113 against zero. Generate the CC flags after the loop.
23115 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23117 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23118 (dl_section_ref): New.
23119 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23120 On AIX, append an expression to subtract the size of the
23121 section length to dl_section_ref.
23123 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23125 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23126 for early expansion of vector absolute builtins.
23128 2017-06-02 Richard Biener <rguenther@suse.de>
23130 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23131 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23133 2017-06-02 Richard Biener <rguenther@suse.de>
23135 PR tree-optimization/80948
23136 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23137 stmt order as well.
23139 2017-06-02 Richard Biener <rguenther@suse.de>
23141 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23143 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23144 for inductions as relevant.
23146 2017-06-02 Richard Biener <rguenther@suse.de>
23148 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23149 (vectorizable_induction): ... this. Remove dead code.
23151 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23153 * builtins. (expand_builtin_alloca): Remove second parameter and
23154 infer its value from the first parameter instead.
23155 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23157 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23159 PR rtl-optimization/80903
23160 * loop-doloop.c (add_test): Unshare sequence.
23162 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23164 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23166 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23168 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23170 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23171 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23172 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23173 (xlogue_layout::get_stub_rtx): Make static.
23174 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23175 (xlogue_layout::compute_stub_managed_regs): Rename to...
23176 (xlogue_layout::count_stub_managed_regs): ...this.
23177 (xlogue_layout::is_stub_managed_reg): New function.
23178 (xlogue_layout::m_stub_names): Rename to...
23179 (xlogue_layout::s_stub_names): ...this, make static.
23180 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23181 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23182 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23183 xlogue_layout::s_stub_names): Instantiate statics.
23184 (stub_managed_regs): Remove.
23185 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23186 (disable_call_ms2sysv_xlogues): Rename to...
23187 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23188 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23190 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23191 change after reload_completed.
23192 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23194 (ix86_expand_prologue): Likewise.
23195 (ix86_expand_epilogue): Likewise.
23196 (ix86_expand_split_stack_prologue): Likewise.
23197 (ix86_compute_frame_layout): Remove frame parameter ...
23198 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23199 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23201 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23202 (ix86_frame): Move from here ...
23203 * config/i386/i386.h (ix86_frame): ... to here.
23204 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23205 complete ix86_frame data structure instead. Remove some_ld_name.
23207 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23209 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23210 symbols that hold a DECL_VALUE_EXPR.
23212 2017-06-01 Martin Jambor <mjambor@suse.cz>
23214 PR tree-optimization/80898
23215 * tree-sra.c (process_subtree_disqualification): Removed.
23216 (disqualify_candidate): Do not acll
23217 process_subtree_disqualification.
23218 (subtree_mark_written_and_enqueue): New function.
23219 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23220 RHS has been disqualified and re-queue LHS if necessary. Apart
23221 from that, ignore disqualified RHS.
23223 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23225 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23226 address fetch for z10 or later.
23228 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23230 * config/arc/arc.md (tst_movb): Add guard when splitting.
23232 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23234 * config/arc/arc.c (arc_can_eliminate): Test against
23235 arc_frame_pointer_needed.
23237 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23239 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23240 to prevent store reordering.
23241 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23242 (type): Add block type.
23243 (stack_tie): Define special instruction to be used in
23246 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23248 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23249 constraint. It is not valid for the pattern.
23250 (noncommutative_binary_comparison): Likewise.
23252 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23254 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23257 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23259 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23260 be used by the reg-alloc.
23262 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23264 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23265 reg-alloc when having mul64 or mul32x16 instructions.
23266 (mulsidi3): Likewise.
23267 (umulsidi3): Likewise.
23268 (mulsi32x16): New pattern.
23269 (mulsi64): Likewise.
23270 (mulsidi64): Likewise.
23271 (umulsidi64): Likewise.
23272 (MUL32x16_REG): Define.
23273 (mul64_600): Use MUL32x16_REG.
23274 (mac64_600): Likewise.
23275 (umul64_600): Likewise.
23276 (umac64_600): Likewise.
23278 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23280 * config/arc/arc.md (mulsi3_700): Make it commutative.
23282 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23284 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23286 (*sign_extendsidi2_insn): Likewise for movstosw.
23288 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23290 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23291 the type of the input discriminant value. Convert the
23292 discriminant value of signedness vary.
23294 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23296 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23297 Add to -Wall section.
23299 2017-06-01 Richard Biener <rguenther@suse.de>
23301 PR middle-end/66313
23302 * fold-const.c (fold_plusminus_mult_expr): If the factored
23303 factor may be zero use a wrapping type for the inner operation.
23304 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23305 and handle moved defs.
23306 (process_assignment): Properly guard the unary op case. Return a
23307 tri-state indicating that moving the stmt before the call may allow
23308 to continue. Pass through to_move.
23309 (find_tail_calls): Handle moving unrelated defs before
23312 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23315 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23316 splitter result in the canonical way.
23318 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23320 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23321 also for 32bit target. Update insn attributes.
23322 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23324 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23326 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23327 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23328 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23329 (_mm_maskz_min_ss): New intrinsics.
23331 2017-05-31 Martin Liska <mliska@suse.cz>
23333 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23334 Change comment style to one we normally use.
23335 (vectorizable_reduction): Likewise.
23336 (vectorizable_induction): Likewise.
23337 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23338 (vectorizable_call): Likewise.
23339 (vectorizable_simd_clone_call): Likewise.
23340 (vectorizable_conversion): Likewise.
23341 (vectorizable_assignment): Likewise.
23342 (vectorizable_shift): Likewise.
23343 (vectorizable_operation): Likewise.
23344 (vectorizable_store): Likewise.
23345 (vectorizable_load): Likewise.
23346 * tree-vectorizer.h: Likewise.
23348 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23350 * passes.c (emergency_dump_function): New.
23351 * tree-pass.h (emergency_dump_function): Declare.
23352 * plugin.c (plugins_internal_error_function): Remove.
23353 * plugin.h (plugins_internal_error_function): Remove declaration.
23354 * toplev.c (internal_error_function): New static function. Use it...
23355 (general_init): ...here.
23357 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23359 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23360 (arc_rtx_costs): Add costs for new patterns.
23361 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23362 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23364 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23366 * tree-ssa-strlen.c (get_next_strinfo): New function.
23367 (get_stridx_plus_constant): Use it.
23368 (zero_length_string): Likewise.
23369 (adjust_related_strinfos): Likewise.
23370 (adjust_last_stmt): Likewise.
23372 2017-05-31 Richard Biener <rguenther@suse.de>
23375 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23376 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23378 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23380 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23381 loop_vinfo argument and use of dependence distance vectors.
23382 Check instead whether the two references differ only in their
23383 initial value and assume that they have the same alignment if the
23384 difference is a multiple of the vector alignment.
23385 (vect_analyze_data_refs_alignment): Update call accordingly.
23387 2017-05-31 Martin Liska <mliska@suse.cz>
23390 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23392 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23394 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23395 (create_intersect_range_checks): Move from ...
23396 * tree-data-ref.c (create_intersect_range_checks_index)
23397 (create_intersect_range_checks): ... to here.
23398 (create_runtime_alias_checks): New function factored from ...
23399 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23400 here. Call above function.
23401 * tree-data-ref.h (create_runtime_alias_checks): New function.
23403 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23405 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23406 segment length for dr_b and compute it in wide_int.
23408 2017-05-31 Richard Biener <rguenther@suse.de>
23410 PR tree-optimization/80906
23411 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23412 and pass through iv_map.
23413 (copy_bb_and_scalar_dependences): Adjust.
23414 (translate_pending_phi_nodes): Likewise.
23415 (copy_loop_close_phi_args): Handle code-generating IVs instead
23418 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23420 * diagnostic-color.c (color_dict): Add "type-diff".
23421 (parse_gcc_colors): Update comment.
23422 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23423 -fdiagnostics-show-template-tree and -fno-elide-type.
23424 (GCC_COLORS): Add type-diff to example.
23426 (-fdiagnostics-show-template-tree): New.
23427 (-fno-elide-type): New.
23428 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23429 the pp_format_decoder callback. Call any m_format_postprocessor's
23431 (pretty_printer::pretty_printer): Initialize
23432 m_format_postprocessor.
23433 (pretty_printer::~pretty_printer): Delete any
23434 m_format_postprocessor.
23435 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23436 (class format_postprocessor): New class.
23437 (struct pretty_printer::format_decoder): Document the new parameters.
23438 (struct pretty_printer::m_format_postprocessor): New field.
23439 * tree-diagnostic.c (default_tree_printer): Update for new
23440 bool and const char ** params.
23441 * tree-diagnostic.h (default_tree_printer): Likewise.
23443 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23445 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23446 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23447 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23448 rs6000_gen_cell_microcode code.
23449 (rs6000_final_prescan_insn): Delete.
23450 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23451 "warn-cell-microcode" entries.
23452 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23453 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23454 throughout. Change cc_reg_not_micro_cr0_operand to
23455 cc_reg_not_cr0_operand throughout.
23456 (*extendhi<mode>2_noload): Delete.
23457 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23458 (mwarn-cell-microcode): Delete.
23459 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23460 -mgen-cell-microcode and -mwarn-cell-microcode.
23462 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23465 * config/i386/constraints.md (Yd): New constraint.
23467 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23468 and (?Yd, r) alternatives. Update insn attributes.
23469 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23470 and (?*Yd, r) alternatives. Update insn attributes.
23471 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23473 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23475 * gimplify.c (gimplify_modify_expr): Don't create a
23476 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23479 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23481 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23483 2017-05-30 Richard Biener <rguenther@suse.de>
23485 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23486 and reduc_def fields.
23487 (STMT_VINFO_REDUC_TYPE): New define.
23488 (STMT_VINFO_REDUC_DEF): Likewise.
23489 (vect_force_simple_reduction): Adjust prototype.
23490 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23491 (vect_is_simple_reduction): Remove check_reduction argument.
23492 (vect_force_simple_reduction): Adjust and set
23493 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23494 (vectorizable_reduction): Do not re-do reduction analysis
23495 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23496 * tree-parloops.c (gather_scalar_reductions): Adjust.
23498 2017-05-30 Richard Biener <rguenther@suse.de>
23500 PR middle-end/80901
23501 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23504 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23506 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23507 Introduce unknown_misalignment parameter and remove vf.
23508 (vect_peeling_hash_get_lowest_cost):
23509 Pass unknown_misalignment parameter.
23510 (vect_enhance_data_refs_alignment):
23511 Fix unsupportable data ref treatment.
23513 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23515 * tree-vect-data-refs.c (vect_get_data_access_cost):
23516 Workaround for SLP handling.
23517 (vect_enhance_data_refs_alignment):
23518 Compute costs for doing no peeling at all, compare to the best
23519 peeling costs so far and avoid peeling if cheaper.
23521 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23523 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23524 Return peeling info and set costs to zero for unlimited cost
23526 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23527 with unknown misalignment. Compute and costs for unknown
23528 misalignment, compare them to the costs for known misalignment
23529 and choose the cheapest for peeling.
23531 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23533 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23534 (vect_get_peeling_costs_all_drs): Create function.
23535 (vect_peeling_hash_get_lowest_cost):
23536 Use vect_get_peeling_costs_all_drs.
23537 (vect_peeling_supportable): Create function.
23538 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23540 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23542 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23543 DR_HAS_NEGATIVE_STEP.
23544 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23545 (vect_enhance_data_refs_alignment): Use.
23546 (vect_duplicate_ssa_name_ptr_info): Use.
23547 * tree-vectorizer.h (dr_misalignment): Use.
23548 (known_alignment_for_access_p): Use.
23550 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23553 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23555 (has_section_name): New function.
23557 2017-05-30 Martin Liska <mliska@suse.cz>
23560 * auto-profile.c (get_function_decl_from_block): Fix
23563 2017-05-30 Richard Biener <rguenther@suse.de>
23565 PR middle-end/80876
23566 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23568 2017-05-30 Martin Liska <mliska@suse.cz>
23570 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23571 * dumpfile.h (struct dump_file_info): Remove ctors.
23573 2017-05-30 Martin Liska <mliska@suse.cz>
23575 * predict.def: Fix GNU coding style.
23577 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23579 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23580 Mark 'to' argument with ATTRIBUTE_UNUSED.
23582 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23584 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23585 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23586 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23589 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23591 * doc/install.texi (Options specification): Restore entry of
23592 --enable-sjlj-exceptions.
23594 2017-05-27 Michael Eager <eager@eagercon.com>
23597 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23599 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23601 * config/microblaze/microblaze.h
23602 (FIXED_REGISTERS): Update in macro.
23603 (CALL_USED_REGISTERS): Update in macro.
23605 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23607 * doc/install.texi: Add links to macOS binary distributions.
23609 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23613 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23615 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23617 2017-05-26 Martin Liska <mliska@suse.cz>
23619 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23621 2017-05-26 Martin Liska <mliska@suse.cz>
23623 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23624 always leading ';; '.
23625 (dump_bb_info): Likewise.
23626 (brief_dump_cfg): Likewise.
23627 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23628 * dumpfile.c: Remove usage of TDF_VERBOSE.
23629 * dumpfile.h (enum dump_kind): Likewise.
23630 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23631 * print-tree.c (debug_verbose): Remove.
23632 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23633 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23634 * tree-diagnostic.c (default_tree_printer): Replace
23635 TDF_DIAGNOSTIC with TDF_SLIM.
23637 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23639 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23640 in parameter loop, rather than loop_vinfo.
23641 (create_intersect_range_checks): Ditto.
23642 (vect_create_cond_for_alias_checks): Update call to above functions.
23644 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23646 PR tree-optimization/80815
23647 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23648 for merging runtime alias checks. Handle negative DR_STEPs.
23650 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23652 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23654 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23655 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23656 out code pruning runtime alias checks.
23657 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23658 factored out from above.
23659 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23661 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23663 (prune_runtime_alias_test_list): New decalaration.
23665 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23667 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23668 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23669 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23670 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23671 (operator==, comp_dr_with_seg_len_pair): Ditto.
23672 (vect_prune_runtime_alias_test_list): Ditto.
23674 2017-05-26 Martin Liska <mliska@suse.cz>
23677 * params.def: Bound partial-inlining-entry-probability param.
23679 2017-05-26 Marek Polacek <polacek@redhat.com>
23682 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23685 2017-05-26 Richard Biener <rguenther@suse.de>
23687 PR tree-optimization/80842
23688 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23691 2017-05-26 Richard Biener <rguenther@suse.de>
23693 PR tree-optimization/80844
23694 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23696 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23698 * doc/md.texi (Machine Constraints): Update x86 family
23699 machine constraints section to match 'config/i386/constraints.md'.
23701 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23703 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23705 2017-05-25 Nathan Sidwell <nathan@acm.org>
23707 * doc/invoke.texi (--enable-languages): Update documentation.
23709 2017-05-25 Martin Liska <mliska@suse.cz>
23711 * dumpfile.c: Add TDF_FOLDING.
23712 * dumpfile.h (enum dump_kind): Likewise.
23713 * genmatch.c (dt_simplify::gen_1): Use it.
23715 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23717 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23719 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23721 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23722 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23724 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23726 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23727 * match.pd (X == C): Rewrite it here.
23728 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23729 with_certain_nonzero_bits2): New predicates.
23730 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23732 2017-05-24 Nathan Sidwell <nathan@acm.org>
23734 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23737 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23740 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23742 * config/powerpcspe: New port. Files are copied from the rs6000
23743 port, with "rs6000" in filenames replaced by "powerpcspe".
23745 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23747 PR rtl-optimization/80754
23748 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23750 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23752 * config/sparc/sparc.md (length): Return the correct value for -mflat
23753 sibcalls to match output_sibcall.
23755 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23759 * config/rs6000/rs6000.c (struct machine_function): Add new field
23761 (rs6000_get_separate_components): Init that field, use it.
23762 (rs6000_components_for_bb): Use the field.
23764 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23766 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23768 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23770 PR middle-end/80823
23771 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23773 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23776 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23777 address against address_operand predicate.
23778 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23780 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23782 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23783 parameters passed indirectly.
23785 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23787 * config/i386/i386.md (*movdi_internal): Remove SSE4
23788 alternative 18 (?r, *v). Update insn attributes.
23789 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23790 Update insn attributes.
23791 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23792 Update insn attributes.
23793 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23794 alternative 1 (r, v). Remove isa attribute.
23795 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23796 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23797 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23799 2017-05-23 Tom de Vries <tom@codesourcery.com>
23801 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23804 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23806 * cgraphunit.c (symbol_table::process_new_functions): Update.
23807 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23808 (inline_generate_summary): Rename to ...
23809 (ipa_fn_summary_generate): ... this one.
23810 (inline_read_summary): Rename to ...
23811 (ipa_fn_summary_read): ... this one.
23812 (inline_write_summary): Rename to ...
23813 (ipa_fn_summary_write): ... this one.
23814 (inline_free_summary): Rename to ...
23815 (ipa_free_fn_summary): ... this one.
23816 (pass_data_local_fn_summary, pass_local_fn_summary,
23817 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23818 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23819 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23820 make_pass_ipa_fn_summary): New.
23821 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23822 inline_write_summary, inline_free_summary): Remove.
23823 (ipa_free_fn_summary) : New.
23824 * ipa-inline.c (ipa_inline): Update.
23825 (pass_ipa_inline): Do not generate summaries.
23826 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23828 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23829 and add pass_ipa_fn_summary.
23830 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23832 (make_pass_inline_parameters): Remove.
23834 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23836 * omp-low.c (struct omp_context): Remove "default_kind" member.
23839 * omp-offload.c (execute_oacc_device_lower): Remove the
23840 parallelism dimensions function attributes for unparallelized
23841 OpenACC kernels constructs.
23843 2017-05-23 Martin Liska <mliska@suse.cz>
23845 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23847 (cgraph_edge::make_speculative): Likewise.
23848 (cgraph_edge::resolve_speculation): Likewise.
23849 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23850 (cgraph_node::dump): Likewise.
23851 * cgraph.h: Likewise.
23852 * cgraphunit.c (analyze_functions): Likewise.
23853 (symbol_table::compile): Likewise.
23854 * ipa-cp.c (print_all_lattices): Likewise.
23855 (determine_versionability): Likewise.
23856 (initialize_node_lattices): Likewise.
23857 (ipcp_verify_propagated_values): Likewise.
23858 (estimate_local_effects): Likewise.
23859 (update_profiling_info): Likewise.
23860 (create_specialized_node): Likewise.
23861 (perhaps_add_new_callers): Likewise.
23862 (decide_about_value): Likewise.
23863 (decide_whether_version_node): Likewise.
23864 (identify_dead_nodes): Likewise.
23865 (ipcp_store_bits_results): Likewise.
23866 * ipa-devirt.c (dump_targets): Likewise.
23867 (ipa_devirt): Likewise.
23868 * ipa-icf.c (sem_item::dump): Likewise.
23869 (sem_function::equals): Likewise.
23870 (sem_variable::equals): Likewise.
23871 (sem_item_optimizer::read_section): Likewise.
23872 (sem_item_optimizer::execute): Likewise.
23873 (congruence_class::dump): Likewise.
23874 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23875 (dump_inline_summary): Likewise.
23876 (estimate_node_size_and_time): Likewise.
23877 (inline_analyze_function): Likewise.
23878 * ipa-inline-transform.c (inline_call): Likewise.
23879 * ipa-inline.c (report_inline_failed_reason): Likewise.
23880 (want_early_inline_function_p): Likewise.
23881 (edge_badness): Likewise.
23882 (update_edge_key): Likewise.
23883 (inline_small_functions): Likewise.
23884 * ipa-profile.c (ipa_profile): Likewise.
23885 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23886 (ipa_make_edge_direct_to_target): Likewise.
23887 (remove_described_reference): Likewise.
23888 (ipa_impossible_devirt_target): Likewise.
23889 (propagate_controlled_uses): Likewise.
23890 (ipa_print_node_params): Likewise.
23891 (ipcp_transform_function): Likewise.
23892 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23893 (propagate_pure_const): Likewise.
23894 * ipa-reference.c (generate_summary): Likewise.
23895 (read_write_all_from_decl): Likewise.
23896 (propagate): Likewise.
23897 (ipa_reference_read_optimization_summary): Likewise.
23898 * ipa-utils.c (ipa_merge_profiles): Likewise.
23899 * ipa.c (walk_polymorphic_call_targets): Likewise.
23900 (symbol_table::remove_unreachable_nodes): Likewise.
23901 (ipa_single_use): Likewise.
23902 * passes.c (execute_todo): Likewise.
23903 * predict.c (drop_profile): Likewise.
23904 * symtab.c (symtab_node::get_dump_name): New function.
23905 (symtab_node::dump_name): Likewise.
23906 (symtab_node::dump_asm_name): Likewise.
23907 (symtab_node::dump_references): Likewise.
23908 (symtab_node::dump_referring): Likewise.
23909 (symtab_node::dump_base): Likewise.
23910 (symtab_node::debug_symtab): Likewise.
23911 * tree-sra.c (convert_callers_for_node): Likewise.
23912 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23913 * value-prof.c (init_node_map): Likewise.
23915 2017-05-23 Martin Liska <mliska@suse.cz>
23917 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23918 and symtab_node::debug_symtab to symbol_table::debug.
23919 * cgraphunit.c (analyze_functions): Use the renamed function.
23920 (symbol_table::compile): Likewise.
23921 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23922 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23923 * passes.c (execute_todo): Likewise.
23924 * symtab.c (symbol_table::dump): New function.
23925 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23927 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23929 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23930 that nonconst implies exec.
23932 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23934 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23935 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23936 (inline_edge_summary_vec): Turn into ...
23937 (ipa_call_summaries): ... this one.
23938 (redirect_to_unreachable, edge_set_predicate,
23939 evaluate_properties_for_edge, inline_summary_alloc,
23940 reset_ipa_call_summary, reset_inline_summary,
23941 inline_summary_t::duplicate): Update.
23942 (inline_edge_duplication_hook): Turn to ...
23943 (ipa_call_summary_t::duplicate): ... this one.
23944 (inline_edge_removal_hook): Turn to ...
23945 (ipa_call_summary_t::remove): ... this one.
23946 (dump_inline_edge_summary): Turn to ...
23947 (dump_ipa_call_summary): ... this one.
23948 (estimate_function_body_sizes): Update.
23949 (inline_update_callee_summaries): Update.
23950 (remap_edge_change_prob): Update.
23951 (remap_edge_summaries): Update.
23952 (inline_merge_summary): Update.
23953 (do_estimate_edge_time): Update.
23954 (inline_generate_summary): Update.
23955 (inline_read_section): Update.
23956 (inline_read_summary): Update.
23957 (inline_free_summary): Update.
23958 * ipa-inline.c (can_inline_edge_p): Update.
23959 (compute_inlined_call_time): Update.
23960 (want_inline_small_function_p): Update.
23961 (edge_badness): Update.
23962 (early_inliner): Update.
23963 * ipa-inline.h (inline_edge_summary): Turn to ...
23964 (ipa_call_summary): ... this one.
23965 (ipa_call_summary_t): New class.
23966 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23967 (ipa_call_summaries): New.
23968 (inline_edge_summary): Remove.
23969 (estimate_edge_growth): Update.
23970 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23971 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23972 * ipa-split.c (execute_split_functions): Update.
23973 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23975 2017-05-23 Tom de Vries <tom@codesourcery.com>
23977 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23978 attributes): Document rdrand effective target.
23980 2017-05-23 Tom de Vries <tom@codesourcery.com>
23982 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23983 attributes): Sort alphabetically.
23985 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23987 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23989 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23992 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23993 V2DF/V2DI splat into two separate patterns, one that handles
23994 registers, and the other that only handles memory. Drop support
23995 for splatting from a GPR on ISA 2.07 and then splitting the
23996 splat into direct move and splat.
23997 (vsx_splat_<mode>_reg): Likewise.
23998 (vsx_splat_<mode>_mem): Likewise.
24000 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
24002 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
24004 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24006 PR middle-end/80809
24007 * omp-low.c (finish_taskreg_remap): New function.
24008 (finish_taskreg_scan): If unit size of ctx->record_type
24009 is non-constant, unshare the size expression and replace
24010 decls in it with possible outer var refs.
24012 PR middle-end/80809
24013 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24014 GOVD_SHARED rather than GOVD_PRIVATE with it.
24015 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24016 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24018 PR middle-end/80853
24019 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24020 as last argument to build_outer_var_ref for pointer bases of array
24021 section reductions.
24023 2017-05-19 Martin Sebor <msebor@redhat.com>
24025 * print-tree.c (print_node): Print DECL_READ_P flag.
24027 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24029 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24030 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24031 * cgraph.c: Likewise.
24032 * cgraphunit.c: Likewise.
24033 * gengtype.c: Likewise.
24034 * ipa-cp.c: Likewise.
24035 * ipa-devirt.c: Likewise.
24036 * ipa-icf.c: Likewise.
24037 * ipa-predicate.c: Likewise.
24038 * ipa-profile.c: Likewise.
24039 * ipa-prop.c: Likewise.
24040 * ipa-split.c: Likewise.
24042 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24043 edge_predicate_pool, dump_inline_hints,
24044 inline_summary::account_size_time, redirect_to_unreachable,
24045 edge_set_predicate, set_hint_predicate,
24046 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24047 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24048 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24049 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24050 ipa_call_summary_t::remove, initialize_growth_caches,
24051 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24052 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24053 mark_modified, unmodified_parm_1, unmodified_parm,
24054 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24055 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24056 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24057 will_be_nonconstant_predicate, record_modified_bb_info,
24058 get_minimal_bb, record_modified, param_change_prob,
24059 phi_result_unknown_predicate, predicate_for_phi_result,
24060 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24061 estimate_function_body_sizes, compute_inline_parameters,
24062 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24063 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24064 inline_update_callee_summaries, remap_edge_change_prob,
24065 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24066 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24067 inline_analyze_function, inline_summary_t::insert,
24068 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24069 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24070 inline_free_summary): Move to ipa-fnsummary.h
24071 (predicate_t): Remove.
24072 * ipa-fnsummary.c: New file.
24073 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24074 (enum inline_hints_vals, inline_hints, agg_position_info,
24075 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24076 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24077 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24078 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24079 inline_read_summary, inline_write_summary, inline_free_summary,
24080 inline_analyze_function, initialize_inline_failed,
24081 inline_merge_summary, inline_update_overall_summary,
24082 compute_inline_parameters): Move to ipa-fnsummary.h
24083 * ipa-fnsummary.h: New file.
24084 * ipa-inline-transform.h: Include ipa-inline.h.
24085 * ipa-inline.c: LIkewise.
24087 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24089 * ipa-inline.c (edge_badness): Use inlined_time instead of
24090 inline_summaries->get.
24092 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24094 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24096 2017-05-22 Nathan Sidwell <nathan@acm.org>
24098 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24099 (fdump-lang): Document 'raw' option.
24100 * dumpfile.h (TDI_tu): Delete.
24101 * dumpfile.c (dump_files): Remove translation-unit.
24102 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24104 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24106 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24107 command option from $(AWK) call.
24108 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24109 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24110 [FORMAT]: Remove handling of variable.
24111 * config/avr/t-multilib: Regenerate.
24113 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24115 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24117 (dump_inline_summary): Do not print self_time.
24118 (estimate_function_body_sizes): Do not set self_time.
24119 (compute_inline_parameters): Likewise.
24120 (inline_read_section, inline_write_summary): Do not stream self_time.
24121 * ipa-inline.h (inline_summary): Drop self_time.
24123 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24125 * ipa-inline-analysis.c (account_size_time): Rename to ...
24126 (inline_summary::account_size_time): ... this one.
24127 (reset_ipa_call_summary): Turn to ...
24128 (ipa_call_summary::reset): ... this one.
24129 (reset_inline_summary): Turn to ...
24130 (inline_summary::reset): ... this one.
24131 (inline_summary_t::remove): Update.
24132 (inline_summary_t::duplicate): Update.
24133 (ipa_call_summary_t::remove): Update.
24134 (dump_inline_summary): Update.
24135 (estimate_function_body_sizes): Update.
24136 (compute_inline_parameters): Update.
24137 (estimate_node_size_and_time): Update.
24138 (inline_merge_summary): Update.
24139 (inline_update_overall_summary): Update.
24140 (inline_read_section): Update.
24141 (inline_write_summary): Update.
24142 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24143 add account_size_time and reset member functions.
24144 (ipa_call_summary): Add reset function.
24145 * ipa-predicate.h (predicate::operator &): Constify.
24147 2017-05-22 Richard Biener <rguenther@suse.de>
24149 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24151 2017-05-19 Jason Merrill <jason@redhat.com>
24153 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24155 2017-05-19 Marek Polacek <polacek@redhat.com>
24158 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24159 TYPE_OVERFLOW_WRAPS checks.
24161 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24163 * tree-core.h (enum omp_clause_default_kind): Add
24164 "OMP_CLAUSE_DEFAULT_PRESENT".
24165 * tree-pretty-print.c (dump_omp_clause): Handle it.
24166 * gimplify.c (enum gimplify_omp_var_data): Add
24167 "GOVD_MAP_FORCE_PRESENT".
24168 (gimplify_adjust_omp_clauses_1): Map it to
24169 "GOMP_MAP_FORCE_PRESENT".
24170 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24172 * gimplify.c (oacc_default_clause): Clarify.
24174 2017-05-19 Nathan Sidwell <nathan@acm.org>
24176 LANG_HOOK_REGISTER_DUMPS
24177 * toplev.c (general_init): Call register dump lang hook.
24178 * doc/invoke.texi: Document -fdump-lang option family.
24179 * dumpfile.c (dump_files): Remove class dump here.
24180 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24181 * dumpfile.h (tree_dump_index): Remove TDI_class.
24182 * langhooks-def.h (lhd_register_dumps): Declare.
24183 (LANG_HOOKS_REGISTER_DUMPS): Define.
24184 (LANG_HOOKS_INITIALIZER): Add it.
24185 * langhooks.c (lhd_register_dumps): Define.
24186 * langhooks.h (struct lang_hooks): Add register_dumps.
24188 2017-05-19 Nathan Sidwell <nathan@acm.org>
24190 * context.h (context::set_passes): New.
24191 * context.c (context::context): Do not create pass manager.
24192 * toplev.c (general_init): Create pass manager here.
24194 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24196 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24197 use this splitter if two add or or instructions would also work for
24198 the constant we want to generate.
24200 2017-05-19 Richard Biener <rguenther@suse.de>
24203 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24204 predicate evaluation.
24206 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24208 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24210 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24213 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24215 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24216 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24217 (inline_edge_summary_vec): Turn into ...
24218 (ipa_call_summaries): ... this one.
24219 (redirect_to_unreachable, edge_set_predicate,
24220 evaluate_properties_for_edge, inline_summary_alloc,
24221 reset_ipa_call_summary, reset_inline_summary,
24222 inline_summary_t::duplicate): Update.
24223 (inline_edge_duplication_hook): Turn to ...
24224 (ipa_call_summary_t::duplicate): ... this one.
24225 (inline_edge_removal_hook): Turn to ...
24226 (ipa_call_summary_t::remove): ... this one.
24227 (dump_inline_edge_summary): Turn to ...
24228 (dump_ipa_call_summary): ... this one.
24229 (estimate_function_body_sizes): Update.
24230 (inline_update_callee_summaries): Update.
24231 (remap_edge_change_prob): Update.
24232 (remap_edge_summaries): Update.
24233 (inline_merge_summary): Update.
24234 (do_estimate_edge_time): Update.
24235 (inline_generate_summary): Update.
24236 (inline_read_section): Update.
24237 (inline_read_summary): Update.
24238 (inline_free_summary): Update.
24239 * ipa-inline.c (can_inline_edge_p): Update.
24240 (compute_inlined_call_time): Update.
24241 (want_inline_small_function_p): Update.
24242 (edge_badness): Update.
24243 (early_inliner): Update.
24244 * ipa-inline.h (inline_edge_summary): Turn to ...
24245 (ipa_call_summary): ... this one.
24246 (ipa_call_summary_t): New class.
24247 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24248 (ipa_call_summaries): New.
24249 (inline_edge_summary): Remove.
24250 (estimate_edge_growth): Update.
24251 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24252 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24253 * ipa-split.c (execute_split_functions): Update.
24254 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24256 2017-05-19 Richard Biener <rguenther@suse.de>
24258 PR middle-end/80764
24259 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24261 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24263 * config/rs6000/rs6000.c (struct machine_function): Add field
24264 fpr_is_wrapped_separately.
24265 (rs6000_get_separate_components): Use 64 components. Handle the
24266 new FPR components.
24267 (rs6000_components_for_bb): Handle the FPR components.
24268 (rs6000_emit_prologue_components): Handle the FPR components.
24269 (rs6000_emit_epilogue_components): Handle the FPR components.
24270 (rs6000_set_handled_components): Handle the FPR components.
24271 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24272 that are already separately shrink-wrapped.
24273 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24274 that are already separately shrink-wrapped.
24276 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24279 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24282 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24283 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24284 cases where the register allocator uses a move and an offsettable
24285 memory operation to/from a FPR register on ISA 2.06/2.07.
24286 (define_peephole2 for Altivec d-form store): Likewise.
24288 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24291 * config/i386/mmx.md (*mov<mode>_internal): Enable
24292 alternatives 11, 12, 13 and 14 also for 32bit targets.
24293 Remove alternatives 15, 16, 17 and 18.
24294 * config/i386/sse.md (vec_concatv2di): Change
24295 alternative (!x, *y) to (x, ?!*Yn).
24297 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24299 * dumpfile.h (enum dump_kind): Remove stray comma.
24301 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24303 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24304 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24305 predicate::num_conditions
24306 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24307 (CHANGED): turn into predicate::changed.
24308 (agg_position_info): Move to ipa-predicate.h
24309 (add_condition, predicate::add_clause, predicate::operator &=,
24310 predicate::or_with, predicate::evaluate, predicate::probability,
24311 dump_condition, dump_clause, predicate::dump,
24312 predicate::remap_after_duplication, predicate::remap_after_inlining,
24313 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24314 (evaluate_conditions_for_known_args): Update.
24315 (set_cond_stmt_execution_predicate): Update.
24316 * ipa-inline.h: Include ipa-predicate.h
24317 (condition, inline_param_summary, conditions, agg_position_info,
24318 predicate): Move to ipa-predicate.h
24319 * ipa-predicate.c: New file.
24320 * ipa-predicate.h: New file.
24322 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24324 * final.c (leaf_function_p): Check we are not in a sequence.
24326 2017-05-18 Martin Liska <mliska@suse.cz>
24328 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24329 * dumpfile.c (dump_register): Use new enum dump_kind.
24330 (get_dump_file_name): Likewise.
24331 (dump_enable_all): Likewise.
24332 (dump_switch_p_1): Likewise.
24333 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24334 * dumpfile.h (enum dump_kind): New enum type.
24335 (struct dump_file_info): Create constructor and
24336 format fields and comments.
24337 * passes.c (pass_manager::register_one_dump_file):
24339 * statistics.c (statistics_early_init): Likewise.
24340 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24341 TDF_TREE with TDF_SLIM.
24342 (gather_memory_references_ref): Likewise.
24344 2017-05-18 Martin Liska <mliska@suse.cz>
24346 * vec.h (struct vnull): Use it.
24348 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24350 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24351 (true_predicate, false_predicate, true_predicate_p,
24352 false_predicate_p): Remove.
24353 (single_cond_predicate, not_inlined_predicate): Turn to member function
24355 (add_condition): Update.
24356 (add_clause): Turn to...
24357 (predicate::add_clause): ... this one; update; allow passing NULL
24359 (and_predicates): Turn to ...
24360 (predicate::operator &=): ... this one.
24361 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24362 (or_predicates): Turn to ...
24363 (predicate::or_with): ... this one.
24364 (evaluate_predicate): Turn to ...
24365 (predicate::evaluate): ... this one.
24366 (predicate_probability): Turn to ...
24367 (predicate::probability): ... this one.
24368 (dump_condition): Update.
24369 (dump_predicate): Turn to ...
24370 (predicate::dump): ... this one.
24371 (account_size_time): Update.
24372 (edge_set_predicate): Update.
24373 (set_hint_predicate): UPdate.
24374 (evaluate_conditions_for_known_args): Update.
24375 (evaluate_properties_for_edge): Update.
24376 (remap_predicate_after_duplication): Turn to...
24377 (predicate::remap_after_duplication): ... this one.
24378 (remap_hint_predicate_after_duplication): Update.
24379 (inline_summary_t::duplicate): UPdate.
24380 (dump_inline_edge_summary): Update.
24381 (dump_inline_summary): Update.
24382 (set_cond_stmt_execution_predicate): Update.
24383 (set_switch_stmt_execution_predicate): Update.
24384 (compute_bb_predicates): Update.
24385 (will_be_nonconstant_expr_predicate): Update.
24386 (will_be_nonconstant_predicate): Update.
24387 (phi_result_unknown_predicate): Update.
24388 (predicate_for_phi_result): Update.
24389 (array_index_predicate): Update.
24390 (estimate_function_body_sizes): Update.
24391 (estimate_node_size_and_time): Update.
24392 (estimate_ipcp_clone_size_and_time): Update.
24393 (remap_predicate): Rename to ...
24394 (predicate::remap_after_inlining): ... this one.
24395 (remap_hint_predicate): Update.
24396 (inline_merge_summary): Update.
24397 (inline_update_overall_summary): Update.
24398 (estimate_size_after_inlining): Update.
24399 (read_predicate): Rename to ...
24400 (predicate::stream_in): ... this one.
24401 (read_inline_edge_summary): Update.
24402 (write_predicate): Rename to ...
24403 (predicate::stream_out): ... this one.
24404 (write_inline_edge_summary): Update.
24405 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24406 (clause_t): Turn to uint32_t
24407 (predicate): Turn to class; implement constructor and operators
24409 (size_time_entry): Update.
24410 (inline_summary): Update.
24411 (inline_edge_summary): Update.
24413 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24415 * fold-const.c (fold_binary_loc): Move transformation...
24416 * match.pd (C - X CMP X): ... here.
24418 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24420 * config/sparc/sparc.c (sparc_option_override): Set function
24421 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24422 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24424 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24426 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24428 2017-05-18 Marek Polacek <polacek@redhat.com>
24431 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24432 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24433 gimple_assign_load_p.
24435 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24437 PR middle-end/80692
24438 * real.c (do_compare): Give decimal_do_compare preference over
24439 comparing just the signs.
24441 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24443 * doc/md.texi (Canonicalization of Instructions): Describe the
24444 canonical form of instructions that inherently set a condition
24447 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24449 PR middle-end/80775
24450 * tree-cfg.c: Move deletion of unreachable case statements to after
24451 the merging of consecutive case labels.
24453 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24455 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24456 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24457 restoring of callee-saved registers.
24459 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24461 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24462 * config/visium/visium.c (single_set_and_flags): Likewise.
24463 * config/visium/visium.md (Substitutions): Likewise.
24465 2017-05-17 Martin Liska <mliska@suse.cz>
24467 * cfg.c: Introduce dump_flags_t type and
24468 use it instead of int type.
24470 * cfghooks.c: Likewise.
24471 * cfghooks.h (struct cfg_hooks): Likewise.
24472 * cfgrtl.c: Likewise.
24473 * cfgrtl.h: Likewise.
24474 * cgraph.c (cgraph_node::get_body): Likewise.
24475 * coretypes.h: Likewise.
24476 * domwalk.c: Likewise.
24477 * domwalk.h: Likewise.
24478 * dumpfile.c (struct dump_option_value_info): Likewise.
24479 (dump_enable_all): Likewise.
24480 (dump_switch_p_1): Likewise.
24481 (opt_info_switch_p): Likewise.
24482 * dumpfile.h (enum tree_dump_index): Likewise.
24483 (struct dump_file_info): Likewise.
24484 * genemit.c: Likewise.
24485 * generic-match-head.c: Likewise.
24486 * gengtype.c (open_base_files): Likewise.
24487 * gimple-pretty-print.c: Likewise.
24488 * gimple-pretty-print.h: Likewise.
24489 * graph.c (print_graph_cfg): Likewise.
24490 * graphite-scop-detection.c (dot_all_sese): Likewise.
24491 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24492 * loop-unroll.c (report_unroll): Likewise.
24493 * passes.c (pass_manager::register_one_dump_file): Likewise.
24494 * print-tree.c: Likewise.
24495 * statistics.c: Likewise.
24496 * tree-cfg.c: Likewise.
24497 * tree-cfg.h: Likewise.
24498 * tree-dfa.c: Likewise.
24499 * tree-dfa.h: Likewise.
24500 * tree-dump.c (dump_function): Likewise.
24501 * tree-dump.h (struct dump_info): Likewise.
24502 * tree-pretty-print.c: Likewise.
24503 * tree-pretty-print.h: Likewise.
24504 * tree-ssa-live.c: Likewise.
24505 * tree-ssa-live.h: Likewise.
24506 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24507 * tree-vect-loop.c: Likewise.
24508 * tree-vect-slp.c: Likewise.
24510 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24511 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24513 PR tree-optimization/80457
24514 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24515 of all arguments to a statement as scalar_to_vec operations.
24516 (vectorizable_call): Adjust call to vect_model_simple_cost for
24518 (vectorizable_conversion): Likewise.
24519 (vectorizable_assignment): Likewise.
24520 (vectorizable_shift): Likewise.
24521 (vectorizable_operation): Likewise.
24522 (vectorizable_comparison): Likewise.
24523 (vect_is_simple_cond): Record the def types for operands.
24524 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24525 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24526 for statement argument count.
24528 2017-05-16 Carl Love <cel@us.ibm.com>
24530 * config/rs6000/rs6000-c: Add support for built-in functions
24531 vector unsigned long long vec_bperm (vector unsigned long long,
24532 vector unsigned char)
24533 vector signed long long vec_mule (vector signed int,
24535 vector unsigned long long vec_mule (vector unsigned int,
24536 vector unsigned int)
24537 vector signed long long vec_mulo (vector signed int,
24539 vector unsigned long long vec_mulo (vector unsigned int,
24540 vector unsigned int)
24541 vector signed char vec_sldw (vector signed char,
24542 vector signed char,
24544 vector unsigned char vec_sldw (vector unsigned char,
24545 vector unsigned char,
24547 vector signed short vec_sldw (vector signed short,
24548 vector signed short,
24550 vector unsigned short vec_sldw (vector unsigned short,
24551 vector unsigned short,
24553 vector signed int vec_sldw (vector signed int,
24556 vector unsigned int vec_sldw (vector unsigned int,
24557 vector unsigned int,
24559 vector signed long long vec_sldw (vector signed long long,
24560 vector signed long long,
24562 vector unsigned long long vec_sldw (vector unsigned long long,
24563 vector unsigned long long,
24565 * config/rs6000/rs6000-c: Add support for built-in functions
24566 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24567 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24568 * doc/extend.texi: Update the built-in documentation for the
24569 new built-in functions.
24571 2017-05-16 Marek Polacek <polacek@redhat.com>
24575 * tree.c (save_expr): Don't fold the expression.
24577 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24579 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24580 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24581 and (?*y,m). Update insn attributes.
24583 2017-05-16 Martin Liska <mliska@suse.cz>
24585 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24586 flags argument of print_gimple_stmt, print_gimple_expr,
24587 print_generic_stmt and print_generic_expr.
24588 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24589 * coretypes.h: Likewise.
24590 * except.c (dump_eh_tree): Likewise.
24591 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24592 * gimple-pretty-print.h: Likewise.
24593 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24594 (backprop::push_to_worklist): Likewise.
24595 (backprop::pop_from_worklist): Likewise.
24596 (backprop::process_use): Likewise.
24597 (backprop::intersect_uses): Likewise.
24598 (note_replacement): Likewise.
24599 * gimple-ssa-store-merging.c
24600 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24601 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24602 (pass_store_merging::execute): Likewise.
24603 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24604 (ssa_base_cand_dump_callback): Likewise.
24605 (dump_incr_vec): Likewise.
24606 (replace_refs): Likewise.
24607 (replace_mult_candidate): Likewise.
24608 (create_add_on_incoming_edge): Likewise.
24609 (create_phi_basis): Likewise.
24610 (insert_initializers): Likewise.
24611 (all_phi_incrs_profitable): Likewise.
24612 (introduce_cast_before_cand): Likewise.
24613 (replace_one_candidate): Likewise.
24614 * gimplify.c (gimplify_expr): Likewise.
24615 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24616 (set_rename): Likewise.
24617 (rename_uses): Likewise.
24618 (copy_loop_phi_nodes): Likewise.
24619 (add_close_phis_to_merge_points): Likewise.
24620 (copy_loop_close_phi_args): Likewise.
24621 (copy_cond_phi_args): Likewise.
24622 (graphite_copy_stmts_from_block): Likewise.
24623 (translate_pending_phi_nodes): Likewise.
24624 * graphite-poly.c (print_pdr): Likewise.
24625 (dump_gbb_cases): Likewise.
24626 (dump_gbb_conditions): Likewise.
24627 (print_scop_params): Likewise.
24628 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24629 (build_cross_bb_scalars_use): Likewise.
24630 (gather_bbs::before_dom_children): Likewise.
24631 * hsa-dump.c (dump_hsa_immed): Likewise.
24632 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24633 (get_replacement_map): Likewise.
24634 * ipa-inline-analysis.c (dump_condition): Likewise.
24635 (estimate_function_body_sizes): Likewise.
24636 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24637 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24638 * ipa-prop.c (ipa_dump_param): Likewise.
24639 (ipa_print_node_jump_functions_for_edge): Likewise.
24640 (ipa_modify_call_arguments): Likewise.
24641 (ipa_modify_expr): Likewise.
24642 (ipa_dump_param_adjustments): Likewise.
24643 (ipa_dump_agg_replacement_values): Likewise.
24644 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24645 * ipa-pure-const.c (check_stmt): Likewise.
24646 (pass_nothrow::execute): Likewise.
24647 * ipa-split.c (execute_split_functions): Likewise.
24648 * omp-offload.c (dump_oacc_loop_part): Likewise.
24649 (dump_oacc_loop): Likewise.
24650 * trans-mem.c (tm_log_emit): Likewise.
24651 (tm_memopt_accumulate_memops): Likewise.
24652 (dump_tm_memopt_set): Likewise.
24653 (dump_tm_memopt_transform): Likewise.
24654 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24655 (print_loop): Likewise.
24656 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24657 (chkp_gather_checks_info): Likewise.
24658 (chkp_get_check_result): Likewise.
24659 (chkp_remove_check_if_pass): Likewise.
24660 (chkp_use_outer_bounds_if_possible): Likewise.
24661 (chkp_reduce_bounds_lifetime): Likewise.
24662 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24663 (chkp_mark_completed_bounds): Likewise.
24664 (chkp_register_incomplete_bounds): Likewise.
24665 (chkp_mark_invalid_bounds): Likewise.
24666 (chkp_maybe_copy_and_register_bounds): Likewise.
24667 (chkp_build_returned_bound): Likewise.
24668 (chkp_get_bound_for_parm): Likewise.
24669 (chkp_build_bndldx): Likewise.
24670 (chkp_get_bounds_by_definition): Likewise.
24671 (chkp_generate_extern_var_bounds): Likewise.
24672 (chkp_get_bounds_for_decl_addr): Likewise.
24673 * tree-chrec.c (chrec_apply): Likewise.
24674 * tree-data-ref.c (dump_data_reference): Likewise.
24675 (dump_subscript): Likewise.
24676 (dump_data_dependence_relation): Likewise.
24677 (analyze_overlapping_iterations): Likewise.
24678 * tree-inline.c (expand_call_inline): Likewise.
24679 (tree_function_versioning): Likewise.
24680 * tree-into-ssa.c (dump_defs_stack): Likewise.
24681 (dump_currdefs): Likewise.
24682 (dump_names_replaced_by): Likewise.
24683 (dump_update_ssa): Likewise.
24684 (update_ssa): Likewise.
24685 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24686 * tree-parloops.c (build_new_reduction): Likewise.
24687 (try_create_reduction_list): Likewise.
24688 (ref_conflicts_with_region): Likewise.
24689 (oacc_entry_exit_ok_1): Likewise.
24690 (oacc_entry_exit_single_gang): Likewise.
24691 * tree-pretty-print.h: Likewise.
24692 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24693 (get_scalar_evolution): Likewise.
24694 (add_to_evolution): Likewise.
24695 (get_loop_exit_condition): Likewise.
24696 (analyze_evolution_in_loop): Likewise.
24697 (analyze_initial_condition): Likewise.
24698 (analyze_scalar_evolution): Likewise.
24699 (instantiate_scev): Likewise.
24700 (number_of_latch_executions): Likewise.
24701 (gather_chrec_stats): Likewise.
24702 (final_value_replacement_loop): Likewise.
24703 (scev_const_prop): Likewise.
24704 * tree-sra.c (dump_access): Likewise.
24705 (disqualify_candidate): Likewise.
24706 (create_access): Likewise.
24707 (reject): Likewise.
24708 (maybe_add_sra_candidate): Likewise.
24709 (create_access_replacement): Likewise.
24710 (analyze_access_subtree): Likewise.
24711 (analyze_all_variable_accesses): Likewise.
24712 (sra_modify_assign): Likewise.
24713 (initialize_constant_pool_replacements): Likewise.
24714 (find_param_candidates): Likewise.
24715 (decide_one_param_reduction): Likewise.
24716 (replace_removed_params_ssa_names): Likewise.
24717 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24718 * tree-ssa-copy.c (dump_copy_of): Likewise.
24719 (copy_prop_visit_cond_stmt): Likewise.
24720 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24721 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24722 (record_equivalences_from_stmt): Likewise.
24723 * tree-ssa-dse.c (compute_trims): Likewise.
24724 (delete_dead_call): Likewise.
24725 (delete_dead_assignment): Likewise.
24726 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24727 (forward_propagate_into_cond): Likewise.
24728 (pass_forwprop::execute): Likewise.
24729 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24730 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24732 (move_computations_worker): Likewise.
24733 (execute_sm): Likewise.
24734 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24735 (remove_exits_and_undefined_stmts): Likewise.
24736 (remove_redundant_iv_tests): Likewise.
24737 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24738 (adjust_iv_update_pos): Likewise.
24739 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24740 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24741 (value_replacement): Likewise.
24742 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24743 * tree-ssa-pre.c (print_pre_expr): Likewise.
24744 (get_representative_for): Likewise.
24745 (create_expression_by_pieces): Likewise.
24746 (insert_into_preds_of_block): Likewise.
24747 (eliminate_insert): Likewise.
24748 (eliminate_dom_walker::before_dom_children): Likewise.
24749 (eliminate): Likewise.
24750 (remove_dead_inserted_code): Likewise.
24751 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24752 * tree-ssa-reassoc.c (get_rank): Likewise.
24753 (eliminate_duplicate_pair): Likewise.
24754 (eliminate_plus_minus_pair): Likewise.
24755 (eliminate_not_pairs): Likewise.
24756 (undistribute_ops_list): Likewise.
24757 (eliminate_redundant_comparison): Likewise.
24758 (update_range_test): Likewise.
24759 (optimize_range_tests_var_bound): Likewise.
24760 (optimize_vec_cond_expr): Likewise.
24761 (rewrite_expr_tree): Likewise.
24762 (rewrite_expr_tree_parallel): Likewise.
24763 (linearize_expr): Likewise.
24764 (break_up_subtract): Likewise.
24765 (linearize_expr_tree): Likewise.
24766 (attempt_builtin_powi): Likewise.
24767 (attempt_builtin_copysign): Likewise.
24768 (transform_stmt_to_copy): Likewise.
24769 (transform_stmt_to_multiply): Likewise.
24770 (dump_ops_vector): Likewise.
24771 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24772 (print_scc): Likewise.
24773 (set_ssa_val_to): Likewise.
24774 (visit_reference_op_store): Likewise.
24775 (visit_use): Likewise.
24776 (sccvn_dom_walker::before_dom_children): Likewise.
24777 (run_scc_vn): Likewise.
24778 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24780 (expr_hash_elt::print): Likewise.
24781 (const_and_copies::pop_to_marker): Likewise.
24782 (const_and_copies::record_const_or_copy_raw): Likewise.
24783 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24784 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24785 (dump_predicates): Likewise.
24786 (find_uninit_use): Likewise.
24787 (warn_uninitialized_phi): Likewise.
24788 (pass_late_warn_uninitialized::execute): Likewise.
24789 * tree-ssa.c (verify_vssa): Likewise.
24790 (verify_ssa): Likewise.
24791 (maybe_optimize_var): Likewise.
24792 * tree-vrp.c (dump_value_range): Likewise.
24793 (dump_all_value_ranges): Likewise.
24794 (dump_asserts_for): Likewise.
24795 (register_edge_assert_for_2): Likewise.
24796 (vrp_visit_cond_stmt): Likewise.
24797 (vrp_visit_switch_stmt): Likewise.
24798 (vrp_visit_stmt): Likewise.
24799 (vrp_visit_phi_node): Likewise.
24800 (simplify_cond_using_ranges_1): Likewise.
24801 (fold_predicate_in): Likewise.
24802 (evrp_dom_walker::before_dom_children): Likewise.
24803 (evrp_dom_walker::push_value_range): Likewise.
24804 (evrp_dom_walker::pop_value_range): Likewise.
24805 (execute_early_vrp): Likewise.
24807 2017-05-16 Richard Biener <rguenther@suse.de>
24809 * dwarf2out.c (loc_list_from_tree_1): Do not create
24810 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24812 2017-05-16 Richard Biener <rguenther@suse.de>
24814 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24816 (note_variable_value_in_expr): If we resolved the decl ref
24817 do not push to the stack.
24819 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24821 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24822 operations in fast-math mode.
24823 (vaddq_f16): Likewise.
24824 (vmul_f16): Likewise.
24825 (vmulq_f16): Likewise.
24826 (vsub_f16): Likewise.
24827 (vsubq_f16): Likewise.
24828 * config/arm/neon.md (add<mode>3): New.
24830 (fma:<VH:mode>3): New. Also remove outdated comment.
24833 2017-05-16 Martin Liska <mliska@suse.cz>
24837 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24838 (odr_types_equivalent_p): Likewise.
24840 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24842 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24844 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24847 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24848 non-interunit SSE move alternatives with '?'.
24849 (zero-extendsidi peephole2): New peephole to skip intermediate
24850 general register in SSE zero-extend sequence.
24852 2017-05-15 Jeff Law <law@redhat.com>
24854 * reorg.c (relax_delay_slots): Create a new variable to hold
24855 the temporary target rather than clobbering TARGET_LABEL.
24857 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24858 missing argument to extract_bit_field call.
24859 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24861 2017-05-15 Martin Liska <mliska@suse.cz>
24864 * gcc.c (process_command): Do not allow empty argument of -o option.
24866 2017-05-15 Renlin Li <renlin.li@arm.com>
24868 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24869 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24870 * config/aarch64/constraints.md (Usf): Add long call check.
24871 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24872 (call_value): Likewise.
24873 (sibcall): Likewise.
24874 (sibcall_value): Likewise.
24876 (call_value_insn): New.
24877 (sibcall_insn): Update rtx pattern.
24878 (sibcall_value_insn): Likewise.
24879 (call_internal): Remove.
24880 (call_value_internal): Likewise.
24881 (sibcall_internal): Likewise.
24882 (sibcall_value_internal): Likewise.
24883 (call_reg): Likewise.
24884 (call_symbol): Likewise.
24885 (call_value_reg): Likewise.
24886 (call_value_symbol): Likewise.
24888 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24891 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24893 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24895 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24896 compatible with CCGOCmode and with CCZmode.
24898 2017-05-14 Martin Sebor <msebor@redhat.com>
24900 PR middle-end/77671
24901 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24902 (gimple_fold_builtin_snprintf): Same.
24903 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24904 (gimple_fold_builtin_snprintf): Same.
24905 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24906 of character types.
24907 (is_call_safe): New function.
24908 (try_substitute_return_value): Call it.
24909 (try_simplify_call): New function.
24910 (pass_sprintf_length::handle_gimple_call): Call it.
24912 2017-05-14 Martin Sebor <msebor@redhat.com>
24914 PR middle-end/80669
24915 * builtins.c (expand_builtin_stpncpy): Simplify.
24917 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24919 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24920 * config/i386/i386.h
24921 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24922 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24923 (struct machine_function): Add new members call_ms2sysv,
24924 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24925 (struct machine_frame_state): New fields sp_realigned and
24926 sp_realigned_offset.
24927 * config/i386/i386.c
24928 (enum xlogue_stub): New enum.
24929 (enum xlogue_stub_sets): New enum.
24930 (class xlogue_layout): New class.
24931 (struct ix86_frame): New fields stack_realign_allocate_offset,
24932 stack_realign_offset and outlined_save_offset. Modify comments to
24933 detail stack layout when using out-of-line stubs.
24934 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24935 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24936 -mcall-ms2sysv-xlogues.
24937 (stub_managed_regs): New static variable.
24938 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24939 registers managed by out-of-line stub.
24940 (disable_call_ms2sysv_xlogues): New function.
24941 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24942 m->call_ms2sysv when appropriate and compute frame layout for
24944 (sp_valid_at, fp_valid_at): New inline functions.
24945 (choose_basereg): New function.
24946 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24948 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24949 Use align parameter of choose_baseaddr to generated aligned SSE movs
24951 (pro_epilogue_adjust_stack): Modify to track
24952 machine_frame_state::sp_realigned.
24953 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24954 (ix86_nsaved_sseregs): Likewise.
24955 (ix86_emit_save_regs): Likewise.
24956 (ix86_emit_save_regs_using_mov): Likewise.
24957 (ix86_emit_save_sse_regs_using_mov): Likewise.
24958 (get_scratch_register_on_entry): Likewise.
24959 (gen_frame_set): New function.
24960 (gen_frame_load): Likewise.
24961 (gen_frame_store): Likewise.
24962 (emit_outlined_ms2sysv_save): Likewise.
24963 (emit_outlined_ms2sysv_restore): Likewise.
24964 (ix86_expand_prologue): Modify stack re-alignment code and call
24965 emit_outlined_ms2sysv_save when appropriate.
24966 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24967 parameter rtx_insn *insn, which allows the function to be used to only
24968 generate the notes.
24969 (ix86_expand_epilogue): Modify validity checks of frame and stack
24970 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24971 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24972 * config/i386/predicates.md
24973 (save_multiple): New predicate.
24974 (restore_multiple): Likewise.
24975 * config/i386/sse.md
24976 (save_multiple<mode>): New pattern.
24977 (save_multiple_realign<mode>): Likewise.
24978 (restore_multiple<mode>): Likewise.
24979 (restore_multiple_and_return<mode>): Likewise.
24980 (restore_multiple_leave_return<mode>): Likewise.
24981 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24983 2017-05-14 Julia Koval <julia.koval@intel.com>
24985 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24986 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24987 (__builtin_ia32_xsetbv): New builtins.
24988 * config/i386/i386.c (ix86_expand_special_args_builtin):
24990 (ix86_expand_builtin): Special expand for new intrinsics.
24991 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24992 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24993 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24995 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24997 * cfganal.c (inverted_post_order_compute): Change argument type
24999 * cfganal.h (inverted_post_order_compute): Adjust prototype.
25000 * df-core.c (rest_of_handle_df_initialize): Adjust.
25001 (rest_of_handle_df_finish): Likewise.
25002 (df_analyze_1): Likewise.
25003 (df_analyze): Likewise.
25004 (loop_inverted_post_order_compute): Change argument to be a vec *.
25005 (df_analyze_loop): Adjust.
25006 (df_get_n_blocks): Likewise.
25007 (df_get_postorder): Likewise.
25008 * df.h (struct df_d): Change field to be a vec.
25009 * lcm.c (compute_laterin): Adjust.
25010 (compute_available): Likewise.
25011 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25012 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25013 * tree-ssa-pre.c (compute_antic): Likewise.
25015 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25017 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25018 (depth_first_search::depth_first_search): Change structure init
25019 function to this constructor.
25020 (depth_first_search::add_bb): Rename function to this member.
25021 (depth_first_search::execute): Likewise.
25022 (flow_dfs_compute_reverse_finish): Adjust.
25024 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25026 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25027 (longest_simple_path): Likewise.
25028 * shrink-wrap.c (spread_components): Likewise.
25029 (disqualify_problematic_components): Likewise.
25030 (emit_common_heads_for_components): Likewise.
25031 (emit_common_tails_for_components): Likewise.
25032 (insert_prologue_epilogue_for_components): Likewise.
25034 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25036 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25039 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25041 * df-core.c (df_set_blocks): Start using auto_bitmap.
25042 (df_compact_blocks): Likewise.
25043 * df-problems.c (df_rd_confluence_n): Likewise.
25044 * df-scan.c (df_insn_rescan_all): Likewise.
25045 (df_process_deferred_rescans): Likewise.
25046 (df_update_entry_block_defs): Likewise.
25047 (df_update_exit_block_uses): Likewise.
25048 (df_entry_block_bitmap_verify): Likewise.
25049 (df_exit_block_bitmap_verify): Likewise.
25050 (df_scan_verify): Likewise.
25051 * lra-constraints.c (lra_constraints): Likewise.
25052 (undo_optional_reloads): Likewise.
25053 (lra_undo_inheritance): Likewise.
25054 * lra-remat.c (calculate_gen_cands): Likewise.
25055 (do_remat): Likewise.
25056 * lra-spills.c (assign_spill_hard_regs): Likewise.
25057 (spill_pseudos): Likewise.
25058 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25059 (bitmap_set_subtract_values): Likewise.
25061 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25063 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25064 management with auto_bitmap.
25065 (fix_inter_tick): Likewise.
25066 (fix_recovery_deps): Likewise.
25067 * ira.c (add_store_equivs): Likewise.
25068 (find_moveable_pseudos): Likewise.
25069 (split_live_ranges_for_shrink_wrap): Likewise.
25070 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25071 (rtx_reuse_manager::seen_def_p): Likewise.
25072 (rtx_reuse_manager::set_seen_def): Likewise.
25073 * print-rtl.h (class rtx_reuse_manager): Likewise.
25075 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25077 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25079 (migrate_btr_def): Likewise.
25080 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25081 * df-core.c (loop_post_order_compute): Likewise.
25082 (loop_inverted_post_order_compute): Likewise.
25083 * hsa-common.h: Likewise.
25084 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25085 * init-regs.c (initialize_uninitialized_regs): Likewise.
25086 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25087 (inline_small_functions): Likewise.
25088 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25089 * ira.c (combine_and_move_insns): Likewise.
25090 (build_insn_chain): Likewise.
25091 * loop-invariant.c (find_invariants): Likewise.
25092 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25093 * predict.c (tree_predict_by_opcode): Likewise.
25094 (predict_paths_leading_to): Likewise.
25095 (predict_paths_leading_to_edge): Likewise.
25096 (estimate_loops_at_level): Likewise.
25097 (estimate_loops): Likewise.
25098 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25099 (spread_components): Likewise.
25100 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25101 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25102 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25103 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25104 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25105 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25106 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25107 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25108 (mark_threaded_blocks): Likewise.
25109 (thread_through_all_blocks): Likewise.
25110 * tree-ssa.c (verify_ssa): Likewise.
25111 (execute_update_addresses_taken): Likewise.
25112 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25114 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25116 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25118 (post_order_compute): Likewise.
25119 (inverted_post_order_compute): Likewise.
25120 (pre_and_rev_post_order_compute_fn): Likewise.
25122 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25124 * genrecog.c (int_set::int_set): Explicitly construct our
25125 auto_vec base class.
25126 * vec.h (auto_vec::auto_vec): New constructor.
25128 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25130 * bitmap.h (class auto_bitmap): New constructor taking
25131 bitmap_obstack * argument.
25133 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25135 * bitmap.h (class auto_bitmap): Change type of m_bits to
25136 bitmap_head, and adjust ctor / dtor and member operators.
25138 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25140 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25141 when returned register mode doesn't match original mode.
25143 2017-05-12 Jeff Law <law@redhat.com>
25144 Jakub Jelinek <jakub@redhat.com>
25146 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25147 we look for cc setter after the compare-elim changes.
25148 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25149 within the vector to match what compare-elim now expects.
25150 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25151 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25153 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25154 after the compare-elim changes.
25155 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25156 the vector to match what compare-elim now expects.
25157 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25158 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25159 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25160 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25161 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25163 * config/visium/visium.c (single_set_and_flags): Fix where
25164 we look for cc setter after the compare-elim changes.
25165 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25166 with the vector to match what compare-elim now expects.
25167 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25168 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25169 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25170 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25171 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25172 (neg<mode>2_insn_set_overflow): Likewise.
25174 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25176 PR middle-end/79794
25177 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25178 maybe_expand_insn call, set ops[0].target. If still set after call,
25179 set alt_rtl. Add extra arg to recursive calls.
25180 (extract_bit_field): Add alt_rtl argument. Pass to
25182 * expmed.h (extract_bit_field): Fix prototype.
25183 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25184 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25185 to extract_bit_field_calls.
25186 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25187 Pass alt_rtl to extract_bit_field calls.
25188 * calls.c (store_unaligned_arguments_into_psuedos)
25189 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25190 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25192 * optabs.h (struct expand_operand): Add target bitfield.
25194 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25196 * compare-elim.c (try_eliminate_compare): Canonicalize
25197 operation with embedded compare to
25198 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25199 (set (reg) (operation)].
25201 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25203 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25206 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25207 cost of adding a carry flag for ADC instruction.
25208 [case MINUS]: Ignore the cost of subtracting a carry flag
25209 for SBB instruction.
25211 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25213 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25215 * config/rs6000/bmiintrin.h: New file.
25216 * config/rs6000/bmi2intrin.h: New file.
25217 * config/rs6000/x86intrin.h: New file.
25219 2017-05-12 Jeff Law <law@redhat.com>
25221 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25224 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25226 PR middle-end/80707
25227 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25229 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25231 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25232 early expansion of vector divide builtins.
25233 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25234 builtins identified as having unsigned arguments.
25236 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25238 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25239 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25240 expansion of vector logical operations (and, andc, or, xor,
25243 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25245 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25246 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25248 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25250 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25251 early GIMPLE expansion of vector multiplies.
25253 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25255 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25256 TARGET_HAVE_MOVT conditional.
25257 (movt splitter): Likewise.
25259 2017-05-12 Richard Biener <rguenther@suse.de>
25261 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25262 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25263 Fold all stmts not inplace.
25265 2017-05-12 Richard Biener <rguenther@suse.de>
25267 PR tree-optimization/80713
25268 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25269 inserted_exprs bit for not removed stmts.
25271 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25273 PR middle-end/69921
25274 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25275 parallelized" attribute for parallelized OpenACC kernels.
25276 * omp-offload.c (execute_oacc_device_lower): Use it.
25278 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25279 Set "oacc kernels" attribute.
25280 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25281 parameter. Adjust all users.
25282 (oacc_fn_attrib_kernels_p): Remove function.
25283 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25284 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25285 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25286 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25287 assert "oacc kernels" attribute is set.
25289 2017-05-11 Carl Love <cel@us.ibm.com>
25291 * config/rs6000/rs6000-c: Add support for built-in functions
25292 vector unsigned char vec_popcnt (vector signed char)
25293 vector unsigned char vec_popcnt (vector unsigned char)
25294 vector unsigned short vec_popcnt (vector signed short)
25295 vector unsigned short vec_popcnt (vector unsigned short)
25296 vector unsigned int vec_popcnt (vector signed int)
25297 vector unsigned int vec_popcnt (vector unsigned int)
25298 vector unsigned long long vec_popcnt (vector signed long long)
25299 vector unsigned long long vec_popcnt (vector unsigned long long)
25300 vector signed long long vec_slo (vector signed long long,
25301 vector signed char)
25302 vector signed long long vec_slo (vector signed long long,
25303 vector unsigned char)
25304 vector unsigned long long vec_slo (vector unsigned long long,
25305 vector signed char)
25306 vector unsigned long long vec_slo (vector unsigned long long,
25307 vector unsigned char)
25308 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25309 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25310 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25311 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25312 * doc/extend.texi: Update the built-in documentation file for the
25313 new built-in functions.
25315 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25317 * attribs.h (sorted_attr_string): Move machine independent
25318 functions for target clone support from the i386 port to common
25319 code. Rename ix86_function_versions to common_function_versions.
25320 Rename make_name to make_unique_name.
25321 (common_function_versions): Likewise.
25322 (make_unique_name): Likewise.
25323 (make_dispatcher_decl): Likewise.
25324 (is_function_default_version): Likewise.
25325 * attribs.c (attr_strcmp): Likewise.
25326 (sorted_attr_string): Likewise.
25327 (common_function_versions): Likewise.
25328 (make_unique_name): Likewise.
25329 (make_dispatcher_decl): Likewise.
25330 (is_function_default_version): Likewise.
25331 * config/i386/i386.c (attr_strcmp): Likewise.
25332 (sorted_attr_string): Likewise.
25333 (ix86_function_versions): Likewise.
25334 (make_name): Likewise.
25335 (make_dispatcher_decl): Likewise.
25336 (is_function_default_version): Likewise.
25337 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25339 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25342 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25343 Account for direct move costs for vec_construct of integer
25346 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25349 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25350 (UNSPEC_STX_ATOMIC): Ditto.
25351 (loaddi_via_sse): New insn.
25352 (storedi_via_sse): Ditto.
25353 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25354 Update corresponding peephole2 patterns.
25355 (atomic_storedi_fpu): Ditto.
25357 2017-05-11 Julia Koval <julia.koval@intel.com>
25359 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25360 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25362 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25363 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25364 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25366 2017-05-11 Nathan Sidwell <nathan@acm.org>
25368 * graphite-poly.c: Include dumpfile.h.
25370 * dumpfle.h (dump_function): Declare here ...
25371 * tree-dump.h (dump_function): ... not here.
25372 * dumpfile.c: #include tree-cfg.h.
25373 (dump_function): Move here from ...
25374 * tree-dump.c (dump_function): ... here.
25375 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25376 * graphite-poly.c: Don't include tree-dump.h.
25377 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25378 * print-tree.c: Likewise.
25379 * stor-layout.c: Likewise.
25380 * tree-nested.c: Likewise.
25382 * dumpfile.c (dump_start): Use TDF_FLAGS.
25383 (dump_enable_all): Fix TDF_KIND check thinko.
25385 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25387 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25388 array entries to represent two legal parameterizations of the
25389 overloaded __builtin_cmpb function, as represented by the
25390 P6_OV_BUILTIN_CMPB constant.
25391 (altivec_resolve_overloaded_builtin): Add special case handling
25392 for the __builtin_cmpb function, as represented by the
25393 P6_OV_BUILTIN_CMPB constant.
25394 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25395 (BU_P6_64BIT_2): New macro.
25396 (BU_P6_OVERLOAD_2): New macro
25397 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25398 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25399 (CMPB): Add overload support to represent both 32-bit and 64-bit
25400 compare-bytes function.
25401 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25402 support for TARGET_CMPB.
25403 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25404 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25405 documentation of the __builtin_cmpb overloaded built-in function.
25407 2017-05-11 Richard Biener <rguenther@suse.de>
25409 PR tree-optimization/80705
25410 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25411 bases are not vectorizable.
25413 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25415 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25416 when counting register pressure.
25418 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25420 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25421 (iv_ca_more_deps): Renamed to ...
25422 (iv_ca_compare_deps): ... this.
25423 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25425 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25427 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25429 (determine_group_iv_costs): ... here.
25430 (find_inv_vars_cb): Record inv var if it's not recorded before.
25432 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25434 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25435 (get_shiftadd_cost): Ditto.
25437 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25439 * tree-ssa-address.c: Include header file.
25440 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25442 (add_to_parts): Refactor.
25443 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25444 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25447 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25449 PR tree-optimization/53090
25450 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25452 (extract_cond_operands): Detect condition with IV on both sides
25453 and return COMP_IV_EXPR_2.
25454 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25455 (rewrite_use_compare): Simplify by removing call to function
25456 extract_cond_operands.
25458 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25460 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25461 (extract_cond_operands): Detect condition comparing against non-
25462 invariant bound and return appropriate enum value.
25463 (find_interesting_uses_cond): Update use of extract_cond_operands.
25464 Handle its return value accordingly.
25465 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25467 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25469 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25470 nonlinear iv_use computation in loop invariant sensitive way.
25472 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25474 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25475 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25477 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25479 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25480 (dump_cand): Support iv_cand.inv_exprs.
25481 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25483 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25486 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25488 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25490 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25491 as local function. Include necessary header files.
25492 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25494 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25496 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25498 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25500 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25501 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25502 RSHIFT_EXPR and BIT_NOT_EXPR.
25504 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25506 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25507 (adjust_setup_cost): New parameter supporting round up adjustment.
25508 (struct address_cost_data): Delete.
25509 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25510 (split_address_cost, ptr_difference_cost): Delete.
25511 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25512 (struct ainc_cost_data): New struct.
25513 (get_address_cost_ainc): New function.
25514 (get_address_cost, get_computation_cost): Reimplement.
25515 (determine_group_iv_cost_address): Record inv_expr for all uses of
25517 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25518 (iv_ca_has_deps): Reimplemented to ...
25519 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25521 (iv_ca_extend): Call iv_ca_more_deps.
25523 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25525 * tree-ssa-address.c (struct mem_address): Move to header file.
25526 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25527 * tree-ssa-address.h (struct mem_address): Move from C file.
25528 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25530 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25532 * tree-affine.h (aff_combination_type): New interface.
25533 (aff_combination_zero_p): Remove static.
25534 (aff_combination_const_p): New interface.
25535 (aff_combination_singleton_var_p): New interfaces.
25537 2017-05-11 Richard Biener <rguenther@suse.de>
25539 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25540 Skip unreachable blocks and destinations.
25541 (eliminate): Move stmt removal and fixup ...
25542 (fini_eliminate): ... here. Skip inserted exprs.
25543 (pass_pre::execute): Move fini_pre after fini_eliminate.
25544 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25545 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25546 PRE to get rid of dead code that has invalid SSA form and
25547 split critical edges again.
25549 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25551 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25553 2017-05-11 Richard Biener <rguenther@suse.de>
25555 * passes.c (execute_function_todo): Verify loops if they are
25556 said to be up-to-date.
25557 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25558 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25560 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25563 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25564 handle calling assemble_external ourself.
25567 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25568 modes with zero size. Enhance comment.
25570 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25572 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25573 built-ins for vec_xl and vec_xst with short and char pointer
25576 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25578 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25579 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25580 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25581 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25582 (_mm_maskz_min_round_ss): New intrinsics.
25583 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25584 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25585 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25586 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25587 (__builtin_ia32_minss_mask_round): New builtins.
25588 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25589 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25590 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25592 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25593 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25595 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25598 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25600 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25601 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25602 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25603 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25604 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25605 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25606 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25607 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25608 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25609 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25610 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25611 (__builtin_ia32_mulss_mask_round): New builtins.
25612 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25613 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25614 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25616 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25617 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25619 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25622 2017-05-10 Julia Koval <julia.koval@intel.com>
25624 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25625 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25626 (_mm256_setr_m128i): New intrinsics.
25628 2017-05-10 Julia Koval <julia.koval@intel.com>
25630 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25631 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25632 (_mm_maskz_rcp14_ss): New intrinsics.
25633 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25634 (__builtin_ia32_rcp14ss_mask): New builtins.
25635 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25637 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25639 PR tree-optimization/51513
25640 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25641 (assert_unreachable_fallthru_edge_p): Use it.
25642 (group_case_labels_stmt): Likewise.
25643 * tree-cfg.h: Prototype it.
25644 * stmt.c: Include cfghooks.h and tree-cfg.h.
25645 (emit_case_dispatch_table) <gap_label>: New local variable.
25646 Use it to fill dispatch table gaps.
25647 Test for default_label before updating probabilities.
25648 (expand_case) <default_label>: Remove unneeded initialization.
25649 Test for unreachable default case statement and remove its edge.
25650 Set default_label accordingly.
25651 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25653 2017-05-10 Carl Love <cel@us.ibm.com>
25655 * config/rs6000/rs6000-c: Add support for built-in functions
25656 vector signed char vec_neg (vector signed char)
25657 vector signed short int vec_neg (vector short int)
25658 vector signed int vec_neg (vector signed int)
25659 vector signed long long vec_neg (vector signed long long)
25660 vector float vec_neg (vector float)
25661 vector double vec_neg (vector double)
25662 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25664 * config/rs6000/altivec.h: Add define for vec_neg
25665 * doc/extend.texi: Update the built-in documentation for the
25666 new built-in functions.
25668 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25670 PR tree-optimization/77644
25671 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25673 2017-05-10 Nathan Sidwell <nathan@acm.org>
25675 * dumpfile.h (TDI_lang_all): New.
25676 (TDF_KIND): New. Renumber others
25677 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25679 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25681 (get_dump_file_name): Adjust suffix generation.
25682 (dump_enable_all): Use TDF_KIND.
25683 * doc/invoke.texi (-fdump-lang-all): Document.
25685 * dumpfile.h: Tabify.
25687 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25690 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25691 Move member access before delete.
25693 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25695 * tree-inline.c (expand_call_inline): Split block at stmt
25698 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25701 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25702 are now unused after splitting mov{sf,sd}_hardfloat.
25703 (f32_lr2): Likewise.
25704 (f32_lm): Likewise.
25705 (f32_lm2): Likewise.
25706 (f32_li): Likewise.
25707 (f32_li2): Likewise.
25708 (f32_lv): Likewise.
25709 (f32_sr): Likewise.
25710 (f32_sr2): Likewise.
25711 (f32_sm): Likewise.
25712 (f32_sm2): Likewise.
25713 (f32_si): Likewise.
25714 (f32_si2): Likewise.
25715 (f32_sv): Likewise.
25716 (f32_dm): Likewise.
25717 (f32_vsx): Likewise.
25718 (f32_av): Likewise.
25719 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25720 For movsf, order stores so the VSX stores occur before the GPR
25721 store which encourages the register allocator to use a traditional
25722 FPR instead of a GPR. For movsd, order the stores so that the GPR
25723 store comes before the VSX stores to allow the power6 to work.
25724 This is due to the power6 not having a 32-bit integer store
25725 instruction from a FPR.
25726 (movsf_hardfloat): Likewise.
25727 (movsd_hardfloat): Likewise.
25729 2017-05-09 Martin Sebor <msebor@redhat.com>
25731 PR translation/80280
25732 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25735 PR translation/80280
25736 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25737 data member added in r247778.
25738 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25740 2017-05-09 Nathan Sidwell <nathan@acm.org>
25742 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25744 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25747 2017-05-09 Marek Polacek <polacek@redhat.com>
25749 * doc/invoke.texi: Fix typo.
25751 2017-05-09 Richard Biener <rguenther@suse.de>
25753 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25754 (vrp_val_is_min): Likewise.
25755 (set_value_range_to_value): Likewise.
25756 (set_value_range_to_nonnegative): Likewise.
25757 (gimple_assign_nonzero_p): Likewise.
25758 (gimple_stmt_nonzero_p): Likewise.
25759 (vrp_int_const_binop): Likewise. Remove unreachable case.
25760 (adjust_range_with_scev): Adjust comments.
25761 (compare_range_with_value): Likewise.
25762 (extract_range_from_phi_node): Likewise.
25763 (test_for_singularity): Likewise.
25765 2017-05-09 Richard Biener <rguenther@suse.de>
25767 * tree-vrp.c (get_single_symbol): Add assert that we don't
25768 get overflowed constants as invariant part.
25769 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25770 checks. Use wi::cmp instead of recursing for integer constants.
25771 (compare_values): Just ignore whether we assumed undefined
25772 overflow instead of failing the compare.
25773 (extract_range_for_var_from_comparison_expr): Add comment before the
25774 TREE_NO_WARNING sets.
25775 (test_for_singularity): Likewise.
25776 (extract_range_from_comparison): Do not disable optimization
25777 when we assumed undefined overflow.
25778 (extract_range_basic): Remove init of unused var.
25780 2017-05-09 Richard Biener <rguenther@suse.de>
25782 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25783 (extract_range_from_multiplicative_op_1): Adjust.
25784 (extract_range_from_binary_expr_1): Use int_const_binop.
25786 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25789 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25790 rs6000_store_data_bypass_p in seven define_bypass directives and
25791 in several comments.
25792 * config/rs6000/rs6000-protos.h: Add prototype for
25793 rs6000_store_data_bypass_p function.
25794 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25795 function implements slightly different (rs6000-specific) semantics
25796 than store_data_bypass_p, returning false rather than aborting
25797 with assertion error when arguments do not satisfy the
25798 requirements of store data bypass.
25799 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25800 rs6000_store_data_bypass_p.
25802 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25804 * config/xtensa/xtensa-protos.h
25805 (xtensa_initial_elimination_offset): New declaration.
25806 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25807 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25808 macro definition, add case for FRAME_POINTER_REGNUM when
25809 FRAME_GROWS_DOWNWARD.
25810 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25811 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25812 xtensa_initial_elimination_offset.
25814 2017-05-08 Nathan Sidwell <nathan@acm.org>
25816 * doc/invoke.texi: Alphabetize -fdump options.
25818 2017-05-08 Martin Sebor <msebor@redhat.com>
25820 PR translation/80280
25821 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25823 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25825 * target.def (compute_frame_layout): New optional target hook.
25826 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25827 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25828 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25830 * reload1.c (verify_initial_elim_offsets): Likewise.
25831 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25832 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25833 (arm_get_frame_offsets): Split up into this ...
25834 (arm_compute_frame_layout): ... and this function.
25836 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25838 * config/aarch64/constraints.md (Usa): New constraint.
25839 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25841 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25843 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25844 with_multilib_list after it has been checked.
25846 2017-05-08 Richard Biener <rguenther@suse.de>
25848 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25849 (bitmap_set_subtract_values): Likewise.
25851 2017-05-08 Richard Biener <rguenther@suse.de>
25853 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25854 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25856 (gimple_stmt_nonzero_warnv_p): Rename to ...
25857 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25859 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25860 (extract_range_basic): Adjust, do not disable propagation on
25861 strict overflow sensitive simplification.
25862 (vrp_visit_cond_stmt): Likewise.
25864 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25866 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25867 body size unconditionally.
25869 2017-05-07 Jeff Law <law@redhat.com>
25872 2017-05-06 Jeff Law <law@redhat.com>
25873 PR tree-optimization/78496
25874 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25877 PR tree-optimization/78496
25878 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25879 (simplify_stmt_using_ranges): Call it.
25880 (vrp_dom_walker::before_dom_children): Extract equivalences
25881 from an ASSERT_EXPR with an equality comparison against a
25884 2017-05-06 Jeff Law <law@redhat.com>
25886 PR tree-optimization/78496
25887 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25890 PR tree-optimization/78496
25891 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25892 (simplify_stmt_using_ranges): Call it.
25893 (vrp_dom_walker::before_dom_children): Extract equivalences
25894 from an ASSERT_EXPR with an equality comparison against a
25897 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25899 * lra-constraints.c (lra_copy_reg_equiv): New function.
25900 (split_reg): Use it to copy equivalence information from the
25901 original register to the spill register.
25903 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25905 PR rtl-optimization/75964
25906 * simplify-rtx.c (simplify_const_relational_operation): Remove
25907 invalid handling of comparisons of integer ABS.
25909 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25911 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25912 initialize to zero.
25913 (init_regs): Remove declaration.
25914 (function_arg_advance_32): Initialize error_p as boolean variable.
25916 2017-05-05 Nathan Sidwell <nathan@acm.org>
25918 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25919 lines. Use for (;;).
25921 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25923 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25924 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25925 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25926 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25927 VF=2 that require versioning.
25929 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25931 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25934 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25936 * diagnostic.h (diagnostic_override_option_index): Convert from
25937 macro to inline function.
25939 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25941 * diagnostic.c (last_module_changed_p): New function.
25942 (set_last_module): New function.
25943 (diagnostic_report_current_module): Convert macro usage to
25944 the above functions.
25945 * diagnostic.h (diagnostic_context::last_module): Strengthen
25946 from const line_map * to const line_map_ordinary *.
25947 (diagnostic_last_module_changed): Delete macro.
25948 (diagnostic_set_last_module): Delete macro.
25950 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25952 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25953 with diagnostic_report_diagnostic.
25954 (diagnostic_n_impl_richloc): Likewise.
25955 * diagnostic.h (report_diagnostic): Delete macro.
25956 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25957 with diagnostic_report_diagnostic.
25958 * substring-locations.c (format_warning_va): Likewise.
25960 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25962 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25963 save/restor of format_spec. Move option-printing code to...
25964 (print_option_information): ...this new function, and
25965 reimplement by simply printing to the pretty_printer,
25966 rather than appending to the format string.
25968 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25970 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25971 handling logic into...
25972 (update_effective_level_from_pragmas): ...this new function.
25974 2017-05-04 Andrew Waterman <andrew@sifive.com>
25976 * config/riscv/riscv.opt (mstrict-align): New option.
25977 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25978 (SLOW_UNALIGNED_ACCESS): Define.
25979 (riscv_slow_unaligned_access): Declare.
25980 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25982 (riscv_slow_unaligned_access): New variable.
25983 (rocket_tune_info): Set slow_unaligned_access to true.
25984 (optimize_size_tune_info): Set slow_unaligned_access to false.
25985 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25986 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25987 (riscv_option_override): Set riscv_slow_unaligned_access.
25988 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25990 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25992 * config/riscv/riscv.md: Unify indentation.
25994 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25999 * config/rs6000/rs6000.md (u code attribute): Add FIX and
26001 (extendsi<mode>2): Add support for doing sign extension via
26002 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
26003 don't have ISA 3.0 instructions.
26004 (extendsi<mode>2 splitter): Likewise.
26005 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26006 generate the normal insns since SImode can now go in vector
26007 registers. Disallow the special UNSPECs needed for previous
26008 machines to hide SImode being used. Add new insns
26009 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26010 (fix_trunc<mode>si2_stfiwx): Likewise.
26011 (fix_trunc<mode>si2_internal): Likewise.
26012 (fixuns_trunc<mode>si2): Likewise.
26013 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26014 (fctiw<u>z_<mode>_smallint): Likewise.
26015 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26016 of floating point to 32-bit integer from doing a direct move to
26017 the GPR registers to do a store.
26018 (fctiwz_<mode>): Break long line.
26020 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26022 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26023 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26024 (addr_list, addr_offset_valid_p): New.
26025 (split_address_groups): Check offset validity with above function.
26026 (gt-tree-ssa-loop-ivopts.h): Include header file.
26028 2017-05-05 Nathan Sidwell <nathan@acm.org>
26030 * config.gcc (arm*-*-*): Add missing 'fi'.
26032 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26034 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26035 included in -fopt-info does not matter.
26036 * doc/optinfo.texi (-fopt-info): Fix description of default
26037 behavour. Explicitly say order of options included in -fopt-info
26040 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26042 * config.gcc: Allow combinations of aprofile and rmprofile values for
26043 --with-multilib-list.
26044 * config/arm/t-multilib: New file.
26045 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26046 variables. Remove setting of ISA and floating-point ABI in
26047 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26048 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26049 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26050 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26052 * config/arm/t-rmprofile: Likewise except for the matches changes.
26053 * doc/install.texi (--with-multilib-list): Document the combination of
26054 aprofile and rmprofile values and warn about pitfalls in doing that.
26056 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26058 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26059 (movdi_aarch64): Likewise.
26061 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26063 PR tree-optimization/80632
26064 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26066 (build_arrays): Initialize it for virtual phis.
26067 (fix_phi_nodes): Use it for virtual phis.
26069 PR tree-optimization/80558
26070 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26071 [x, y] op z into [x op, y op z] for op & or | if conditions
26074 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26075 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26078 * config/arm/arm.md (use_literal_pool): Remove.
26079 (64-bit immediate split): No longer takes cost into consideration
26080 if arm_disable_literal_pool is enabled.
26081 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26082 used when arm_disable_literal_pool is enabled.
26083 (arm_max_const_double_inline_cost): Remove use of
26084 arm_disable_literal_pool.
26085 (push_minipool_fix): Add assert.
26086 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26087 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26088 (no_literal_pool_sf_immediate): New.
26090 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26092 PR tree-optimization/80613
26093 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26094 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26096 2017-05-05 Richard Biener <rguenther@suse.de>
26098 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26100 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26102 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26103 of this flag from insn conditions due to removal from r247495.
26105 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26107 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26109 (arm_early_store_addr_dep_ptr): Likewise.
26110 * config/arm/aarch-common-protos.h
26111 (arm_early_load_addr_dep_ptr): Add prototype.
26112 (arm_early_store_addr_dep_ptr): Likewise.
26113 * config/arm/cortex-a53.md: Add new bypasses.
26115 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26117 * tree.c (next_type_uid): Change type to unsigned.
26118 (type_hash_canon): Decrement back next_type_uid if
26119 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26120 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26123 2017-05-04 Martin Sebor <msebor@redhat.com>
26125 * builtins.c: Fix a trivial typo in a comment.
26127 PR middle-end/79234
26128 * builtins.c (check_sizes): Adjust to handle reading past the end.
26129 Avoid printing excessive upper bound of ranges. Use %E to print
26130 tree nodes instead of converting them to %wu.
26131 (expand_builtin_memchr): New function.
26132 (compute_dest_size): Rename...
26133 (compute_objsize): ...to this.
26134 (expand_builtin_memcpy): Adjust.
26135 (expand_builtin_mempcpy): Adjust.
26136 (expand_builtin_strcat): Adjust.
26137 (expand_builtin_strcpy): Adjust.
26138 (check_strncat_sizes): Adjust.
26139 (expand_builtin_strncat): Adjust.
26140 (expand_builtin_strncpy): Adjust and simplify.
26141 (expand_builtin_memset): Adjust.
26142 (expand_builtin_bzero): Adjust.
26143 (expand_builtin_memcmp): Adjust.
26144 (expand_builtin): Handle memcmp.
26145 (maybe_emit_chk_warning): Check strncat just once.
26147 2017-05-04 Martin Sebor <msebor@redhat.com>
26149 PR preprocessor/79214
26150 PR middle-end/79222
26151 PR middle-end/79223
26152 * builtins.c (check_sizes): Add inlining context and issue
26153 warnings even when -Wno-system-headers is set.
26154 (check_strncat_sizes): Same.
26155 (expand_builtin_strncat): Same.
26156 (expand_builtin_memmove): New function.
26157 (expand_builtin_stpncpy): Same.
26158 (expand_builtin): Handle memmove and stpncpy.
26160 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26162 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26163 which is not used any more.
26165 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26167 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26169 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26171 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26172 (cortexa53_tunings): Likewise.
26173 (cortexa57_tunings): Likewise.
26174 (cortexa72_tunings): Likewise.
26175 (cortexa73_tunings): Likewise.
26177 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26179 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26180 Set loop alignment to 8.
26182 2017-05-04 Martin Sebor <msebor@redhat.com>
26184 PR translation/80280
26185 * builtins.c (expand_builtin_object_size): Add missing quoting to
26186 %D and like directives.
26187 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26188 (hsa_type_for_tree_type): Same.
26189 (verify_function_arguments): Same.
26190 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26191 * varasm.c (get_section): Same.
26194 2017-05-04 Martin Sebor <msebor@redhat.com>
26196 PR translation/80280
26197 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26199 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26201 * config/aarch64/aarch64.c (generic_addrcost_table):
26202 Change HI/TI mode setting.
26204 2017-05-04 Martin Jambor <mjambor@suse.cz>
26206 PR tree-optimization/80622
26207 * tree-sra.c (comes_initialized_p): New function.
26208 (build_accesses_from_assign): Only set write lazily when
26209 comes_initialized_p is false.
26210 (analyze_access_subtree): Use comes_initialized_p.
26211 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26212 instead of testing for PARM_DECL.
26214 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26216 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26217 constraint on operand 0 to allow more general addressing modes.
26218 Adjust output template.
26219 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26221 * config/aarch64/aarch64-protos.h
26222 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26223 * config/aarch64/constraints.md (Dp): New address constraint.
26224 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26227 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26229 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26230 update use of estimate_ipcp_clone_size_and_time.
26231 (estimate_local_effects): Update use of
26232 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26233 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26234 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26235 Return nonspecialized time.
26237 2017-05-04 Richard Biener <rguenther@suse.de>
26239 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26240 for the last VUSE which def dominates the PHI. Directly call
26242 (get_continuation_for_phi_1): Remove.
26244 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26246 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26247 to explain the use of truncating division. Cap the number of
26248 iterations to the maximum given by nb_iterations_upper_bound,
26251 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26253 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26254 * configure: Regenerate.
26255 * config.in: Regenerate.
26256 * config/i386/driver-mingw32.c: new file.
26257 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26258 * config.host: Link driver-mingw32.o on MinGW host.
26259 * doc/install.texi: Document new --enable-mingw-wildcard configure
26262 2017-05-04 Marek Polacek <polacek@redhat.com>
26264 PR tree-optimization/80612
26265 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26267 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26268 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26270 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26271 (movt splitter): Likewise.
26272 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26273 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26274 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26275 block for Thumb-1 with MOVT.
26276 (thumb2_legitimate_address_p): Move code block ...
26277 (can_avoid_literal_pool_for_label_p): ... into this new function.
26278 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26280 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26281 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26282 "M-profile targets with the MOVT instruction".
26284 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26286 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26287 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26288 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26290 2017-05-04 Martin Liska <mliska@suse.cz>
26292 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26293 variable cond_code.
26295 2017-05-04 Richard Biener <rguenther@suse.de>
26297 * tree.c (array_at_struct_end_p): Handle arrays at struct
26298 end with flexarrays more conservatively. Refactor and treat
26299 arrays of arrays or aggregates more strict. Fix
26300 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26301 * tree.c (array_at_struct_end_p): Adjust prototype.
26302 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26303 * gimple-fold.c (get_range_strlen): Likewise.
26304 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26306 2017-05-04 Richard Biener <rguenther@suse.de>
26308 PR tree-optimization/31130
26309 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26311 (supports_overflow_infinity): Likewise.
26312 (is_negative_overflow_infinity): Likewise.
26313 (is_positive_overflow_infinity): Likewise.
26314 (is_overflow_infinity): Likewise.
26315 (stmt_overflow_infinity): Likewise.
26316 (overflow_infinity_range_p): Likewise.
26317 (usable_range_p): Remove as always returning true.
26318 (make_overflow_infinity): Remove.
26319 (negative_overflow_infinity): Likewise.
26320 (positive_overflow_infinity): Likewise.
26321 (avoid_overflow_infinity): Likewise.
26322 (set_value_range): Adjust accordingly.
26323 (set_value_range_to_nonnegative): Likewise, remove now unused
26324 overflow_infinity arg.
26325 (vrp_operand_equal_p): Adjust.
26326 (update_value_range): Likewise.
26327 (range_int_cst_singleton_p): Likewise.
26328 (operand_less_p): Likewise.
26329 (compare_values_warnv): Likewise.
26330 (extract_range_for_var_from_comparison_expr): Likewise.
26331 (vrp_int_const_binop): Likewise.
26332 (zero_nonzero_bits_from_vr): Likewise.
26333 (extract_range_from_multiplicative_op_1): Likewise.
26334 (extract_range_from_binary_expr_1): Likewise.
26335 (extract_range_from_unary_expr): Likewise.
26336 (extract_range_from_comparison): Likewise.
26337 (extract_range_basic): Likewise.
26338 (adjust_range_with_scev): Likewise.
26339 (compare_ranges): Likewise.
26340 (compare_range_with_value): Likewise.
26341 (dump_value_range): Likewise.
26342 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26344 (simplify_cond_using_ranges): Adjust.
26346 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26348 * brig-builtins.def: Added a builtin for class_f64.
26349 * builtin-types.def: Added a builtin type needed by class_f64.
26351 2017-05-03 Jason Merrill <jason@redhat.com>
26353 * timevar.def: Add TV_CONSTEXPR.
26355 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26357 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26359 2017-05-03 Martin Jambor <mjambor@suse.cz>
26361 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26362 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26363 * ipa-cp.c (ipcp_propagate_stage): Do not call
26364 ipa_update_after_lto_read.
26365 * ipa-inline.c (ipa_inline): Likewise.
26367 2017-05-03 Martin Jambor <mjambor@suse.cz>
26369 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26370 tag. Added a default constructor and a destructor.
26371 (ipa_edge_args_sum_t): New class;
26372 (ipa_edge_args_sum): Declare.
26373 (ipa_edge_args_vector): Remove declaration.
26374 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26375 (ipa_free_edge_args_substructures): Remove declaration.
26376 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26377 (ipa_edge_args_info_available_for_edge_p): Likewise.
26378 * ipa-prop.c (ipa_edge_args_vector): Removed.
26379 (edge_removal_hook_holder): Likewise.
26380 (edge_duplication_hook_holder): Likewise.
26381 (ipa_edge_args_sum): New variable.
26382 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26383 ipa_edge_args_vector.
26384 (ipa_free_edge_args_substructures): Likewise.
26385 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26386 ipa_edge_args_vector.
26387 (ipa_edge_removal_hook): Turned into method
26388 ipa_edge_args_sum_t::remove.
26389 (ipa_edge_duplication_hook): Turned into method
26390 ipa_edge_args_sum_t::duplicate.
26391 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26392 registering edge hooks.
26393 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26394 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26395 ipa_edge_args_sum instead of ipa_edge_args_vector.
26396 * ipa-profile.c (ipa_profile): Likewise.
26398 2017-05-03 Martin Jambor <mjambor@suse.cz>
26400 * symbol-summary.h (function_summary): New method exists.
26401 (function_summary::symtab_removal): Deallocate through release.
26402 (call_summary): New class.
26403 (gt_ggc_mx): New overload.
26404 (gt_pch_nx): Likewise.
26405 (gt_pch_nx): Likewise.
26407 2017-05-03 Jeff Law <law@redhat.com>
26409 PR tree-optimization/78496
26410 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26411 from simplify_cond_using_ranges. Split off code to walk
26412 backwards through casts into ...
26413 (simplify_cond_using_ranges_2): New function.
26414 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26415 (execute_vrp): After identifying jump threads, call
26416 simplify_cond_using_ranges_2.
26418 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26421 * ipa-inline.h (inline_summary): Add ctor.
26422 (create_ggc): Do not use ggc_cleared_alloc.
26424 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26425 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26427 * gcc.c (handle_braces): Support escaping in switch matching
26429 * doc/invoke.texi (Spec Files): Document it.
26430 Remove superfluous @code markup in items.
26432 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26434 * diagnostic-show-locus.c (struct column_range): New struct.
26435 (get_affected_columns): New function.
26436 (get_printed_columns): New function.
26437 (struct correction): New struct.
26438 (correction::ensure_capacity): New function.
26439 (correction::ensure_terminated): New function.
26440 (struct line_corrections): New struct.
26441 (line_corrections::~line_corrections): New dtor.
26442 (line_corrections::add_hint): New function.
26443 (layout::print_trailing_fixits): Reimplement in terms of the new
26445 (selftest::test_overlapped_fixit_printing): New function.
26446 (selftest::diagnostic_show_locus_c_tests): Call it.
26448 2017-05-03 Nathan Sidwell <nathan@acm.org>
26450 Canonicalize canonical type hashing
26451 * tree.h (type_hash_canon_hash): Declare.
26452 * tree.c (type_hash_list, attribute_hash_list): Move into
26453 type_hash_canon_hash.
26454 (build_type_attribute_qual_variant): Break out hash code calc into
26455 type_hash_canon_hash.
26456 (type_hash_canon_hash): New. Generic type hash computation.
26457 (build_range_type_1, build_array_type_1, build_function_type,
26458 build_method_type_directly, build_offset_type, build_complex_type,
26459 make_vector_type): Call it.
26461 2017-05-03 Richard Biener <rguenther@suse.de>
26463 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26464 When all DRs have unknown misaligned do not always peel
26465 when there is a store but apply the same costing model as if
26466 there were only loads.
26468 2017-05-03 Richard Biener <rguenther@suse.de>
26471 PR tree-optimization/80492
26472 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26473 compare_base_decls returning dont-know properly.
26475 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26477 * config/arm/iterators.md (CCSI): New mode iterator.
26478 (arch): New mode attribute.
26479 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26480 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26481 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26482 code iterator for success result mode.
26483 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26484 the corresponding new insn generators.
26486 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26489 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26490 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26492 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26494 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26495 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26496 (DDR_A): Wrap DDR argument in brackets.
26497 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26498 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26499 (DDR_REVERSED_P): Likewise.
26501 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26503 PR tree-optimization/79472
26504 * tree-switch-conversion.c (struct switch_conv_info): Add
26505 contiguous_range and default_case_nonstandard fields.
26506 (collect_switch_conv_info): Compute contiguous_range and
26507 default_case_nonstandard fields, don't clear final_bb if
26508 contiguous_range and only the default case doesn't have the required
26510 (check_all_empty_except_final): Set default_case_nonstandard instead
26511 of failing if contiguous_range and the default case doesn't have empty
26513 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26514 and only the default case doesn't have the required constants. Skip
26516 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26517 if default_case_nonstandard.
26518 (build_constructors): Build constant 1 just once. Assert that default
26519 values aren't inserted in between cases if contiguous_range. Skip
26521 (build_arrays): Skip virtual phis.
26522 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26523 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26524 Handle virtual phis.
26525 (gen_inbound_check): Handle default_case_nonstandard case.
26526 (process_switch): Adjust check_final_bb caller. Call
26527 gather_default_values with the first non-default case instead of
26528 default case if default_case_nonstandard.
26530 2017-05-02 Nathan Sidwell <nathan@acm.org>
26532 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26533 check. Fix formatting.
26535 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26537 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26538 errors when comparing specialized and unspecialized times.
26540 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26542 * diagnostic-show-locus.c
26543 (layout::should_print_annotation_line_p): Make private.
26544 (layout::print_annotation_line): Make private.
26545 (layout::annotation_line_showed_range_p): Make private.
26546 (layout::show_ruler): Make private.
26547 (layout::print_source_line): Make private. Pass in line and
26548 line_width, rather than calling location_get_source_line. Drop
26550 (layout::print_leading_fixits): New method.
26551 (layout::print_any_fixits): Rename to...
26552 (layout::print_trailing_fixits): ...this, and make private.
26553 Don't print newline fixits.
26554 (diagnostic_show_locus): Move logic for printing one row into...
26555 (layout::print_line): ...this new function. Move the
26556 location_get_source_line call and error-handling from
26557 print_source_line to here. Call print_leading_fixits, and rename
26558 print_any_fixits to print_trailing_fixits.
26559 (selftest::test_fixit_insert_containing_newline): Update now that
26560 newlines are partially supported.
26561 (selftest::test_fixit_insert_containing_newline_2): New test.
26562 (selftest::test_fixit_replace_containing_newline): Update comments.
26563 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26564 * edit-context.c (class added_line): New class.
26565 (class edited_line): Describe newline handling in comment.
26566 (edited_line::actually_edited_p): New method.
26567 (edited_line::print_content): Delete redundant decl.
26568 (edited_line::m_predecessors): New field.
26569 (edited_file::print_content): Call edited_line::print_content.
26570 (edited_file::print_diff): Update to support newlines.
26571 (edited_file::print_diff_hunk): Likewise.
26572 (edited_file::print_run_of_changed_lines): New function.
26573 (edited_file::print_diff_line): Convert to...
26574 (print_diff_line): ...this.
26575 (edited_file::get_effective_line_count): New function.
26576 (edited_line::edited_line): Initialize new field m_predecessors.
26577 (edited_line::~edited_line): Clean up m_predecessors.
26578 (edited_line::apply_fixit): Handle newlines.
26579 (edited_line::get_effective_line_count): New function.
26580 (edited_line::print_content): New function.
26581 (edited_line::print_diff_lines): New function.
26582 (selftest::test_applying_fixits_insert_containing_newline): New
26584 (selftest::test_applying_fixits_replace_containing_newline): New
26586 (selftest::insert_line): New function.
26587 (selftest::test_applying_fixits_multiple_lines): Add example of
26589 (selftest::edit_context_c_tests): Call the new tests.
26591 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26593 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26594 parameter cand. Update dump information.
26595 (get_computation_cost): Update uses.
26597 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26599 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26600 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26601 (get_computation_at, rewrite_use_address): Update use of
26602 get_computation_aff.
26604 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26606 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26607 (get_computation): Delete.
26608 (get_computation_cost): Implement like get_computation_cost_at.
26609 Use get_computation_at.
26610 (get_computation_cost_at): Delete.
26611 (rewrite_use_nonlinear_expr): Use get_computation_at.
26612 (rewrite_use_compare, remove_unused_ivs): Ditto.
26614 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26616 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26618 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26620 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26621 (ivopts_global_cost_for_size): Rename parameter and update uses.
26622 (iv_ca_recount_cost): Update uses.
26623 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26624 candidates seperately in n_invs and n_cands.
26625 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26627 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26629 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26630 (find_inv_vars_cb): New.
26631 (find_depends): Renamed to ...
26632 (find_inv_vars): ... this.
26633 (add_candidate_1, force_var_cost): Call find_inv_vars.
26634 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26636 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26638 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26639 inv_vars. Add inv_exprs.
26640 (struct iv_cand): Rename depends_on to inv_vars.
26641 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26642 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26643 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26644 (dump_cand): Dump inv_vars.
26645 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26646 (record_invariant, find_depends, add_candidate_1): Ditto.
26647 (set_group_iv_cost, force_var_cost): Ditto.
26648 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26649 (get_computation_cost_at, get_computation_cost): Ditto.
26650 (determine_group_iv_cost_generic): Ditto.
26651 (determine_group_iv_cost_address): Ditto.
26652 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26653 (determine_group_iv_costs): Ditto.
26654 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26655 (iv_ca_set_remove_invariants): Renamed to ...
26656 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26657 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26658 (iv_ca_set_add_invariants): Renamed to ...
26659 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26660 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26661 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26662 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26663 (create_new_ivs): Remove useless dump.
26665 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26667 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26669 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26670 (iv_ca_set_no_cp, create_new_iv): Ditto.
26672 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26674 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26676 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26678 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26679 function tree_check2.
26681 2017-05-02 Martin Liska <mliska@suse.cz>
26683 * doc/gcov.texi: Add missing preposition.
26684 * gcov.c (function_info::function_info): Properly fill up
26685 all member variables.
26687 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26689 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26691 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26693 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26695 2017-05-02 Martin Liska <mliska@suse.cz>
26698 * lto-streamer-in.c (lto_read_tree_1): Remove
26699 LTO_STREAMER_DEBUG.
26700 * lto-streamer.c (struct tree_hash_entry): Likewise.
26701 (struct tree_entry_hasher): Likewise.
26702 (tree_entry_hasher::hash): Likewise.
26703 (tree_entry_hasher::equal): Likewise.
26704 (lto_streamer_init): Likewise.
26705 (lto_orig_address_map): Likewise.
26706 (lto_orig_address_get): Likewise.
26707 (lto_orig_address_remove): Likewise.
26708 * lto-streamer.h: Likewise.
26709 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26710 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26712 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26714 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26715 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26716 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26717 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26718 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26719 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26720 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26721 (mm_maskz_sub_ss): New intrinsics.
26722 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26723 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26724 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26725 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26726 (__builtin_ia32_subss_mask_round): New builtins.
26727 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26728 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26729 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26731 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26732 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26734 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26737 2017-05-02 Martin Jambor <mjambor@suse.cz>
26739 PR tree-optimization/78687
26740 * tree-sra.c (access): New field parent.
26741 (process_subtree_disqualification): New function.
26742 (disqualify_candidate): Call it.
26743 (build_accesses_from_assign): Reset write flag if creating an
26745 (build_access_subtree): Fill in parent field and also prpagate
26746 down grp_write flag.
26747 (create_artificial_child_access): New parameter set_grp_write, set
26748 grp_write to its value.
26749 (propagate_subaccesses_across_link): Also propagate grp_write flag
26751 (propagate_all_subaccesses): Push the closest parent back to work
26752 queue if add_access_to_work_queue returned true.
26754 2017-05-02 Richard Biener <rguenther@suse.de>
26756 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26757 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26758 -fstrict-overflow documentation.
26759 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26760 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26761 flag_strict_overflow.
26762 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26763 * lto-opts.c (lto_write_options): Do not stream it.
26764 * lto-wrapper.c (merge_and_complain): Do not handle it.
26765 * opts.c (default_options_table): Do not set -fstrict-overflow.
26766 (finish_options): Likewise do not clear it when sanitizing.
26767 * simplify-rtx.c (simplify_const_relational_operation): Do not
26768 test flag_strict_overflow.
26770 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26772 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26773 using enabled attribute.
26774 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26775 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26776 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26777 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26778 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26779 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26780 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26781 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26782 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26783 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26785 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26787 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26789 2017-05-02 Richard Biener <rguenther@suse.de>
26791 PR tree-optimization/80591
26793 2017-04-10 Richard Biener <rguenther@suse.de>
26795 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26798 2017-05-02 Richard Biener <rguenther@suse.de>
26800 PR tree-optimization/80549
26801 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26802 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26803 headers if they do not have a preheader.
26805 2017-05-02 Martin Liska <mliska@suse.cz>
26808 * common.opt: Fix typo.
26809 * doc/invoke.texi: Likewise.
26811 2017-05-01 Jan Beulich <jbeulich@suse.com>
26813 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26814 swapping, add (x,x,m,x,n) alternative.
26816 2017-05-01 Nathan Sidwell <nathan@acm.org>
26818 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26819 unnecessary unadjusted_alignment check.
26821 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26824 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26826 * gimplify.c (gimplify_cilk_detach): New function.
26827 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26828 * tree-core.h: Document EXPR_CILK_SPAWN.
26829 * tree.h (EXPR_CILK_SPAWN): Define.
26831 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26833 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26834 to use new fixit_hint representation, using the "replace" logic.
26835 (get_line_span_for_fixit_hint): Likewise.
26836 (layout::print_any_fixits): Likewise.
26837 (selftest::test_one_liner_many_fixits): Rename to...
26838 (selftest::test_one_liner_many_fixits_1): ...this, and update
26839 comment and expected output to reflect that the multiple fix-it
26840 hints are now consolidated into one insertion.
26841 (selftest::test_one_liner_many_fixits_2): New test.
26842 (selftest::test_diagnostic_show_locus_one_liner): Update for
26844 (selftest::test_fixit_consolidation): Update for fix-it API
26846 * diagnostic.c (print_parseable_fixits): Likewise.
26847 * edit-context.c (edited_line::m_line_events): Convert from
26848 auto_vec <line_event *> to auto_vec <line_event>.
26849 (class line_event): Convert from abstract base class to a concrete
26850 class, taking over the role of replace_event.
26851 (class insert_event): Delete.
26852 (class replace_event): Rename to class line_event. Convert to
26854 (edit_context::add_fixits): Reimplement.
26855 (edit_context::apply_insert): Delete.
26856 (edit_context::apply_replace): Rename to...
26857 (edit_context::apply_fixit): ...this. Convert to half-open range.
26858 (edited_file::apply_insert): Delete.
26859 (edited_file::apply_replace): Rename to...
26860 (edited_file::apply_fixit): ...this.
26861 (edited_line::~edited_line): Drop deletion of events.
26862 (edited_line::apply_insert): Delete.
26863 (edited_line::apply_replace): Rename to...
26864 (edited_line::apply_fixit): ...this. Convert to half-open range.
26865 Update for change to type of m_line_events.
26866 * edit-context.h (edit_context::apply_insert): Delete.
26867 (edit_context::apply_replace): Rename to...
26868 (edit_context::apply_fixit): ...this.
26870 2017-05-01 Martin Sebor <msebor@redhat.com>
26872 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26875 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26878 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26879 __get_cpuid_max returns 0.
26880 (__get_cpuid_count): Ditto.
26882 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26884 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26885 replacement expression is another instance of one of its arguments.
26887 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26890 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26891 check for stack push/pop autoinc.
26892 * config/i386/i386.c (ix86_agi_dependent): Return false
26893 if the only reason why modified_in_p returned true is that
26894 addr is SP based and set_insn is a push or pop.
26896 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26898 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26901 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26904 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26905 (account_size_time): Use two predicates - exec_pred and
26907 (evaluate_conditions_for_known_args): Compute both clause and
26909 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26910 (inline_summary_t::duplicate): Update.
26911 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26913 (compute_inline_parameters): Likewise.
26914 (estimate_edge_size_and_time): Update caluclation of time.
26915 (estimate_node_size_and_time): Compute both time and nonspecialized
26917 (estimate_ipcp_clone_size_and_time): Update.
26918 (inline_merge_summary): Update.
26919 (do_estimate_edge_time): Update.
26920 (do_estimate_edge_size): Update.
26921 (do_estimate_edge_hints): Update.
26922 (inline_read_section, inline_write_summary): Stream both new predicates.
26923 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26925 (compute_inlined_call_time): Cleanup.
26926 (big_speedup_p): Update.
26927 (edge_badness): Update.
26928 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26929 (size_time_entry): Replace predicate by exec_predicate and
26930 nonconst_predicate.
26931 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26932 (estimate_edge_time): Return also nonspec_time.
26933 (reset_edge_growth_cache): Update.
26935 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26937 PR rtl-optimization/80491
26938 * ifcvt.c (noce_process_if_block): When looking for x setter
26939 with missing else_bb, don't check only the insn right before
26940 cond_earliest, but look for the last insn that x is modified in
26941 within the same bb.
26943 PR rtl-optimization/80491
26944 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26946 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26948 PR tree-optimization/80487
26949 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26951 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26953 PR tree-optimization/79697
26954 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26955 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26956 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26958 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26959 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26961 2017-04-28 Martin Sebor <msebor@redhat.com>
26963 PR tree-optimization/80523
26964 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26965 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26967 (maybe_warn, format_directive, parse_directive): Use new functions.
26968 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26970 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26972 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26974 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26976 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26977 target_header_dir): Set correctly.
26978 * configure: Regenerated.
26979 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26980 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26981 instead of SYSTEM_HEADER_DIR.
26983 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26985 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26986 (estimate_local_effects): Likewise.
26987 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26988 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26989 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26990 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26991 do_estimate_edge_time, estimate_edge_time): Likewise.
26992 * ipa-inline-analysis.c (estimate_node_size_and_time,
26993 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26994 (estimate_time_after_inlining): Remove.
26996 2017-04-28 Martin Liska <mliska@suse.cz>
26998 * doc/gcov.texi: Enhance documentation of gcov.
27000 2017-04-28 Martin Liska <mliska@suse.cz>
27002 * doc/gcov.texi: Sort options in alphabetic order.
27003 * doc/gcov-dump.texi: Likewise.
27004 * doc/gcov-tool.texi: Likewise.
27005 * gcov.c (print_usage): Likewise.
27006 * gcov-dump.c (print_usage): Likewise.
27007 * gcov-tool.c (print_merge_usage_message): Likewise.
27008 (print_rewrite_usage_message): Likewise.
27009 (print_overlap_usage_message): Likewise.
27011 2017-04-28 Martin Liska <mliska@suse.cz>
27013 PR gcov-profile/53915
27014 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27016 2017-04-28 Martin Liska <mliska@suse.cz>
27018 PR gcov-profile/79891
27019 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27020 is marked by compiler as living on a line.
27021 (get_cycles_count): Remove usage of the union.
27022 (output_intermediate_file): Likewise.
27023 (find_source): Fix GNU coding style.
27024 (accumulate_line_counts): Remove old non-all block mode.
27025 (output_lines): Remove usage of the union.
27026 * profile.c (output_location): Include all BBs, even if
27027 belonging to a same line (and file) as a previous BB.
27029 2017-04-28 Martin Liska <mliska@suse.cz>
27031 * gcov.c (process_args): Handle new argument 'w'.
27032 (read_graph_file): Assign ID to BBs.
27033 (output_branch_count): Display BB # if verbose flag is set.
27034 (output_lines): Likewise for arcs.
27035 (print_usage): Add '--verbose' option help.
27036 * doc/gcov.texi: Document --verbose (-w) option.
27038 2017-04-28 Martin Liska <mliska@suse.cz>
27040 * gcov.c (struct block_location_info): New struct.
27041 (process_file): Fill up the new structure.
27042 (read_graph_file): Replace usage of encoding by the newly added
27044 (add_line_counts): Likewise.
27045 (accumulate_line_counts): Remove usage of the union.
27046 (function_info::function_info): New function.
27047 (function_info::~function_info): Likewise.
27048 (process_file): Call delete instead of release_function.
27049 (release_function): Release the function.
27050 (release_structures): Call delete instead of release_function.
27051 (solve_flow_graph): Replace usage of num_blocks.
27052 (find_exception_blocks): Likewise.
27053 (output_lines): Fix GNU coding style.
27055 2017-04-28 Martin Liska <mliska@suse.cz>
27058 * coverage.c (coverage_remove_note_file): New function.
27059 * coverage.h: Declare the function.
27060 * toplev.c (finalize): Clean if an error has been seen.
27062 2017-04-28 Martin Liska <mliska@suse.cz>
27064 PR gcov-profile/80031
27065 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27066 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27067 * gcov.c (read_graph_file): Read just number of blocks.
27068 * profile.c (branch_prob): Do not stream 0 flags per a basic
27071 2017-04-28 Martin Liska <mliska@suse.cz>
27073 * gcov-dump.c (tag_*): Add new argument to declarations.
27074 (dump_gcov_file): Likewise.
27075 (tag_blocks): Add and use new argument depth.
27076 (tag_arcs): Likewise.
27077 (tag_lines): Likewise.
27078 (tag_counters): Likewise.
27079 (tag_summary): Likewise.
27080 (dump_working_sets): Use depth to do a proper indentation.
27082 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27085 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27086 * symtab.c (symtab_node::debug_symtab): Move definition here.
27088 2017-04-28 Richard Biener <rguenther@suse.de>
27090 * lto-streamer.h (LTO_major_version): Bump to 7.
27092 2017-04-28 Richard Biener <rguenther@suse.de>
27094 * tree-vrp.c (assert_info): New struct.
27095 (add_assert_info): New helper.
27096 (register_edge_assert_for_2): Refactor to add asserts to a vector
27098 (register_edge_assert_for_1): Likewise.
27099 (register_edge_assert_for): Likewise.
27100 (finish_register_edge_assert_for): New helper actually registering
27101 asserts where live on edge.
27102 (find_conditional_asserts): Adjust.
27103 (find_switch_asserts): Likewise.
27104 (evrp_dom_walker::try_find_new_range): Generalize.
27105 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27107 2017-04-27 Marek Polacek <polacek@redhat.com>
27110 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27111 arg10 and arg11 to itype.
27113 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27115 * doc/extend.texi (Object Size Checking): Improve grammar.
27117 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27120 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27121 that the logic for permitting reciprocal estimates matches that
27124 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27127 * tree.c (type_cache_hasher::equal): Only compare
27128 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27129 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27130 non-aggregate element types.
27131 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27132 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27134 2017-04-27 Richard Biener <rguenther@suse.de>
27136 PR middle-end/80533
27137 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27138 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27139 keeping a reference to a trailing array.
27141 2017-04-27 Richard Biener <rguenther@suse.de>
27143 PR middle-end/80539
27144 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27145 being in loop-closed SSA form conservatively.
27146 (chrec_fold_multiply_poly_poly): Likewise.
27148 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27150 PR middle-end/79665
27151 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27152 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27154 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27157 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27158 (aarch64_function_arg_alignment): Return unsigned int again, but still
27159 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27160 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27161 Don't emit -Wpsabi note.
27162 (aarch64_function_arg_boundary): Likewise.
27163 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27166 2017-04-26 Nathan Sidwell <nathan@acm.org>
27168 * tree.h (crc32_unsigned_n): Declare.
27169 (crc32_unsigned, crc32_unsigned): Make inline.
27170 * tree.c (crc32_unsigned_bits): Replace with ...
27171 (crc32_unsigned_n): ... this.
27172 (crc32_unsigned, crc32_byte): Remove.
27173 (crc32_string): Remove unnecessary braces.
27175 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27177 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27178 * ipa-inline-analysis.c (MAX_TIME): Remove.
27179 (account_size_time): Use sreal for time.
27180 (dump_inline_summary): Update.
27181 (estimate_function_body_sizes): Update.
27182 (estimate_edge_size_and_time): Update.
27183 (estimate_calls_size_and_time): Update.
27184 (estimate_node_size_and_time): Update.
27185 (inline_merge_summary): Update.
27186 (inline_update_overall_summary): Update.
27187 (estimate_time_after_inlining): Update.
27188 (inline_read_section): Update.
27189 (inline_write_summary): Update.
27190 * ipa-inline.c (compute_uninlined_call_time): Update.
27191 (compute_inlined_call_time): Update.
27192 (recursive_inlining): Update.
27193 (inline_small_functions): Update.
27194 (dump_overall_stats): Update.
27195 * ipa-inline.h: Include sreal.h.
27196 (size_time_entry): Turn time to sreal.
27197 (inline_summary): Turn self_time nad time to sreal.
27199 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27201 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27203 (sreal::stream_out, sreal::stream_in): New.
27204 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27206 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27208 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27211 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27214 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27216 (dimode_scalar_chain::compute_convert_gain): Ditto.
27217 (dimode_scalar_chain::make_vector_copies): Ditto.
27218 (dimode_scalar_chain::convert_reg): Ditto.
27219 (dimode_scalar_chain::convert_insn): Ditto.
27220 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27221 (VI248_AVX512BW_1): New mode iterator.
27222 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27223 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27226 2017-04-25 Martin Sebor <msebor@redhat.com>
27228 PR tree-optimization/80497
27229 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27230 constants are representable in HOST_WIDE_INT.
27231 (parse_directive): Ditto.
27233 2017-04-25 Martin Sebor <msebor@redhat.com>
27236 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27237 (new_zero_array): Adjust signature.
27238 (dom_info::dom_init): Used unsigned rather that size_t.
27239 (dom_info::dom_info): Same.
27241 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27242 Jakub Jelinek <jakub@redhat.com>
27245 * config/arm/arm.c: Include gimple.h.
27246 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27247 returns negative, increment ncrn only if it returned positive.
27248 (arm_needs_doubleword_align): Return int instead of bool,
27249 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27250 members, but if there is any such non-FIELD_DECL
27251 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27252 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27253 returns negative, increment nregs only if it returned positive.
27254 (arm_setup_incoming_varargs): Likewise.
27255 (arm_function_arg_boundary): Emit -Wpsabi note if
27256 arm_needs_doubleword_align returns negative, return
27257 DOUBLEWORD_ALIGNMENT only if it returned positive.
27259 2017-04-25 Marek Polacek <polacek@redhat.com>
27262 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27263 first argument to type.
27265 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27268 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27269 type checks to test for compatibility instead of equality.
27271 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27272 Jakub Jelinek <jakub@redhat.com>
27275 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27277 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27278 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27279 the alignment computation, but return their maximum in warn_alignment.
27280 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27281 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27283 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27284 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27287 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27289 * config/arc/simdext.md (dmpyh): Fix typo.
27291 2017-04-25 Richard Biener <rguenther@suse.de>
27293 PR tree-optimization/80492
27294 * alias.c (compare_base_decls): Handle registers with asm
27295 specification conservatively.
27296 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27297 compare_base_decls returning dont-know properly.
27299 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27301 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27302 (legitimate_offset_address_p): New function.
27303 (arc_legitimate_address_p): Use above function.
27305 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27307 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27309 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27311 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27312 ACCH registers whenever they are available.
27314 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27316 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27317 double regs fix when not used.
27319 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27321 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27323 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27324 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27326 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27328 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27329 when emitting short ADD instructions.
27331 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27333 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27335 (cmpsi_cc_c_insn): Likewise.
27336 (cbranchsi4_scratch): Compute proper instruction length using
27337 compact_hreg_operand.
27338 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27340 2017-04-25 Richard Biener <rguenther@suse.de>
27342 PR middle-end/80509
27343 * passes.c (pass_manager::pass_manager): Initialize
27344 m_name_to_pass_map.
27346 2017-04-25 Richard Biener <rguenther@suse.de>
27348 PR tree-optimization/79201
27349 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27351 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27354 * config/s390/vector.md: Split MEM->GPR vector moves for
27355 non-s_operand addresses.
27357 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27360 * config/s390/predicates.md (reload_const_wide_int_operand): New
27362 * config/s390/s390.md ("movti"): Remove d/P alternative.
27363 ("movti_bigconst"): New pattern definition.
27365 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27368 * s390-protos.h (s390_expand_cs_hqi): Removed.
27369 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27370 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27371 modes as well as CCZ1mode and CCZmode.
27372 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27373 signature of s390_emit_compare_and_swap.
27374 (s390_expand_cs_hqi): Likewise, make static.
27375 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27376 compare-and-swap, in some cases.
27377 (s390_expand_cs): Wrapper function.
27378 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27380 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27381 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27382 patterns for small and large integers. Forbid symref memory operands.
27383 Move expander to s390.c. Require cc register.
27384 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27385 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27386 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27387 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27388 symref memory operands. Remove CC mode and call s390_match_ccmode
27390 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27392 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27394 * config/s390/s390.md (define_peephole2): New peephole to help
27395 combining the load-and-test pattern with volatile memory.
27397 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27399 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27400 with CCZmode for TARGET_Z196.
27402 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27404 PR rtl-optimization/80501
27405 * combine.c (make_compound_operation_int): Set subreg_code to SET
27406 even for AND with mask of the sign bit of mode.
27408 PR rtl-optimization/80500
27409 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27410 sum's initial value.
27412 2017-04-25 Julian Brown <julian@codesourcery.com>
27413 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27415 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27417 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27419 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27421 2017-04-25 Julian Brown <julian@codesourcery.com>
27422 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27424 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27425 (thunderx2t99_sha): New Reservation.
27427 2017-04-25 Julian Brown <julian@codesourcery.com>
27428 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27430 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27431 type for 1-element load.
27433 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27435 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27437 2017-04-24 Martin Jambor <mjambor@suse.cz>
27439 PR tree-optimization/80293
27440 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27441 char arrays not totally scalarizable if it is false.
27442 (analyze_all_variable_accesses): Pass correct value in the new
27443 parameter. Add a statistics counter.
27445 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27447 PR middle-end/79931
27448 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27450 2017-04-24 Richard Biener <rguenther@suse.de>
27452 PR tree-optimization/80494
27453 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27454 out for complex types.
27456 2017-04-24 Richard Biener <rguenther@suse.de>
27458 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27459 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27460 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27461 (DFS): Adjust and never fail.
27462 (sccvn_dom_walker::fail): Remove.
27463 (sccvn_dom_walker::before_dom_children): Adjust.
27464 (run_scc_vn): Likewise and never fail.
27465 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27466 (pass_fre::execute): Likewise.
27468 2017-04-24 Richard Biener <rguenther@suse.de>
27470 PR tree-optimization/79725
27471 * tree-ssa-sink.c (statement_sink_location): Return whether
27472 failure reason was zero uses. Move that check later.
27473 (sink_code_in_bb): Deal with zero uses by removing the stmt
27476 2017-04-24 Richard Biener <rguenther@suse.de>
27479 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27480 pointer-based references.
27482 2017-04-24 Richard Biener <rguenther@suse.de>
27485 * pass_manager.h (pass_manager::operator new): Remove.
27486 (pass_manager::operator delete): Likewise.
27487 * passes.c (pass_manager::operator new): Remove.
27488 (pass_manager::operator delete): Likewise.
27489 (pass_manager::pass_manager): Zero individual pass members.
27491 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27494 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27495 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27496 Check "XEXP (src, 1)" operand here.
27497 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27498 Check "XEXP (src, 1)" operand here.
27499 (dimode_scalar_chain::make_vector_copies): Detect count register
27500 of a shift instruction. Zero extend count register from QImode
27501 to DImode to satisfy vector shift pattern count operand predicate.
27502 Substitute vector shift count operand with a DImode copy.
27503 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27506 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27508 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27509 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27510 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27511 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27512 (UNSPEC_NOREX_MEM): Remove definition.
27514 2017-04-21 Richard Biener <rguenther@suse.de>
27516 PR tree-optimization/79547
27517 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27518 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27519 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27520 without any constraints.
27522 2017-04-21 Richard Biener <rguenther@suse.de>
27524 PR tree-optimization/78847
27525 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27527 2017-04-21 Richard Biener <rguenther@suse.de>
27529 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27530 (build_distinct_type_copy): Likewise.
27531 (build_variant_type_copy): Likewise.
27532 * tree.c (build_qualified_type): Pass down mem-stat info.
27533 (build_distinct_type_copy): Likewise.
27534 (build_variant_type_copy): Likewise.
27536 2017-04-21 Richard Biener <rguenther@suse.de>
27538 PR tree-optimization/80237
27539 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27541 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27542 for a simplified result.
27544 2016-04-21 Richard Biener <rguenther@suse.de>
27546 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27547 sth as strict as a simple_iv but a chrec without symbols and an
27548 operand defined in the loop we are peeling (and not some subloop).
27549 (propagate_constants_for_unrolling): Propagate all constants.
27551 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27554 * config/i386/i386.c (print_reg): Remove assert for disalowed
27555 regno values, call output_operand_lossage instead.
27557 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27560 * config/i386/constraints.md (Yc): New register constraint.
27561 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27562 Use Yc constraint for alternative 2 of operand 0. Remove
27563 preferred_for_speed attribute.
27565 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27567 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27568 lastprivate clauses in SIMT case.
27570 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27572 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27574 2017-04-20 Richard Biener <rguenther@suse.de>
27576 PR tree-optimization/57796
27577 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27578 as N scalar stores.
27579 (vect_model_load_cost): Cost gathers as N scalar loads.
27581 2017-04-20 Richard Biener <rguenther@suse.de>
27583 * ggc-page.c (ggc_allocated_p): Rename to ...
27584 (safe_lookup_page_table_entry): ... this and return the lookup
27586 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27588 2017-04-20 Richard Biener <rguenther@suse.de>
27590 PR tree-optimization/80453
27591 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27592 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27593 from the conditions.
27594 (vn_phi_eq): Pass them down.
27595 (vn_phi_lookup): Record them.
27596 (vn_phi_insert): Likewise.
27598 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27600 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27601 uninitialized variable warning to avoid buffer overrun.
27603 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27606 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27607 is suppressed for '{ 0 }' in C.
27609 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27611 * BASE-VER: Set to 8.0.0.
27613 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27615 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27616 priority .init_array and .fini_array section with SECTION_NOTYPE
27619 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27621 PR middle-end/80423
27622 * tree.h (build_array_type): Add typeless_storage default argument.
27623 * tree.c (type_cache_hasher::equal): Also compare
27624 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27625 (build_array_type): Add typeless_storage argument, set
27626 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27628 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27629 (build_array_type): Likewise. Add typeless_storage argument.
27631 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27632 Jakub Jelinek <jakub@redhat.com>
27634 PR tree-optimization/80426
27635 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27636 operation on symbolic operands, also compute the overflow for the
27637 invariant part when the operation degenerates into a negation.
27639 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27642 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27643 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27646 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27648 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27651 * config/avr/avr.c (tree.h): Include it.
27652 (cgraph.h): Include it.
27653 (avr_encode_section_info): Don't warn for uninitialized progmem
27654 variable if it's just an alias.
27656 2017-04-19 Richard Biener <rguenther@suse.de>
27659 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27660 when needed by AutoPGO.
27662 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27665 * doc/lto.texi: Remove an extra 'that'.
27667 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27669 PR rtl-optimization/80429
27670 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27671 are only used in debug insns.
27673 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27674 Vladimir Makarov <vmakarov@redhat.com>
27676 * config/sparc/predicates.md (input_operand): Add comment. Return
27677 true for any memory operand when LRA is in progress.
27678 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27680 2017-04-18 Jeff Law <law@redhat.com>
27683 * mips.md ({return,simple_return}_internal): Do not overwrite
27686 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27688 PR tree-optimization/80443
27689 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27690 instead of adding 1, subtract -1 and similarly instead of subtracting
27693 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27695 PR rtl-optimization/80357
27696 * haifa-sched.c (tmp_bitmap): New variable.
27697 (model_recompute): Handle duplicate use records.
27698 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27699 (free_global_sched_pressure_data): Free it.
27701 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27704 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27705 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27706 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27707 instead of SYSTEM_HEADER_DIR.
27709 2017-04-18 Jeff Law <law@redhat.com>
27711 PR middle-end/80422
27712 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27713 predecessors after walking up the insn chain.
27715 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27718 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27719 sizetype type into debug info.
27721 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27724 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27725 unneeded test for TARGET_UPPER_REGS_SF.
27726 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27728 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27731 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27732 instead of gsi_after_labels.
27734 2017-04-18 Jeff Law <law@redhat.com>
27736 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27740 2017-04-13 Jeff Law <law@redhat.com>
27741 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27742 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27744 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27747 * config/avr/avr.c (intl.h): Include it.
27748 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27750 2017-04-18 Martin Liska <mliska@suse.cz>
27752 PR gcov-profile/78783
27753 * gcov-tool.c (gcov_output_files): Validate that destination
27754 file is either removed by the tool or by a user.
27756 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27757 Guy Benyei <guybe@mellanox.com>
27759 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27760 block, and do not negate it, the stored id is already negative.
27762 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27764 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27766 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27769 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27770 masks of options that should be turned off if the VSX vector
27771 options are turned off.
27772 (OTHER_P8_VECTOR_MASKS): Likewise.
27773 (OTHER_VSX_VECTOR_MASKS): Likewise.
27774 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27775 rs6000_disable_incompatible_switches to validate no type switches
27777 (rs6000_incompatible_switch): New function to disallow turning on
27778 other vector options if -mno-vsx, -mno-power8-vector, or
27779 -mno-power9-vector are specified.
27781 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27783 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27785 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27787 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27788 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27789 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27790 (ARG_POINTER_CFA_OFFSET): Likewise.
27792 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27794 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27795 conditions to take advantage of various optimizations.
27797 2017-04-13 Jeff Law <law@redhat.com>
27799 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27800 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27801 (zero_extendsidi2_dext): Likewise.
27803 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27806 * fold-const.c (fold_ternary_loc): Revert
27807 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27810 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27812 PR rtl-optimization/80343
27813 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27814 new scratch pseudo.
27816 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27819 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27820 to ubsan_encode_value.
27822 2017-04-13 Jeff Law <law@redhat.com>
27824 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27825 appearing in DEBUG_INSNs.
27827 2017-04-13 Martin Liska <mliska@suse.cz>
27829 PR gcov-profile/80413
27830 * gcov-io.c (gcov_write_string): Copy to buffer just when
27831 allocated size is greater than zero.
27833 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27836 * dwarf2out.c (decls_for_scope): Ignore declarations of
27837 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27839 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27842 * ipa-visibility.c (non_local_p): Fix typos.
27843 (localize_node): When localizing symbol in same comdat group,
27844 dissolve the group only when we know external symbols are going
27846 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27848 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27850 PR tree-optimization/79390
27851 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27852 order does not result in usable sequence, retry with reversed operand
27858 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27859 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27860 op0 instead of fold_convert_loc (loc, type, arg0).
27862 2017-04-12 Jeff Law <law@redhat.com>
27864 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27865 has a delay slot in the generated code.
27867 * config/cris/cris.md (cris_preferred_reload_class): Return
27868 GENNONACR_REGS rather than GENERAL_REGS.
27870 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27873 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27874 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27875 signedness of the result type.
27877 2017-04-12 Richard Biener <rguenther@suse.de>
27878 Jeff Law <law@redhat.com>
27880 PR tree-optimization/80359
27881 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27882 trim stores to TARGET_MEM_REFs.
27884 2017-04-12 Richard Biener <rguenther@suse.de>
27886 PR tree-optimization/79390
27887 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27888 threading case even more.
27890 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27893 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27894 for quad_address_p for TImode, instead of just not indexed_address.
27896 2017-04-12 Richard Biener <rguenther@suse.de>
27897 Bernd Edlinger <bernd.edlinger@hotmail.de>
27899 PR middle-end/79671
27900 * alias.c (component_uses_parent_alias_set_from): Handle
27901 TYPE_TYPELESS_STORAGE.
27902 (get_alias_set): Likewise.
27903 * tree-core.h (tree_type_common): Add typeless_storage flag.
27904 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27905 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27906 for types containing members with TYPE_TYPELESS_STORAGE.
27907 (place_field): Likewise.
27908 (layout_type): Likewise for ARRAY_TYPE.
27909 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27910 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27911 TYPE_TYPELESS_STORAGE.
27912 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27914 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27917 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27918 first argument to type.
27920 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27924 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27925 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27926 (rs6000_expand_binop_builtin): Likewise.
27927 (rs6000_expand_ternop_builtin): Likewise; also add missing
27928 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27929 vshasigma built-ins.
27930 * doc/extend.texi: Document that vec_xxpermdi's third argument
27931 must be a constant.
27933 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27935 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27936 Use shift_const cost parameter when calculating gain of STV shifts.
27938 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27940 PR rtl-optimization/70478
27941 * lra-constraints.c (process_alt_operands): Check memory for
27942 disfavoring memory insn operand.
27944 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27946 PR middle-end/80100
27947 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27948 left shift in unsigned HOST_WIDE_INT type.
27950 PR rtl-optimization/80385
27951 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27952 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27955 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27956 if they have any depend clauses.
27958 2017-04-11 Martin Liska <mliska@suse.cz>
27961 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27962 * ipa-split.c (split_function): Create a local comdat symbol
27963 if caller is in a comdat group.
27965 2017-04-11 Martin Liska <mliska@suse.cz>
27968 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27971 2017-04-11 Martin Sebor <msebor@redhat.com>
27973 PR middle-end/80364
27974 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27975 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27977 (directive::set_width, directive::set_precision, format_character):
27979 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27982 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27985 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27986 conflict, set target->arch_name instead of target->cpu_name.
27988 2017-04-11 Richard Biener <rguenther@suse.de>
27990 PR tree-optimization/80374
27991 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27992 build_zero_cst, remove fold_convertible_p check again.
27994 2017-04-11 Martin Liska <mliska@suse.cz>
27997 * ubsan.c (instrument_object_size): Do not instrument register
28000 2017-04-11 Jakub Jelinek <jakub@redhat.com>
28003 * config/i386/i386-builtin-types.def
28004 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28005 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28006 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28007 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28008 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28009 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28010 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28011 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28012 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28013 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28014 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28015 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28016 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28017 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28018 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28019 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28020 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28021 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28022 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28023 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28024 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28025 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28026 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28027 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28028 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28029 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28030 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28031 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28032 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28033 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28034 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28035 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28036 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28037 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28038 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28039 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28040 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28041 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28042 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28043 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28044 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28045 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28046 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28047 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28048 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28049 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28051 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28052 flag to second_arg_count, handle 4 argument function type _COUNT
28053 aliases, handle second_arg_count on second argument rather than last.
28055 2017-04-10 Jeff Law <law@redhat.com>
28057 PR tree-optimization/80374
28058 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28059 record anything if we can not convert integer_zero_node to the
28062 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28065 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28066 Enhance special handling given to the TARGET_P9_MINMAX option in
28067 relation to certain other options.
28069 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28071 PR tree-optimization/80153
28072 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28073 remove POINTER_PLUS_EXPR's base part directly, rather than through
28076 2017-04-10 Richard Biener <rguenther@suse.de>
28077 Bin Cheng <bin.cheng@arm.com>
28079 PR tree-optimization/80153
28080 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28081 the first element of pointer type aff_tree. Build result expr in
28083 (add_elt_to_tree): Convert to type unconditionally. Remove other
28084 fold_convert calls.
28085 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28086 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28088 2017-04-10 Richard Biener <rguenther@suse.de>
28090 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28093 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28095 PR rtl-optimization/70478
28096 * lra-constraints.c (curr_small_class_check): New.
28097 (update_and_check_small_class_inputs): New.
28098 (process_alt_operands): Update curr_small_class_check. Disfavor
28099 alternative insn memory operands. Check available regs for small
28102 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28105 * config/mips/mips.opt (-mvirt): Update description.
28106 * doc/invoke.texi (-mvirt): Likewise.
28108 2017-04-10 Richard Biener <rguenther@suse.de>
28110 PR middle-end/80362
28111 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28112 looking for NEGATE_EXPR in -A / -B to A / B folding.
28114 2017-04-10 Martin Liska <mliska@suse.cz>
28116 PR gcov-profile/80224
28117 * gcov.c (print_usage): Fix usage string.
28118 (get_gcov_intermediate_filename): Remove.
28119 (output_gcov_file): Use both for normal and intermediate format.
28120 (generate_results): Do not initialize special file for
28121 intermediate format.
28123 2017-04-10 Richard Biener <rguenther@suse.de>
28125 PR tree-optimization/80304
28126 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28129 2017-04-10 Nathan Sidwell <nathan@acm.org>
28132 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28133 (rs6000_init_builtins): Use it.
28135 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28137 * config/arm/arm.md (<mrc>): Add mode to SET source.
28138 (<mrrc>): Likewise.
28140 2017-04-10 Richard Biener <rguenther@suse.de>
28142 PR middle-end/80344
28143 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28145 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28148 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28149 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28150 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28151 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28152 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28153 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28154 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28155 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28156 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28157 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28158 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28159 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28160 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28161 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28162 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28163 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28164 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28165 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28166 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28167 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28168 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28169 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28170 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28172 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28174 PR rtl-optimization/70478
28175 * lra-constraints.c: Reverse the last patch.
28177 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28179 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28180 Add comment for WCHAR_T.
28182 2017-04-08 Martin Liska <mliska@suse.cz>
28185 2017-04-07 Martin Liska <mliska@suse.cz>
28188 * ipa-split.c (split_function): Add function part to a same comdat
28191 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28194 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28196 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28198 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28199 (vec_pairing): ...to this.
28200 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28201 (rs6000_sched_init): Adjust for name change.
28202 (struct rs6000_sched_context): Likewise.
28203 (rs6000_init_sched_context): Likewise.
28204 (rs6000_set_sched_context): Likewise.
28206 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28212 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28214 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28215 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28216 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28218 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28220 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28222 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28224 PR rtl-optimization/70703
28225 * ira-color.c (update_conflict_hard_regno_costs): Use
28226 int64_t instead of HOST_WIDE_INT.
28228 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28230 PR rtl-optimization/70478
28231 * lra-constraints.c (process_alt_operands): Disfavor alternative
28232 insn memory operands.
28234 2017-04-07 Jeff Law <law@redhat.com>
28236 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28237 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28239 2017-04-07 Martin Liska <mliska@suse.cz>
28242 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28243 Show error message instead of an ICE.
28245 2017-04-07 Martin Liska <mliska@suse.cz>
28248 * ipa-split.c (split_function): Add function part to a same comdat
28251 2017-04-07 Richard Biener <rguenther@suse.de>
28253 PR middle-end/80341
28254 * tree.c (get_unwidened): Also handle ! for_type case for
28256 * convert.c (do_narrow): Split out from ...
28257 (convert_to_integer_1): ... here. Do not pass final truncation
28258 type to get_unwidened for TRUNC_DIV_EXPR.
28260 2017-04-07 Richard Biener <rguenther@suse.de>
28262 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28264 (aff_combination_const): Adjust.
28265 (aff_combination_scale): Likewise.
28266 (aff_combination_add_elt): Likewise.
28267 (aff_combination_add_cst): Likewise.
28268 (aff_combination_convert): Likewise.
28269 (add_elt_to_tree): Likewise. Remove unused argument.
28270 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28272 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28274 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28276 * config/arm/arm.c (arm_default_short_enums): Use
28277 ARM_DEFAULT_SHORT_ENUMS.
28278 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28280 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28283 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28284 members with redundant out-of-class redeclaration.
28286 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28289 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28290 * config/i386/i386.md (*zero_extendsidi2):
28291 Add (?*x,*x) and (?*v,*v) alternatives.
28293 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28296 * config/i386/i386.c (ix86_expand_builtin)
28297 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28298 mode from insn data. Convert operands to insn operand mode.
28299 Copy operands that don't satisfy insn predicate to a register.
28301 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28303 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28306 2017-04-06 Richard Biener <rguenther@suse.de>
28308 PR tree-optimization/80334
28309 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28310 preserve alignment of accesses.
28312 2017-04-06 Richard Biener <rguenther@suse.de>
28314 PR tree-optimization/80262
28315 * tree-sra.c (build_ref_for_offset): Preserve address-space
28317 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28318 Drop useless address-space information on MEM_REF offsets.
28320 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28322 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28324 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28326 PR rtl-optimization/70703
28327 * ira-color.c (update_conflict_hard_regno_costs): Use
28328 HOST_WIDE_INT instead of long.
28330 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28333 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28334 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28336 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28337 for x86_64 target. Handle -m3dnowa option.
28339 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28341 PR rtl-optimization/70703
28342 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28343 (update_conflict_hard_regno_costs): Use long instead of unsigned
28344 arithmetic for cost calculation.
28346 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28347 Bernd Edlinger <bernd.edlinger@hotmail.de>
28350 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28353 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28356 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28357 ptr_mode with Pmode throughout.
28358 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28359 into probe_stack_range and use DImode.
28361 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28364 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28365 call_eh_return is true.
28367 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28369 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28370 Initialize last_match_fntype_index.
28372 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28375 * tree-nvr.c: Include internal-fn.h.
28376 (pass_return_slot::execute): Ignore internal calls without
28379 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28380 Richard Biener <rguenther@suse.de>
28383 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28384 captures used multiple times, except for the last use.
28385 * generic-match-head.c: Include gimplify.h.
28387 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28389 PR tree-optimization/79390
28390 * target.h (struct noce_if_info): Declare.
28391 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28392 * target.def (noce_conversion_profitable_p): New target hook.
28393 * ifcvt.h (struct noce_if_info): New type, moved from ...
28394 * ifcvt.c (struct noce_if_info): ... here.
28395 (noce_conversion_profitable_p): Renamed to ...
28396 (default_noce_conversion_profitable_p): ... this. No longer
28398 (noce_try_store_flag_constants, noce_try_addcc,
28399 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28400 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28401 instead of noce_conversion_profitable_p.
28402 * config/i386/i386.c: Include ifcvt.h.
28403 (ix86_option_override_internal): Don't override
28404 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28405 (ix86_noce_conversion_profitable_p): New function.
28406 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28407 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28408 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28409 * doc/tm.texi: Regenerated.
28411 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28413 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28416 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28419 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28420 instructions for small multiply cores.
28422 2017-04-04 Jeff Law <law@redhat.com>
28424 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28426 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28427 that are not set by the loop over the elements.
28429 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28432 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28433 int mode, convert_modes it to mode as unsigned, otherwise use
28434 lowpart_subreg to mode rather than SImode.
28435 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28436 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28437 Use DImode instead of SImode for the shift count operand.
28438 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28441 2017-04-04 Richard Biener <rguenther@suse.de>
28443 PR middle-end/80281
28444 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28445 arithmetic done for the negate or the plus. Simplify.
28446 (A - (-B) -> A + B): Likewise.
28447 * fold-const.c (split_tree): Make sure to not negate pointers.
28449 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28451 PR rtl-optimization/60818
28452 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28453 a compare of comparisons with the thing compared if this results
28454 in a different machine mode.
28456 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28458 * alias.c (base_alias_check): Fix typo in comment.
28459 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28460 * cgraphunit.c (symbol_table::compile): Likewise.
28461 * collect2.c (maybe_run_lto_and_relink): Likewise.
28462 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28463 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28464 * config/avr/avr.c (avr_map_op_t): Likewise.
28465 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28466 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28467 * config/epiphany/epiphany.md (movcc): Likewise.
28468 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28469 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28471 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28472 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28473 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28474 * config/sh/sh.c (sh_rtx_costs): Likewise.
28475 * fold-const.c (fold_truth_andor): Likewise.
28476 * genautomata.c (collapse_flag): Likewise.
28477 * gengtype.h (struct type::u::s): Likewise.
28478 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28479 * input.c (FORMAT_AMOUNT): Likewise.
28480 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28481 (known_aggs_to_agg_replacement_list): Likewise.
28482 * ipa-inline-analysis.c: Likewise.
28483 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28484 * ipa-polymorphic-call.c
28485 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28486 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28487 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28489 * modulo-sched.c (apply_reg_moves): Likewise.
28490 * omp-expand.c (build_omp_regions_1): Likewise.
28491 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28492 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28493 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28494 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28495 * value-prof.c: Likewise.
28496 * var-tracking.c (val_reset): Likewise.
28498 2017-04-03 Richard Biener <rguenther@suse.de>
28500 PR tree-optimization/80275
28501 * fold-const.c (split_address_to_core_and_offset): Handle
28504 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28506 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28507 descriptors is at least equal to that of functions.
28509 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28511 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28513 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28516 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28517 (mov<IMOD4:mode>): New expander.
28518 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28520 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28522 PR rtl-optimization/79405
28523 * fwprop.c (propagations_left): New variable.
28524 (forward_propagate_into): Decrement it.
28525 (fwprop_init): Initialize it.
28526 (fw_prop): If the variable has reached zero, stop propagating.
28527 (fwprop_addr): Ditto.
28529 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28532 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28533 a FUNCTION_DECL, pass it as decl instead of origin to
28536 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28538 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28541 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28544 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28545 TARGET_VSX_SMALL_INTEGER.
28547 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28549 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28550 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28552 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28554 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28555 extraction from odd-numbered MSA register.
28557 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28559 PR middle-end/80173
28560 * expmed.c (store_bit_field_1): Don't attempt to create
28561 a word subreg out of hard registers wider than word if they
28562 have HARD_REGNO_NREGS of 1 for their mode.
28564 PR middle-end/80163
28565 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28566 conversions to integer types wider than word and pointer.
28569 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28570 (rtx_equal_for_cselib_p): Pass 0 to it.
28571 * cselib.c (cselib_hasher::equal): Likewise.
28572 (rtx_equal_for_cselib_1): Add depth argument. If depth
28573 is 128, don't look up VALUE locs and punt. Increment
28574 depth in recursive calls when walking VALUE locs.
28576 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28578 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28579 (make_gcov_file_name): Use the canonical path name for generating
28581 (read_line): Fix handling of files with ascii null bytes.
28583 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28585 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28586 to initialise a vector register instead
28587 of using a const_int.
28589 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28591 PR translation/80189
28592 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28593 diagnostic messages.
28595 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28598 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28599 (dfp_diex_<mode>): Update mode of operand 1.
28600 * doc/extend.texi (dxex, dxexq): Document change to return type.
28601 (diex, diexq): Document change to argument type.
28603 2017-03-30 Martin Jambor <mjambor@suse.cz>
28606 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28607 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28608 it reflects the signature changes performed at the callee side.
28609 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28610 to cgraph_build_function_type_skip_args.
28611 (build_function_decl_skip_args): Adjust call to the above function.
28613 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28616 * config/i386/sse.md
28617 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28618 register as dest whenever it is a MEM not rtx_equal_p to the
28619 corresponding dup operand, and when forcing into reg move the
28620 reg into the memory afterwards.
28621 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28622 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28623 for the force_reg mode.
28624 (avx512vl_vextractf128<mode>): Use register as dest either
28625 always when a MEM, or when it is a MEM not rtx_equal_p to the
28626 corresponding dup operand, or even not when it is a CONST_VECTOR
28627 depending on the mode and lo vs. hi.
28628 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28630 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28631 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28632 Likewise. Require that operands[2] is even.
28633 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28634 Remove extraneous parens. Require that operands[2] is a multiple
28636 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28637 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28638 disallow memory then.
28640 2017-03-30 Richard Biener <rguenther@suse.de>
28642 PR tree-optimization/77498
28643 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28644 to non-constants over backedges.
28646 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28648 PR rtl-optimization/80233
28649 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28650 as last_combined_insn. Do not test for BARRIER_P separately.
28652 2017-03-29 Andreas Schwab <schwab@suse.de>
28655 * calls.c (prepare_call_address): Convert funexp to Pmode before
28656 copying to temp reg.
28658 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28660 PR tree-optimization/80158
28661 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28662 Handle possible future case of more than one alternate
28664 (replace_rhs_if_not_dup): Likewise.
28665 (replace_one_candidate): Likewise.
28667 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28669 PR rtl-optimization/80193
28670 * ira.c (ira): Do not check allocation for LRA.
28672 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28674 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28675 (nvptx_output_simt_exit): Declare.
28676 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28677 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28678 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28679 (nvptx_declare_function_name): ...here. Emit declaration of local
28680 memory space buffer for omp_simt_enter insn.
28681 (nvptx_output_unisimt_switch): New.
28682 (nvptx_output_softstack_switch): New.
28683 (nvptx_output_simt_enter): New.
28684 (nvptx_output_simt_exit): New.
28685 * config/nvptx/nvptx.h (struct machine_function): New fields
28686 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28687 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28688 (UNSPECV_SIMT_EXIT): Ditto.
28689 (omp_simt_enter_insn): New insn.
28690 (omp_simt_enter): New expansion.
28691 (omp_simt_exit): New insn.
28692 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28694 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28695 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28696 (expand_GOMP_SIMT_EXIT): New.
28697 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28698 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28699 (GOMP_SIMT_EXIT): Ditto.
28700 * target-insns.def (omp_simt_enter): New insn.
28701 (omp_simt_exit): Ditto.
28702 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28704 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28705 (lower_rec_input_clauses): Likewise.
28706 (lower_lastprivate_clauses): Handle SIMT privatization.
28708 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28709 (ompdevlow_adjust_simt_enter): New.
28710 (find_simtpriv_var_op): New.
28711 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28712 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28714 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28715 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28716 (copy_decl_for_dup_finish): Ditto.
28718 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28720 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28723 * config/i386/i386.c (ix86_option_override_internal): Always
28724 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28726 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28728 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28730 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28732 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28733 mark new edge's irreducible flag accordign to it.
28734 (vect_do_peeling): Check loop preheader edge's irreducible flag
28735 and pass it to function slpeel_add_loop_guard.
28737 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28739 PR tree-optimization/80218
28740 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28741 Update block frequencies and counts.
28743 2017-03-28 Richard Biener <rguenther@suse.de>
28745 PR tree-optimization/78644
28746 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28747 of a simplification result we may not use it at all.
28749 2017-03-28 Richard Biener <rguenther@suse.de>
28752 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28753 without arguments, generate default definition of a SSA name.
28755 2017-03-28 Richard Biener <rguenther@suse.de>
28757 PR middle-end/80222
28758 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28759 TYPE_REF_CAN_ALIAS_ALL references.
28760 * fold-const.c (fold_indirect_ref_1): Likewise.
28762 2017-03-28 Martin Liska <mliska@suse.cz>
28765 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28766 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28768 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28769 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28771 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28772 (EXTRA_SPECS): Define.
28773 (SUBTARGET_EXTRA_SPECS): Likewise.
28774 (SUBTARGET_CPP_SPEC): Likewise.
28775 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28776 SUBTARGET_EXTRA_SPECS.
28777 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28779 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28781 * config/arc/simdext.md (vst64_insn): Update pattern.
28782 (vld32wh_insn): Likewise.
28783 (vld32wl_insn): Likewise.
28784 (vld64_insn): Likewise.
28785 (vld32_insn): Likewise.
28787 2017-03-28 Marek Polacek <polacek@redhat.com>
28790 * fold-const.c (fold_comparison): Use protected_set_expr_location
28791 instead of SET_EXPR_LOCATION.
28793 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28795 * tree.c (add_expr): Avoid name lookup warning.
28797 2017-03-27 Jeff Law <law@redhat.com>
28799 PR tree-optimization/80216
28800 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28801 function name. Limit recursion depth.
28802 (record_temporary_equivalences): Corresponding changes.
28804 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28806 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28809 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28812 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28814 * cfgcleanup.c (reg_note_cfa_p): New array.
28815 (insns_have_identical_cfa_notes): New function.
28816 (old_insns_match_p): Don't cross-jump in between /f
28817 and non-/f instructions. If both i1 and i2 are frame related,
28818 verify all CFA notes, their order and content.
28820 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28823 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28824 HImode and SImode with zero extend to DImode to one insn.
28825 (bswap<mode>2_extenddi): Likewise.
28826 (bswapsi2_extenddi): Likewise.
28827 (bswaphi2_extendsi): Likewise.
28828 (bswaphi2): Combine bswap HImode and SImode into one insn.
28829 Separate memory insns from swapping register.
28830 (bswapsi2): Likewise.
28831 (bswap<mode>2): Likewise.
28832 (bswaphi2_internal): Delete, no longer used.
28833 (bswapsi2_internal): Likewise.
28834 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28835 store, and gpr<-gpr swap insns.
28836 (bswap<mode>2_store): Likewise.
28837 (bswaphi2_reg): Register only splitter, combine with the splitter.
28838 (bswaphi2 splitter): Likewise.
28839 (bswapsi2_reg): Likewise.
28840 (bswapsi2 splitter): Likewise.
28841 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28842 the insns into load, store, and register/register insns.
28843 (bswapdi2_ldbrx): Likewise.
28844 (bswapdi2_load): Likewise.
28845 (bswapdi2_store): Likewise.
28846 (bswapdi2_reg): Likewise.
28848 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28850 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28851 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28853 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28856 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28858 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28859 special handling for target option conflicts between dform
28860 options (-mpower9-dform, -mpower9-dform-vector,
28861 -mpower9-dform-scalar) and -mno-direct-move.
28863 2017-03-27 Richard Biener <rguenther@suse.de>
28865 PR tree-optimization/80181
28866 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28868 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28870 * config/arc/predicates.md (move_double_src_operand): Replace the
28871 call to move_double_src_operand with a call to address_operand.
28873 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28875 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28876 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28877 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28879 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28881 * config/arc/predicates.md (long_immediate_loadstore_operand):
28882 Consider scaled addresses cases.
28884 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28886 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28887 restored when in interrupt.
28888 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28889 doesn't have delay slot.
28891 2017-03-27 Richard Biener <rguenther@suse.de>
28894 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28895 inlined thunk clones.
28897 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28900 * asan.c (instrument_derefs): Copy over last operand from
28901 original COMPONENT_REF to the new COMPONENT_REF with
28902 DECL_BIT_FIELD_REPRESENTATIVE.
28903 * ubsan.c (instrument_object_size): Likewise.
28905 2017-03-27 Richard Biener <rguenther@suse.de>
28907 PR tree-optimization/80170
28908 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28909 sure DR/SCEV didnt fold in constants we do not see when looking
28910 at the reference base alignment.
28912 2017-03-27 Richard Biener <rguenther@suse.de>
28914 PR middle-end/80171
28915 * gimple-fold.c (fold_ctor_reference): Properly guard against
28916 NULL return value from canonicalize_constructor_val.
28918 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28921 * config/i386/i386.c (ix86_expand_builtin)
28922 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28923 flags reg setting and flags reg using instructions.
28924 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28925 clobbering instructions to zero extend op2.
28927 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28929 * doc/install.texi (Configuration) <--with-aix-soname>:
28930 Update link to AIX ld.
28932 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28934 PR rtl-optimization/80160
28935 PR rtl-optimization/80159
28936 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28937 reg_alternate_class into account.
28939 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28942 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28943 to consider in curr_insn_transform.
28945 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28947 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28948 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28949 and emit_mode_inner.
28951 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28953 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28954 argument to the overloaded builtin variants. Use the new flag to
28955 deprecate certain builtin variants.
28956 * config/s390/s390-builtin-types.def: Add new builtin types.
28957 * config/s390/s390-builtins.h: Support new flags field for
28958 overloaded builtins.
28959 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28960 (s390_macro_to_expand): Enable vector float data type.
28961 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28962 builtins by incrementing the __VEC__ version number.
28963 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28965 (s390_resolve_overloaded_builtin): Emit error messages depending
28966 on the builtin flags.
28967 * config/s390/s390.c (s390_expand_builtin): Support additional
28968 flags argument. Change error message to match the messages
28969 emitted in s390-c.c.
28970 * config/s390/s390.md: New UNSPEC_* constants.
28971 (op_type): Add new instruction types.
28972 * config/s390/vecintrin.h: Add new builtins and test data class
28974 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28975 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28976 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28977 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28978 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28979 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28981 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28982 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28983 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28984 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28986 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28987 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28988 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28989 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28991 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28992 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28993 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28994 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28995 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28996 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28997 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28999 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
29000 ("vec_scatter_element<V_HW_4:mode>_DI")
29001 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
29002 ("vec_fpint<mode>", "vflls")
29003 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
29004 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29005 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29006 ("*vec_cmphe<mode>_cc"): ... these.
29008 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29009 mode constant instead of magic value.
29011 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29013 * config/s390/s390.c (s390_expand_vec_compare): Support other
29014 vector floating point modes than just V2DF.
29015 (s390_expand_vcond): Likewise.
29016 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29017 (s390_cannot_change_mode_class): Prevent mode changes between TF
29018 and V1TF in vector registers.
29019 * config/s390/s390.md (DF, SF): New mode attributes.
29020 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29021 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29022 SFmode support for VRs.
29023 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29025 (VFT, VF_HW): New mode iterators.
29026 (vw, sdx): New mode attributes.
29027 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29028 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29029 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29030 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29031 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29032 also the new vector floating point modes. Renaming to ...
29034 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29035 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29036 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29037 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29038 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29039 ("vec_unordered<mode>"): ... these.
29041 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29042 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29043 ("*vec_extendv2df"): New insn definitions.
29045 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29047 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29048 ("mulditi3_2", "*muldi3_sign"): New patterns.
29049 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29050 rename the pattern definition.
29052 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29054 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29056 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29058 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29060 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29061 instruction if possible.
29062 * config/s390/vector.md (vec_halfnumelts): New mode
29064 ("*vec_vllezlf<mode>"): New pattern.
29066 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29068 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29069 ("popcountv4si2", "popcountv2di2"): Rename to ...
29070 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29071 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29073 ("popcount<mode>2_vxe"): New pattern.
29075 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29077 * common/config/s390/s390-common.c (processor_flags_table): Add
29079 * config.gcc: Add arch12.
29080 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29081 Default to arch12 for unknown CPU model numbers.
29082 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29083 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29084 PROCESSOR_max sanity check.
29085 * config/s390/s390-opts.h (enum processor_type): Add
29087 * config/s390/s390.c (processor_table): Add arch12.
29088 (s390_expand_builtin): Add check for B_VXE flag.
29089 (s390_issue_rate): Add PROCESSOR_ARCH12.
29090 (s390_get_sched_attrmask): Likewise.
29091 (s390_get_unit_mask): Likewise.
29092 (s390_sched_score): Enable z13 scheduling for arch12.
29093 (s390_sched_reorder): Likewise.
29094 (s390_sched_variable_issue): Likewise.
29095 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29097 (s390_tune_attr): Use z13 scheduling also for arch12.
29098 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29099 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29100 (TARGET_VXE_P): New macros.
29101 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29102 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29103 * config/s390/s390.opt: Add arch12 as processor_type.
29105 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29107 * config/s390/s390.md
29108 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29109 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29110 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29112 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29113 Rename expanders to ...
29115 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29116 ("fixuns_truncdddi2_emu"): ... these.
29118 ("fixuns_trunc<mode>si2_emu"): New expander.
29120 ("*fixuns_truncdfdi2_z13"): Rename to ...
29121 ("*fixuns_truncdfdi2_vx"): ... this.
29123 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29125 * config/s390/2964.md: Remove the single element vector compare
29126 instructions which are no longer used.
29127 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29129 (s390_canonicalize_comparison): Remove handling of DFmode
29131 (s390_expand_vec_compare_scalar): Remove function.
29132 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29133 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29135 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29137 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29139 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29141 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29142 will anyway by matched by mov<mode>_64dfp.
29144 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29146 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29147 vlef/vstef. Add missing operand to vleif.
29149 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29151 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29152 pair for all vector types with 64 bit elements.
29153 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29154 * config/s390/vector.md (V_HW_64): ... here.
29155 (V_128_NOSINGLE): New mode iterator.
29156 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29157 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29158 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29159 ("*vec_load_pairv2di"): Change to ...
29160 ("*vec_load_pair<mode>"): ... this one.
29162 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29164 * config/s390/constraints.md: Add comments.
29165 (jKK): Reject element sizes > 8 bytes.
29166 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29168 * config/s390/s390.md: Add the s_operand checks formerly in
29169 s390_split_ok_p to various splitters where they are still
29171 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29172 for 128 bit vectors. Plus two splitters.
29174 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29176 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29179 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29182 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29183 error if the boundary argument is not constant.
29185 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29187 PR rtl-optimization/80112
29188 * loop-doloop.c (doloop_condition_get): Don't check condition
29189 if cmp isn't SET with IF_THEN_ELSE src.
29191 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29193 PR tree-optimization/80158
29194 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29195 replacing a candidate statement, also replace it for the
29196 candidate's alternate interpretation.
29197 (replace_rhs_if_not_dup): Likewise.
29198 (replace_one_candidate): Likewise.
29200 2017-03-24 Richard Biener <rguenther@suse.de>
29202 PR tree-optimization/80167
29203 * graphite-isl-ast-to-gimple.c
29204 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29206 (translate_isl_ast_to_gimple::get_rename): Likewise.
29208 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29210 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29211 handling of certain combinations of target options, including the
29212 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29213 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29215 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29218 * config/arm/arm.md (*load_multiple): Add reload_completed to
29219 matching condition.
29221 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29222 Richard Biener <rguenth@suse.de>
29224 PR tree-optimization/79908
29225 PR tree-optimization/80136
29226 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29227 been cast away, gimplify_and_add suffices.
29229 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29231 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29233 2017-03-23 Richard Biener <rguenther@suse.de>
29235 PR tree-optimization/80032
29236 * gimplify.c (gimple_push_cleanup): Forced unconditional
29237 cleanups still have to go to the conditional_cleanups
29240 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29242 PR tree-optimization/80072
29243 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29245 (next_operand_entry_id): Change type to unsigned int.
29246 (sort_by_operand_rank): Make sure to return the right return value
29247 even if unsigned fields are bigger than INT_MAX.
29248 (struct oecount): Change cnt and id type to unsigned int.
29249 (oecount_hasher::equal): Formatting fix.
29250 (oecount_cmp): Make sure to return the right return value
29251 even if unsigned fields are bigger than INT_MAX.
29252 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29255 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29256 TREE_READONLY on result if writing it more than once.
29259 * doc/invoke.texi (-fsanitize=thread): Document that with
29260 -fnon-call-exceptions atomics are not able to throw
29264 * tsan.c: Include tree-eh.h.
29265 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29266 maybe_clean_or_replace_eh_stmt where needed.
29267 (instrument_memory_accesses): Add cfg_changed argument.
29268 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29269 if it returned true.
29270 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29272 PR rtl-optimization/63191
29273 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29274 wrapper function, moved the whole old content into ...
29275 (ix86_delegitimize_address_1): ... this. New inline function.
29276 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29277 true as last argument instead of ix86_delegitimize_address.
29279 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29281 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29282 cortexa57_branch_cost.
29284 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29286 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29288 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29291 * doc/md.texi (Constraints): Document wA constraint.
29292 * config/rs6000/constraints.md (wA): New.
29293 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29294 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29295 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29296 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29298 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29301 * gimplify.c (is_oacc_declared): New function.
29302 (oacc_default_clause): Use it to set default flags for acc declared
29303 variables inside parallel regions.
29304 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29305 declared variables.
29306 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29307 declare attribute to any decl as necessary.
29309 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29312 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29313 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29314 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29315 (arm_arch_lpae): This.
29316 * config/arm/arm.c (arm_arch7ve): Rename into ...
29317 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29318 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29321 2017-03-22 Martin Liska <mliska@suse.cz>
29324 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29325 error message instead of an ICE.
29327 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29329 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29331 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29333 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29335 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29338 2017-03-21 Martin Sebor <msebor@redhat.com>
29340 * doc/extend.texi: Use "cannot" instead of "can't."
29341 * doc/hostconfig.texi: Same.
29342 * doc/install.texi: Same.
29343 * doc/invoke.texi: Same.
29344 * doc/loop.texi: Same.
29345 * doc/md.texi: Same.
29346 * doc/objc.texi: Same.
29347 * doc/rtl.texi: Same.
29348 * doc/tm.texi: Same.
29349 * doc/tm.texi.in: Same.
29350 * doc/trouble.texi: Same.
29352 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29355 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29356 (collect_checksum_attributes): Set it.
29357 (die_checksum_ordered): Use it.
29359 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29361 PR tree-optimization/79908
29362 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29363 change: For a VA_ARG whose LHS has been cast away, use
29364 force_gimple_operand to construct the side effects.
29366 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29368 PR translation/80001
29369 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29370 more amenable to translation.
29371 (oacc_loop_auto_partitions): Likewise.
29373 2017-03-21 Marek Polacek <polacek@redhat.com>
29374 Martin Sebor <msebor@redhat.com>
29376 PR tree-optimization/80109
29377 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29378 on INTEGRAL_TYPE_P.
29380 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29381 Segher Boessenkool <segher@kernel.crashing.org>
29384 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29385 check reg_used_between_p between insn and one of succ or succ2
29386 depending on if succ is artificial insn not inserted into insn
29389 2017-03-21 Martin Liska <mliska@suse.cz>
29391 PR gcov-profile/80081
29392 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29393 * doc/gcc.texi: Include gcov-dump stuff.
29394 * doc/gcov-dump.texi: New file.
29396 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29398 PR rtl-optimization/79150
29399 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29400 conditional jump, if the jump is the last insn of the loop.
29402 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29403 Richard Biener <rguenth@suse.de>
29405 PR tree-optimization/79908
29406 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29407 been cast away, use force_gimple_operand to construct the side
29410 2017-03-21 Martin Liska <mliska@suse.cz>
29412 PR libfortran/79956
29413 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29416 2017-03-21 Brad Spengler <spender@grsecurity.net>
29419 * plugin.c (htab_hash_plugin): New function.
29420 (add_new_plugin): Use it and adjust.
29421 (parse_plugin_arg_opt): Adjust.
29422 (init_one_plugin): Likewise.
29424 2017-03-21 Richard Biener <rguenther@suse.de>
29426 PR tree-optimization/80032
29427 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29428 if set force the cleanup to happen unconditionally.
29429 (gimplify_target_expr): Push inserted clobbers with force_uncond
29430 to avoid them being removed by control-dependent DCE.
29432 2017-03-21 Richard Biener <rguenther@suse.de>
29434 PR tree-optimization/80122
29435 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29436 va_arg_pack_len when the inlined call stmt requires pack
29438 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29440 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29443 * tsan.c (instrument_builtin_call): If the memory model argument
29444 is not a constant, assume it is valid.
29447 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29450 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29452 PR rtl-optimization/79910
29453 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29454 if its dest is used by an insn before I2 (other than the combined
29455 insns themselves, which are properly handled already).
29457 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29460 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29462 * combine.c (record_used_regs): New static function.
29463 (try_combine): Handle situations where there is an additional
29464 instruction between I2 and I3 which needs to have a LOG_LINK
29468 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29470 * combine.c (try_combine): Delete redundant i1 test. Call
29471 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29473 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29476 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29477 alternatives 13/14.
29479 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29481 PR tree-optimization/80054
29482 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29483 the optimization if a PHI or any of its arguments is not dominated
29484 by the candidate's basis. Use gphi* rather than gimple* as
29486 (replace_profitable_candidates): Clean up a gimple* variable that
29487 should be a gphi* variable.
29489 2017-03-20 Martin Sebor <msebor@redhat.com>
29492 * doc/extend.texi (attribute constructor): Document present limitation.
29494 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29497 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29498 __POWER9_VECTOR__ #ifdef control, change template definition to
29499 use Power9-specific built-in function.
29500 (vec_any_eq): Likewise.
29501 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29502 to control outcomes from this test.
29503 (vector_ae_<mode>p): For VEC_F modes, likewise.
29505 2017-03-20 Ian Lance Taylor <iant@google.com>
29507 * config/i386/i386.c (ix86_function_regparm): Save an extra
29508 register for -fsplit-stack with DECL_STATIC_CHAIN.
29510 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29513 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29514 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29516 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29518 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29520 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29523 2017-03-20 Marek Polacek <polacek@redhat.com>
29526 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29528 2017-03-20 Richard Biener <rguenther@suse.de>
29530 PR tree-optimization/80113
29531 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29532 allocate extra SSA name for PHI def.
29533 (add_close_phis_to_outer_loops): Likewise.
29534 (add_close_phis_to_merge_points): Likewise.
29535 (copy_loop_close_phi_args): Likewise.
29536 (copy_cond_phi_nodes): Likewise.
29538 2017-03-20 Martin Liska <mliska@suse.cz>
29540 PR middle-end/79753
29541 * tree-chkp.c (chkp_build_returned_bound): Do not build
29542 returned bounds for a LHS that's not a BOUNDED_P type.
29544 2017-03-20 Martin Liska <mliska@suse.cz>
29548 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29549 COMPLEX_CST and VECTOR_CST.
29551 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29554 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29555 target operand. A new splitter adds the clobber statement in case
29556 the target operand is dead anyway.
29558 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29560 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29561 to age-old versions of binutils and glibc.
29563 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29565 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29567 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29569 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29571 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29573 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29574 requirement for binutils 2.13.
29576 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29578 * combine.c (try_combine): Delete redundant i1 test. Call
29579 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29581 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29583 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29584 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29586 <riscv64-*-elf>: Re-arrange section
29587 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29588 <riscv32-*-linux>: Likewise.
29589 <riscv64-*-elf>: Likewise
29590 <riscv64-*-linux>: Likewise.
29592 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29595 * aarch64.opt(verbose-cost-dump): Fix typo.
29597 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29600 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29601 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29603 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29605 * reload.c (find_reloads): When reloading a nonoffsettable address,
29606 use RELOAD_OTHER for it and its address reloads.
29608 PR rtl-optimization/79910
29609 * combine.c (record_used_regs): New static function.
29610 (try_combine): Handle situations where there is an additional
29611 instruction between I2 and I3 which needs to have a LOG_LINK
29614 2017-03-17 Jeff Law <law@redhat.com>
29616 PR tree-optimization/71437
29617 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29618 conditional in the hash table first.
29619 (vrp_dom_walker::before_dom_children): Extract condition from
29620 ASSERT_EXPR. Record condition, its inverion and any implied
29621 conditions as well.
29623 2017-03-17 Marek Polacek <polacek@redhat.com>
29624 Markus Trippelsdorf <markus@trippelsdorf.de>
29626 PR tree-optimization/80079
29627 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29630 2017-03-17 Richard Biener <rguenther@suse.de>
29632 PR middle-end/80075
29633 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29634 Properly verify the LHS before the RHS possibly claims to be
29636 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29639 2017-03-17 Martin Jambor <mjambor@suse.cz>
29641 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29642 (List of -O2 options): Likewise.
29643 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29646 2017-03-17 Tom de Vries <tom@codesourcery.com>
29648 * gcov-dump.c (print_usage): Print bug_report_url.
29650 2017-03-17 Richard Biener <rguenther@suse.de>
29652 PR middle-end/80050
29653 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29654 (parser::peek): Likewise.
29656 2017-03-17 Richard Biener <rguenther@suse.de>
29658 PR tree-optimization/80048
29659 * sese.c (free_sese_info): Properly release rename_map and
29660 copied_bb_map elements.
29662 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29664 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29665 Add linked-list forward and backlinks. Insert on
29666 construction, remove on destruction.
29667 (class pass_store_merging): Add m_stores_head field.
29668 (pass_store_merging::terminate_and_process_all_chains):
29669 Iterate over m_stores_head list.
29670 (pass_store_merging::terminate_all_aliasing_chains):
29672 (pass_store_merging::execute): Check for debug stmts first.
29673 Push new chains onto the m_stores_head stack.
29675 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29678 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29679 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29680 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29682 2017-03-16 Jeff Law <law@redhat.com>
29684 PR tree-optimization/71437
29685 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29686 member function. Implementation moved into after_dom_children
29687 member function and into the threader's thread_outgoing_edges
29689 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29690 some code into new thread_outgoing_edges.
29691 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29692 definition. Simplify marker handling (do it here). Assume we always
29693 have the available expression and the const/copies tables.
29694 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29696 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29697 * tree-vrp.c (equiv_stack): No longer file scoped.
29698 (vrp_dom_walker): New class.
29699 (vrp_dom_walker::before_dom_children): New member function.
29700 (vrp_dom_walker::after_dom_children): Likewise.
29701 (identify_jump_threads): Setup domwalker. Use it rather than
29702 walking edges in a random order by hand. Simplify setup/finalization.
29703 (finalize_jump_threads): Remove.
29704 (vrp_finalize): Do not call identify_jump_threads here.
29705 (execute_vrp): Do it here instead and call thread_through_all_blocks
29708 PR tree-optimization/71437
29709 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29711 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29713 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29714 (dom_opt_dom_walker::thread_across_edge): Remove
29715 handle_dominating_asserts argument. All callers changed.
29716 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29717 changes. Remove calls to lhs_of_dominating_assert. Other
29718 uses of handle_dominating_asserts turn into unconditional code
29719 (simplify_control_stmt_condition_1): Likewise.
29720 (simplify_control_stmt_condition): Likewise.
29721 (thread_through_normal_block, thread_across_edge): Likewise.
29722 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29723 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29724 object if it is not an SSA_NAME.
29725 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29726 before calling into the VRP specific simplifiers.
29727 (identify_jump_threads): Remove handle_dominating_asserts
29730 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29733 * tree-diagnostic.c (default_tree_printer): No longer static.
29734 * tree-diagnostic.h (default_tree_printer): New prototype.
29736 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29738 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29739 Change ins into fmov.
29741 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29743 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29744 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29745 Use h_con constraint for operand 1.
29746 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29747 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29749 2017-03-15 Jeff Law <law@redhat.com>
29751 PR tree-optimization/71437
29752 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29753 (record_temporary_equivalences): Use it.
29755 PR tree-optimization/71437
29756 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29757 tree-ssa-scopedtables.
29758 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29759 (record_conditions, record_cond, vuse_eq): Likewise.
29760 (record_edge_info): Adjust to API tweak of record_conditions.
29761 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29762 (record_temporary_equivalences, optimize_stmt): Likewise.
29763 (eliminate_redundant_computations): Likewise.
29764 (record_equivalences_from_stmt): Likewise.
29765 * tree-ssa-scopedtables.c: Include options.h and params.h.
29766 (vuse_eq): New function, moved from tree-ssa-dom.c
29767 (build_and_record_new_cond): Likewise.
29768 (record_conditions): Likewise. Accept vector of conditions rather
29769 than edge_equivalence structure for first argument.
29770 for the first argument.
29771 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29772 from tree-ssa-dom.c.
29773 (avail_exprs_stack::record_cond): Likewise.
29774 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29775 from tree-ssa-dom.c.
29776 (avail_exprs_stack): Add new member functions lookup_avail_expr
29778 (record_conditions): Declare.
29780 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29783 * lra-constraints.c (process_alt_operands): Increase reject for
29784 reloading an input/output operand.
29786 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29789 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29790 insns to convert from signed/unsigned char/short to IEEE 128-bit
29792 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29794 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29797 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29798 subreg of inner mode for values already in registers.
29800 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29802 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29803 iteration reg is used after the loop.
29805 2017-03-14 Martin Sebor <msebor@redhat.com>
29807 PR tree-optimization/79800
29808 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29809 precision in negative-positive range.
29810 (format_floating): Call non-const overload with adjusted precision.
29812 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29815 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29818 2017-03-14 Martin Sebor <msebor@redhat.com>
29820 PR middle-end/80020
29821 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29822 * builtins.def (aligned_alloc): Use it.
29825 * Makefile.in (GTFILES): Add calls.c.
29826 * calls.c: Include "gt-calls.h".
29828 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29830 PR rtl-optimization/79728
29831 * regs.h (struct target_regs): New field
29832 x_contains_allocatable_regs_of_mode.
29833 (contains_allocatable_regs_of_mode): New macro.
29834 * reginfo.c (init_reg_sets_1): Initialize it, and change
29835 contains_reg_of_mode so it includes global regs as well.
29836 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29837 rather than contains_regs_of_mode.
29839 2017-03-14 Martin Liska <mliska@suse.cz>
29841 * doc/invoke.texi: Document options that can't be combined with
29842 -fcheck-pointer-bounds.
29844 2017-03-14 Martin Liska <mliska@suse.cz>
29846 PR middle-end/79831
29847 * doc/invoke.texi (-Wchkp): Document the option.
29849 2017-03-14 Martin Liska <mliska@suse.cz>
29851 * Makefile.in: Install gcov-dump.
29853 2017-03-14 Martin Liska <mliska@suse.cz>
29855 * multiple_target.c (expand_target_clones): Bail out for
29856 an invalid attribute.
29858 2017-03-14 Richard Biener <rguenther@suse.de>
29860 * alias.c (struct alias_set_entry): Pack properly.
29861 * cfgloop.h (struct loop): Likewise.
29862 * cse.c (struct set): Likewise.
29863 * ipa-utils.c (struct searchc_env): Likewise.
29864 * loop-invariant.c (struct invariant): Likewise.
29865 * lra-remat.c (struct cand): Likewise.
29866 * recog.c (struct change_t): Likewise.
29867 * rtl.h (struct address_info): Likewise.
29868 * symbol-summary.h (function_summary): Likewise.
29869 * tree-loop-distribution.c (struct partition): Likewise.
29870 * tree-object-size.c (struct object_size_info): Likewise.
29871 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29872 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29873 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29874 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29875 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29876 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29877 (struct _stmt_vec_info): Likewise.
29879 2017-03-14 Martin Liska <mliska@suse.cz>
29882 * multiple_target.c (create_dispatcher_calls): Check that
29883 a target can create a function dispatcher.
29885 2017-03-14 Martin Liska <mliska@suse.cz>
29888 * multiple_target.c (expand_target_clones): Drop local.local
29889 flag for default implementation.
29891 2017-03-14 Richard Biener <rguenther@suse.de>
29893 PR tree-optimization/80030
29894 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29896 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29898 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29899 gcc_fallthrough() instead of __attribute__((fallthrough));
29901 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29903 * doc/gcc.texi: Remove "up" link to (DIR).
29904 * doc/gccint.texi: Ditto.
29906 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29908 * doc/install.texi (Specific) <avr>: Remove reference to
29911 2017-03-13 Jeff Law <law@redhat.com>
29913 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29914 attribute rather than comments.
29916 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29917 match_scratch operand is highest.
29919 2017-03-13 Martin Liska <mliska@suse.cz>
29921 PR middle-end/78339
29922 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29923 is a CHKP clone, use original declaration.
29925 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29927 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29928 (arc_conditional_register_usage): Use a different allocation order
29929 when optimizing for size.
29930 * common/config/arc/arc-common.c (arc_option_optimization_table):
29931 Section anchors default on when optimizing for size.
29933 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29935 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29937 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29939 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29940 * config/arc/arc.md (cpu_facility): Add cd variant.
29941 (*movqi_insn): Add code density variant.
29942 (*movhi_insn): Likewise.
29943 (*movqi_insn): Likewise.
29944 (*addsi3_mixed): Likewise.
29945 (subsi3_insn): Likewise.
29947 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29949 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29951 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29953 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29954 expressions with MINUS and UNARY ops.
29956 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29959 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29961 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29962 between vec_select and vector argument.
29963 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29964 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29965 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29966 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29967 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29968 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29970 2017-03-13 Richard Biener <rguenther@suse.de>
29973 * params.def (vect-max-peeling-for-alignment): Fix typo.
29975 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29977 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29978 issue that only occurred with binutils below 2.18.
29980 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29982 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29983 refer to binutils 2.11/2.12 minimum.
29985 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29987 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29988 ftp.kernel.org and simplify binutils requirement.
29990 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29992 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29994 (Optimize Options): Ditto. Also remove redundancy.
29996 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29998 PR translation/79848
29999 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
30001 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
30002 to G_ to avoid double translation.
30004 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30006 PR translation/79923
30007 * auto-profile.c (get_combined_location): Convert leading
30008 character of diagnostics to lower case and remove trailing period.
30009 (read_profile): Likewise for various diagnostics.
30010 * config/arm/arm.c (arm_option_override): Remove trailing period
30011 from various diagnostics.
30012 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30013 (msp430_expand_delay_cycles): Likewise.
30015 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30018 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30019 full command-line argument, rather than just "str".
30020 (aarch64_validate_march): Likewise.
30021 (aarch64_validate_mtune): Likewise.
30023 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30025 PR rtl-optimization/78911
30026 * lra-assigns.c (must_not_spill_p): New function.
30027 (spill_for): Use it.
30029 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30031 PR tree-optimization/79981
30032 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30033 ATOMIC_COMPARE_EXCHANGE ifn result.
30034 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30035 IFN_ATOMIC_COMPARE_EXCHANGE.
30037 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30040 * opts.c (parse_sanitizer_options): Add missing question mark to
30041 "did you mean" message.
30043 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30045 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30047 (VMULEUH_UNS): Likewise.
30048 (VMULOUB_UNS): Likewise.
30049 (VMULOUH_UNS): Likewise.
30050 * config/rs6000/rs6000.c (builtin_function_type): Remove
30051 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30053 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30056 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30057 x with result of extra_parsing_for_operand_code_0.
30058 (function_reader::extra_parsing_for_operand_code_0): Convert
30059 return type from void to rtx, returning x. When reading
30060 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30061 larger size containing struct block_symbol.
30063 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30065 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30066 -mfloat128-hardware without -m64.
30068 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30071 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30072 entries to the case statement that marks unsigned arguments to
30073 overloaded functions.
30075 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30077 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30078 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30080 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30083 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30084 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30086 2017-03-10 Martin Liska <mliska@suse.cz>
30090 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30091 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30094 2017-03-10 Olivier Hainque <hainque@adacore.com>
30096 * tree-switch-conversion (array_value_type): Start by resetting
30097 candidate type to it's main variant.
30099 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30101 PR rtl-optimization/79909
30102 * combine.c (try_combine): Use simplify_replace_rtx on individual
30103 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30104 of the whole CALL_INSN_FUNCTION_USAGE.
30106 PR tree-optimization/79972
30107 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30108 get_range_info on SSA_NAMEs. Formatting fixes.
30110 2017-03-10 Richard Biener <rguenther@suse.de>
30111 Jakub Jelinek <jakub@redhat.com>
30113 PR tree-optimization/77975
30114 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30115 edge to be constant.
30116 (get_val_for): For constant x return it. Formatting fix.
30117 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30118 would use the same bases as the current one.
30120 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30122 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30123 instead of vec_select for V1TImode.
30124 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30126 (VSX_LE_128): Add V1TI to this mode iterator.
30127 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30128 (*vsx_le_perm_store_<mode>): Likewise.
30129 (pre-reload splitter for VSX stores): Likewise.
30130 (post-reload splitter for VSX stores): Likewise.
30131 (*vsx_xxpermdi2_le_<mode>): Likewise.
30132 (*vsx_lxvd2x2_le_<mode>): Likewise.
30133 (*vsx_stxvd2x2_le_<mode>): Likewise.
30135 2017-03-09 Michael Eager <eager@eagercon.com>
30137 Correct failures with --enable-checking=yes,rtl.
30139 * config/microblaze/microblaze.c (microblaze_expand_shift):
30140 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30141 test for const0_rtx.
30142 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30143 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30145 2017-03-09 Richard Biener <rguenther@suse.de>
30147 PR tree-optimization/79977
30148 * graphite-scop-detection.c (scop_detection::merge_sese):
30149 Handle the case of extra exits to blocks dominating the entry.
30151 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30153 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30156 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30158 PR rtl-optimization/79949
30159 * lra-constraints.c (process_alt_operands): Check memory when
30160 trying to predict a cycle. Print about the overall increase.
30162 2017-03-09 Richard Biener <rguenther@suse.de>
30164 PR middle-end/79971
30165 * gimple-expr.c (useless_type_conversion_p): Preserve
30166 TYPE_SATURATING for fixed-point types.
30168 2017-03-09 Richard Biener <rguenther@suse.de>
30171 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30172 alignment of BLKmode params.
30174 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30177 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30178 (VALL_NO_V2Q): Likewise.
30180 * config/aarch64/aarch64-simd.md
30181 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30183 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30184 VALL_NO_V2Q mode iterator.
30185 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30187 2017-03-09 Martin Liska <mliska@suse.cz>
30189 PR tree-optimization/79631
30190 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30191 tree_int_cst_sign_bit just for INTEGER constants.
30193 2017-03-09 Martin Liska <mliska@suse.cz>
30197 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30200 2017-03-09 Marek Polacek <polacek@redhat.com>
30203 * tree.c (inchash::add_expr): Handle TREE_VEC.
30205 2017-03-09 Martin Liska <mliska@suse.cz>
30208 (chkp_narrow_size_and_offset): New function.
30209 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30210 (void chkp_parse_bit_field_ref): New function.
30211 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30212 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30214 2017-03-09 Martin Liska <mliska@suse.cz>
30217 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30218 (chkp_find_bounds_1): Remove gcc_unreachable.
30220 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30223 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30224 BUILT_IN_SYNC*, determine the access type from the size suffix and
30225 always build a MEM_REF with that type. Handle forgotten
30226 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30229 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30230 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30231 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30232 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30233 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30234 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30235 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30236 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30237 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30238 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30239 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30240 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30241 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30242 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30243 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30244 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30245 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30246 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30247 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30248 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30249 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30250 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30251 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30252 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30253 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30254 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30255 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30256 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30257 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30258 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30259 definitions outside of __OPTIMIZE__ guarded section.
30262 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30263 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30264 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30265 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30268 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30270 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30271 ("vfenez<mode>"): Add missing constraints.
30273 2017-03-08 Martin Sebor <msebor@redhat.com>
30276 * config/nds32/nds32.c (nds32_option_override):
30277 Fix misspelled diagnostic.
30279 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30282 * gimplify.c (gimplify_omp_for): Replace index var in outer
30283 taskloop statement with an artificial variable and add
30284 OMP_CLAUSE_PRIVATE clause for it.
30286 2017-03-08 Richard Biener <rguenther@suse.de>
30288 PR tree-optimization/79955
30289 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30290 for accesses that are completely outside of the variable.
30292 2017-03-08 Andrew Haley <aph@redhat.com>
30294 PR tree-optimization/79943
30295 * tree-ssa-loop-split.c (compute_new_first_bound): When
30296 calculating the new upper bound, (END-BEG) should be added, not
30299 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30301 * config/avr/avr.md (setmemhi): Make sure match_dup
30302 operand number comes before match_scratch.
30304 2017-03-08 Richard Biener <rguenther@suse.de>
30306 PR tree-optimization/79920
30307 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30308 with ncopies == 1 to ...
30309 (vect_transform_slp_perm_load): ... here. Properly compute
30310 all element loads by iterating VF times over the group. Do
30311 not handle ncopies (computed in a broken way) in
30312 vect_create_mask_and_perm.
30314 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30317 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30318 is a uniform vector, use uniform_vector_p return value instead of
30319 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30321 2017-03-07 Marek Polacek <polacek@redhat.com>
30323 PR middle-end/79809
30324 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30325 (alloca_call_type): Likewise.
30327 2017-03-07 Martin Liska <mliska@suse.cz>
30329 * gcov.c (process_args): Put comment to correct location.
30331 2017-03-07 Martin Liska <mliska@suse.cz>
30333 PR middle-end/68270
30334 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30335 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30336 (chkp_narrow_bounds_for_field): Likewise.
30337 (chkp_parse_array_and_component_ref): Pass one more argument to
30340 2017-03-07 Richard Biener <rguenther@suse.de>
30342 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30345 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30347 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30348 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30350 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30353 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30354 to end of description.
30355 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30357 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30359 PR rtl-optimization/79901
30360 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30362 (*avx512f_<code><mode>3<mask_name>): ... this.
30363 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30364 iterator instead of VI8_AVX2_AVX512BW.
30366 PR rtl-optimization/79901
30367 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30368 min/max expander, expand it using expand_vec_cond_expr.
30371 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30374 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30377 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30378 to void * for PCH reasons.
30379 * dwarf2out.c (output_loc_operands, output_die): Cast
30380 v.val_vec.array to unsigned char *.
30382 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30385 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30388 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30390 PR rtl-optimization/79571
30391 * lra-constraints.c (process_alt_operands): Calculate static
30392 reject and subtract it from overall when only addresses will be
30395 2017-03-06 Julia Koval <julia.koval@intel.com>
30398 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30399 incoming stack boundary to 128 for 64-bit targets.
30401 2017-03-06 Richard Biener <rguenther@suse.de>
30403 PR tree-optimization/79894
30404 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30405 to NULL after folding it.
30407 2017-03-06 Richard Biener <rguenther@suse.de>
30409 PR tree-optimization/79824
30410 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30411 check disabling peeling for gaps.
30413 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30415 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30416 attributes): Document gettimeofday.
30418 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30420 * config/s390/s390.c (s390_option_override_internal): Set
30421 PARAM_MIN_VECT_LOOP_BOUND
30423 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30425 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30426 * config/s390/s390.md: Likewise.
30428 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30431 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30432 (<avx2_avx512>_perm<mode>): Rename to ...
30433 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30435 (<avx512>_perm<mode>_mask): Rename to ...
30436 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30438 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30439 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30440 instead of VI8F_256_512.
30441 (avx512f_perm<mode>): New define_expand.
30442 (avx512f_perm<mode>_mask): Likewise.
30443 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30444 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30446 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30448 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30449 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30451 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30453 2017-03-06 Martin Liska <mliska@suse.cz>
30456 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30457 when having a SSA NAME w/o VAR_DECL assigned to it.
30459 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30461 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30462 msa_dpsub_<su>_d): Fix MODE for vec_select.
30464 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30466 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30468 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30470 2017-03-06 Richard Biener <rguenther@suse.de>
30472 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30473 * plugin.c (register_plugin_info): Likewise.
30474 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30476 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30478 * config/i386/sse.md (sse_storehps, sse_storelps,
30479 avx_<castmode><avxsizesuffix>_<castmode>,
30480 avx512f_<castmode><avxsizesuffix>_<castmode>,
30481 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30482 in condition that at least one operand is not a MEM.
30484 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30486 PR middle-end/79805
30487 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30488 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30490 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30491 gimple_call_nothrow_p flag based on whether original builtin can throw.
30492 If it can, emit following stmts on the fallthrough edge.
30493 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30494 don't create new bb if inserting just debug stmts on the edge, try to
30495 insert them on the fallthru bb or just reset debug stmts.
30497 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30500 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30501 restore recog_data (including the operand rtxes inside it) around
30502 the call to get_insn_template.
30504 2017-03-03 Martin Sebor <msebor@redhat.com>
30506 PR tree-optimization/79699
30507 * context.c (context::~context): Free MPFR caches to avoid
30508 a memory leak on program exit.
30510 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30512 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30513 Use wide_int::ulow () instead of .elt (0).
30515 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30517 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30518 (*pushxf): Limit oF constraint to 32bit targets and add oC
30519 constraint for 64bit targets.
30520 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30521 (*pushdf): Change rmF constraint to rmC.
30523 2017-03-03 Martin Liska <mliska@suse.cz>
30525 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30526 Remove unused variable.
30528 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30531 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30532 is a memory operand, increase num_memory.
30533 (ix86_expand_args_builtin): Likewise.
30535 2017-03-03 Jan Hubicka <jh@suse.cz>
30538 * ipa-devirt.c (maybe_record_node): Properly handle
30539 __cxa_pure_virtual visibility.
30541 2017-03-03 Martin Liska <mliska@suse.cz>
30543 PR tree-optimization/79803
30544 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30546 (pass_loop_prefetch::execute): Disabled optimization if an
30547 assumption about L1 cache size is not met.
30549 2017-03-03 Martin Liska <mliska@suse.cz>
30551 PR rtl-optimization/79574
30552 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30553 (hash_scan_set): Likewise.
30554 (dump_hash_table): Likewise.
30555 (hoist_code): Likewise.
30557 2017-03-03 Richard Biener <rguenther@suse.de>
30559 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30560 in place of uhigh ().
30561 (fixed_convert_from_real): Likewise.
30563 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30566 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30568 2017-03-03 Richard Biener <rguenther@suse.de>
30570 PR middle-end/79818
30571 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30572 TYPE_OVERFLOW_UNDEFINED check.
30574 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30576 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30578 (vector_ae_<mode>_p): Likewise.
30579 (vector_nez_<mode>_p): Likewise.
30580 (vector_ne_v2di_p): Likewise.
30581 (vector_ae_v2di_p): Likewise.
30582 (vector_ne_<mode>_p): Likewise.
30583 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30585 (vsx_tsqrt<mode>2_fe): Likewise.
30587 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30590 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30592 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30594 PR rtl-optimization/79780
30595 * cprop.c (one_cprop_pass): When second and further conditional trap
30596 in a single basic block is turned into an unconditional trap, turn it
30597 into a deleted note to avoid RTL verification failures.
30599 2017-03-02 Richard Biener <rguenther@suse.de>
30601 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30603 2017-03-02 Richard Biener <rguenther@suse.de>
30605 PR tree-optimization/79345
30607 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30608 param and abort the walk, returning -1 if it is hit.
30609 (walk_aliased_vdefs): Take a limit param and pass it on.
30610 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30611 defaulting to 0 and return a signed int.
30612 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30613 (check_defs): New helper.
30614 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30615 about uninitialized memory.
30616 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30617 bogus uninitialized warning.
30618 (fixed_convert_from_real): Likewise.
30620 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30622 PR tree-optimization/66768
30623 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30624 iv_use if base object can't be determined.
30626 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30628 PR tree-optimization/79345
30629 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30630 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30631 (get_pattern_stats): Initialize it.
30632 * genemit.c (gen_expand): Verify match_scratch numbers come after
30633 match_operand/match_dup numbers.
30634 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30635 match_scratch numbers.
30636 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30638 * config/s390/s390.md (trunctdsd2): Likewise.
30640 2017-03-02 Richard Biener <rguenther@suse.de>
30642 * wide-int.h (wide_int_storage::operator=): Implement in terms
30645 2017-03-02 Richard Biener <rguenther@suse.de>
30647 PR tree-optimization/79777
30648 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30649 the to insert expression to sth existing.
30651 2017-03-01 Martin Sebor <msebor@redhat.com>
30653 PR middle-end/79692
30654 * gimple-ssa-sprintf.c
30655 (directive::known_width_and_precision): New function.
30656 (format_integer): Use it.
30657 (get_mpfr_format_length): Consider the full range of precision
30658 when computing %g output with the # flag. Set the likely byte
30659 count to 3 rather than 1 when precision is indeterminate.
30660 (format_floating): Correct the lower bound of precision.
30662 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30664 * doc/invoke.texi: Document default code model for 64-bit Linux.
30666 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30669 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30670 udiv rather than div since input pattern is unsigned.
30672 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30674 * config/i386/i386.c (print_reg): Warn for values of
30675 unsupported size in integer register.
30677 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30680 * config/rs6000/predicates.md (current_file_function_operand): Do
30681 not allow self calls to be local if the function is replaceable.
30683 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30686 * config/rs6000/altivec.h (vec_ctz and others): Change the
30687 preprocessor macro that controls conditional compilation from
30688 _ARCH_PWR9 to __POWER9_VECTOR__.
30689 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30690 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30691 control (instead of _ARCH_PWR9 control) so that template
30692 definition uses power9-specific function.
30693 (vec_any_eq): Likewise.
30694 (vec_all_ne): Change macro definition to use a power9-specific
30695 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30696 _ARCH_PWR9 control).
30697 (vec_any_eq) Likewise.
30698 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30699 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30700 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30701 support for xvcmpnedp instruction.
30702 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30703 macro expansion so that Power9 implementation of vec_all_ne does
30704 not use the AltiVec predicate framework.
30705 (VCMPNEH_P): Likewise.
30706 (VCMPNEW_P): Likewise.
30707 (VCMPNED_P): Likewise.
30708 (VCMPNEFP_P): Likewise.
30709 (VCMPNEDP_P): Likewise.
30710 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30711 implementation of vec_any_eq to not use AltiVec predicate
30713 (VCMPAEH_P): Likewise.
30714 (VCMPAEW_P): Likewise.
30715 (VCMPAED_P): Likewise.
30716 (VCMPAEFP_P): Likewise.
30717 (VCMPAEDP_P): Likewise.
30718 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30719 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30720 not use the AltiVec predicate framework.
30721 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30722 of vec_any_eq to not use AltiVec predicate framework.
30723 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30724 support for predefined __POWER9_VECTOR__ macro to indicate that
30725 Power9 instruction selection is enabled.
30726 (altivec_overloaded_builtins): Remove extraneous
30727 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30728 function argument types RS6000_BTI_bool_V16QI and
30729 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30730 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30731 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30732 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30733 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30734 Power9 for implementations of vec_cmpne. Change the signature for
30735 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30736 (representing vec_all_ne) to remove the previously described first
30737 argument of type RS6000_BTI_INTSI, as this was an artifact of
30738 reliance on the AltiVec predicate framework, which is no longer
30739 used in the implementation of these functions. Add
30740 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30741 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30742 since, unlike the AltiVec predicate framework implementation, we
30743 do not share function descriptors between vec_alle and vec_anyeq.
30744 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30745 set of modes that receive special treatment even when
30746 TARGET_P9_VECTOR is true. The special treatment emits code that
30747 does not depend on Power9 instructions.
30748 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30749 define_expand to not rely on AltiVec predicate framework.
30750 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30752 (vector_ne_v2di_p): Change this define_expand to not rely on
30753 AltiVec predicate framework.
30754 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30756 (vector_ne_<mode>_p): Change this define_expand to not rely on
30757 AltiVec predicate framework.
30758 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30760 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30761 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30762 define_insn pattern.
30763 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30764 define_insn pattern because the xvcmpne<VSs>. instruction is not
30766 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30767 instruction is not supported.
30769 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30771 * config/nvptx/nvptx.c: Include intl.h.
30773 2017-03-01 Martin Jambor <mjambor@suse.cz>
30776 * ipa-prop.h (ipa_bits): Removed field known.
30777 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30778 to pointers. Adjusted their comments to warn about their sharing.
30779 (ipcp_transformation_summary): Change bits to a vector of pointers.
30780 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30781 (ipa_get_ipa_bits_for_value): Declare.
30782 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30783 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30784 (ipa_bits_hash_table): Likewise.
30785 (ipa_vr_ggc_hash_traits): Likewise.
30786 (ipa_vr_hash_table): Likewise.
30787 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30788 being pointers and vr_known being removed.
30789 (ipa_set_jf_unknown): Likewise.
30790 (ipa_get_ipa_bits_for_value): New function.
30791 (ipa_set_jfunc_bits): Likewise.
30792 (ipa_get_value_range): New overloaded functions.
30793 (ipa_set_jfunc_vr): Likewise.
30794 (ipa_compute_jump_functions_for_edge): Use the above functions to
30795 construct bits and vr parts of jump functions.
30796 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30797 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30799 (ipcp_grow_transformations_if_necessary): Also allocate
30800 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30802 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30803 them. Fix too long lines.
30804 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30805 vr_known being removed.
30806 (ipa_read_jump_function): Use new setter functions to construct bits
30807 and vr parts of jump functions or set them to NULL.
30808 (write_ipcp_transformation_info): Adjust for bits being pointers.
30809 (read_ipcp_transformation_info): Likewise.
30810 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30812 Include gt-ipa-prop.h.
30813 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30815 (ipcp_store_bits_results): Likewise.
30816 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30817 Do not write to existing jump functions but use a temporary instead.
30819 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30822 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30823 attempt to use its first operand as BIT_FIELD_REF base.
30825 2017-03-01 Richard Biener <rguenther@suse.de>
30827 PR middle-end/79721
30828 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30829 interpolating formula in wrapping arithmetic.
30830 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30832 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30834 PR tree-optimization/79734
30835 * tree-vect-generic.c (expand_vector_condition): Optimize
30836 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30837 Handle VEC_COND_EXPR where comparison has different inner width from
30838 type's inner width.
30840 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30842 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30843 markup, and similar issues. Remove @opindex entries for things
30844 that aren't options. Add missing -mmpy-option entries.
30846 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30848 PR tree-optimization/79737
30849 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30850 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30851 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30852 instead of byte_size. Formatting fix.
30853 (shift_bytes_in_array_right): Formatting fix.
30855 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30858 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30859 condition on optimize for the leaf function test.
30861 2017-02-28 Martin Liska <mliska@suse.cz>
30864 * read-rtl-function.c (function_reader::handle_unknown_directive):
30865 Bail out when one uses -flto.
30867 2017-02-28 Martin Liska <mliska@suse.cz>
30869 * common.opt: Replace space with tabular for options of <number>
30871 * config/i386/i386.opt: Show <number> value for
30872 -mlarge-data-threshold.
30873 * opts.c (print_filtered_help): Do not display number in hexadecimal
30876 2017-02-28 Martin Liska <mliska@suse.cz>
30878 * common.opt: Fix --help=option -Q for options which are of
30881 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30883 * config/i386/i386.c (print_reg): Error out for values
30884 of 8-bit size in invalid integer register.
30886 2017-02-28 Martin Sebor <msebor@redhat.com>
30888 PR tree-optimization/79691
30889 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30891 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30894 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30895 gcc_unreachable with output_operand_lossage.
30897 2017-02-28 Richard Biener <rguenther@suse.de>
30899 PR tree-optimization/79740
30900 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30902 (visit_nary_op): Insert the nary into the hashtable if we
30903 pattern-matched sth.
30904 * tree-ssa-pre.c (eliminate_insert): Robustify.
30906 2017-02-28 Richard Biener <rguenther@suse.de>
30908 PR middle-end/79731
30909 * fold-const.c (decode_field_reference): Reject out-of-bound
30912 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30914 * config/i386/i386.c: Include intl.h.
30915 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30916 instead of just cond ? "..." : "...".
30917 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30918 * coverage.c (read_counts_file): Likewise.
30919 * omp-offload.c: Include intl.h.
30920 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30921 of just cond ? "..." : "...".
30922 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30923 of just cond ? "..." : "...".
30925 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30928 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30930 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30931 'tune for' CPU name.
30932 * config/arm/arm-cpu-data.h: Regenerated.
30934 2017-02-28 Richard Biener <rguenther@suse.de>
30936 PR tree-optimization/79732
30937 * tree-inline.c (expand_call_inline): Do not shadow var.
30939 2017-02-28 Richard Biener <rguenther@suse.de>
30941 PR tree-optimization/79723
30942 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30943 address-space properly.
30945 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30947 * doc/optinfo.texi (Optimization groups): Fix option used for
30949 * doc/invoke.texi (-fopt-info): Document "omp".
30950 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30951 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30952 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30953 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30954 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30956 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30958 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30959 OPTGROUP_OMP with "omp".
30961 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30964 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30965 for arithmetic shift of unsigned V2DI.
30967 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30969 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30970 arc/linux.h headers.
30971 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30972 (LINK_SPEC): Likewise.
30973 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30974 (EXTRA_SPECS): Likewise.
30975 (STARTFILE_SPEC): Likewise.
30976 (ENDFILE_SPEC): Likewise.
30977 (LIB_SPEC): Likewise.
30978 (TARGET_SDATA_DEFAULT): Likewise.
30979 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30980 (MULTILIB_DEFAULTS): Likewise.
30981 (DWARF2_UNWIND_INFO): Likewise.
30982 * config/arc/big.h: New file.
30983 * config/arc/elf.h: Likewise.
30984 * config/arc/linux.h: Likewise.
30985 * config/arc/t-uClibc: Remove.
30987 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30989 PR tree-optimization/77536
30990 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30991 (tree_transform_and_unroll_loop): Use above function to compute the
30992 estimated niter of unrolled loop and use it when scaling profile.
30993 Also use count info rather than frequency if it's non-zero.
30994 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30995 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30996 (vect_transform_loop): Call above function.
30998 2017-02-27 Richard Biener <rguenther@suse.de>
31000 PR tree-optimization/45397
31001 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
31002 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
31003 (visit_nary_op): Add pattern matching for CSEing sign-changed
31004 or truncated operations with wider ones.
31006 2017-02-27 Richard Biener <rguenther@suse.de>
31008 PR tree-optimization/79690
31009 * tree-vect-stmts.c (vectorizable_store): Use vector type
31010 built from the DR with address-space.
31012 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31014 * doc/invoke.texi (Optimize Options): Refine the description
31015 of asan-use-after-return.
31017 2017-02-25 Alan Modra <amodra@gmail.com>
31019 PR rtl-optimization/79584
31020 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31021 base, not ad->base_term, the reg within base. Remove assertion
31022 that ad->base == ad->base_term. Replace gen_int_mode using
31023 bogus mode with const0_rtx.
31025 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31027 PR middle-end/79396
31028 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31029 FMA_EXPR like tcc_binary or tcc_unary.
31031 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31034 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31036 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31037 (output_loc_operands): Handle DW_OP_call_ref and
31038 DW_OP_GNU_variable_value.
31039 (struct variable_value_struct): New type.
31040 (struct variable_value_hasher): Likewise.
31041 (variable_value_hash): New variable.
31042 (string_types): Remove.
31043 (copy_loc_descr): New function.
31044 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31045 (prepend_loc_descr_to_each): New function.
31046 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31047 instead of add_loc_descr_to_each if the first argument is single
31048 location list and the second has multiple.
31049 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31050 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31051 when looking for variable value which doesn't have other location info.
31052 (loc_list_from_tree): Formatting fix.
31053 (gen_array_type_die): Simplify DW_AT_string_length handling.
31054 (adjust_string_types): Remove.
31055 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31056 string_types. Call resolve_variable_values.
31057 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31058 (resolve_addr_in_expr): Likewise. Add A argument.
31059 (copy_deref_exprloc): Remove deref argument. Adjust for the
31060 original expression being DW_OP_GNU_variable_value with optionally
31061 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31062 optionally after it.
31063 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31064 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31065 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31066 (variable_value_hasher::hash, variable_value_hasher::equal): New
31068 (resolve_variable_value_in_expr, resolve_variable_value,
31069 resolve_variable_values, note_variable_value_in_expr,
31070 note_variable_value): New functions.
31071 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31074 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31077 * opts.h (handle_generated_option): Add GENERATED_P argument.
31078 * opts-common.c (handle_option): Adjust function comment.
31079 (handle_generated_option): Add GENERATED_P argument, pass it to
31081 (control_warning_option): Pass false to handle_generated_option
31083 * opts.c (maybe_default_option): Pass true to handle_generated_option
31085 * optc-gen.awk: Likewise.
31087 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31089 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31090 a REG, look at the REG it is a SUBREG of.
31091 (splitter for cmpeqsi_t): Ditto.
31093 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31095 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31096 the special USEs with the pattern of the insn, not the insn itself.
31098 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31101 * doc/invoke.texi: Document -mload-store-pairs.
31103 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31104 Sandra Loosemore <sandra@codesourcery.com>
31106 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31107 argument isn't a CONST_INT.
31108 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31109 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31110 (nios2_validate_compare): Bypass alternate compare logic if *op2
31111 is not a CONST_INT.
31112 (ldstwm_operation_p): Return false if first_base is not a REG or
31113 if first_offset is not a CONST_INT.
31115 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31117 * config/cris/cris.md: Use correct operand in a define_peephole2.
31119 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31121 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31123 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31125 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31126 this_insn if it is an INSN or JUMP_INSN.
31127 (force_offsettable): Look at base, not at addr.
31128 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31129 on things that aren't necessarily CONST_INTs.
31131 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31133 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31134 -mfpmath=sse is the default also for x86-32 targets with SSE2
31135 instruction set when @option{-ffast-math} is enabled
31137 2017-02-24 Jeff Law <law@redhat.com>
31139 PR rtl-optimizatoin/79286
31140 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31143 2017-02-24 Richard Biener <rguenther@suse.de>
31145 PR tree-optimization/79389
31146 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31149 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31151 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31152 function comment to reflect reality.
31153 (loop_exits_before_overflow): Fix typo in function description.
31155 2017-02-24 Richard Biener <rguenther@suse.de>
31157 PR tree-optimization/79389
31158 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31159 properly that a threading opportunity exists. Detect conditional
31160 copy/constant propagation opportunities.
31162 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31164 * config/visium/visium.md (type): Add trap.
31165 (b): New mode attribute.
31166 (*btst): Rename into...
31167 (*btst<mode>): ...this and adjust.
31168 (*cbranchsi4_btst_insn): Rename into...
31169 (*cbranch<mode>4_btst_insn): ...this and adjust.
31170 (trap): New define_insn.
31172 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31174 PR tree-optimization/79389
31175 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31176 (noce_reversed_cond_code): New function.
31177 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31178 reversed_comparison_code. Formatting fix.
31179 (noce_try_store_flag): Test rev_cond != NULL in addition to
31180 reversed_comparison_code.
31181 (noce_try_store_flag_constants): Likewise.
31182 (noce_try_store_flag_mask): Likewise.
31183 (noce_try_addcc): Use rev_cond if non-NULL instead of
31184 reversed_comparison_code.
31185 (noce_try_cmove_arith): Likewise. Formatting fixes.
31186 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31187 (noce_find_if_block): Initialize rev_cond.
31188 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31189 instead of false as last argument never attempt to reverse it
31192 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31194 PR tree-optimization/79663
31195 * tree-predcom.c (combine_chains): Process refs in reverse order
31196 only for ZERO length chains, and add explaining comment.
31198 2017-02-23 Jeff Law <law@redhat.com>
31200 PR tree-optimization/79578
31201 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31202 in call to operand_equal_p.
31204 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31207 * config/i386/cpuid.h: Fix another undefined behavior.
31209 2017-02-23 Richard Biener <rguenther@suse.de>
31211 PR tree-optimization/79683
31212 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31213 vector types for data-refs.
31215 2017-02-23 Martin Liska <mliska@suse.cz>
31217 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31219 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31221 PR middle-end/79665
31222 * internal-fn.c (get_range_pos_neg): Moved to ...
31223 * tree.c (get_range_pos_neg): ... here. No longer static.
31224 * tree.h (get_range_pos_neg): New prototype.
31225 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31226 are known to be in between 0 and signed maximum inclusive, try to
31227 expand both unsigned and signed divmod and use the cheaper one from
31230 2017-02-22 Jeff Law <law@redhat.com>
31232 PR tree-optimization/79578
31233 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31234 to compare base operands.
31236 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31239 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31240 gpc_reg_operand instead of fpr_reg_operand.
31242 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31244 * config/mips/mips.c (mips_return_in_memory): Force FP
31245 vector types to be returned in memory for o32 ABI.
31247 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31249 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31250 instead of DW_TAG_member for static data member declarations and don't
31251 set no_linkage_name for static inline data members.
31252 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31255 2017-02-22 Martin Liska <mliska@suse.cz>
31257 * doc/invoke.texi: Replace inequality signs with square brackets
31260 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31263 * lra-constraints.c (simplify_operand_subreg): Handle
31264 WORD_REGISTER_OPERATIONS targets.
31266 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31269 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31270 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31271 elimination by swapping fld*.
31273 2017-02-22 Richard Biener <rguenther@suse.de>
31275 PR tree-optimization/79673
31276 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31277 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31278 irrelevant address-space qualifiers and avoiding a
31279 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31281 2017-02-22 Richard Biener <rguenther@suse.de>
31283 PR tree-optimization/79666
31284 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31285 to not symbolically negate if that may introduce undefined
31288 2017-02-22 Martin Liska <mliska@suse.cz>
31291 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31292 * data-streamer-out.c (streamer_write_gcov_count_stream):
31294 * value-prof.c (stream_out_histogram_value): Make assert more
31295 precise based on type of counter.
31297 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31300 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31301 Use nonimmediate_operand instead of memory_operand for operand 1.
31302 (float-extend standard_x87sse_constant_load splitter): Ditto.
31304 2017-02-21 Jeff Law <law@redhat.com>
31306 PR tree-optimization/79621
31307 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31308 blocks with edges to themselves.
31310 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31313 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31314 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31315 Use gimple_call_builtin_p.
31318 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31319 on temporarily removed DEBUG_INSNs.
31321 PR tree-optimization/79649
31322 * tree-loop-distribution.c (classify_partition): Give up on
31323 non-generic address space loads/stores.
31325 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31327 * doc/loop.texi (Loop manipulation): Remove nonexistent
31328 tree_ssa_loop_version from the documentation.
31329 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31331 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31334 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31335 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31336 * config/rs6000/rs6000.c: Include except.h.
31337 (rs6000_expand_split_stack_prologue): Call
31338 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31340 2017-02-21 Martin Jambor <mjambor@suse.cz>
31343 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31344 have been analyzed.
31346 2017-02-21 Martin Jambor <mjambor@suse.cz>
31348 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31349 for backward compatibility only.
31350 * doc/invoke.texi (Option Summary): Remove all references to
31351 -fipa-cp-alignment.
31353 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31357 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31359 * lra-constraints.c (curr_insn_transform): Handle
31360 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31362 2017-02-21 Martin Liska <mliska@suse.cz>
31364 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31366 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31368 PR translation/79638
31369 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31371 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31374 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31375 (arm_function_ok_for_sibcall): Return false for an indirect call by
31376 descriptor if all the argument registers are used.
31377 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31378 alignment of the function.
31380 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31382 PR tree-optimization/61441
31383 * simplify-rtx.c (simplify_const_unary_operation): For
31384 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31385 the sNaN unmodified.
31387 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31389 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31390 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31391 instead of SYSTEM_HEADER_DIR.
31393 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31394 Martin Liška <mliska@suse.cz>
31396 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31397 Fix typos and grammar, use active voice, and clarify.
31399 2017-02-20 Marek Polacek <polacek@redhat.com>
31401 PR middle-end/79537
31402 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31405 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31407 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31410 * config/i386/i386.c (ix86_expand_builtin): Handle
31411 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31412 ix86_builtins_isa[fcode].isa as a requirement of those
31413 flags and any other flag in the bitmask.
31414 (ix86_init_mmx_sse_builtins): Use 0 instead of
31415 ~OPTION_MASK_ISA_64BIT as mask.
31416 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31417 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31418 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31419 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31421 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31424 * lra-constraints.c (split_reg): Check requested split mode
31425 is supported by the register.
31427 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31429 * lra-constraints.c (simplify_operand_subreg): Remove early
31432 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31435 * lra-constraints.c (curr_insn_transform): Tighten condition
31436 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31438 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31441 * lra-constraints.c (curr_insn_transform): Handle
31442 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31444 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31447 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31449 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31451 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31454 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31457 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31459 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31462 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31464 * final.c (last_columnnum, override_columnnum): New variables.
31465 (final_start_function): Set last_columnnum, pass it to begin_prologue
31466 hook and pass 0 to dwarf2out_begin_prologue.
31467 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31468 to source_line debug hook.
31469 (notice_source_line): Compute last_columnnum and for debug_column_info
31470 return true on column changes.
31471 * debug.h (struct gcc_debug_hooks): Add column argument to
31472 source_line and begin_prologue hooks.
31473 (debug_nothing_int_charstar_int_bool): Remove prototype.
31474 (debug_nothing_int_int_charstar,
31475 debug_nothing_int_int_charstar_int_bool): New prototypes.
31476 (dwarf2out_begin_prologue): Add column argument.
31477 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31478 begin_prologue hooks.
31479 (debug_nothing_int_charstar_int_bool): Remove.
31480 (debug_nothing_int_int_charstar,
31481 debug_nothing_int_int_charstar_int_bool): New functions.
31482 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31483 through to dwarf2out_source_line.
31484 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31485 (dwarf2out_source_line): Add column argument, emit it if requested.
31486 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31488 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31489 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31490 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31491 through to dwarf2out_begin_prologue.
31492 (vmsdbgout_source_line): Add column argument, pass it through to
31493 dwarf2out_source_line.
31494 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31495 dbxout_source_line caller.
31496 (dbxout_source_line): Add column argument.
31498 * common.opt (gno-column-info, gcolumn-info): New options.
31499 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31500 (check_die): Also test for multiple DW_AT_decl_column attributes.
31501 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31502 DW_AT_decl_column if requested.
31503 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31505 (gen_variable_die): Likewise.
31506 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31507 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31510 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31511 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31512 (ix86_handle_option): Handle OPT_m3dnowa.
31513 * doc/invoke.texi (-m3dnowa): Document.
31514 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31515 -m3dnowa instead of -m3dnow -march=athlon.
31518 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31519 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31521 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31524 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31525 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31526 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31527 generator for vsx_xxpermdi_<mode>_be.
31528 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31529 force big-endian semantics.
31530 (vsx_xxpermdi_<mode>_be): New define_expand with same
31531 implementation as previous version of vsx_xxpermdi_<mode>.
31533 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31535 PR tree-optimization/79327
31536 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31537 variable, its initialization and use.
31539 2017-02-17 Julia Koval <julia.koval@intel.com>
31541 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31542 (OPTION_MASK_ISA_PKU_UNSET): New.
31543 (ix86_handle_option): Handle -mrdpid.
31544 * config/i386/cpuid.h (bit_RDPID): New.
31545 * config/i386/driver-i386.c (host_detect_local_cpu):
31546 Detect RDPID feature.
31547 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31548 * config/i386/i386-c.c (ix86_target_macros_internal):
31550 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31551 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31552 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31553 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31554 * config/i386/i386.md (define_insn "rdpid"): New.
31555 * config/i386/i386.opt Add -mrdpid.
31556 * config/i386/immintrin.h (_rdpid_u32): New.
31558 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31560 PR rtl-optimization/79541
31561 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31562 instead of transforming it into USE.
31564 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31566 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31567 If HONOR_SNANS (SFmode) force the input to a register.
31568 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31569 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31570 an frsp or similar insn.
31572 2017-02-17 Martin Liska <mliska@suse.cz>
31574 PR rtl-optimization/79577
31575 * params.def (selsched-max-sched-times): Increase minimum to 1.
31577 2017-02-17 Martin Liska <mliska@suse.cz>
31579 PR rtl-optimization/79574
31580 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31582 2017-02-17 Martin Liska <mliska@suse.cz>
31584 PR tree-optimization/79529
31585 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31586 ssa_defined_default_def_p to handle cases which are implicitly
31588 * tree-ssa.c (ssa_defined_default_def_p): New function.
31589 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31590 which are implicitly defined.
31591 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31593 2017-02-17 Richard Biener <rguenther@suse.de>
31595 PR middle-end/79576
31596 * params.def (max-ssa-name-query-depth): Limit to 10.
31598 2017-02-17 Richard Biener <rguenther@suse.de>
31600 PR tree-optimization/79552
31601 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31604 2017-02-17 Richard Biener <rguenther@suse.de>
31607 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31609 2017-02-17 Marek Polacek <polacek@redhat.com>
31611 PR middle-end/79536
31612 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31613 (fold_negate_expr): New wrapper.
31615 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31617 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31618 Correct terminology and de-emphasize pre-standard behavior.
31620 2017-02-16 Alan Modra <amodra@gmail.com>
31622 PR rtl-optimization/79286
31623 * ira.c (def_dominates_uses): New function.
31624 (update_equiv_regs): Don't create an equivalence for insns that
31625 may trap where the register def does not dominate the use.
31627 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31629 PR rtl-optimization/78127
31630 * lra.c (lra): Call lra_eliminate before finish the loop after
31633 2017-02-16 Richard Biener <rguenther@suse.de>
31635 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31637 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31638 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31639 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31640 (isl_val_int_from_wi): New function.
31641 (extract_affine_gmp): Rename to ...
31642 (extract_affine_wi): ... this, take a widest_int.
31643 (extract_affine_int): Just wrap extract_affine_wi.
31644 (add_param_constraints): Use isl_val_int_from_wi.
31645 (add_loop_constraints): Likewise, and extract_affine_wi.
31647 2017-02-15 Jeff Law <law@redhat.com>
31649 PR middle-end/79521
31650 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31651 ira_init_register_move_cost_if_necessary.
31653 2017-02-15 Martin Sebor <msebor@redhat.com>
31655 PR middle-end/32003
31656 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31657 removed in a prior commit.
31659 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31661 PR tree-optimization/79347
31662 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31663 counters during peeling.
31665 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31667 * Makefile.in (site.exp): Remove "set ISLVER".
31669 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31672 * real.c (real_from_integer): Call real_convert even for decimal.
31674 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31677 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31679 2017-02-14 Andrew Pinski <apinski@cavium.com>
31681 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31682 cores and change the partno/implementer to be correct.
31683 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31684 the 'B" as the implementer.
31685 * config/aarch64/aarch64-tune.md: Regenerate.
31687 2017-02-14 Carl Love <cel@us.ibm.com>
31689 * config/rs6000/rs6000.c: Add case statement entry to make the
31690 xvcvuxdsp built-in argument unsigned.
31691 * config/rs6000/vsx.md: Fix the source and return operand types so they
31692 match the instruction definitions from the ISA document. Fix typo
31693 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31696 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31699 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31700 member early_clobber_alts.
31701 * lra-lives.c (reg_early_clobber_p): New.
31702 (process_bb_lives): Use it.
31703 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31704 (debug_operand_data): Initialize early_clobber_alts.
31705 (setup_operand_alternative): Set up early_clobber_alts.
31706 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31707 alternatives to new_insn_reg.
31708 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31710 (lra_update_insn_regno_info): Pass the new arg.
31712 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31714 PR middle-end/79505
31715 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31716 (new_oacc_loop_raw): Don't clear already cleared fields.
31719 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31720 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31721 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31723 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31726 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31728 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31731 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31732 the extra instruction to the right place to store 128-bit constant
31735 2017-02-14 Martin Sebor <msebor@redhat.com>
31737 PR middle-end/79448
31738 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31739 warning for strings of unknown length.
31741 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31743 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31745 2017-02-14 Jeff Law <law@redhat.com>
31748 * ira-costs.c (scan_one_insn): Initialize register move costs
31749 for pseudos seen in USE/CLOBBER insns.
31751 PR tree-optimization/79095
31752 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31753 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31754 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31755 if the operands are known to be not equal, then the resulting range
31757 (intersect_ranges): If the new range is ~[0,0] and the old range is
31758 wide, then prefer ~[0,0].
31759 * tree-vrp.c (overflow_comparison_p_1): New function.
31760 (overflow_comparison_p): New function.
31761 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31762 if NAME is used in an overflow test.
31763 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31764 overflow check that can be expressed as an equality test, then adjust
31765 ops to be that equality test.
31767 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31769 * config/s390/s390-builtin-types.def: Remove flags argument.
31770 * config/s390/s390.c (s390_init_builtins): Likewise.
31772 2017-02-14 Martin Liska <mliska@suse.cz>
31774 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31775 vector. Fix trailing white spaces.
31777 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31779 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31782 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31784 PR rtl-optimization/68664
31785 * config/arm/arm.c (arm_sched_can_speculate_insn):
31786 New function. Declare prototype.
31787 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31789 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31791 PR rtl-optimization/68664
31792 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31794 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31796 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31798 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31799 max skip bytes for function, loop and jump.
31801 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31803 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31804 ABS_EXPR for gimple dump.
31806 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31809 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31811 PR tree-optimization/79408
31812 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31813 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31814 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31815 also if rhs1 is INTEGER_CST.
31817 2017-02-14 Richard Biener <rguenther@suse.de>
31819 PR middle-end/79432
31820 * tree-into-ssa.c (insert_phi_nodes): When the function can
31821 have abnormal edges rewrite SSA names with broken use-def
31822 dominance out of SSA and register them for PHI insertion.
31824 2017-02-13 Martin Sebor <msebor@redhat.com>
31826 PR middle-end/79496
31827 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31828 clearing info.nowrite flag when snprintf size argument is a range.
31830 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31832 * cprop.c (cprop_jump): Add missing space in string literal.
31833 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31834 (get_constraint_for_component_ref): Likewise.
31835 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31836 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31837 * lra-constraints.c (process_alt_operands): Likewise.
31838 * ipa-inline.c (inline_small_functions): Likewise.
31839 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31840 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31841 * trans-mem.c (diagnose_tm_1_op): Likewise.
31842 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31843 (grid_parallel_clauses_gridifiable): Likewise.
31845 * config/nvptx/mkoffload.c (process): Add space in between
31848 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31849 "MOD4_SSE_REGS" and "ALL_REGS".
31851 * spellcheck.c (test_data): Add , in between "foo" and "food".
31853 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31856 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31857 boundary crossing check and subsequent code generation agree.
31859 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31861 * config/aarch64/aarch64.c (has_memory_op): Delete.
31862 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31865 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31867 PR rtl-optimization/79388
31868 PR rtl-optimization/79450
31869 * combine.c (distribute_notes): When removing TEM_INSN for which
31870 corresponding dest has last value recorded, invalidate that last
31873 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31875 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31876 of explicit '@'. Add missing assembly comment marker on branch costs
31879 2017-02-13 Nathan Sidwell <nathan@acm.org>
31881 * gengtype-lex.l (<in_struct>): Add '/'.
31883 2017-02-13 Martin Liska <mliska@suse.cz>
31886 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31888 2017-02-13 Richard Biener <rguenther@suse.de>
31890 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31892 * configure: Re-generate.
31893 * config.in: Likewise.
31894 * graphite-dependences.c: Simplify as if
31895 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31896 * graphite-isl-ast-to-gimple.c: Likewise.
31897 * graphite-optimize-isl.c: Likewise.
31898 * graphite-poly.c: Likewise.
31899 * graphite-sese-to-poly.c: Likewise.
31900 * graphite.h: Likewise.
31901 * toplev.c: Include isl/version.h and use isl_version () for
31902 printing the ISL version.
31903 * doc/install.texi: Update ISL requirement.
31905 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31907 * doc/standards.texi (Standards): Update reference to
31910 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31912 * doc/extend.texi (Named Address Spaces): sourceware.org now
31914 * doc/install.texi (Binaries): Ditto.
31917 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31919 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31920 standard terminology "stringize"/"stringizing" throughout.
31921 * doc/cppinternals.texi: Likewise.
31923 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31925 * doc/extend.texi: Fix some spelling mistakes and typos.
31926 * doc/invoke.texi: Likewise.
31928 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31931 * params.def (inline-min-speedup) Change from 10 to 8.
31933 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31935 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31938 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31941 * ipa-inline-analysis.c (get_minimal_bb): New function.
31942 (record_modified): Use it.
31943 (remap_edge_change_prob): Handle also ancestor functions.
31945 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31947 * doc/contrib.texi (Contributors): Remove broken link into
31948 the Mauve CVS repository.
31950 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31952 PR middle-end/79454
31953 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31954 result computation whenever lhs doesn't have vector mode, not
31955 just when it has BLKmode.
31957 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31959 * doc/makefile.texi (profiledbootstrap): Refer to the
31960 installation instructions only in textual form.
31962 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31965 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31967 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31969 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31970 (Specific): Update mingw-w64 reference.
31972 (Specific): Remove broken link to Renesas RX processor.
31974 2017-02-10 Richard Biener <rguenther@suse.de>
31976 * toplev.c (process_options): Do not mention obsolete graphite
31977 options when printing sorry message about missing graphite support.
31978 Mention -floop-nest-optimize.
31980 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31982 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31983 (vtst_p16): Likewise.
31984 (vtstq_p8): Likewise.
31985 (vtstq_p16): Likewise.
31987 (vtstq_p64): Likewise.
31988 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31989 (vset_lane_p64): New.
31990 (vsetq_lane_p64): New.
31992 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31994 PR tree-optimization/79411
31995 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31996 stmt operands are SSA_NAMEs used in abnormal phis.
31997 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
32000 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32003 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32006 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32008 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32010 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32012 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32016 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32017 not arbitrary TREE_CONSTANT.
32020 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32021 "omp declare target link" attribute unless is_global_var.
32022 * omp-offload.c (find_link_var_op): Likewise.
32024 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32025 Chung-Lin Tang <cltang@codesourcery.com>
32027 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32029 (gimplify_adjust_omp_clauses): Don't delete TILE.
32030 (gimplify_omp_for): Deal with TILE.
32031 * internal-fn.c (expand_GOACC_TILE): New function.
32032 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32034 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32035 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32037 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32038 avoid DIV for outermost collapse var.
32039 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32040 Remove out of date comments, fix whitespace.
32041 * omp-general.c (omp_extract_for_data): Deal with tiling.
32042 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32043 adjust OLF_DIM_BASE value.
32044 (struct omp_for_data): Add tiling field.
32045 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32046 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32047 for auto loops. Remove default auto determining, moved to
32048 oacc_loop_fixed_partitions.
32049 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32050 stmts, add e_mask field.
32051 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32052 (oacc_thread_numbers): Use oacc_dim_call.
32053 (oacc_xform_tile): New.
32054 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32055 (finish_oacc_loop): Adjust for ifns vector.
32056 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32057 add case for GOACC_TILE fns.
32058 (oacc_loop_xform_loop): Delete.
32059 (oacc_loop_process): Iterate over call list directly, and add
32060 handling for GOACC_TILE fns.
32061 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32063 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32064 vector partitioning to outer loops. Assign 2 partitions to loops
32065 when available. Add TILE handling.
32066 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32067 (execite_oacc_device_lower): Process GOACC_TILE fns,
32068 ignore unknown specs.
32069 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32070 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32071 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32073 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32075 * configure.ac (ACX_BUGURL): Update.
32076 * configure: Regenerate.
32078 2017-02-09 Richard Biener <rguenther@suse.de>
32080 PR tree-optimization/69823
32081 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32082 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32084 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32086 * config/arc/arc-c.def: Add __NPS400__ definition.
32087 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32088 (TARGET_NPS400): Define.
32090 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32092 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32094 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32095 pointer, arch_info.
32096 (arc_cpu_types): Fill the arch_info field with a pointer into the
32097 arc_arch_types table.
32098 (arc_selected_cpu): Declare.
32099 * config/arc/arc.c (arc_selected_cpu): Make global.
32100 (arc_selected_arch): Delete.
32101 (arc_base_cpu): Delete.
32102 (arc_override_options): Remove references to deleted variables,
32103 update access to arch information.
32104 (ARC_OPT): Update access to arch information.
32105 (ARC_OPTX): Likewise.
32106 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32107 (TARGET_ARC600): Update access to arch information.
32108 (TARGET_ARC601): Likewise.
32109 (TARGET_ARC700): Likewise.
32110 (TARGET_EM): Likewise.
32111 (TARGET_HS): Likewise.
32112 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32115 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32118 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32119 condition/operands for integer GE/LE/GEU/LEU operations.
32121 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32123 PR translation/79397
32124 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32127 2017-02-08 Martin Jambor <mjambor@suse.cz>
32130 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32131 whether allocation happened.
32132 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32133 nothing was allocated.
32135 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32137 PR tree-optimization/79408
32138 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32139 constant, but SSA_NAME with a known integer range, use the minimum
32140 of that range instead of op1 to determine if modulo can be replaced
32141 with its first operand.
32143 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32145 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32147 2017-02-08 Richard Biener <rguenther@suse.de>
32149 PR tree-optimization/71824
32150 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32151 Check all loops contained in the merged region.
32153 2017-02-07 Andrew Pinski <apinski@cavium.com>
32155 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32157 2017-02-07 Andrew Pinski <apinski@cavium.com>
32159 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32160 (thunderxt88): Likewise.
32161 (thunderxt81): Disable LSE and change v8.1 to v8.
32162 (thunderxt83): Likewise.
32164 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32165 Richard Biener <rguenther@suse.de>
32167 PR middle-end/79399
32168 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32169 type from int to size_t.
32170 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32172 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32174 PR rtl-optimization/79386
32175 * cprop.c (bypass_conditional_jumps): Initialize
32176 bypass_last_basic_block already before splitting bbs after
32177 unconditional traps...
32178 (bypass_conditional_jumps): ... rather than here.
32181 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32182 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32183 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32184 fix -masm=intel patterns.
32186 2017-02-07 Richard Biener <rguenther@suse.de>
32188 PR tree-optimization/79256
32189 PR middle-end/79278
32190 * builtins.c (get_object_alignment_2): Use min_align_of_type
32191 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32192 and ADJUST_FIELD_ALIGN.
32194 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32196 * doc/tm.texi: Regenerate.
32197 * stor-layout.c (layout_decl): Adjust.
32198 (update_alignment_for_field): Likewise.
32199 (place_field): Likewise.
32200 (min_align_of_type): Likewise.
32201 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32202 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32203 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32204 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32205 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32206 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32207 * config/i386/i386.c (x86_field_alignment): Likewise.
32208 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32209 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32210 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32211 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32212 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32213 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32217 2017-01-30 Richard Biener <rguenther@suse.de>
32219 PR tree-optimization/79256
32220 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32221 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32224 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32226 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32227 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32228 builtins to SImode and emit a zero-extend, if necessary.
32230 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32232 * docs/invoke.texi (RISC-V Options): Alphabetize.
32234 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32236 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32239 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32241 * config/riscv/riscv.c: New file.
32242 * common/config/riscv/riscv-common.c: Likewise.
32243 * config.gcc: Likewise.
32244 * config/riscv/constraints.md: Likewise.
32245 * config/riscv/elf.h: Likewise.
32246 * config/riscv/generic.md: Likewise.
32247 * config/riscv/linux.h: Likewise.
32248 * config/riscv/multilib-generator: Likewise.
32249 * config/riscv/peephole.md: Likewise.
32250 * config/riscv/pic.md: Likewise.
32251 * config/riscv/predicates.md: Likewise.
32252 * config/riscv/riscv-builtins.c: Likewise.
32253 * config/riscv/riscv-c.c: Likewise.
32254 * config/riscv/riscv-ftypes.def: Likewise.
32255 * config/riscv/riscv-modes.def: Likewise.
32256 * config/riscv/riscv-opts.h: Likewise.
32257 * config/riscv/riscv-protos.h: Likewise.
32258 * config/riscv/riscv.h: Likewise.
32259 * config/riscv/riscv.md: Likewise.
32260 * config/riscv/riscv.opt: Likewise.
32261 * config/riscv/sync.md: Likewise.
32262 * config/riscv/t-elf-multilib: Likewise.
32263 * config/riscv/t-linux: Likewise.
32264 * config/riscv/t-linux-multilib: Likewise.
32265 * config/riscv/t-riscv: Likewise.
32266 * configure.ac: Likewise.
32267 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32268 Waterman as RISC-V maintainers.
32269 * doc/install.texi: Add RISC-V entries.
32270 * doc/invoke.texi: Add RISC-V options section.
32271 * doc/md.texi: Add RISC-V constraints section.
32272 * configure: Regenerated.
32274 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32277 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32278 false values to be constant vectors with all 0 or all 1 bits set.
32279 (vcondu<mode><mode>): Likewise.
32280 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32282 (fpmask_comparison_operator): Update comment.
32283 (vecint_comparison_operator): New predicate.
32284 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32285 vector conditionals when the true and false values are constant
32286 vectors with all 0 bits or all 1 bits set.
32288 2017-02-06 Martin Sebor <msebor@redhat.com>
32290 PR tree-optimization/79376
32291 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32293 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32295 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32296 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32297 to simplify split condition.
32299 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32301 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32302 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32305 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32307 PR rtl-optimization/68664
32308 * target.def (can_speculate_insn): New hook.
32309 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32310 * doc/tm.texi: Regenerate.
32311 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32312 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32313 (rs6000_sched_can_speculate_insn): New function.
32315 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32317 PR tree-optimization/79284
32318 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32319 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32320 vectorizable_mask_load_store, vectorizable_operation,
32321 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32322 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32323 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32324 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32325 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32326 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32327 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32328 is_gimple_assign (stmt). Replace another such test with
32329 is_gimple_assign (stmt).
32331 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32334 * config/avr/avr.c (rtl-iter.h): Include it.
32335 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32336 (avr_legitimate_combined_insn): ...and implementation.
32338 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32340 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32341 * config/s390/s390.c (s390_const_operand_ok)
32342 (s390_canonicalize_comparison, s390_extract_part)
32343 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32344 (s390_contiguous_bitmask_p, s390_rtx_costs)
32345 (legitimize_pic_address): Likewise.
32346 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32347 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32348 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32349 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32350 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32352 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32354 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32355 REGNO($0) == REGNO($1).
32357 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32359 * config/s390/linux.h(SIZE_TYPE): Add comment.
32361 2017-02-06 Julian Brown <julian@codesourcery.com>
32362 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32363 Virendra Pathak <virendra.pathak@broadcom.com>
32365 * config/aarch64/aarch64-cores.def: Change the scheduler
32367 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32368 * config/aarch64/thunderx2t99.md: New file.
32370 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32372 * doc/standards.texi (Go Language): Update link to language
32375 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32377 * tree-eh.c (lower_resx): Sanitize profile.
32378 (cleanup_empty_eh_move_lp): Likewise.
32380 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32383 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32385 * cfgloopmanip.h (loop_version): Update prototype.
32386 * modulo-sched.c (sms_schedule): Update call of loop_version.
32387 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32388 * tree-parloops.c (gen_parallel_loop): Likewise.
32389 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32390 * tree-ssa-loop-split.c (split_loop): Likewise.
32391 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32392 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32394 2017-02-05 Martin Liska <mliska@suse.cz>
32397 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32399 (print_operand_address): Initialize a struct to zero.
32401 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32403 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32404 garbage collector only in textual form.
32406 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32408 * doc/extend.texi (x86 specific memory model extensions for
32409 transactional memory): Simplify a phrase.
32411 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32414 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32415 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32416 (atomic_storedi_1): Likewise.
32418 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32420 PR tree-optimization/79338
32421 * tree-parloops.c (gather_scalar_reductions): Don't call
32422 vect_analyze_loop_form for loop->inner before destroying loop's
32425 2017-02-03 Martin Sebor <msebor@redhat.com>
32427 PR tree-optimization/79327
32428 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32429 when precision has resulted in leading zeros.
32430 (format_integer): Adjust the likely counter to assume an unknown
32431 argument that may be zero is non-zero.
32433 2017-02-03 Jason Merrill <jason@redhat.com>
32436 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32437 avoid copying non-taken branch.
32439 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32441 PR tree-optimization/79340
32442 * tree-vect-loop.c (vectorizable_reduction): Release
32443 vec_defs elements after safe_splicing them into other vectors.
32446 PR tree-optimization/79327
32447 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32448 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32450 (format_integer): Use wide_int_to_tree instead of build_int_cst
32451 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32452 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32453 of shortest and longest sequence.
32455 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32457 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32458 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32460 2017-02-03 Walter Lee <walt@tilera.com>
32463 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32464 after initial stackframe link reg save.
32465 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32467 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32470 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32471 wu for stxssp alternative.
32473 2017-02-03 Martin Sebor <msebor@redhat.com>
32475 PR tree-optimization/79352
32476 * gimple-fold.c (get_range_strlen): Add argument.
32477 (get_range_strlen): Change return type to bool.
32478 (get_maxval_strlen): Pass in a dummy argument.
32479 * gimple-fold.h (get_range_strlen): Change return type to bool.
32480 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32481 * tree.h (array_at_struct_end_p): Add argument.
32482 * tree.c (array_at_struct_end_p): Handle it.
32484 2017-02-03 Martin Liska <mliska@suse.cz>
32487 * multiple_target.c (create_dispatcher_calls): Redirect edge
32488 from a caller of a dispatcher.
32489 (expand_target_clones): Make the clones local.
32490 (ipa_target_clone): Do both target clones and resolvers.
32491 (ipa_dispatcher_calls): Remove the pass.
32492 (pass_dispatcher_calls::gate): Likewise.
32493 (make_pass_dispatcher_calls): Likewise.
32494 * passes.def (pass_target_clone): Put as very first IPA early
32497 2017-02-03 Martin Liska <mliska@suse.cz>
32499 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32500 in case of a function with ifunc attribute.
32502 2017-02-03 Martin Liska <mliska@suse.cz>
32504 * cgraph.c (cgraph_node::dump): Dump function version info.
32505 * symtab.c (symtab_node::dump_base): Add missing new line.
32507 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32509 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32510 (ifcombine_ifandif): Use it.
32512 2017-02-03 Martin Liska <mliska@suse.cz>
32514 * doc/invoke.texi: Document default value for
32515 use-after-scope-direct-emission-threshold.
32517 2017-02-03 Martin Liska <mliska@suse.cz>
32519 PR tree-optimization/79339
32520 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32521 (format_floating): Likewise.
32523 2017-02-03 Martin Liska <mliska@suse.cz>
32526 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32528 (ipa_node_params_t::remove): Likewise.
32529 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32530 initialization from removed ipa_node_params_t::insert.
32531 (ipa_node_params::~ipa_node_params): Move from removed
32532 ipa_node_params_t::release.
32533 * symbol-summary.h (symbol_summary::m_released): New member.
32534 Do not release a summary twice. Do not allow to call finalizer
32535 for types of a summary that live in GGC memory.
32537 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32539 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32542 2017-02-02 Martin Sebor <msebor@redhat.com>
32544 PR middle-end/79275
32545 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32546 (format_string): Tighten up the range of output for non-constant
32547 strings and correct the expected range for wide non-constant strings.
32549 2017-02-02 Martin Sebor <msebor@redhat.com>
32551 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32553 PR middle-end/32003
32554 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32556 (-fdump-tree-@var): Add to index and document how to come up
32557 with pass-specific option and dump file names.
32558 (-fdump-passes): Clarify where to look for output.
32560 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32562 PR middle-end/77445
32563 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32564 statistics of the analyzed path; allow threading for speed when
32565 any of BBs along the path are optimized for speed.
32567 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32569 PR middle-end/78468
32570 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32571 settings of the virtual registers.
32574 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32576 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32577 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32580 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32582 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32583 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32585 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32587 * config/s390/s390.md: Add missing comments with the expanded
32589 * config/s390/vector.md: Likewise.
32590 * config/s390/vx-builtins.md: Likewise.
32592 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32595 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32596 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32597 conditions on a single line.
32599 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32601 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32602 __S390_VX__ to __VX__.
32604 2017-02-01 Andrew Pinski <apinski@cavium.com>
32606 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32607 stmt_info to record_stmt_cost.
32608 (vect_get_known_peeling_cost): Pass stmt_info if known to
32610 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32611 cpu_vector_cost field into
32612 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32613 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32614 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32615 splitting of scalar_stmt_cost and vec_stmt_cost.
32616 (thunderx_vector_cost): Likewise.
32617 (cortexa57_vector_cost): LIkewise.
32618 (exynosm1_vector_cost): Likewise.
32619 (xgene1_vector_cost): Likewise.
32620 (thunderx2t99_vector_cost): Improve after the splitting of the two
32622 (aarch64_builtin_vectorization_cost): Update for the splitting of
32623 scalar_stmt_cost and vec_stmt_cost.
32625 2017-02-01 Torvald Riegel <triegel@redhat.com>
32626 Richard Henderson <rth@redhat.com>
32628 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32629 conditional on existance of a fast atomic load.
32630 * optabs-query.c (can_atomic_load_p): New function.
32631 * optabs-query.h (can_atomic_load_p): Declare it.
32632 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32633 no fast atomic load is available for the particular size of access.
32634 (expand_atomic_compare_and_swap): Likewise.
32635 (expand_atomic_load): Likewise.
32636 (expand_atomic_store): Likewise.
32637 (expand_atomic_fetch_op): Likewise.
32638 * testsuite/lib/target-supports.exp
32639 (check_effective_target_sync_int_128): Remove x86 because it provides
32640 no fast atomic load.
32641 (check_effective_target_sync_int_128_runtime): Likewise.
32643 2017-02-01 Richard Biener <rguenther@suse.de>
32645 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32646 (graphite_transform_loops): Provide opt-info for optimized nests.
32647 * tree-parloop.c (parallelize_loops): Provide opt-info for
32648 parallelized loops.
32650 2017-02-01 Richard Biener <rguenther@suse.de>
32652 PR middle-end/79315
32653 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32654 was not set before.
32656 2017-02-01 Richard Biener <rguenther@suse.de>
32658 PR tree-optimization/71824
32659 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32660 Verify the loops are valid in the merged SESE region.
32661 (scop_detection::can_represent_loop_1): Check analyzing the
32662 evolution of the number of iterations in the region succeeds.
32664 2017-01-31 Ian Lance Taylor <iant@golang.org>
32666 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32667 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32669 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32671 PR preprocessor/79210
32672 * input.c (get_substring_ranges_for_loc): Replace line_width
32673 assertion with error-handling.
32675 2017-01-31 Richard Biener <rguenther@suse.de>
32677 PR tree-optimization/77318
32678 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32679 (create_pw_aff_from_tree): Take loop parameter.
32680 (add_condition_to_pbb): Pass loop of the condition to
32681 create_pw_aff_from_tree.
32683 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32685 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32686 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32688 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32692 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32694 (convert_int_to_float128): Likewise.
32695 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32696 (convert_int_to_float128): Likewise.
32697 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32698 (UNSPEC_IEEE128_CONVERT): Likewise.
32699 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32700 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32701 Use local variables for IBM extended format.
32702 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32703 (fix_trunc<mode>si2_fprs): Likewise.
32704 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32705 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32706 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32707 to know that we can now have integers of all sizes in vector
32709 (fix<uns>_<mode>di2_hw): Likewise.
32710 (float<uns>_<mode>si2_hw): Likewise.
32711 (fix_<mode>si2_hw): Likewise.
32712 (fixuns_<mode>si2_hw): Likewise.
32713 (float<uns>_<mode>di2_hw): Likewise.
32714 (float_<mode>di2_hw): Likewise.
32715 (float_<mode>si2_hw): Likewise.
32716 (floatuns_<mode>di2_hw): Likewise.
32717 (floatuns_<mode>si2_hw): Likewise.
32718 (xscvqp<su>wz_<mode>): Delete, no longer used.
32719 (xscvqp<su>dz_<mode>): Likewise.
32720 (xscv<su>dqp_<mode>): Likewise.
32721 (ieee128_mfvsrd_64bit): Likewise.
32722 (ieee128_mfvsrd_32bit): Likewise.
32723 (ieee128_mfvsrwz): Likewise.
32724 (ieee128_mtvsrw): Likewise.
32725 (ieee128_mtvsrd_64bit): Likewise.
32726 (ieee128_mtvsrd_32bit): Likewise.
32728 2017-01-31 Martin Liska <mliska@suse.cz>
32731 * ipa-prop.c (ipa_free_all_node_params): Call release method
32732 instead of ~sumbol_summary to not to trigger double times
32735 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32737 PR tree-optimization/71691
32738 * bitmap.h (class auto_bitmap): New.
32739 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32740 is_maybe_undefined instead of ssa_undefined_value_p.
32742 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32744 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32745 __S390_ARCH_LEVEL__ to __ARCH__.
32747 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32749 PR tree-optimization/79267
32750 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32751 if should_remove_lhs_p is true.
32753 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32756 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32757 (add_alignment_attribute): New.
32758 (base_type_die): Add alignment attribute.
32759 (subrange_type_die): Likewise.
32760 (modified_type_die): Likewise.
32761 (gen_array_type_die): Likewise.
32762 (gen_descr_array_type_die: Likewise.
32763 (gen_enumeration_type_die): Likewise.
32764 (gen_subprogram_die): Likewise.
32765 (gen_variable_die): Likewise.
32766 (gen_field_die): Likewise.
32767 (gen_ptr_to_mbr_type_die): Likewise.
32768 (gen_struct_or_union_type_die): Likewise.
32769 (gen_subroutine_type_die): Likewise.
32770 (gen_typedef_die): Likewise.
32771 (base_type_cmp): Compare alignment attribute.
32773 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32776 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32777 (setb_unsigned) New pattern for setb with CCUNS.
32778 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32779 subfc./subfe sequence to avoid overflow problems. Generate a
32780 shorter sequence with cmpld/setb for power9.
32781 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32782 for generating subfc. instruction.
32783 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32784 now uses this instruction.
32786 2017-01-30 Ian Lance Taylor <iant@google.com>
32789 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32790 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32792 2017-01-30 Martin Sebor <msebor@redhat.com>
32794 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32795 Move constant to the right of a relational operator.
32796 (get_mpfr_format_length, format_character, format_string): Ditto.
32797 (should_warn_p, maybe_warn): Same.
32799 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32801 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32804 * asan.c (get_translation_unit_decl): Remove function.
32805 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32807 2017-01-30 Martin Liska <mliska@suse.cz>
32809 PR gcov-profile/79259
32810 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32811 -fprofile-generate.
32813 2017-01-30 Martin Liska <mliska@suse.cz>
32816 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32817 Initialize variables with NULL value.
32819 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32822 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32824 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32826 2017-01-30 Richard Biener <rguenther@suse.de>
32828 PR tree-optimization/79276
32829 * tree-vrp.c (process_assert_insertions): Properly adjust common
32830 when removing a duplicate.
32832 2017-01-30 Richard Biener <rguenther@suse.de>
32834 PR tree-optimization/79256
32835 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32836 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32838 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32840 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32843 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32844 ("*r<noxa>sbg_<mode>_sll_bitmask")
32845 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32846 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32847 Use contiguous_bitmask_nowrap_operand.
32849 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32852 * config/rs6000/altivec.h (vec_xl): Revise #define.
32853 (vec_xst): Likewise.
32855 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32857 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32859 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32861 PR rtl-optimization/79194
32862 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32863 traps before call to bypass_conditional_jumps.
32865 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32867 PR tree-optimization/71374
32868 * lra-constraints.c (check_conflict_input_operands): New.
32869 (match_reload): Use it.
32871 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32874 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32875 account to calculate conflict_set.
32877 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32879 PR rtl-optimization/78559
32880 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32881 other_insn in combine.
32883 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32885 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32886 uint16_type_node for BT_UINT16.
32888 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32890 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32891 "RTL Tests" to menu.
32892 (GIMPLE Tests): New node.
32893 (RTL Tests): New node.
32895 2017-01-27 Richard Biener <rguenther@suse.de>
32897 PR tree-optimization/79245
32898 * tree-loop-distribution.c (distribute_loop): Apply cost
32899 modeling also to detected patterns.
32901 2017-01-27 Richard Biener <rguenther@suse.de>
32903 PR tree-optimization/71433
32904 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32905 (compare_assert_loc): New function.
32906 (process_assert_insertions): Sort and optimize assert locations
32907 to remove duplicates and push down identical assertions on
32908 edges to their destination block.
32910 2017-01-27 Richard Biener <rguenther@suse.de>
32912 PR tree-optimization/79244
32913 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32914 out SSA names even if abnormal.
32916 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32918 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32919 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32920 instead of MPFR_RNDN.
32922 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32925 * arm.c (arm_option_override): Don't call build_target_option_node
32926 until after doing all option overrides.
32927 (arm_valid_target_attribute_tree): Likewise.
32929 2017-01-27 Martin Liska <mliska@suse.cz>
32931 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32932 for {cd}tors and C++ {cd}tors.
32934 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32936 * config/s390/s390.md ("*setmem_long_and")
32937 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32939 2017-01-26 Martin Sebor <msebor@redhat.com>
32941 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32944 2017-01-26 Martin Sebor <msebor@redhat.com>
32946 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32947 HAVE_DFmode before using XFmode or DFmode.
32948 (parse_directive): Avoid using the z length modifier to avoid
32949 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32951 PR middle-end/78703
32952 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32953 to accept adjustment as an array.
32954 (get_int_range): New function.
32955 (struct directive): Make width and prec arrays.
32956 (directive::set_width, directive::set_precision): Call get_int_range.
32957 (format_integer, format_floating): Handle width and precision ranges.
32958 (format_string, parse_directive): Same.
32960 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32963 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32964 comdat_type_p set, just clone them, but keep the children in the
32968 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32969 which have direct callers with -fvar-tracking-assignments enabled
32971 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32972 inside of type units.
32974 2017-01-26 Martin Sebor <msebor@redhat.com>
32976 PR middle-end/78703
32977 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32979 (struct format_result): Replace number_chars, number_chars_min,
32980 and number_chars_max with a single member of struct result_range.
32982 (format_result::operator+=): Adjust.
32983 (struct fmtresult): Remove bounded. Handle likely and unlikely
32985 (fmtresult::adjust_for_width_or_precision): New function.
32986 (fmtresult:type_max_digits): New function.
32987 (bytes_remaining): Handle likely and unlikely counters.
32988 (min_bytes_remaining): Remove.
32989 (format_percent): Simplify.
32990 (format_integer, format_floating): Set likely and unlikely counters.
32991 (get_string_length, format_character, format_string): Same.
32992 (format_plain, should_warn_p): New function.
32993 (maybe_warn): Call should_warn_p. Update diagnostic messages
32994 and handle those for all directives, including plain strings.
32995 (format_directive): Handle likely and unlikely counters.
32996 Remove unnecessary quoting from diagnostics. Add an informational
32998 (add_bytes): Remove.
32999 (pass_sprintf_length::compute_format_length): Simplify.
33000 (try_substitute_return_value): Handle likely and unlikely counters.
33002 2017-01-26 Carl Love <cel@us.ibm.com>
33004 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33005 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33007 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33010 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33011 endianess for subregs into account.
33012 * lra-constraints.c (lra_constraints): Do risky transformations
33013 always on the first iteration.
33014 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33015 last_call_used_reg_set.
33016 (process_bb_lives): Define and use last_call_used_reg_set.
33017 * lra.c (lra): Always continue after lra_constraints on the first
33020 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33022 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33024 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33026 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33028 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33029 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33030 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33031 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33032 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33033 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33034 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33035 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33036 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33038 2017-01-26 Marek Polacek <polacek@redhat.com>
33041 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33042 for the third operand.
33044 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33046 PR middle-end/79236
33047 * omp-low.c (struct omp_context): Add simt_stmt field.
33048 (scan_omp_for): Return omp_context *.
33049 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33050 context to the _simt_ SIMD stmt.
33051 (lower_omp_for): For combined SIMD with sibling _simt_
33052 SIMD, make sure to use the same decls in _looptemp_
33053 clauses as in the sibling.
33055 2017-01-26 David Sherwood <david.sherwood@arm.com>
33057 PR middle-end/79212
33058 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33061 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33064 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33065 emit fld b; fld a; if possible.
33067 * brig-builtins.def: Update copyright years.
33068 * config/arm/arm_acle_builtins.def: Update copyright years.
33070 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33073 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33074 constraint instead of o for the stxsd instruction.
33076 2017-01-25 Carl Love <cel@us.ibm.com>
33078 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33079 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33081 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33083 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33085 2017-01-25 Richard Biener <rguenther@suse.de>
33087 PR tree-optimization/69264
33088 * target.def (vector_alignment_reachable): Improve documentation.
33089 * doc/tm.texi: Regenerate.
33090 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33092 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33093 earlier changes with respect to TYPE_USER_ALIGN.
33094 (vector_alignment_reachable_p): Likewise. Improve dumping.
33096 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33099 * config/arm/arm.md (xordi3): Force constant operand into a register
33102 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33104 * doc/invoke.texi (-fstore-merging): Correct default optimization
33105 levels at which it is enabled.
33106 (-O): Move -fstore-merging from list to...
33109 2017-01-25 Richard Biener <rguenther@suse.de>
33112 * omp-expand.c: Include debug.h.
33113 (expand_omp_taskreg): Make sure to generate early debug before
33114 outlining anything from a function.
33115 (expand_omp_target): Likewise.
33116 (grid_expand_target_grid_body): Likewise.
33118 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33121 * asan.c (get_translation_unit_decl): New function.
33122 (asan_add_global): Extract modules file name from globals
33123 TRANSLATION_UNIT_DECL name.
33125 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33128 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33129 for long calls with APCS frame and VFP.
33131 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33133 * cfg.c (original_copy_tables_initialized_p): New function.
33134 * cfg.h (original_copy_tables_initialized_p): New decl.
33135 * cfgrtl.c (relink_block_chain): Guard the call to
33136 free_original_copy_tables with a call to
33137 original_copy_tables_initialized_p.
33138 * cgraph.h (symtab_node::native_rtl_p): New decl.
33139 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33140 (symtab_node::needed_p): Don't assert for early assembly output
33141 for __RTL functions.
33142 (cgraph_node::finalize_function): Set "force_output" for __RTL
33144 (cgraph_node::analyze): Bail out early for __RTL functions.
33145 (analyze_functions): Update assertion to support __RTL functions.
33146 (cgraph_node::expand): Bail out early for __RTL functions.
33147 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33149 * function.h (struct function): Update comment for field
33151 * gimple-expr.c: Include "tree-pass.h".
33152 (gimple_has_body_p): Return false for __RTL functions.
33153 * Makefile.in (OBJS): Add run-rtl-passes.o.
33154 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33156 (gcc::pass_manager::get_clean_slate): New accessor.
33157 * passes.c: Include "insn-addr.h".
33158 (should_skip_pass_p): Add logging. Update logic for running
33159 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33160 property-provider override so it is only done for gimple passes.
33162 (skip_pass): New function.
33163 (execute_one_pass): Call skip_pass when skipping passes.
33164 * read-md.c (md_reader::read_char): Support filtering
33165 the input to a subset of line numbers.
33166 (md_reader::md_reader): Initialize fields
33167 m_first_line and m_last_line.
33168 (md_reader::read_file_fragment): New function.
33169 * read-md.h (md_reader::read_file_fragment): New decl.
33170 (md_reader::m_first_line): New field.
33171 (md_reader::m_last_line): New field.
33172 * read-rtl-function.c (function_reader::create_function): Only
33173 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33174 curr_properties. Set DECL_INITIAL to a dummy block.
33175 (read_rtl_function_body_from_file_range): New function.
33176 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33178 * run-rtl-passes.c: New file.
33179 * run-rtl-passes.h: New file.
33181 2017-01-24 Jeff Law <law@redhat.com>
33183 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33186 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33188 PR tree-optimization/79159
33189 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33190 (record_nonwrapping_iv): Improve boundary using above function if no
33191 value range information.
33193 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33194 Martin Jambor <mjambor@suse.cz>
33196 * brig-builtins.def: New file.
33197 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33198 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33199 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33200 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33201 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33202 * builtin-types.def (BT_INT8): New.
33203 (BT_INT16): Likewise.
33204 (BT_UINT8): Likewise.
33205 (BT_UINT16): Likewise.
33206 (BT_FN_ULONG): Likewise.
33207 (BT_FN_UINT_INT): Likewise.
33208 (BT_FN_UINT_ULONG): Likewise.
33209 (BT_FN_UINT_LONG): Likewise.
33210 (BT_FN_UINT_PTR): Likewise.
33211 (BT_FN_ULONG_PTR): Likewise.
33212 (BT_FN_INT8_FLOAT): Likewise.
33213 (BT_FN_INT16_FLOAT): Likewise.
33214 (BT_FN_UINT32_FLOAT): Likewise.
33215 (BT_FN_UINT16_FLOAT): Likewise.
33216 (BT_FN_UINT8_FLOAT): Likewise.
33217 (BT_FN_UINT64_FLOAT): Likewise.
33218 (BT_FN_UINT16_UINT32): Likewise.
33219 (BT_FN_UINT32_UINT16): Likewise.
33220 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33221 (BT_FN_INT_PTR_INT): Likewise.
33222 (BT_FN_UINT_PTR_UINT): Likewise.
33223 (BT_FN_LONG_PTR_LONG): Likewise.
33224 (BT_FN_ULONG_PTR_ULONG): Likewise.
33225 (BT_FN_VOID_UINT64_UINT64): Likewise.
33226 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33227 (BT_FN_INT8_INT8_INT8): Likewise.
33228 (BT_FN_INT16_INT16_INT16): Likewise.
33229 (BT_FN_INT_INT_INT): Likewise.
33230 (BT_FN_UINT_FLOAT_UINT): Likewise.
33231 (BT_FN_FLOAT_UINT_UINT): Likewise.
33232 (BT_FN_ULONG_UINT_UINT): Likewise.
33233 (BT_FN_ULONG_UINT_PTR): Likewise.
33234 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33235 (BT_FN_UINT_UINT_UINT): Likewise.
33236 (BT_FN_VOID_UINT_PTR): Likewise.
33237 (BT_FN_UINT_UINT_PTR: Likewise.
33238 (BT_FN_UINT32_UINT64_PTR): Likewise.
33239 (BT_FN_INT_INT_UINT_UINT): Likewise.
33240 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33241 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33242 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33243 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33244 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33245 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33246 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33247 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33248 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33249 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33250 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33251 * doc/frontends.texi: List BRIG FE.
33252 * doc/install.texi (Testing): Add BRIG tesring requirements.
33253 * doc/invoke.texi (Overall Options): Mention BRIG.
33254 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33256 2017-01-24 Richard Biener <rguenther@suse.de>
33258 PR translation/79208
33259 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33261 2017-01-24 Martin Jambor <mjambor@suse.cz>
33264 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33265 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33266 and known_contexts.
33268 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33270 PR middle-end/79123
33271 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33272 casts from signed to unsigned really don't have a range.
33274 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33276 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33277 GMP_RNDx for compatiblity.
33279 2017-01-24 Martin Liska <mliska@suse.cz>
33282 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33283 that would prevent us to call alloca with -1 as argument.
33285 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33287 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33288 Avoid -Wformat-security warning.
33290 2017-01-23 Andrew Pinski <apinski@cavium.com>
33292 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33295 2017-01-23 Martin Sebor <msebor@redhat.com>
33297 PR middle-end/78703
33298 * gimple-ssa-sprintf.c (warn_level): New global.
33299 (format_integer): Use it here and throughout the rest of the file.
33300 Use the same switch to compute sign as base.
33301 (maybe_warn): New function.
33302 (format_directive): Factor out warnings into maybe_warn.
33303 Add debugging output. Use warn_level.
33304 (add_bytes): Use warn_level.
33305 (pass_sprintf_length::compute_format_length): Add debugging output.
33306 (try_substitute_return_value): Same.
33307 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33309 PR middle-end/78703
33310 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33311 (struct fmtresult, format_integer, format_floating): Adjust.
33312 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33313 (get_string_length, format_string,format_directive): Same.
33314 (pass_sprintf_length::compute_format_length): Same.
33315 (try_substitute_return_value): Simplify slightly.
33317 PR middle-end/78703
33318 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33319 (fmtresult::operator+=): Outlined.
33320 (struct fmtresult): Add ctors.
33321 (struct conversion_spec): Rename...
33322 (struct directive): ...to this. Add and remove data members.
33323 (directive::set_width, directive::set_precision): New functions.
33324 (format_percent): Use fmtresult ctor.
33325 (get_width_and_precision): Remove.
33326 (format_integer): Make naming changes. Avoid computing width and
33328 (format_floating): Same. Adjust indentation.
33329 (format_character, format_none): New functions.
33330 (format_string): Moved character handling to format_character.
33331 (format_directive): Remove arguments, change return type.
33332 (parse_directive): New function.
33333 (pass_sprintf_length::compute_format_length): Move directive
33334 parsing to parse_directive.
33336 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33338 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33339 (assign_assembler_name_if_needed): ... this.
33340 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33341 (assign_assembler_name_if_needed): ... this.
33342 (free_lang_data_in_cgraph): Adjust callers.
33343 * cgraphunit.c (cgraph_node::analyze): Likewise.
33344 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33347 2017-01-23 Richard Biener <rguenther@suse.de>
33349 PR tree-optimization/79088
33350 PR tree-optimization/79188
33351 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33352 resetting loop bounds after last path deletion. Reset loop
33353 bounds of the target loop, make code match the comments.
33354 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33355 Make sure loops need no fixups.
33357 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33359 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33360 exponent support with double type for first argument.
33361 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33362 type returned by __builtin_vec_extract_sig,
33363 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33364 functions from "vector int" to "vector unsigned int" or from
33365 "vector long long int" to "vector unsigned long long int".
33366 Changed type returned by __builtin_vec_extract_exp,
33367 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33368 functions from "vector int" to "vector unsigned int" or from
33369 "vector long long int" to "vector unsigned long long int".
33370 Changed return type of __builtin_vec_test_data_class,
33371 __builtin_vec_test_data_class_sp, and
33372 __builtin_vec_test_data_class_dp from "vector int" to
33373 "vector bool int" or from "vector long long int" to "vector bool
33374 long long int" and changed second argument type from "unsigned
33375 int" to "int". Added new overloaded function forms "vector float
33376 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33377 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33378 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33379 double, vector unsigned long long int)" and "vector double
33380 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33381 long int)". Changed return type of
33382 __builtin_scalar_test_data_class and
33383 __builtin_scalar_test_data_class_sp and
33384 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33385 int" and changed second argument from "unsigned int" to "int".
33386 Changed type returned by __builtin_scalar_test_neg,
33387 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33388 from "int" to "bool int". Added new overloaded function form
33389 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33390 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33391 exponent double-precision with floating point first argument.
33392 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33393 documentation of scalar_test_data_class, scalar_test_neg,
33394 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33395 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33396 vec_test_data_class built-in functions to reflect refinements in
33397 their type signatures.
33399 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33401 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33403 (aarch64_elf_asm_destructor): Likewise.
33405 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33407 PR rtl-optimization/78634
33408 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33409 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33410 * ifcvt.c (noce_try_cmove): Add missing cost check.
33412 PR rtl-optimization/71724
33413 * combine.c (if_then_else_cond): Look for situations where it is
33414 beneficial to undo the work of one of the recursive calls.
33416 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33418 PR tree-optimization/70754
33419 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33420 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33421 combined stmt before it if not NULL.
33422 (combine_chains): Process refs reversely and compute dominance point
33425 2017-01-23 Martin Liska <mliska@suse.cz>
33427 PR tree-optimization/79196
33428 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33429 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33431 (strlen_optimize_stmt): Call the renamed function.
33433 2017-01-23 Michael Matz <matz@suse.de>
33435 PR tree-optimization/78384
33436 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33438 2017-01-23 Richard Biener <rguenther@suse.de>
33440 PR tree-optimization/79186
33441 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33442 both incoming edges before moving an assert.
33444 2017-01-23 Martin Jambor <mjambor@suse.cz>
33446 * ipa-prop.c (load_from_param_1): Removed.
33447 (load_from_unmodified_param): Bits from load_from_param_1 put back
33449 (load_from_param): Removed.
33450 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33451 with stmt. Reverted back to use of load_from_unmodified_param.
33453 2017-01-23 Martin Jambor <mjambor@suse.cz>
33456 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33457 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33458 field a pointer to garbage collected vector, mark lattices and
33459 ipcp_orig_node with GTY((skip)).
33460 (ipa_get_param_count): Adjust to descriptors being a pointer.
33461 (ipa_get_param): Likewise.
33462 (ipa_get_type): Likewise.
33463 (ipa_get_param_move_cost): Likewise.
33464 (ipa_set_param_used): Likewise.
33465 (ipa_get_controlled_uses): Likewise.
33466 (ipa_set_controlled_uses): Likewise.
33467 (ipa_is_param_used): Likewise.
33468 (ipa_node_params_t): Move into garbage collector. New methods insert
33470 (ipa_node_params_sum): Annotate wth GTY(()).
33471 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33473 (ipa_load_from_parm_agg): Adjust declaration.
33474 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33475 * ipa-profile.c (ipa_profile): Likewise.
33476 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33477 (ipa_populate_param_decls): Make descriptors parameter garbage
33479 (ipa_dump_param): Adjust to descriptors being a pointer.
33480 (ipa_alloc_node_params): Likewise.
33481 (ipa_initialize_node_params): Likewise.
33482 (load_from_param_1): Make descriptors parameter garbage collected.
33483 (load_from_unmodified_param): Likewise.
33484 (load_from_param): Likewise.
33485 (ipa_load_from_parm_agg): Likewise.
33486 (ipa_node_params::~ipa_node_params): Removed.
33487 (ipa_free_all_node_params): Remove call to delete operator.
33488 (ipa_node_params_t::insert): New.
33489 (ipa_node_params_t::remove): Likewise.
33490 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33491 copy known_csts and known_contexts vectors.
33492 (ipa_read_node_info): Adjust to descriptors being a pointer.
33493 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33495 (ipcp_transform_function): Make descriptors variable garbage
33498 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33500 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33501 * config/i386/avx512dqintrin.h: Ditto.
33502 * config/i386/avx512fintrin.h: Ditto.
33503 * config/i386/i386.c: Handle new builtins.
33504 * config/i386/i386-builtin.def: Add new builtins.
33505 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33506 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33508 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33509 Martin Liska <mliska@suse.cz>
33511 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33512 * asan.c (asan_expand_poison_ifn): Support stores and use
33513 appropriate ASAN report function.
33514 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33515 * internal-fn.def (ASAN_POISON_USE): Declare.
33516 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33517 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33518 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33519 ASAN_POISON calls w/o LHS.
33520 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33521 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33522 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33523 * gimplify.c (asan_poison_variables): Add attribute
33524 use_after_scope_memory to variables that really needs to live
33526 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33527 having the attribute.
33529 2017-01-23 Martin Liska <mliska@suse.cz>
33531 * asan.c (create_asan_shadow_var): New function.
33532 (asan_expand_poison_ifn): Likewise.
33533 * asan.h (asan_expand_poison_ifn): New declaration.
33534 * internal-fn.c (expand_ASAN_POISON): Likewise.
33535 * internal-fn.def (ASAN_POISON): New builtin.
33536 * sanopt.c (pass_sanopt::execute): Expand
33537 asan_expand_poison_ifn.
33538 * tree-inline.c (copy_decl_for_dup_finish): Make function
33540 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33541 * tree-ssa.c (is_asan_mark_p): New function.
33542 (execute_update_addresses_taken): Rewrite local variables
33543 (identified just by use-after-scope as addressable) into SSA.
33545 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33547 * doc/install.texi (Specific): opensource.apple.com uses https
33548 now. Remove trailing slash.
33550 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33552 * README.Portability: Remove note on an Irix compatibility issue.
33554 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33556 * gcov.c (INCLUDE_ALGORITHM): Define.
33557 (INCLUDE_VECTOR): Define.
33558 No longer include <vector> and <algorithm> directly.
33560 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33562 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33564 * doc/invoke.texi (Code Gen Options): Ditto.
33566 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33569 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33571 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33573 rtl-optimization/79125
33574 * cprop.c (local_cprop_pass): Handle cases where we make an
33575 unconditional trap.
33577 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33581 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33582 read from, for big endian.
33584 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33586 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33587 register pauth builtins for LP64 only.
33589 2017-01-20 Marek Polacek <polacek@redhat.com>
33592 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33595 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33597 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33599 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33600 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33601 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33603 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33606 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33607 in big-endian mode when they are not a single duplicated value.
33609 2017-01-20 Richard Biener <rguenther@suse.de>
33611 * BASE-VER: Bump to 7.0.1.
33613 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33615 * omp-low.c (omplow_simd_context): New struct. Use it...
33616 (lower_rec_simd_input_clauses): ...here and...
33617 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33618 references to idx, lane, max_vf, is_simt.
33620 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33622 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33625 2017-01-20 Martin Jambor <mjambor@suse.cz>
33627 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33628 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33630 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33631 (GTFILES): Rename hsa.c to hsa-common.c.
33632 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33633 * hsa-dump.c: Likewise.
33634 * hsa-gen.c: Likewise.
33635 * hsa-regalloc.c: Likewise.
33636 * ipa-hsa.c: Likewise.
33637 * omp-expand.c: Likewise.
33638 * omp-low.c: Likewise.
33639 * toplev.c: Likewise.
33641 2017-01-20 Marek Polacek <polacek@redhat.com>
33644 * doc/invoke.texi: Document -Wduplicated-branches.
33645 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33646 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33647 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33648 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33649 return 0 only when not OEP_LEXICOGRAPHIC.
33650 (fold_build_cleanup_point_expr): Use the expression
33651 location when building CLEANUP_POINT_EXPR.
33652 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33653 * tree.c (add_expr): Handle error_mark_node.
33655 2017-01-20 Martin Liska <mliska@suse.cz>
33658 * tree-profile.c (init_ic_make_global_vars): Do not call
33660 (gimple_init_gcov_profiler): Likewise.
33662 2017-01-20 Martin Liska <mliska@suse.cz>
33665 * cgraph.h (maybe_create_reference): Remove argument and
33667 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33669 * ipa-cp.c (create_specialized_node): Likewise.
33670 * symtab.c (symtab_node::maybe_create_reference): Handle
33671 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33673 2017-01-20 Martin Liska <mliska@suse.cz>
33675 * read-rtl-function.c (function_reader::create_function): Use
33676 build_decl instread of build_decl_stat.
33678 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33680 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33681 * config/i386/avx512dqintrin.h: Ditto.
33682 * config/i386/avx512fintrin.h: Ditto.
33683 * config/i386/i386-builtin-types.def: Add new types.
33684 * config/i386/i386.c: Handle new types.
33685 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33686 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33687 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33688 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33689 (__builtin_ia32_kshiftridi): New.
33690 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33692 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33696 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33697 define to rs6000_init_stack_protect_guard.
33698 (rs6000_init_stack_protect_guard): New function.
33700 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33701 Yunqiang Su <yunqiang.su@imgtec.com>
33703 * config.gcc (supported_defaults): Add madd4.
33704 (with_madd4): Add validation.
33705 (all_defaults): Add madd4.
33706 * config/mips/mips.opt (mmadd4): New option.
33707 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33709 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33711 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33712 (ISA_HAS_FUSED_MADD4): Likewise.
33713 * doc/invoke.texi (-mmadd4): Document the new option.
33714 * doc/install.texi (--with-madd4): Document the new option.
33716 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33718 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33719 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33720 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33721 (aarch64_init_pauth_hint_builtins): New.
33722 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33723 (aarch64_expand_builtin): Expand new builtins.
33725 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33727 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33728 * combine-stack-adj.c (no_unhandled_cfa): Handle
33729 REG_CFA_TOGGLE_RA_MANGLE.
33730 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33731 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33732 info for return address signing.
33733 (aarch64_expand_epilogue): Likewise.
33735 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33737 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33738 * config/aarch64/aarch64-protos.h
33739 (aarch64_return_address_signing_enabled): New declaration.
33740 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33742 (aarch64_expand_prologue): Sign return address before it's pushed onto
33744 (aarch64_expand_epilogue): Authenticate return address fetched from
33746 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33747 (aarch64_attributes): New function attributes for "sign-return-address".
33748 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33749 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33750 ("*do_return"): Generate combined instructions according to key index.
33751 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33752 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33754 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33755 * config/aarch64/aarch64.opt (msign-return-address=): New.
33756 * doc/extend.texi (AArch64 Function Attributes): Documents
33757 "sign-return-address=".
33758 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33760 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33762 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33763 overall option summary.
33765 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33767 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33768 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33769 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33770 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33772 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33774 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33775 -mpower9-minmax by default for -mcpu=power9.
33776 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33777 128-bit floating point.
33779 2017-01-20 Alan Modra <amodra@gmail.com>
33781 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33782 optimizing for size.
33784 2017-01-20 Alan Modra <amodra@gmail.com>
33787 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33788 for strcmp and strncmp from corresponding builtin decl.
33790 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33792 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33793 instead of i386/rtems-64.h.
33794 * config/i386/rtems-64.h: Remove.
33796 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33800 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33802 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33804 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33806 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33807 Change int to HOST_WIDE_INT.
33808 * config/aarch64/aarch64-protos.h
33809 (aarch64_simd_gen_const_vector_dup): Likewise.
33810 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33812 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33814 * langhooks-def.h (lhd_type_for_size): New decl.
33815 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33816 * langhooks.c (lhd_type_for_size): New function, taken from
33819 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33821 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33822 define_bypass for CR latency.
33823 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33824 (power9-alu2): Add define_bypass for CR latency.
33826 (power9-mul): Update insn latency.
33827 (power9-mul-compare): Update insn latency, bypass latency and remove
33830 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33832 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33834 * config/aarch64/aarch64.md
33835 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33836 aarch64_nopcrelative_literal_loads.
33837 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33839 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33841 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33842 TARGET_LOONGSON_3A.
33843 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33845 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33848 * config.gcc (supported_defaults): Add lxc1-sxc1.
33849 (with_lxc1_sxc1): Add validation.
33850 (all_defaults): Add lxc1-sxc1.
33851 * config/mips/mips.opt (mlxc1-sxc1): New option.
33852 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33854 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33855 __mips_no_lxc1_sxc1.
33856 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33857 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33858 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33860 2017-01-19 Richard Biener <rguenther@suse.de>
33862 PR tree-optimization/72488
33863 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33864 sure to restore SSA info.
33865 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33867 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33869 PR rtl-optimization/79121
33870 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33871 of the inner type when shifting an extended value.
33873 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33876 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33877 interposable aliases.
33879 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33882 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33883 Use the evmergelohi instruction.
33884 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33885 (mov_sitf_e500_subreg8_2_be): Likewise.
33886 (mov_sitf_e500_subreg12_2_le): Likewise.
33887 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33888 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33889 (mov_sitf_e500_subreg8_2_le): Likewise.
33890 (mov_sitf_e500_subreg12_2_be): Likewise.
33892 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33894 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33895 attribute from vecsimple to vecperm.
33896 (altivec_vbpermq2): Likewise.
33898 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33901 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33903 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33904 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33905 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33906 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33907 case where N arg is SIZE_MAX.
33908 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33909 (cmpstrsi): Add pattern.
33911 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33913 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33914 __builtin_vec_revb builtins.
33915 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33916 built-in functions to support generation of the ISA 3.0 XXBR<x>
33917 vector byte reverse instructions.
33918 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33919 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33920 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33921 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33922 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33923 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33924 (P9V_BUILTIN_VEC_REVB): Likewise.
33925 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33926 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33927 (p9_xxbrq_v16qi): Likewise.
33928 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33929 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33930 (p9_xxbrh_v8hi): Likewise.
33931 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33932 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33933 vec_revb built-in functions.
33935 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33937 PR rtl-optimization/78952
33938 * config/i386/i386.md (any_extract): New code iterator.
33939 (*insvqi_2): Use any_extract for source operand.
33940 (*insvqi_3): Use any_shiftrt for source operand.
33942 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33944 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33946 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33948 2017-01-18 Matthias Klose <doko@ubuntu.com>
33950 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33952 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33954 * config/rs6000/altivec.h (vec_bperm): Change #define.
33955 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33956 (altivec_vbpermq2): New define_insn.
33957 (altivec_vbpermd): Likewise.
33958 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33959 function interface.
33960 (VBPERMD): Likewise.
33961 (VBPERM): New polymorphic function interface.
33962 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33963 Add entries for P9V_BUILTIN_VEC_VBPERM.
33964 * doc/extend.texi: Add interfaces for vec_bperm.
33966 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33968 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33969 first letter of error messages.
33970 (s390_resolve_overloaded_builtin): Likewise.
33971 * config/s390/s390.c (s390_expand_builtin): Likewise.
33972 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33973 (s390_valid_target_attribute_inner_p): Likewise.
33974 * config/s390/s390.md ("tabort"): Likewise.
33976 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33978 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33979 (ISA_AVOID_DIV_HILO): New macro.
33980 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33981 (ISA_HAS_DDIV): Likewise.
33983 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33985 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33987 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33989 * doc/invoke.texi (fabi-version): Spelling fix.
33991 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33994 * doc/invoke.texi (fabi-version): Mention mangling fix for
33997 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
34000 * doc/invoke.texi (fabi-version): Document discriminator mangling.
34002 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34005 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34006 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34008 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34010 (stack_protect_test): Ditto.
34011 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34012 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34014 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34015 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34016 -mstack-protector-guard-offset=.
34017 (RS/6000 and PowerPC Options): Ditto.
34019 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34021 * config/i386/i386.h (MASK_CLASS_P): New define.
34022 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34023 there are no registers from different register sets also when
34024 mask registers are used. Update function comment.
34025 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34026 to (*k/*r) and (*k/*km) alternatives.
34028 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34030 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34031 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34032 (EH_RETURN_HANDLER_RTX): New define.
34033 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34034 Force frame pointer in EH return functions.
34035 (aarch64_expand_epilogue): Add barrier for eh_return.
34036 (aarch64_final_eh_return_addr): Remove.
34037 (aarch64_eh_return_handler_rtx): New function.
34038 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34040 (aarch64_eh_return_handler_rtx): New prototype.
34042 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34044 * config/rs6000/altivec.h (vec_rlmi): New #define.
34045 (vec_vrlnm): Likewise.
34046 (vec_rlnm): Likewise.
34047 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34048 (UNSPEC_VRLNM): Likewise.
34049 (VIlong): New mode iterator.
34050 (altivec_vrl<VI_char>mi): New define_insn.
34051 (altivec_vrl<VI_char>nm): Likewise.
34052 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34054 (VRLDNM): Likewise.
34055 (RLNM): New polymorphic function entry.
34056 (VRLWMI): New monomorphic function entry.
34057 (VRLDMI): Likewise.
34058 (RLMI): New polymorphic function entry.
34059 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34060 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34061 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34064 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34067 * dwarf2out.c (field_byte_offset): Restore the
34068 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34069 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34070 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34073 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34076 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34078 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34081 * dwarf2out.c (add_data_member_location_attribute): For constant
34082 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34083 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34084 DW_AT_byte_size attributes.
34086 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34088 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34089 after forcing to constant memory when the code model is medium.
34091 2017-01-17 Julia Koval <julia.koval@intel.com>
34094 * config/i386/avx512fintrin.h
34095 (_mm512_i32gather_ps): Change __addr type to void const*.
34096 (_mm512_mask_i32gather_ps): Ditto.
34097 (_mm512_i32gather_pd): Ditto.
34098 (_mm512_mask_i32gather_pd): Ditto.
34099 (_mm512_i64gather_ps): Ditto.
34100 (_mm512_mask_i64gather_ps): Ditto.
34101 (_mm512_i64gather_pd): Ditto.
34102 (_mm512_mask_i64gather_pd): Ditto.
34103 (_mm512_i32gather_epi32): Ditto.
34104 (_mm512_mask_i32gather_epi32): Ditto.
34105 (_mm512_i32gather_epi64): Ditto.
34106 (_mm512_mask_i32gather_epi64): Ditto.
34107 (_mm512_i64gather_epi32): Ditto.
34108 (_mm512_mask_i64gather_epi32): Ditto.
34109 (_mm512_i64gather_epi64): Ditto.
34110 (_mm512_mask_i64gather_epi64): Ditto.
34111 (_mm512_i32scatter_ps): Change __addr type to void*.
34112 (_mm512_mask_i32scatter_ps): Ditto.
34113 (_mm512_i32scatter_pd): Ditto.
34114 (_mm512_mask_i32scatter_pd): Ditto.
34115 (_mm512_i64scatter_ps): Ditto.
34116 (_mm512_mask_i64scatter_ps): Ditto.
34117 (_mm512_i64scatter_pd): Ditto.
34118 (_mm512_mask_i64scatter_pd): Ditto.
34119 (_mm512_i32scatter_epi32): Ditto.
34120 (_mm512_mask_i32scatter_epi32): Ditto.
34121 (_mm512_i32scatter_epi64): Ditto.
34122 (_mm512_mask_i32scatter_epi64): Ditto.
34123 (_mm512_i64scatter_epi32): Ditto.
34124 (_mm512_mask_i64scatter_epi32): Ditto.
34125 (_mm512_i64scatter_epi64): Ditto.
34126 (_mm512_mask_i64scatter_epi64): Ditto.
34127 * config/i386/avx512pfintrin.h
34128 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34129 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34130 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34131 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34132 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34133 (_mm512_prefetch_i32scatter_ps): Ditto.
34134 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34135 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34136 (_mm512_prefetch_i64scatter_pd): Ditto.
34137 (_mm512_prefetch_i64scatter_ps): Ditto.
34138 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34139 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34140 * config/i386/avx512vlintrin.h
34141 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34142 (_mm_mmask_i32gather_ps): Ditto.
34143 (_mm256_mmask_i32gather_pd): Ditto.
34144 (_mm_mmask_i32gather_pd): Ditto.
34145 (_mm256_mmask_i64gather_ps): Ditto.
34146 (_mm_mmask_i64gather_ps): Ditto.
34147 (_mm256_mmask_i64gather_pd): Ditto.
34148 (_mm_mmask_i64gather_pd): Ditto.
34149 (_mm256_mmask_i32gather_epi32): Ditto.
34150 (_mm_mmask_i32gather_epi32): Ditto.
34151 (_mm256_mmask_i32gather_epi64): Ditto.
34152 (_mm_mmask_i32gather_epi64): Ditto.
34153 (_mm256_mmask_i64gather_epi32): Ditto.
34154 (_mm_mmask_i64gather_epi32): Ditto.
34155 (_mm256_mmask_i64gather_epi64): Ditto.
34156 (_mm_mmask_i64gather_epi64): Ditto.
34157 (_mm256_i32scatter_ps): Change __addr type to void*.
34158 (_mm256_mask_i32scatter_ps): Ditto.
34159 (_mm_i32scatter_ps): Ditto.
34160 (_mm_mask_i32scatter_ps): Ditto.
34161 (_mm256_i32scatter_pd): Ditto.
34162 (_mm256_mask_i32scatter_pd): Ditto.
34163 (_mm_i32scatter_pd): Ditto.
34164 (_mm_mask_i32scatter_pd): Ditto.
34165 (_mm256_i64scatter_ps): Ditto.
34166 (_mm256_mask_i64scatter_ps): Ditto.
34167 (_mm_i64scatter_ps): Ditto.
34168 (_mm_mask_i64scatter_ps): Ditto.
34169 (_mm256_i64scatter_pd): Ditto.
34170 (_mm256_mask_i64scatter_pd): Ditto.
34171 (_mm_i64scatter_pd): Ditto.
34172 (_mm_mask_i64scatter_pd): Ditto.
34173 (_mm256_i32scatter_epi32): Ditto.
34174 (_mm256_mask_i32scatter_epi32): Ditto.
34175 (_mm_i32scatter_epi32): Ditto.
34176 (_mm_mask_i32scatter_epi32): Ditto.
34177 (_mm256_i32scatter_epi64): Ditto.
34178 (_mm256_mask_i32scatter_epi64): Ditto.
34179 (_mm_i32scatter_epi64): Ditto.
34180 (_mm_mask_i32scatter_epi64): Ditto.
34181 (_mm256_i64scatter_epi32): Ditto.
34182 (_mm256_mask_i64scatter_epi32): Ditto.
34183 (_mm_i64scatter_epi32): Ditto.
34184 (_mm_mask_i64scatter_epi32): Ditto.
34185 (_mm256_i64scatter_epi64): Ditto.
34186 (_mm256_mask_i64scatter_epi64): Ditto.
34187 (_mm_i64scatter_epi64): Ditto.
34188 (_mm_mask_i64scatter_epi64): Ditto.
34189 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34190 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34191 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34192 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34193 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34194 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34195 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34196 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34197 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34198 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34199 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34200 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34201 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34202 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34203 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34204 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34205 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34206 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34207 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34208 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34209 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34210 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34211 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34212 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34213 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34214 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34215 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34216 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34217 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34218 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34219 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34220 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34221 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34222 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34223 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34224 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34225 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34226 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34227 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34228 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34229 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34230 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34231 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34232 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34233 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34234 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34235 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34236 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34237 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34238 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34239 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34240 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34241 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34242 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34243 definitions accordingly.
34245 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34246 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34249 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34252 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34255 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34256 hard regs for inner regno.
34258 2017-01-17 Martin Liska <mliska@suse.cz>
34261 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34262 assumption and add comment.
34264 2017-01-17 Nathan Sidwell <nathan@acm.org>
34266 * ipa-visibility.c (localize_node): New function, broken out of ...
34267 (function_and_variable_visibility): ... here. Call it.
34269 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34271 PR middle-end/77445
34272 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34273 correctly set frequency of oudgoing edge.
34274 (duplicate_thread_path): Fix profile updating.
34276 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34279 * configure.ac: Add GCC_BASE_VER.
34280 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34281 version from BASE-VER file.
34282 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34283 (gcc.o): Depend on $(BASEVER).
34284 * common.opt (dumpfullversion): New option.
34285 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34286 * doc/invoke.texi: Document -dumpfullversion.
34287 * doc/install.texi: Document --with-gcc-major-version-only.
34288 * configure: Regenerated.
34290 2017-01-17 Richard Biener <rguenther@suse.de>
34292 PR tree-optimization/71433
34293 * tree-vrp.c (register_new_assert_for): Merge same asserts
34294 on all incoming edges.
34295 (process_assert_insertions_for): Handle insertions at the
34298 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34300 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34301 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34303 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34306 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34309 2017-01-17 Alan Modra <amodra@gmail.com>
34312 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34313 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34314 symbolic stack limit when pic.
34316 2017-01-16 Martin Sebor <msebor@redhat.com>
34318 PR tree-optimization/78608
34319 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34321 2017-01-16 Jeff Law <law@redhat.com>
34324 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34325 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34326 for several include directories that may be relative to sysroot.
34327 * config/i386/x-mingw32 (gplus_includedir): Define.
34328 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34329 (native_system_includedir): Likewise.
34330 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34331 override if TARGET_SYSTEM_ROOT is defined.
34332 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34334 PR tree-optimization/79090
34335 PR tree-optimization/33562
34336 PR tree-optimization/61912
34337 PR tree-optimization/77485
34338 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34339 and computed trims into the dump file.
34341 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34343 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34345 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34348 * gimplify.c (gimplify_init_constructor): If want_value and
34349 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34353 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34354 sequence. Formatting fixes.
34355 (doloop_optimize): Formatting fixes.
34358 * gcc.c (debug_level_greater_than_spec_func): New function.
34359 (static_spec_functions): Add debug-level-gt spec function.
34360 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34362 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34363 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34364 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34365 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34366 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34367 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34369 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34371 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34372 QImode fixups to general and mask registers only.
34374 2017-01-16 Carl Love <cel@us.ibm.com>
34376 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34377 for built-in functions
34378 vector signed char vec_nabs (vector signed char)
34379 vector signed short vec_nabs (vector signed short)
34380 vector signed int vec_nabs (vector signed int)
34381 vector signed long long vec_nabs (vector signed long long)
34382 vector float vec_nabs (vector float)
34383 vector double vec_nabs (vector double)
34384 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34386 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34387 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34388 * doc/extend.texi: Update the documentation file for the new built-in
34391 2017-01-16 Martin Sebor <msebor@redhat.com>
34393 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34396 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34398 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34399 UNSPEC_VSX__XXSPLTD to require special splat handling.
34401 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34404 * system.h: Poison strndup.
34406 2017-01-16 Alan Modra <amodra@gmail.com>
34409 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34412 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34414 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34416 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34418 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34419 call recog here. Assert that INSN_CODE (insn) is non-negative.
34421 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34424 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34426 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34427 in the currently scheduled RTL fragment.
34429 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34431 PR rtl-optimization/78751
34432 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34435 2017-01-14 Jeff Law <law@redhat.com>
34437 PR tree-optimization/79090
34438 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34439 variable length stores.
34440 (compute_trims): Delete dead assignment to *trim_tail.
34441 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34444 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34446 PR rtl-optimization/78626
34447 PR rtl-optimization/78727
34448 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34449 of a block, and split such blocks after everything else is finished.
34451 2017-01-14 Alan Modra <amodra@gmail.com>
34454 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34455 target legitimate_combined_insn.
34456 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34457 (rs6000_legitimate_combined_insn): New function.
34458 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34460 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34461 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34462 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34464 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34466 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34468 2017-01-13 Jeff Law <law@redhat.com>
34470 PR tree-optimization/33562
34471 PR tree-optimization/61912
34472 PR tree-optimization/77485
34473 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34475 (delete_dead_assignment): Likewise.
34476 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34477 statement to delete_dead_call and delete_dead_assignment.
34479 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34482 * substring-locations.c (format_warning_va): Strengthen case 1 so
34483 that both endpoints of the substring must be within the format
34484 range for just the substring to be printed.
34486 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34488 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34489 * config/i386/i386.c (ix86_target_string): Add missing options
34490 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34491 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34492 flags_other and ix86_target_other to flags2_other. Display unknown
34494 (ix86_valid_target_attribute_inner_p): Add missing options and
34495 reorder options by implied ISAs, as in ix86_target_string.
34497 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34499 * hash-table.h (hash_table::too_empty_p): New function.
34500 (hash_table::expand): Use it.
34501 (hash_table::traverse): Likewise.
34502 (hash_table::empty_slot): Use sizeof (value_type) instead of
34503 sizeof (PTR) to convert bytes to elements. Shrink the table
34504 if the current size is excessive for the current number of
34507 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34509 * ira-costs.c (record_reg_classes): Break from the inner loop
34510 early once alt_fail is known to be true. Update outer loop
34511 handling accordingly.
34513 2017-01-13 Jeff Law <law@redhat.com>
34515 * tree-ssa-dse.c (decrement_count): New function.
34516 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34517 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34518 when we know the partially dead statement is a mem* function.
34520 PR tree-optimization/61912
34521 PR tree-optimization/77485
34522 * tree-ssa-dse.c: Include expr.h.
34523 (maybe_trim_constructor_store): New function.
34524 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34526 PR tree-optimization/33562
34527 PR tree-optimization/61912
34528 PR tree-optimization/77485
34529 * doc/invoke.texi: Document new dse-max-object-size param.
34530 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34531 * tree-ssa-dse.c: Include params.h.
34532 (dse_store_status): New enum.
34533 (initialize_ao_ref_for_dse): New, partially extracted from
34535 (valid_ao_ref_for_dse, normalize_ref): New.
34536 (setup_live_bytes_from_ref, compute_trims): Likewise.
34537 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34538 (maybe_trim_partially_dead_store): Likewise.
34539 (maybe_trim_complex_store): Likewise.
34540 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34541 Track what bytes live from the original store. Return tri-state
34542 for dead, partially dead or live.
34543 (dse_dom_walker): Add constructor, destructor and new private members.
34544 (delete_dead_call, delete_dead_assignment): New extracted from
34546 (dse_optimize_stmt): Make a member of dse_dom_walker.
34547 Use initialize_ao_ref_for_dse.
34549 PR tree-optimization/33562
34550 PR tree-optimization/61912
34551 PR tree-optimization/77485
34552 * sbitmap.h (bitmap_count_bits): Prototype.
34553 (bitmap_clear_range, bitmap_set_range): Likewise.
34554 * sbitmap.c (bitmap_clear_range): New function.
34555 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34557 2017-01-13 Martin Liska <mliska@suse.cz>
34560 * function.c (set_cfun): Add new argument force.
34561 * function.h (set_cfun): Likewise.
34562 * ipa-inline-transform.c (inline_call): Use the function when
34563 strict alising from is dropped for function we inline to.
34565 2017-01-13 Richard Biener <rguenther@suse.de>
34567 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34568 for dumping GIMPLE INTEGER_CSTs.
34570 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34572 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34573 to 201112L since C++17.
34575 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34578 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34579 if -fsanitize=kernel-address is present.
34581 2017-01-13 Richard Biener <rguenther@suse.de>
34583 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34584 as _Literal ( type ) number in case usual suffixes do not
34585 preserve all information.
34587 2017-01-13 Richard Biener <rguenther@suse.de>
34589 PR tree-optimization/77283
34590 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34591 and ssa-iterators.h.
34592 (is_feasible_trace): Implement a cost model based on joiner
34595 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34598 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34599 char or short to __float128/_Float128 directly.
34601 2017-01-12 Martin Sebor <msebor@redhat.com>
34603 to -Wformat-overflow.
34604 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34605 (min_bytes_remaining): Same.
34606 (get_string_length): Same.
34607 (format_string): Same.
34608 (format_directive): Same.
34610 (pass_sprintf_length::handle_gimple_call): Same.
34612 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34614 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34615 info.nowrite calls with no lhs that can't throw. Return bool
34616 whether gsi_remove has been called or not.
34617 (pass_sprintf_length::handle_gimple_call): Return bool whether
34618 try_substitute_return_value called gsi_remove. Formatting fix.
34619 (pass_sprintf_length::execute): Don't use gsi_remove if
34620 handle_gimple_call returned true.
34623 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34624 be removed due to side-effects, don't remove following barrier nor
34625 turn the successor edge into fallthru edge.
34627 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34630 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34631 element-reversing loads and stores as not swappable.
34633 2017-01-12 Nathan Sidwell <nathan@acm.org>
34634 Nicolai Stange <nicstange@gmail.com>
34636 * combine.c (try_combine): Don't ignore result of overlap checking
34637 loop. Combine overlap & asm check into single loop.
34639 2017-01-12 Richard Biener <rguenther@suse.de>
34641 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34642 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34644 2017-01-12 Richard Biener <rguenther@suse.de>
34646 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34647 and TS_TARGET_OPTION directly derive from TS_BASE.
34648 * tree-core.h (tree_optimization_option): Derive from tree_base.
34649 (tree_target_option): Likewise.
34651 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34653 * config/i386/i386.c (memory_address_length): Increase len
34654 only when rip_relative_addr_p returns false.
34656 2017-01-11 Julia Koval <julia.koval@intel.com>
34658 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34659 (OPTION_MASK_ISA_SGX_SET): New.
34660 (ix86_handle_option): Handle OPT_msgx.
34661 * config.gcc: Added sgxintrin.h.
34662 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34663 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34664 * config/i386/i386.c (ix86_target_string): Add -msgx.
34666 (ix86_option_override_internal): Handle new options.
34667 (ix86_valid_target_attribute_inner_p): Add sgx.
34668 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34669 * config/i386/i386.opt: Add msgx.
34670 * config/i386/sgxintrin.h: New file.
34671 * config/i386/x86intrin.h: Add sgxintrin.h.
34673 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34676 * fold-const.c (maybe_nonzero_address): Return 1 for function
34678 (tree_single_nonzero_warnv_p): Don't handle function local objects
34682 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34685 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34688 * opts.c: Include "spellcheck.h"
34689 (struct string_fragment): New struct.
34690 (struct edit_distance_traits<const string_fragment &>): New
34692 (get_closest_sanitizer_option): New function.
34693 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34695 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34697 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34699 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34700 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34701 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34702 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34703 for initial die_offset if dwarf_split_debug_info.
34704 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34705 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34707 (output_skeleton_debug_sections): Formatting fix. Use
34708 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34709 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34711 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34713 * config/arm/cortex-a53.md: Add bypasses for
34714 cortex_a53_r2f_cvt.
34715 (cortex_a53_r2f): Only use for transfers.
34716 (cortex_a53_f2r): Likewise.
34717 (cortex_a53_r2f_cvt): Add reservation for conversions.
34718 (cortex_a53_f2r_cvt): Likewise.
34720 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34722 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34723 to all inlined functions, change static to extern.
34725 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34728 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34730 (arm_assemble_integer): Likewise.
34732 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34734 * config.gcc: Use new awk script to check CPU, FPU and architecture
34735 parameters for --with-... options.
34736 * config/arm/parsecpu.awk: New file
34737 * config/arm/arm-cpus.in: New file.
34738 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34740 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34742 * config/arm/t-arm: Update dependency rules.
34743 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34744 of processing .def files.
34745 * config/arm/genopt.sh: Deleted.
34746 * config/arm/gentune.sh: Deleted.
34747 * config/arm/arm-cores.def: Deleted.
34748 * config/arm/arm-arches.def: Deleted.
34749 * config/arm/arm-fpus.def: Deleted.
34750 * config/arm/arm-tune.md: Regenerated.
34751 * config/arm/arm-tables.opt: Regenerated.
34752 * config/arm/arm-cpu.h: New generated file.
34753 * config/arm/arm-cpu-data.h: New generated file.
34754 * config/arm/arm-cpu-cdata.h: New generated file.
34756 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34759 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34761 (input_varpool_node): Unpack dynamically_initialized bit.
34763 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34765 PR rtl-optimization/79032
34766 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34767 the alignment of the adjusted memory reference against that of MODE,
34768 instead of the alignment of the original memory reference.
34770 2017-01-11 Martin Jambor <mjambor@suse.cz>
34772 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34774 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34775 decorated functions.
34777 2017-01-11 Richard Biener <rguenther@suse.de>
34779 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34780 set range/nonnull info for PHI results. Do not set it on
34781 stmts marked for removal.
34783 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34785 * expr.c (store_field): In the bitfield case, fetch the return value
34786 from the registers before applying a single big-endian adjustment.
34787 Always do a final load for a BLKmode value not larger than a word.
34789 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34792 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34793 that we correctly handle column numbers greater than
34794 LINE_MAP_MAX_COLUMN_NUMBER.
34796 2017-01-10 Martin Sebor <msebor@redhat.com>
34798 PR middle-end/78245
34799 * gimple-ssa-sprintf.c (get_destination_size): Call
34800 {init,fini}object_sizes.
34801 * tree-object-size.c (addr_object_size): Adjust.
34802 (pass_through_call): Adjust.
34803 (pass_object_sizes::execute): Adjust.
34804 * tree-object-size.h (fini_object_sizes): Declare.
34806 2017-01-10 Martin Sebor <msebor@redhat.com>
34808 PR tree-optimization/78775
34809 * builtins.c (get_size_range): Move...
34810 * calls.c: ...to here.
34811 (alloc_max_size): Accept zero argument.
34812 (operand_signed_p): Remove.
34813 (maybe_warn_alloc_args_overflow): Call get_size_range.
34814 * calls.h (get_size_range): Declare.
34816 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34818 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34819 from TI's devices.csv file as of September 2016.
34820 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34822 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34824 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34825 * doc/invoke.texi: Likewise.
34826 * doc/md.texi: Likewise.
34827 * doc/objc.texi: Likewise.
34829 2017-01-10 Joshua Conner <joshconner@google.com>
34831 * config/arm/fuchsia-elf.h: New file.
34832 * config/fuchsia.h: New file.
34833 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34834 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34836 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34838 2016-01-10 Richard Biener <rguenther@suse.de>
34840 PR tree-optimization/79034
34841 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34842 Propagate out degenerate PHIs in the joiner.
34844 2017-01-10 Martin Liska <mliska@suse.cz>
34846 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34847 (sort_congruence_classes_by_decl_uid): Likewise.
34848 (sort_congruence_class_groups_by_decl_uid): Likewise.
34849 (sem_item_optimizer::merge_classes): Sort class, groups in these
34850 classes and members in the groups by DECL_UID of declarations.
34851 This would make merge operations stable.
34853 2017-01-10 Martin Liska <mliska@suse.cz>
34855 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34856 usage of m_classes_vec.
34857 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34858 (sem_item_optimizer::get_group_by_hash): Likewise.
34859 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34860 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34861 (sem_item_optimizer::verify_classes): Likewise.
34862 (sem_item_optimizer::process_cong_reduction): Likewise.
34863 (sem_item_optimizer::dump_cong_classes): Likewise.
34864 (sem_item_optimizer::merge_classes): Likewise.
34865 * ipa-icf.h (congruence_class_hash): Rename from
34866 congruence_class_group_hash. Remove declaration of m_classes_vec.
34868 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34870 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34871 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34872 * config.gcc: Add avx512vpopcntdqintrin.h.
34873 * config/i386/avx512vpopcntdqintrin.h: New.
34874 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34875 * config/i386/i386-builtin-types.def: Add new types.
34876 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34877 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34878 __builtin_ia32_vpopcountq_v8di_mask): New.
34879 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34880 __AVX512VPOPCNTDQ__.
34881 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34882 (PTA_AVX512VPOPCNTDQ): Define.
34883 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34884 TARGET_AVX512VPOPCNTDQ_P): Define.
34885 * config/i386/i386.opt: Add mavx512vpopcntdq.
34886 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34887 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34889 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34891 PR middle-end/77484
34892 * predict.def (PRED_CALL): Set to 67.
34894 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34896 * expr.c (store_field): In the bitfield case, if the value comes from
34897 a function call and is of an aggregate type returned in registers, do
34898 not modify the field mode; extract the value in all cases if the mode
34899 is BLKmode and the size is not larger than a word.
34901 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34904 * config/i386/cpuid.h: Fix undefined behavior.
34906 2017-01-04 Jeff Law <law@redhat.com>
34908 PR tree-optimization/79007
34909 PR tree-optimization/67955
34910 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34911 conservative for pt.null when flag_non_call_exceptions is on.
34913 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34915 PR translation/79019
34916 PR translation/79020
34917 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34918 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34919 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34921 * config/avr/avr.opt (maccumulate-args): Likewise.
34922 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34923 * common.opt (freport-bug): Likewise.
34924 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34925 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34926 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34927 translatable string.
34928 * config/i386/i386.c (function_value_32): Likewise.
34929 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34930 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34932 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34933 * common/config/msp430/msp430-common.c (msp430_handle_option):
34935 * symtab.c (symtab_node::verify_base): Likewise.
34936 * opts.c (set_debug_level): Likewise.
34937 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34938 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34939 missing whitespace to translatable strings.
34940 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34941 * config/sh/superh.h: Likewise.
34942 * config/i386/xopintrin.h: Likewise.
34943 * config/i386/znver1.md: Likewise.
34944 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34945 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34946 * double-int.h (struct double_int): Likewise.
34947 * double-int.c (div_and_round_double): Likewise.
34948 * wide-int.cc: Likewise.
34949 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34950 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34951 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34952 (crossjumps_occurred): ... this.
34953 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34956 PR tree-optimization/78899
34957 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34958 returning bool return struct loop *, NULL for failure and the new
34960 (versionable_outer_loop_p): Don't version outer loop if it has
34961 dont_vectorized bit set.
34962 (tree_if_conversion): When versioning outer loop, ensure
34963 tree_if_conversion is performed also on the inner loop of the
34964 non-vectorizable outer loop copy.
34965 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34966 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34967 prevent vectorization of it.
34968 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34969 the outer loop vectorization of the non-scalar version is attempted
34970 before vectorization of the inner loop in scalar version. If
34971 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34972 vectorization of its inner loop.
34973 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34974 has 2 inner loops, rename also on edges from bb whose single pred
34975 is outer_loop->header. Fix typo in function comment.
34977 2017-01-09 Martin Sebor <msebor@redhat.com>
34980 * asan.c (asan_emit_stack_protection): Increase local buffer size
34981 to avoid snprintf truncation warning.
34983 2017-01-09 Andrew Pinski <apinski@cavium.com>
34985 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34986 to reference thunderx2t99 for the tuning structure
34987 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34989 (thunderx2t99_extra_costs): This.
34990 * config/aarch64/aarch64-tune.md: Regenerate.
34991 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34992 (vulcan_addrcost_table): This.
34993 (vulcan_regmove_cost): Rename to ...
34994 (thunderx2t99_regmove_cost): This.
34995 (vulcan_vector_cost): Rename to ...
34996 (thunderx2t99_vector_cost): this.
34997 (vulcan_branch_cost): Rename to ...
34998 (thunderx2t99_branch_cost): This.
34999 (vulcan_tunings): Rename to ...
35000 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
35001 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
35003 2017-01-09 Martin Jambor <mjambor@suse.cz>
35007 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35008 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35009 (propagate_vr_accross_jump_function): Use the above function for all
35010 value range computations for pass-through jump functions and type
35011 converasion from explicit value range values.
35012 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35013 parameters from TYPE_ARG_TYPES.
35014 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35015 (ipa_write_node_info): Stream type of the actual argument.
35016 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35018 2017-01-09 Martin Liska <mliska@suse.cz>
35021 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35022 (lookup_compiler): Do not show error message with have_E.
35024 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35026 PR tree-optimization/78938
35027 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35028 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35029 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35030 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35033 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35035 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35038 2017-01-09 Richard Biener <rguenther@suse.de>
35040 PR tree-optimization/78997
35041 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35042 name condition properly.
35044 2017-01-09 Richard Biener <rguenther@suse.de>
35047 * dwarf2out.c (is_cxx): New overload with context.
35048 (is_naming_typedef_decl): Use it.
35050 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35052 * invoke.texi (Option Summary): Correct spacing in option lists
35053 and add line breaks to fix over-long lines.
35055 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35057 PR middle-end/17660
35059 * extend.texi (Common Variable Attributes): Add xref to GCC
35060 Internals manual to explain mode attribute keywords.
35062 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35065 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35066 and Preprocessor Options.
35067 (Options for Linking): Document -pthread here....
35068 (RS/6000 and PowerPC Options): ...not here.
35069 (Solaris 2 Options): ...or here.
35070 * doc/cppopts.texi: Document -pthread.
35072 2017-01-08 Martin Sebor <msebor@redhat.com>
35074 PR middle-end/77708
35075 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35076 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35077 New member functions.
35078 (format_directive): Used them.
35080 (pass_sprintf_length::handle_gimple_call): Same.
35081 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35082 to avoid truncation for any argument.
35083 (extract_affine_mul): Same.
35084 * tree.c (get_file_function_name): Same.
35086 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35088 PR middle-end/77484
35089 * predict.def (PRED_INDIR_CALL): Set to 86.
35091 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35093 PR preprocessor/54124
35094 * doc/cppopts.texi: Reformat -d subtable to list the full name
35095 of the options. Add cross-reference to the docs for the general
35096 compiler -d options.
35097 * doc/invoke.texi (Developer Options): Add cross-reference to the
35098 preprocessor-specific -d option documentation.
35100 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35102 PR preprocessor/13498
35103 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35104 redudant material, and reflect new command-line options.
35105 (System Headers): Likewise.
35107 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35109 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35110 -isystem, and -idirafter. Copy-edit.
35111 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35112 default for -ftrack-macro-expansion. Delete obsolete and
35113 badly-formatted implementation details about -fdebug-cpp output.
35114 * doc/cppwarnopts.texi: Copy-edit.
35116 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35119 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35120 that the transition from a max line width >= 1<<10 to narrower
35121 lines works correctly.
35123 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35125 * doc/options.texi (PerFunction): New.
35126 * opt-functions.awk (switch_flags): Map both Optimization and
35127 PerFunction to CL_OPTIMIZATION.
35128 * opth-gen.awk: Test for PerFunction flag along with
35130 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35131 it only when the latter is present. Skip those that don't in
35132 the hash function generator.
35133 * common.opt (fvar-tracking): Mark as PerFunction instead of
35135 (fvar-tracking-assignments): Likewise.
35136 (fvar-tracking-assignments-toggle): Likewise.
35137 (fvar-tracking-uninit): Likewise.
35139 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35141 PR translation/79018
35142 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35145 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35148 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35149 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35150 TARGET_LONG_JUMP_TABLE_OFFSETS.
35151 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35152 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35153 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35154 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35155 * config/m68k/m68k.md (tablejump expander): Likewise.
35156 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35157 TARGET_LONG_JUMP_TABLE_OFFSETS.
35158 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35159 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35161 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35162 David Holsgrove <david.holsgrove@xilinx.com>
35164 * common/config/microblaze/microblaze-common.c
35165 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35166 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35168 * config/microblaze/microblaze.c (microblaze_must_save_register)
35169 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35171 (microblaze_eh_return): New function.
35172 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35173 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35174 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35175 * config/microblaze/microblaze.md (eh_return): New pattern.
35177 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35179 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35180 GCC_DIAGNOSTIC_STRINGIFY): Define.
35182 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35184 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35186 * config/arm/arm.md (<mcrr>): New.
35188 * config/arm/arm.c (arm_arch5te): New.
35189 (arm_option_override): Set arm_arch5te.
35190 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35192 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35193 (arm_mcrr_qualifiers): ... this. New.
35194 (MRRC_QUALIFIERS): Define to...
35195 (arm_mrrc_qualifiers): ... this. New.
35196 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35198 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35199 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35200 (MRRCI, mrrc, MRRC): New.
35201 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35202 VUNSPEC_MRRC2): New.
35204 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35206 * config/arm/arm.md (<mcr>): New.
35208 * config/arm/arm.c (arm_coproc_builtin_available): Add
35209 support for mcr, mrc, mcr2 and mrc2.
35210 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35211 (arm_mcr_qualifiers): ... this. New.
35212 (MRC_QUALIFIERS): Define to ...
35213 (arm_mrc_qualifiers): ... this. New.
35214 (MCR_QUALIFIERS): Define to ...
35215 (arm_mcr_qualifiers): ... this. New.
35216 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35218 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35219 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35220 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35221 VUNSPEC_MRC2): New.
35223 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35225 * config/arm/arm.md (*ldc): New.
35229 * config/arm/arm.c (arm_coproc_builtin_available): Add
35230 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35231 (arm_coproc_ldc_stc_legitimate_address): New.
35232 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35233 'qualifier_const_pointer'.
35234 (LDC_QUALIFIERS): Define to...
35235 (arm_ldc_qualifiers): ... this. New.
35236 (STC_QUALIFIERS): Define to...
35237 (arm_stc_qualifiers): ... this. New.
35238 * config/arm/arm-protos.h
35239 (arm_coproc_ldc_stc_legitimate_address): New.
35240 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35241 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35242 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35243 stc2, stcl, stc2l): New.
35244 * config/arm/constraints.md (Uz): New.
35245 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35246 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35247 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35248 VUNSPEC_STC2L): New.
35250 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35252 * config/arm/arm.md (<cdp>): New.
35253 * config/arm/arm.c (neon_const_bounds): Rename this ...
35254 (arm_const_bounds): ... this.
35255 (arm_coproc_builtin_available): New.
35256 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35257 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35258 (CDP_QUALIFIERS): Define to...
35259 (arm_cdp_qualifiers): ... this. New.
35261 (arm_expand_builtin_args): Add case for 6 arguments.
35262 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35263 (arm_const_bounds): ... this.
35264 (arm_coproc_builtin_available): New.
35265 * config/arm/arm_acle.h (__arm_cdp): New.
35267 * config/arm/arm_acle_builtins.def (cdp): New.
35269 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35270 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35271 'arm_const_bounds'.
35272 * config/arm/types.md (coproc): New.
35273 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35274 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35275 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35276 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35278 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35280 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35281 (UBINOP_QUALIFIERS): New.
35283 (acle_builtin_data): New. Change comment.
35284 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35285 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35286 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35287 arm_acle_builtins.def.
35288 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35289 (arm_init_acle_builtins): New.
35290 (CRC32_BUILTIN): Remove.
35291 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35292 crc32cb, crc32ch and crc32cw.
35293 (arm_init_crc32_builtins): Remove.
35294 (arm_init_builtins): Use arm_init_acle_builtins rather
35295 than arm_init_crc32_builtins.
35296 (arm_expand_acle_builtin): New.
35297 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35298 * config/arm/arm_acle_builtins.def: New.
35300 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35302 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35303 (arm_builtin_datum): ... this.
35304 (arm_init_neon_builtin): Rename to ...
35305 (arm_init_builtin): ... this. Add a new parameters PREFIX
35306 and USE_SIG_IN_NAME.
35307 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35308 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35309 'arm_builtin_datum'.
35310 (arm_init_vfp_builtins): Likewise.
35311 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35312 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35313 (arm_expand_neon_args): Rename to ...
35314 (arm_expand_builtin_args): ... this. Rename builtin_arg
35315 enum values and differentiate between ARG_BUILTIN_MEMORY
35316 and ARG_BUILTIN_NEON_MEMORY.
35317 (arm_expand_neon_builtin_1): Rename to ...
35318 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35319 values, arm_expand_builtin_args and add bool parameter NEON.
35320 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35321 (arm_expand_vfp_builtin): Likewise.
35322 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35324 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35326 PR middle-end/77484
35327 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35328 * predict.c (tree_estimate_probability_bb): Reverse direction of
35329 polymorphic call predictor.
35331 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35333 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35334 (determine_pass_name_match): ...this new function and...
35335 (should_skip_pass_p): ...this new function.
35337 2017-01-06 Nathan Sidwell <nathan@acm.org>
35339 * ipa-visibility.c (function_and_variable_visibility): Reformat
35340 comments and long lines. Remove extrneous if.
35341 * symtab.c (symtab_node::make_decl_local): Fix code format.
35342 (symtab_node::set_section_for_node): Fix comment typo.
35344 2017-01-06 Martin Liska <mliska@suse.cz>
35347 * lra-constraints.c: Rename invariant to lra_invariant.
35348 * predict.c (set_even_probabilities): Initialize e to NULL.
35350 2017-01-05 Martin Sebor <msebor@redhat.com>
35352 PR tree-optimization/78910
35353 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35354 (format_integer): Correct off-by-one error in the handling
35355 of precision with negative numbers in signed conversions..
35357 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35359 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35361 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35363 PR tree-optimization/71016
35364 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35365 factor_out_conditional_conversion. Formatting fix.
35366 (factor_out_conditional_conversion): Add cond_stmt argument.
35367 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35368 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35371 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35373 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35374 read-rtl-function.o, and selftest-rtl.o.
35375 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35376 (selftest::aarch64_test_loading_full_dump): New function.
35377 (selftest::aarch64_run_selftests): New function.
35378 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35379 selftest::aarch64_run_selftests.
35380 * config/i386/i386.c
35381 (selftest::ix86_test_loading_dump_fragment_1): New function.
35382 (selftest::ix86_test_loading_call_insn): New function.
35383 (selftest::ix86_test_loading_full_dump): New function.
35384 (selftest::ix86_test_loading_unspec): New function.
35385 (selftest::ix86_run_selftests): Call the new functions.
35386 * emit-rtl.c (maybe_set_max_label_num): New function.
35387 * emit-rtl.h (maybe_set_max_label_num): New decl.
35388 * function.c (instantiate_decls): Guard call to
35389 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35390 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35392 * gensupport.c (gen_reader::gen_reader): Pass "false"
35393 for new "compact" param of rtx_reader.
35394 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35395 rather than an empty string for NULL strings.
35396 * read-md.c: Potentially include config.h rather than bconfig.h.
35397 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35398 (have_error): New global, copied from errors.c.
35399 (md_reader::read_name): Rename to...
35400 (md_reader::read_name_1): ...this, adding "out_loc" param,
35401 and converting "missing name or number" to returning false, rather
35403 (md_reader::read_name): Reimplement in terms of read_name_1.
35404 (md_reader::read_name_or_nil): New function.
35405 (md_reader::read_string): Handle "(nil)" by returning NULL.
35406 (md_reader::md_reader): Add new param "compact".
35407 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35408 (md_reader::read_file): New method.
35409 * read-md.h (md_reader::md_reader): Add new param "compact".
35410 (md_reader::read_file): New method.
35411 (md_reader::is_compact): New accessor.
35412 (md_reader::read_name): Convert return type from void to file_location.
35413 (md_reader::read_name_or_nil): New decl.
35414 (md_reader::read_name_1): New decl.
35415 (md_reader::m_compact): New field.
35416 (noop_reader::noop_reader): Pass "false" for new "compact" param
35418 (rtx_reader::rtx_reader): Add new "compact" param.
35419 (rtx_reader::read_rtx_operand): Make virtual and convert return
35420 type from void to rtx.
35421 (rtx_reader::read_until): New decl.
35422 (rtx_reader::handle_any_trailing_information): New virtual function.
35423 (rtx_reader::postprocess): New virtual function.
35424 (rtx_reader::finalize_string): New virtual function.
35425 (rtx_reader::m_in_call_function_usage): New field.
35426 (rtx_reader::m_reuse_rtx_by_id): New field.
35427 * read-rtl-function.c: New file.
35428 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35429 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35430 (selftest::verify_three_block_rtl_cfg): New decl.
35431 * read-rtl-function.h: New file.
35432 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35433 For host, include function.h, memmodel.h, and emit-rtl.h.
35434 (one_time_initialization): New function.
35435 (struct compact_insn_name): New struct.
35436 (compact_insn_names): New array.
35437 (find_code): Handle insn codes in compact dumps.
35438 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35439 (bind_subst_iter_and_attr): Likewise.
35440 (add_condition_to_string): Likewise.
35441 (add_condition_to_rtx): Likewise.
35442 (apply_attribute_uses): Likewise.
35443 (add_current_iterators): Likewise.
35444 (apply_iterators): Likewise.
35445 (initialize_iterators): Guard usage of apply_subst_iterator with
35446 #ifdef GENERATOR_FILE.
35447 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35448 (md_reader::read_mapping): Likewise.
35449 (add_define_attr_for_define_subst): Likewise.
35450 (add_define_subst_attr): Likewise.
35451 (read_subst_mapping): Likewise.
35452 (check_code_iterator): Likewise.
35453 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35455 (one_time_initialization): New function.
35456 (rtx_reader::read_until): New method.
35457 (read_flags): New function.
35458 (parse_reg_note_name): New function.
35459 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35460 Handle reuse_rtx ids.
35461 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35462 Add parsing support for RTL dumps, mirroring the special-cases in
35463 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35464 values, and calling handle_any_trailing_information.
35465 (rtx_reader::read_rtx_operand): Convert return type from void
35466 to rtx, returning return_rtx. Handle case 'e'. Call
35467 finalize_string on XSTR and XTMPL fields.
35468 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35469 "(nil)" values were omitted. Call the postprocess vfunc on the
35471 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35472 class ctor. Initialize m_in_call_function_usage. Call
35473 one_time_initialization.
35474 * rtl-tests.c (selftest::test_uncond_jump): Call
35475 set_new_first_and_last_insn.
35476 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35477 * selftest-rtl.c: New file.
35478 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35479 (selftest::get_insn_by_uid): New decl.
35480 * selftest-run-tests.c (selftest::run_tests): Call
35481 read_rtl_function_c_tests.
35482 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35483 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35486 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35488 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35489 operands in a special way. Assert that pos+len <= mode precision.
35491 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35493 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35494 3 argument Alias with unlimited for the negative form.
35495 (fno-vect-cost-model): Removed.
35497 2017-01-05 Martin Liska <mliska@suse.cz>
35499 * hsa-gen.c (gen_hsa_divmod): New function.
35500 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35502 2017-01-05 Martin Liska <mliska@suse.cz>
35505 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35508 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35510 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35511 small constant length operands.
35513 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35515 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35516 between loop iterations.
35518 2017-01-05 Martin Liska <mliska@suse.cz>
35521 * gimplify.c (gimplify_decl_expr): Compare to
35522 asan_poisoned_variables instread of checking flags.
35523 (gimplify_target_expr): Likewise.
35524 (gimplify_expr): Likewise.
35525 (gimplify_function_tree): Conditionally initialize
35526 asan_poisoned_variables.
35528 2017-01-04 Jeff Law <law@redhat.com>
35530 PR tree-optimizatin/78812
35531 * rtl.h (contains_mem_rtx_p): Prototype.
35532 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35533 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35534 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35535 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35536 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35538 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35540 * input.c (assert_char_at_range): Default-initialize actual_range.
35542 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35544 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35545 to match the caller.
35547 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35549 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35550 insns after final jump in test to emit dummy move.
35552 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35554 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35555 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35557 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35559 * multiple_target.c (create_dispatcher_calls): Init e_next.
35560 * tree-ssa-loop-split.c (split_loop): Init border.
35561 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35564 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35569 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35570 (altivec_register_operand): Do not return true if the operand
35571 contains a SUBREG mixing SImode and SFmode.
35572 (vsx_register_operand): Likewise.
35573 (vsx_reg_sfsubreg_ok): New predicate.
35574 (vfloat_operand): Do not return true if the operand contains a
35575 SUBREG mixing SImode and SFmode.
35576 (vint_operand): Likewise.
35577 (vlogical_operand): Likewise.
35578 (gpc_reg_operand): Likewise.
35579 (int_reg_operand): Likewise.
35580 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35581 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35582 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35584 (rs6000_emit_move_si_sf_subreg): New helper function.
35585 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35586 fixup SUBREGs involving SImode and SFmode.
35587 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35588 numbers for the new peephole2 optimization.
35589 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35590 the GLIBC math library that do AND/IOR/XOR operations on single
35591 precision floating point.
35592 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35593 target macros to say whether we need to avoid SUBREGs mixing
35595 (TARGET_ALLOW_SF_SUBREG): Likewise.
35596 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35597 (UNSPEC_SI_FROM_SF): Likewise.
35598 (iorxor): Change spacing.
35599 (and_ior_xor): New iterator for AND, IOR, and XOR.
35600 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35601 (movdi_from_sf_zero_ext): Likewise.
35602 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35603 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35604 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35605 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35606 (fms<mode>4): Likewise.
35607 (fnma<mode>4): Likewise.
35608 (fnms<mode>4): Likewise.
35609 (nfma<mode>4): Likewise.
35610 (nfms<mode>4): Likewise.
35612 2017-01-04 Marek Polacek <polacek@redhat.com>
35615 * doc/invoke.texi: Document -Wpointer-compare.
35617 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35619 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35622 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35623 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35626 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35629 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35630 documentation of the powerpc_popcntb_ok attribute.
35631 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35632 code to issue warning messages if a requested CPU configuration is
35633 not supported by the binary (assembler and loader) toolchain.
35634 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35635 made to define a built-in function that has been disabled.
35636 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35637 made to define a built-in function that has been disabled.
35638 (altivec_init_builtins): Add comment explaining why definition
35639 of the DST built-in functions is not preceded by an assertion
35640 check. Add assertions to prevent ICE if attempts are made to
35641 define an altivec predicate or an abs* built-in function that has
35643 (htm_init_builtins): Add comment explaining why definition of the
35644 htm built-in functions is not preceded by an assertion check.
35646 2017-01-04 Jeff Law <law@redhat.com>
35648 PR tree-optimizatin/67955
35649 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35650 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35651 the points-to solution does not include pt_null. Use DECL_PT_UID
35654 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35656 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35657 Use gen_int_mode instead of gen_lopwart for const_int operands.
35659 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35661 PR tree-optimization/71563
35662 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35663 out of range value - has low bits known to be zero.
35665 2017-01-04 Alan Modra <amodra@gmail.com>
35667 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35668 * configure: Regenerate.
35669 * config.in: Regenerate.
35671 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35674 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35675 a substring of the message, but strcmp with the whole message. Ifdef
35676 ENABLE_NLS, translate the message first using dgettext.
35678 2017-01-03 Jeff Law <law@redhat.com>
35680 PR tree-optimizatin/78856
35681 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35682 (mark_threaded_blocks): Remove code to truncate thread paths that
35683 cross multiple loop headers. Instead invalidate the cached loop
35684 iteration information and handle case of a thread path walking
35685 into an irreducible region.
35687 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35690 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35691 assertions. Add support for doing the signbit if the IEEE 128-bit
35692 floating point value is in a GPR.
35693 * config/rs6000/rs6000.md (Fsignbit): Delete.
35694 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35695 Update the length attribute if the value is in a GPR.
35696 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35697 the sign or zero extension instruction, since the value is always 0/1.
35698 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35701 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35702 extracting SImode to a GPR register so that we can generate a
35703 store, limit the vector to be in a traditional Altivec register
35704 for the vextuwrx instruction.
35706 2017-01-03 Ian Lance Taylor <iant@google.com>
35708 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35710 2017-01-03 Martin Sebor <msebor@redhat.com>
35712 PR tree-optimization/78696
35713 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35714 precision. Use MPFR for %f for greater fidelity. Correct handling
35716 (pass_sprintf_length::compute_format_length): Set width and precision
35717 specified by asrerisk to void_node for vararg functions.
35718 (try_substitute_return_value): Adjust dump output.
35720 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35722 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35724 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35726 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35727 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35728 -mlra/-mno-lra was passed to the compiler.
35730 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35732 PR rtl-optimization/65618
35733 * emit-rtl.c (try_split): Move initialization of "before" and
35734 "after" to just before the call to emit_insn_after_setloc.
35736 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35738 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35740 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35742 * dwarf2out.c (gen_enumeration_type_die): When
35743 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35745 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35747 PR tree-optimization/78965
35748 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35749 Change first argument from const call_info & to call_info &. For %n
35750 set info.nowrite to false.
35752 PR middle-end/78901
35753 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35754 possibly throwing calls.
35756 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35757 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35758 and fns handling, rather than in a separate case SSA_NAME.
35760 2017-01-02 Jeff Law <law@redhat.com>
35762 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35763 fixes for first_period and second_period variables.
35765 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35768 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35769 (*insvqi_1): New insn pattern.
35770 (*insvqi_1_mem_rex64): Ditto.
35771 (*insvqi_2): Ditto.
35772 (*insvqi_3): Rename from *insvqi.
35774 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35776 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35778 * doc/cfg.texi (Edges): Remove reference to Java.
35779 (Maintaining the CFG): Ditto.
35781 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35783 PR middle-end/77674
35784 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35785 transparent aliases.
35787 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35789 PR middle-end/77484
35790 * predict.def (PRED_CALL): Update hitrate.
35791 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35792 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35793 into direct/indirect/polymorphic variants.
35795 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35797 Update copyright years.
35799 * gcc.c (process_command): Update copyright notice dates.
35800 * gcov-dump.c (print_version): Ditto.
35801 * gcov.c (print_version): Ditto.
35802 * gcov-tool.c (print_version): Ditto.
35803 * gengtype.c (create_file): Ditto.
35804 * doc/cpp.texi: Bump @copying's copyright year.
35805 * doc/cppinternals.texi: Ditto.
35806 * doc/gcc.texi: Ditto.
35807 * doc/gccint.texi: Ditto.
35808 * doc/gcov.texi: Ditto.
35809 * doc/install.texi: Ditto.
35810 * doc/invoke.texi: Ditto.
35812 Copyright (C) 2017 Free Software Foundation, Inc.
35814 Copying and distribution of this file, with or without modification,
35815 are permitted in any medium without royalty provided the copyright
35816 notice and this notice are preserved.