1 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3 * vec.h (debug_helper): New function.
4 (DEFINE_DEBUG_VEC): New macro.
5 * hash-set.h (debug_helper): New function.
6 (DEFINE_DEBUG_HASH_SET): New macro.
7 * cfg.c (debug_slim (edge)): New function.
8 Call DEFINE_DEBUG_VEC for edges.
9 Call DEFINE_DEBUG_HASH_SET for edges.
10 * cfghooks.c (debug_slim (basic_block)): New function.
11 Call DEFINE_DEBUG_VEC for basic blocks.
12 Call DEFINE_DEBUG_HASH_SET for basic blocks.
13 * print-tree.c (debug_slim): New function to handle trees.
14 Call DEFINE_DEBUG_VEC for trees.
15 Call DEFINE_DEBUG_HASH_SET for trees.
16 (debug (vec<tree, va_gc>) &): Remove.
17 (debug (<vec<tree, va_gc>) *): Remove.
18 * print-rtl.c (debug_slim): New function to handle const_rtx.
19 Call DEFINE_DEBUG_VEC for rtx_def.
20 Call DEFINE_DEBUG_VEC for rtx_insn.
21 Call DEFINE_DEBUG_HASH_SET for rtx_def.
22 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
23 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
24 (debug (vec<rtx_insn *> *ptr): Remove.
25 (debug_insn_vector): Remove.
26 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
28 2017-11-20 Tom de Vries <tom@codesourcery.com>
30 PR rtl-optimization/82020
31 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
32 IF_THEN_ELSE condition.
34 2017-11-19 Jeff Law <law@redhat.com>
36 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
37 of degenerates resulting from ignoring an edge.
39 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
42 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
44 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
47 * profile-count.c (profile_count::to_sreal_scale): Fix return value
48 for uninitialied counts.
50 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
53 * tree-inline.c (estimate_num_insns): Set to 1 at least.
55 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
58 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane
61 2017-11-19 Tom de Vries <tom@codesourcery.com>
63 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
66 2017-11-19 Tom de Vries <tom@codesourcery.com>
68 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
71 2017-11-19 Tom de Vries <tom@codesourcery.com>
73 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
76 2017-11-19 Tom de Vries <tom@codesourcery.com>
78 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
80 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
82 2017-11-19 Tom de Vries <tom@codesourcery.com>
84 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
86 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
87 semicolon after MCORE_EXPORT_NAME call.
89 2017-11-19 Tom de Vries <tom@codesourcery.com>
92 * vmsdbgout.c (vmsdbgout_early_finish): New function.
93 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
95 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
97 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
98 (cgraph_edge::clone): Cleanup updating of profile.
99 * ipa-cp.c (update_profiling_info): Likewise.
100 * ipa-inline-transform.c (inline_transform): Likewise.
101 * ipa-inline.c (inline_small_functions): Add missing space to dump.
102 * ipa-split.c (execute_split_functions): Do not split when function
104 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
105 * profile-count.c (profile_count::dump): Add global0.
106 (profile_count::to_cgraph_frequency): Do not ICE when entry is
108 (profile_count::to_sreal_scale): Likewise.
109 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
110 (profile_count::combine_with_ipa_count): New function.
111 * profile-count.h (profile_guessed_global0adjusted): New.
112 (profile_count::adjusted_zero): New.
113 (profile_count::global0adjusted): New.
114 (profile_count::combine_with_ipa_count): New.
115 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
116 correct profile of return block of split functions.
117 (copy_cfg_body): Remove unused profile_count.
118 (copy_body): Likewise.
119 (expand_call_inline): Update.
120 (tree_function_versioning): Update.
122 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
124 * hash-set.h (hash_set::empty): New.
125 * tree-ssa-threadbackward.h: Delete.
126 * tree-ssa-threadbackward.c (class thread_jumps): New.
127 Move max_threaded_paths into class.
128 (fsm_find_thread_path): Remove arguments that are now in class.
129 (profitable_jump_thread_path): Rename to...
130 (thread_jumps::profitable_jump_thread_path): ...this.
131 (convert_and_register_jump_thread_path): Rename to...
132 (thread_jumps::convert_and_register_current_path): ...this.
133 (check_subpath_and_update_thread_path): Rename to...
134 (thread_jumps::check_subpath_and_update_thread_path): ...this.
135 (register_jump_thread_path_if_profitable): Rename to...
136 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
137 (handle_phi): Rename to...
138 (thread_jumps::handle_phi): ...this.
139 (handle_assignment): Rename to...
140 (thread_jumps::handle_assignment): ...this.
141 (fsm_find_control_statement_thread_paths): Rename to...
142 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
143 (find_jump_threads_backwards): Rename to...
144 (thread_jumps::find_jump_threads_backwards): ...this.
145 Initialize path local data.
146 (pass_thread_jumps::execute): Call find_jump_threads_backwards
147 from within thread_jumps class.
148 (pass_early_thread_jumps::execute): Same.
150 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
152 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
154 2017-11-17 Steve Ellcey <sellcey@cavium.com>
156 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
157 to canonical location.
159 2017-11-17 Steve Ellcey <sellcey@cavium.com>
162 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
164 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
166 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
168 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
169 rcrt1.o%s/grcrt1.o%s for -static-pie.
171 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
173 * i386.c (ix86_multiplication_cost, ix86_division_cost,
174 ix86_shift_rotate_cost): Break out from ...
175 (ix86_rtx_costs): ... here.
176 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
179 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
181 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
182 when entry block was promoted unlikely.
183 (estimate_bb_frequencies): Increase frequency scale.
184 * profile-count.h (profile_count): Export precision info.
186 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
188 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
189 disturbin profile of entry block.
191 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
193 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
196 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
198 * ipa-cp.c (update_profiling_info): Handle conversion to local
200 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
202 2017-11-17 Jeff Law <law@redhat.com>
204 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
205 from evrp_dom_walker class. Various methods moved into new class.
206 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
207 (evrp_range_analyzer::enter): New method.
208 (evrp_range_analyzer::leave): New method.
209 (evrp_dom_walker): Remove delegators no longer needed by this class.
210 Replace vr_values data member with evrp_range_analyzer
212 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
213 method extracted from evrp_dom_walker::before_dom_children.
214 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
215 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
217 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
218 Add private copy constructor and move assignment operators.
219 Privatize methods and class data where trivially possible.
220 (evrp_dom_walker::cleanup): New function, extracted from
221 execute_early_vrp. Simplify access to class data.
223 * vr-values.h (get_output_for_vrp): Prototype.
224 * vr-values.c (get_output_for_vrp): New function extracted from
225 vrp_visit_assignment_or_call and extract_range_from_stmt.
226 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
228 2017-11-17 Luis Machado <luis.machado@linaro.org>
230 * config/aarch64/aarch64.c
231 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
232 (qdf24xx_tunings) <autoprefetcher_model>: Set to
233 tune_params::AUTOPREFETCHER_WEAK.
235 2017-11-17 Tamar Christina <tamar.christina@arm.com>
238 * config/arm/arm.c (arm_valid_target_attribute_rec):
239 Parse "arch=" and "+<ext>".
240 (arm_valid_target_attribute_tree): Re-init global options.
241 (arm_option_override): Make non-static.
242 (arm_options_perform_arch_sanity_checks): Make errors fatal.
243 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
244 (__ARM_FEATURE_CRC32): Support undef.
245 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
246 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
248 2017-11-17 David Malcolm <dmalcolm@redhat.com>
250 * gdbinit.in (break-on-diagnostic): New command.
252 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
254 * config/i386/i386.c (ix86_expand_epilogue): Change simple
255 return to indirect jump for EH return if control-flow
256 protection is enabled. Change explicit 'false' argument in
257 pro_epilogue_adjust_stack with a value of
259 * config/i386/i386.md (simple_return_indirect_internal):
260 Remove SImode restriction to support 64-bit.
262 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
264 * combine.c (added_notes_insn): New.
265 (try_combine): Handle added_notes_insn like added_links_insn.
266 Rewrite return value code.
267 (distribute_notes): Set added_notes_insn to the earliest insn we added
270 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
272 PR rtl-optimization/82621
273 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
274 dest of one of those SETs is unused.
276 2017-11-17 Richard Biener <rguenther@suse.de>
279 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
280 * tree-pretty-print.c (dump_generic_node): Handle
281 annot_expr_parallel_kind.
282 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
283 * gimplify.c (gimple_boolify): Likewise.
285 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
287 * config.gcc (extra_headers): Add cet.h for x86 targets.
288 * config/i386/cet.h: New file.
289 * doc/install.texi: Add --enable-cet/--disable-cet.
291 2017-11-17 Richard Biener <rguenther@suse.de>
293 PR tree-optimization/83017
294 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
295 (gen_parallel_loop): Properly count iterations.
296 (parallelize_loops): Handle loop->can_be_parallel independent
297 of flag_loop_parallelize_all. Make static profitability test match
299 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
300 * invoke.texi (parloops-min-per-thread): Document.
302 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
304 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
305 upstreaming review comments.
307 2017-11-17 Tamar Christina <tamar.christina@arm.com>
309 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
310 with fast unaligned access.
311 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
313 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
315 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
316 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
318 2017-11-17 Richard Biener <rguenther@suse.de>
320 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
321 folding of references.
323 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
326 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
327 of replacing call to strncmp with corresponding call to strcmp when
330 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
332 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
333 option prefer-avx256 for skylake-avx512 configuration.
334 * config/i386/i386.c (ix86_option_override_internal): Ditto.
335 (get_builtin_code_for_version): Ditto.
337 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
338 Monk Chiang <sh.chiang04@gmail.com>
340 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
341 (FIXED_REGISTERS): Reserve more register numbers.
342 (CALL_USED_REGISTERS): Likewise.
343 (REG_ALLOC_ORDER): Likewise.
344 (REG_CLASS_CONTENTS): Likewise.
345 (REGISTER_NAMES): Likewise.
347 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
348 Kito Cheng <kito.cheng@gmail.com>
350 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
352 * config/nds32/iterators.md: Add vector mode iterators and attributes.
354 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
356 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
357 parameter list for vec_splats.
359 2017-11-16 Joseph Myers <joseph@codesourcery.com>
361 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
363 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option
366 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
369 * doc/install.texi: Document incompatibility of Perl >=5.6.26
370 with the required version of automake 1.11.6.
372 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
374 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
376 (DU_C2_3_power9): Correct reservation combinations.
377 (FP_DIV_power9, VEC_DIV_power9): New.
378 (power9-alu): Split out rotate/shift...
379 (power9-rot): ...to here, correct dispatch resource.
380 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
382 (power9-fp): Correct latency.
383 (power9-sdiv): Add div/sqrt resource.
384 (power9-ddiv): Correct latency, add div/sqrt resource.
385 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
386 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
388 (power9-qpdiv, power9-qpmul): Adjust resource usage.
390 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
392 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
393 switch statement mapping KF built-ins to TF built-ins if we don't
394 have the proper ISA 3.0 assembler support.
396 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
398 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
399 (lower_emutls_function_body): Do not compute it.
401 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
403 * ipa-split.c (split_bb_info): Turn time to sreal.
404 (split_point): Likewise.
405 (dump_split_point): Likewise.
406 (fine_split_points): Likewise.
407 (execute_split_functions): Only zero split_bbs; turn time to sreals.
409 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
411 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
414 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
416 * predict.c (combine_predictions_for_bb): Preserve zero predicted
418 (expensive_function_p): Remove useless assert.
419 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
421 2017-11-16 Martin Sebor <msebor@redhat.com>
423 PR tree-optimization/82588
424 PR tree-optimization/82583
425 * tree-vrp.c (check_array_ref): Handle flexible array members,
426 string literals, and inner indices.
427 (search_for_addr_array): Add detail to diagnostics.
429 2017-11-16 Nathan Sidwell <nathan@acm.org>
433 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
434 (SET_DECL_ASSEMBLER_NAME): Forward to
435 overwrite_decl_assembler_name.
436 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
437 (overwrite_decl_assembler_name): Declare.
438 * tree.c (overwrite_decl_assembler_name): New.
439 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
440 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
441 (LANG_HOOKS_INITIALIZER): Add it.
442 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
443 * langhooks.c (lhd_set_decl_assembler_name): Use
444 SET_DECL_ASSEMBLER_NAME.
445 (lhd_overwrite_decl_assembler_name): Default implementation.
447 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
448 Jackson Woodruff <jackson.woodruff@arm.com>
450 PR tree-optimization/71026
451 * match.pd: Canonicalize constant multiplies in division.
453 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
455 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
456 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
457 Remove OPT_fomit_frame_pointer entry.
458 * common/config/arc/arc-common.c: Likewise.
459 * common/config/arm/arm-common.c: Likewise.
460 * common/config/avr/avr-common.c: Likewise.
461 * common/config/c6x/c6x-common.c: Likewise.
462 * common/config/cr16/cr16-common.c: Likewise.
463 * common/config/cris/cris-common.c: Likewise.
464 * common/config/epiphany/epiphany-common.c: Likewise.
465 * common/config/fr30/fr30-common.c: Likewise.
466 * common/config/frv/frv-common.c: Likewise.
467 * common/config/ia64/ia64-common.c: Likewise.
468 * common/config/iq2000/iq2000-common.c: Likewise.
469 * common/config/lm32/lm32-common.c: Likewise.
470 * common/config/m32r/m32r-common.c: Likewise.
471 * common/config/mcore/mcore-common.c: Likewise.
472 * common/config/microblaze/microblaze-common.c: Likewise.
473 * common/config/mips/mips-common.c: Likewise.
474 * common/config/mmix/mmix-common.c: Likewise.
475 * common/config/mn10300/mn10300-common.c: Likewise.
476 * common/config/nios2/nios2-common.c: Likewise.
477 * common/config/pa/pa-common.c: Likewise.
478 * common/config/pdp11/pdp11-common.c: Likewise.
479 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
480 * common/config/riscv/riscv-common.c: Likewise.
481 * common/config/rs6000/rs6000-common.c: Likewise.
482 * common/config/rx/rx-common.c: Likewise.
483 * common/config/s390/s390-common.c: Likewise.
484 * common/config/sh/sh-common.c: Likewise.
485 * common/config/sparc/sparc-common.c: Likewise.
486 * common/config/tilegx/tilegx-common.c: Likewise.
487 * common/config/tilepro/tilepro-common.c: Likewise.
488 * common/config/v850/v850-common.c: Likewise.
489 * common/config/visium/visium-common.c: Likewise.
490 * common/config/xstormy16/xstormy16-common.c: Likewise.
491 * common/config/xtensa/xtensa-common.c: Likewise.
492 * invoke.texi (-fomit-frame-pointer): Update documentation.
494 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
496 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
498 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
500 * cfg.c (scale_bbs_frequencies_int,
501 cale_bbs_frequencies_gcov_type): Remove.
502 * cfg.h (scale_bbs_frequencies_int,
503 cale_bbs_frequencies_gcov_type): Remove.
505 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
507 * tree-ssa-loop-manip.c
508 (scale_dominated_blocks_in_loop): Update to profile counts.
509 (tree_transform_and_unroll_loop): Likewise.
511 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
513 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
514 scale_bbs_frequencies_int.
516 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
518 * final.c (compute_alignments): Use counts rather than frequencies.
520 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
522 * cfgloopanal.c: Include sreal.h
523 (average_num_loop_insns): Use counts and sreal for accounting.
525 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
527 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
530 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
531 Kito Cheng <kito.cheng@gmail.com>
533 * config/nds32/constraints.md: Provide more constraints.
534 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
535 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
536 support constraints usage.
538 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
540 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
542 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
543 Kito Cheng <kito.cheng@gmail.com>
545 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
546 * config/nds32/nds32.opt: Refine the layout.
547 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
548 TARGET_EXT_STRING): Support new options.
549 * config/nds32/nds32.h: Likewise.
550 * config/nds32/nds32.md: Likewise.
551 * config/nds32/nds32-predicates.c: Likewise.
552 * config/nds32/constraints.md: Likewise.
553 * common/config/nds32/nds32-common.c: Likewise.
555 2017-11-16 Julia Koval <julia.koval@intel.com>
558 * config/i386/gfniintrin.h: Add sse check.
559 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
561 2017-11-16 Julia Koval <julia.koval@intel.com>
563 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
564 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
565 (ix86_handle_option): Handle -mavx512vbmi2.
566 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
567 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
568 * config/i386/i386-c.c (__AVX512VBMI2__): New.
569 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
570 (ix86_valid_target_attribute_inner_p): Ditto.
571 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
572 * config/i386/i386.opt (mavx512vbmi2): New option.
573 * doc/invoke.texi: Add new option.
575 [2017-11-16 Julia Koval <julia.koval@intel.com>
577 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
578 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
579 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
580 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
581 _mm512_gf2p8mul_epi8): New intrinsics.
582 * config/i386/i386-builtin-types.def
583 (V64QI_FTYPE_V64QI_V64QI): New type.
584 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
585 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
586 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
587 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
588 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
589 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
591 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
593 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
594 explicitly as a stream of bytes.
596 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
598 * config/rs6000/altivec.h (vec_xst_be): New #define.
599 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
600 and externalize from *altivec_vperm_<mode>_internal.
601 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
603 (XL_BE_V8HI): Likewise.
604 (XL_BE_V4SI): Likewise.
605 (XL_BE_V4SI): Likewise.
606 (XL_BE_V2DI): Likewise.
607 (XL_BE_V4SF): Likewise.
608 (XL_BE_V2DF): Likewise.
609 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
610 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
611 all array entries with these keys: VSX_BUILTIN_VEC_XL,
612 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
613 VSX_BUILTIN_VEC_XST_BE.
614 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
615 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
617 (altivec_init_builtins): Replace conditional calls to def_builtin
618 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
619 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
620 with unconditional calls. Remove calls to def_builtin for
621 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
622 __builtin_vec_xst_be.
623 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
624 to define_expand, and add alternate RTL generation for P8.
625 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
627 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
628 add alternate RTL generation for P8.
629 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
630 vsx_ld_elemrev_v16qi.
631 (vsx_st_elemrev_v8hi): Convert define_insn
632 to define_expand, and add alternate RTL generation for P8.
633 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
635 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
636 add alternate RTL generation for P8.
637 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
638 vsx_st_elemrev_v16qi.
640 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
643 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
644 TARGET_AVX512ER check.
645 (ix86_option_override_internal): Set MASK_VZEROUPPER if
646 neither -mzeroupper nor -mno-zeroupper is used and
647 TARGET_EMIT_VZEROUPPER is set.
648 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
649 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
651 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
653 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
654 folding of vector compares.
655 (fold_build_vec_cmp): New helper function.
656 (fold_compare_helper): New helper function.
657 (builtin_function_type): Add compare builtins to the list of functions
658 having unsigned arguments. Cosmetic updates to comment indentation.
659 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
660 the not+eq combination.
662 2017-11-15 Bin Cheng <bin.cheng@arm.com>
664 PR tree-optimization/82726
665 PR tree-optimization/70754
666 * tree-predcom.c (order_drefs_by_pos): New function.
667 (combine_chains): Move code setting has_max_use_after to...
668 (try_combine_chains): ...here. New parameter. Sort combined chains
669 according to position information.
670 (tree_predictive_commoning_loop): Update call to above function.
671 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
673 2017-11-15 Bin Cheng <bin.cheng@arm.com>
675 PR tree-optimization/82726
677 2017-01-23 Bin Cheng <bin.cheng@arm.com>
679 PR tree-optimization/70754
680 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
681 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
682 combined stmt before it if not NULL.
683 (combine_chains): Process refs reversely and compute dominance point
687 2017-02-23 Bin Cheng <bin.cheng@arm.com>
689 PR tree-optimization/79663
690 * tree-predcom.c (combine_chains): Process refs in reverse order
691 only for ZERO length chains, and add explaining comment.
693 2017-11-15 Tamar Christina <tamar.christina@arm.com>
695 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
696 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
697 * doc/invoke.texi (ARM Options): Add armv8.3-a.
699 2017-11-15 Tamar Christina <tamar.christina@arm.com>
701 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
703 2017-11-15 Martin Liska <mliska@suse.cz>
705 * tree-cfg.c (pass_warn_function_return::execute):
706 Compare warn_return_type for greater than zero.
708 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
712 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
713 to return true on Xeon and not on Xeon Phi.
714 (ix86_check_avx256_register): Changed to ...
715 (ix86_check_avx_upper_register): ... this. Add extra check for
716 VALID_AVX512F_REG_OR_XI_MODE.
717 (ix86_avx_u128_mode_needed): Changed
718 ix86_check_avx256_register to ix86_check_avx_upper_register.
719 (ix86_check_avx256_stores): Changed to ...
720 (ix86_check_avx_upper_stores): ... this. Changed
721 ix86_check_avx256_register to ix86_check_avx_upper_register.
722 (ix86_avx_u128_mode_after): Changed
723 avx_reg256_found to avx_upper_reg_found. Changed
724 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
725 (ix86_avx_u128_mode_entry): Changed
726 ix86_check_avx256_register to ix86_check_avx_upper_register.
727 (ix86_avx_u128_mode_exit): Ditto.
728 * config/i386/i386.h: (host_detect_local_cpu): New define.
730 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
732 * config/arm/xgene1.md (xgene1): Split into automatons
733 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
734 (xgene1_f_load): Adjust reservations and/or types.
735 (xgene1_f_store): Likewise.
736 (xgene1_load_pair): Likewise.
737 (xgene1_store_pair): Likewise.
738 (xgene1_fp_load1): Likewise.
739 (xgene1_load1): Likewise.
740 (xgene1_store1): Likewise.
741 (xgene1_move): Likewise.
742 (xgene1_alu): Likewise.
743 (xgene1_simd): Likewise.
744 (xgene1_bfm): Likewise.
745 (xgene1_neon_load1): Likewise.
746 (xgene1_neon_store1): Likewise.
747 (xgene1_neon_logic): Likewise.
748 (xgene1_neon_st1): Likewise.
749 (xgene1_neon_ld1r): Likewise.
750 (xgene1_alu_cond): Added.
751 (xgene1_shift_reg): Likwise.
752 (xgene1_bfx): Likewise.
753 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
755 2017-11-15 Jakub Jelinek <jakub@redhat.com>
758 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
760 (can_widen_mult_without_libcall): New function.
761 (expand_mul_overflow): If only checking unsigned mul overflow,
762 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
763 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
764 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
765 (expand_DIVMOD): Formatting fix.
766 * expmed.h (expand_mult): Add NO_LIBCALL argument.
767 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
768 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
770 PR tree-optimization/82977
771 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
772 constructed temporary to strlen_to_stridx.put.
774 2017-11-15 Martin Liska <mliska@suse.cz>
776 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
777 * configure: Regenerate.
779 2017-11-15 Martin Liska <mliska@suse.cz>
782 * config/sh/sh-mem.cc: Use proper probability for
785 2017-11-14 Jeff Law <law@redhat.com>
787 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
788 the red zone for stack_clash_protection_final_dynamic_probe targets
789 when the total dynamic stack size is zero bytes.
791 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
792 blocks is post order.
794 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
796 * dumpfile.h (TDF_COMPARE_DEBUG): New.
797 * final.c (rest_of_clean_state): Set it for the
798 -fcompare-debug dump.
799 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
800 class when TDF_COMPARE_DEBUG is set.
802 * dwarf2out.c (gen_producer_string): Discard
805 2017-11-15 Joseph Myers <joseph@codesourcery.com>
808 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
809 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
810 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
811 (__TGMATH_REAL_2_3): Remove macros.
812 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
813 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
815 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
817 (remquo): Define using __TGMATH_REAL_3.
819 2017-11-14 Jeff Law <law@redhat.com>
821 * vr-values.c: New file with contents extracted from tree-vrp.c.
822 * Makefile.in (OBJS): Add vr-values.o
823 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
824 (set_value_range, set_and_canonicalize_value_range): Likewise.
825 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
826 (value_range_constant_singleton, symbolic_range_p): Likewise.
827 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
828 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
829 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
830 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
831 (find_case_label_range, find_case_label_index): Likewise.
832 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
833 (range_int_cst_singleton_p, value_inside_range): Likewise.
834 (get_single_symbol): Likewise.
835 (switch_update): Move structure definition here.
836 (to_remove_edges, to_update_switch_stmts): Provide externs.
837 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
838 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
839 (vrp_val_is_min, set_value_range): Likewise.
840 (set_and_canonicalize_value_range, copy_value_range): Likewise.
841 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
842 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
843 (range_is_nonnull, range_int_cst_p): Likewwise.
844 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
845 (get_single_symbol, operand_less_p): Likewise
846 (compare_values_warnv, compare_values): Likewise.
847 (value_inside_range, value_range_constant_singleton): Likewise.
848 (zero_nonzero_bitgs_from_vr): Likewise.
849 (extract_range_from_binary_expr_1): Likewise.
850 (overflow_comparison_p): Likewise.
851 (to_remove_edges, to_update_switch_stmts): Likewise.
852 (find_case_label-index, find_case_label_range): Likewise.
853 (switch_update, set_value_range_to_nonnegative): Remove.
854 (set_value_range_to_truthvalue): Likewise.
855 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
856 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
857 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
858 (find_case_label_ranges, test_for_singularity): Likewise.
859 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
860 (x_vr_values): Move to its remaining use site.
862 2017-11-10 Jeff Law <law@redhat.com>
864 * vr-values.h (VR_INITIALIZER): Move #define here.
865 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
866 * Makefile.in (OBJS): Add tree-evrp.o
867 * tree-vrp.h (assert_info): Move structure definition here.
868 (set_value_range_to_varying): Prototype.
869 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
870 (infer_value_range, register_edge_assert_for): Likewise.
871 (stmt_interesting_for_vrp): Likewise.
872 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
873 (set_value_range_to_varying): No longer static.
874 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
875 (infer_value_range, register_edge_assert_for): Likewise.
877 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
879 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
880 generate the XXBRD instruction.
882 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
883 (rs6000_builtin_type_compatible): Treat _Float128 and long double
884 as being compatible if -mabi=ieeelongdouble.
885 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
886 to setup float128 built-ins with hardware support.
887 (BU_FLOAT128_HW_2): Likewise.
888 (BU_FLOAT128_HW_3): Likewise.
889 (BU_FLOAT128_HW_VSX_1): Likewise.
890 (BU_FLOAT128_HW_VSX_2): Likewise.
891 (scalar_extract_expq): Change float128 built-in functions to
892 accommodate having both KFmode and TFmode functions. Use the
893 KFmode variant as the default.
894 (scalar_extract_sigq): Likewise.
895 (scalar_test_neg_qp): Likewise.
896 (scalar_insert_exp_q): Likewise.
897 (scalar_insert_exp_qp): Likewise.
898 (scalar_test_data_class_qp): Likewise.
899 (sqrtf128_round_to_odd): Delete processing the round to odd
900 built-in functions as special built-in functions, and define them
901 as float128 built-ins. Use the KFmode variant as the default.
902 (truncf128_round_to_odd): Likewise.
903 (addf128_round_to_odd): Likewise.
904 (subf128_round_to_odd): Likewise.
905 (mulf128_round_to_odd): Likewise.
906 (divf128_round_to_odd): Likewise.
907 (fmaf128_round_to_odd): Likewise.
908 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
909 support for KFmode and TFmode xststdcqp calls.
910 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
911 point, switch the built-in handlers for the get/set float128
912 exponent, get float128 mantissa, float128 test built-ins, and the
913 float128 round to odd built-in functions. Eliminate creating the
914 float128 round to odd built-in functions as special built-ins.
915 (rs6000_init_builtins): Eliminate special creation of the float128
916 round to odd built-in functions.
917 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
918 function insns to support both TFmode and KFmode variants.
919 (xsxsigqp_<mode>): Likewise.
920 (xsiexpqpf_<mode>): Likewise.
921 (xsiexpqp_<mode>): Likewise.
922 (xststdcqp_<mode>): Likewise.
923 (xststdcnegqp_<mode>): Likewise.
924 (xststdcqp_<mode>): Likewise.
926 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
928 * tree-ssa-threadupdate.c (compute_path_counts): Remove
929 unused path_in_freq_ptr parameter.
930 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
932 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
934 * ipa-inline.c (edge_badness): Dump sreal frequency.
935 (compute_inlined_call_time): Match natural implementaiton ...
936 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
937 forgotten division by CGRAPH_FREQ_BASE.
939 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
941 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
942 Solaris 11. Update comment.
943 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
945 * config/sol2.h (STARTFILE_SPEC): Likewise.
946 * configure: Regenerate.
948 2017-11-14 Carl Love <cel@us.ibm.com>
950 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
951 le_ and be_ prefixes to swap* variables. Remove
952 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
955 2017-11-14 Jason Merrill <jason@redhat.com>
957 Support GTY((cache)) on hash_map.
958 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
959 (ggc_cache_remove): Override it instead of ggc_mx.
960 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
961 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
962 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
963 (simple_cache_map_traits): Override maybe_mx.
964 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
965 (hash_map): Friend gt_cleare_cache.
966 (gt_cleare_cache): New.
967 * tree.h (tree_cache_traits): New hash_map traits class.
968 (tree_cache_map): New typedef.
970 2017-11-14 Richard Biener <rguenther@suse.de>
972 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
973 paramter and handling.
974 (cleanup_control_flow_bb): Likewise.
975 (cleanup_control_flow_pre): New helper performing a DFS walk
976 to call cleanup_control_flow_bb in PRE order.
977 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
978 via cleanup_control_flow_pre.
980 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
982 * config/aarch64/aarch64-simd.md
983 (aarch64_simd_bsl<mode>_internal): Remove DImode.
984 (*aarch64_simd_bsl<mode>_alt): Likewise.
985 (aarch64_simd_bsldi_internal): New.
986 (aarch64_simd_bsldi_alt): Likewise.
988 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
990 * tracer.c (better_p): Do not compare frequencies.
991 * reg-stack.c (better_edge): Likewise.
992 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
995 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
997 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
998 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
999 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
1000 * lto-streamer-in.c (input_function): Use update_max_bb_count.
1001 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
1002 * predict.c (maybe_hot_frequency_p): Inline to ...
1003 (maybe_hot_count_p): ... here; rewrite to counts.
1004 (counts_to_freqs): Rename to ...
1005 (update_max_bb_count): ... this one.
1006 (expensive_function_p): Use counts.
1007 (estimate_bb_frequencies): Update.
1008 (rebuild_frequencies): Update.
1009 * predict.h (counts_to_freqs): Rename to ...
1010 (update_max_bb_count): ... this one.
1011 * profile.c (compute_branch_probabilities): Add debug info
1012 * tree-inline.c (expand_call_inline): Update debug info.
1013 (optimize_inline_calls): Use update_max_bb_count..
1014 (tree_function_versioning): Use update_max_bb_count..
1015 * value-prof.c (gimple_value_profile_transformations):
1016 Do not use update_max_bb_count.
1018 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1020 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
1021 always use frequencies.
1023 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1025 * bb-reorder.c: Remove frequencies from comments.
1026 (better_edge_p): Use profile counts.
1027 (find_traces): Dump profile counts.
1028 (rotate_loop): Use profile counts.
1029 (find_traces_1_round): Likewise.
1030 (connect_better_edge_p): Use counts instead of probabilities for
1032 (copy_bb_p): Drop early check for non-0 frequency.
1033 (sanitize_hot_paths): Update comments.
1035 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1037 * ipa-split.c (struct split_point): Add count.
1038 (consider_split): Do not compute incoming frequency; compute incoming
1039 count and store it to split_point.
1040 (split_function): Set count of the call to split part correctly.
1042 2017-11-13 Carl Love <cel@us.ibm.com>
1044 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
1046 2017-11-13 Tom Tromey <tom@tromey.com>
1048 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
1050 2017-11-13 Carl Love <cel@us.ibm.com>
1052 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
1053 Add support for builtins:
1054 unsigned int vec_first_{,miss}_match_{,or_eos}index,
1055 vector {un,}signed {char,int,short},
1056 vector {un,}signed {char,int,short}) arguments.
1057 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
1058 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
1059 Add BU_P9V_AV_2 expansions for the builtins.
1060 * config/rs6000/altivec.h (vec_first_match_index,
1061 vec_first_mismatch_index, vec_first_match_or_eos_index,
1062 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
1063 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
1064 new extern declaration.
1065 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
1066 * config/rs6000/vsx.md (first_match_index_<mode>,
1067 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
1068 first_mismatch_or_eos_index_<mode>): Add define expand.
1069 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
1070 * doc/extend.texi: Update the built-in documenation file for the new
1073 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
1075 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
1076 into the min/max operations for _Float<N> and _Float<N>X types.
1078 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1081 * dwarf2out.c (do_eh_frame): New static variable.
1082 (dwarf2out_begin_prologue): Set it.
1083 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
1085 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1087 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
1089 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
1091 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
1094 * debug.h (dwarf2out_do_eh_frame): Declare.
1095 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
1096 (dwarf2out_do_frame): Use it.
1097 (dwarf2out_do_cfi_asm): Likewise.
1098 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
1099 (dwarf2out_assembly_start): Likewise.
1100 (dwarf2out_begin_prologue): Fix comment.
1101 * toplev.c (compile_file): Always call dwarf2out_frame_finish
1102 if the target needs either debug or unwind DWARF2 info.
1103 * lto-opts.c (lto_write_options): Do not save -fexceptions,
1104 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
1105 -ftrapping-math, -ftrapv and -fwrapv.
1107 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
1109 * cgraph.c (cgraph_edge::sreal_frequency): New function.
1110 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
1111 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
1112 (estimate_edge_size_and_time): Likewise.
1113 (ipa_merge_fn_summary_after_inlining): Likewise.
1114 * ipa-inline.c (cgraph_freq_base_rec): Remove.
1115 (compute_uninlined_call_time): Use sreal_frequency.
1116 (compute_inlined_call_time): Likewise.
1117 (ipa_inline): Do not initialize cgraph_freq_base_rec.
1118 * profile-count.c: Include sreal.h.
1119 (profile_count::to_sreal_scale): New.
1120 * profile-count.h: Forward declare sreal.
1121 (profile_count::to_sreal_scale): Declare.
1123 2017-11-13 Nathan Sidwell <nathan@acm.org>
1125 * diagnostic.c (maybe_line_and_column): New.
1126 (diagnostic_get_location_text): Use it.
1127 (diagnostic_report_current_module): Likewise.
1128 (test_diagnostic_get_location_text): Add tests.
1130 2017-11-13 Luis Machado <luis.machado@linaro.org>
1132 * doc/md.texi (Specifying processor pipeline description): Fix
1133 incorrect latency for the div instruction example.
1135 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1137 PR tree-optimization/78821
1138 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
1139 that bit_not_p is the same.
1140 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1141 (split_group): Count precisely bit_not_p bits in each statement.
1142 (invert_op): New function.
1143 (imm_store_chain_info::output_merged_store): Use invert_op to
1144 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
1145 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
1147 2017-11-13 Martin Liska <mliska@suse.cz>
1149 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
1150 (struct source_info): Likewise.
1151 (add_branch_counts): Likewise.
1152 (add_line_counts): Likewise.
1153 (function_summary): Likewise.
1154 (output_intermediate_line): Likewise.
1155 (generate_results): Likewise.
1157 2017-11-13 Martin Liska <mliska@suse.cz>
1159 * gcov.c (struct block_info): Remove typedef for block_t.
1160 (struct line_info): Likewise.
1161 (line_info::has_block): Likewise.
1162 (EXIT_BLOCK): Likewise.
1163 (unblock): Likewise.
1164 (circuit): Likewise.
1165 (get_cycles_count): Likewise.
1166 (process_file): Likewise.
1167 (read_graph_file): Likewise.
1168 (solve_flow_graph): Likewise.
1169 (find_exception_blocks): Likewise.
1170 (add_line_counts): Likewise.
1171 (accumulate_line_info): Likewise.
1172 (output_line_details): Likewise.
1174 2017-11-13 Martin Liska <mliska@suse.cz>
1176 * gcov.c (struct arc_info): Remove typedef for arc_t.
1177 (struct line_info): Likewise.
1178 (add_branch_counts): Likewise.
1179 (output_branch_count): Likewise.
1180 (function_info::~function_info): Likewise.
1181 (circuit): Likewise.
1182 (output_intermediate_line): Likewise.
1183 (read_graph_file): Likewise.
1184 (solve_flow_graph): Likewise.
1185 (find_exception_blocks): Likewise.
1186 (add_line_counts): Likewise.
1187 (accumulate_line_info): Likewise.
1188 (output_line_details): Likewise.
1189 (output_function_details): Likewise.
1191 2017-11-13 Martin Liska <mliska@suse.cz>
1193 * gcov.c (struct function_info): Remove typedef for function_t.
1194 (struct source_info): Likewise.
1195 (source_info::get_functions_at_location): Likewise.
1196 (solve_flow_graph): Likewise.
1197 (find_exception_blocks): Likewise.
1198 (add_line_counts): Likewise.
1199 (output_intermediate_file): Likewise.
1200 (process_file): Likewise.
1201 (generate_results): Likewise.
1202 (release_structures): Likewise.
1203 (read_graph_file): Likewise.
1204 (read_count_file): Likewise.
1205 (accumulate_line_counts): Likewise.
1206 (output_lines): Likewise.
1208 2017-11-13 Martin Liska <mliska@suse.cz>
1210 * gcov.c (function_info::function_info): Remove num_counts
1211 and add vector<gcov_type>.
1212 (function_info::~function_info): Use the vector.
1213 (process_file): Likewise.
1214 (read_graph_file): Likewise.
1215 (read_count_file): Likewise.
1216 (solve_flow_graph): Likewise.
1218 2017-11-13 Martin Liska <mliska@suse.cz>
1220 * gcov.c (function_info::is_artificial): New function.
1221 (process_file): Erase all artificial early.
1222 (generate_results): Skip as all artificial are already
1225 2017-11-13 Martin Liska <mliska@suse.cz>
1227 * gcov.c (read_graph_file): Store to global vector of functions.
1228 (read_count_file): Iterate the vector.
1229 (process_file): Likewise.
1230 (generate_results): Likewise.
1231 (release_structures): Likewise.
1233 2017-11-13 Jakub Jelinek <jakub@redhat.com>
1235 PR tree-optimization/82954
1236 * gimple-ssa-store-merging.c
1237 (imm_store_chain_info::coalesce_immediate_stores): If
1238 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
1240 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
1242 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
1243 Upddate call to ENDIAN_LANE_N.
1244 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
1245 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
1246 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
1247 and use aarch64_endian_lane_rtx.
1248 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
1250 2017-11-12 Tom de Vries <tom@codesourcery.com>
1252 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
1254 2017-11-12 Tom de Vries <tom@codesourcery.com>
1256 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
1258 2017-11-12 Tom de Vries <tom@codesourcery.com>
1260 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
1262 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1263 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
1264 * defaults.h (ASM_OUTPUT_LABELREF): Same.
1266 2017-11-11 Martin Sebor <msebor@redhat.com>
1269 * doc/extend.texi (attribute nonstring): Remove spurious argument.
1272 * prefic.c (translate_name): Replace strncpy with memcpy to
1273 avoid -Wstringop-truncation.
1275 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1277 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
1279 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1281 * predict.c (maybe_hot_frequency_p): Do not use cfun.
1283 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1285 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
1288 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
1290 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
1291 enable generating XXBRH if the value is in a vector register.
1292 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
1293 value is in a vector register.
1294 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
1295 register to register bswap64's instead of doing the GPR sequence
1296 used on previous machines.
1297 (bswapdi2_xxbrd): New insn.
1298 (bswapdi2_reg): Disallow on ISA 3.0.
1299 (register to register bswap64 splitter): Do not split the insn on
1300 ISA 3.0 systems that use XXBRD.
1302 2017-11-10 Martin Sebor <msebor@redhat.com>
1305 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
1307 (find_subframework_file): Same.
1309 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
1311 * auto-profile.c (afdo_indirect_call): Drop frequency.
1312 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
1313 (cgraph_node::create_edge): Drop frequency argument.
1314 (cgraph_node::create_indirect_edge): Drop frequency argument.
1315 (cgraph_edge::make_speculative): Drop frequency arguments.
1316 (cgraph_edge::resolve_speculation): Do not update frequencies
1317 (cgraph_edge::dump_edge_flags): Do not dump frequency.
1318 (cgraph_node::dump): Check consistency in IPA mode.
1319 (cgraph_edge::maybe_hot_p): Use IPA counter.
1320 (cgraph_edge::verify_count_and_frequency): Rename to ...
1321 (cgraph_edge::verify_count): ... this one; drop frequency checking.
1322 (cgraph_node::verify_node): Update.
1323 * cgraph.h (struct cgraph_edge): Drop frequency.
1324 (cgraph_edge::frequency): New function.
1325 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
1327 (cgraph_edge::rebuild_edges): Likewise.
1328 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
1329 (duplicate_thunk_for_node): Do not pass frequency.
1330 (cgraph_node::create_clone): Scale only counts.
1331 (cgraph_node::create_virtual_clone): Do not pass frequency.
1332 (cgraph_node::create_edge_including_clones): Do not pass frequency.
1333 (cgraph_node::create_version_clone): Do not pass frequency.
1334 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
1335 (cgraph_node::expand_thunk): Do not pass frequency.
1336 (cgraph_node::create_wrapper): Do not pass frequency.
1337 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
1339 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
1340 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
1341 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
1342 (gather_caller_stats): Use frequency function.
1343 (ipcp_cloning_candidate_p): Use frequency function.
1344 (ipcp_propagate_stage): Use frequency function.
1345 (get_info_about_necessary_edges): Use frequency function.
1346 (update_profiling_info): Update only IPA profile.
1347 (update_specialized_profile): Use frequency functoin.
1348 (perhaps_add_new_callers): Update only IPA profile.
1349 * ipa-devirt.c (ipa_devirt): Use IPA profile.
1350 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
1351 (dump_ipa_call_summary): Use frequency function.
1352 (estimate_edge_size_and_time): Use frequency function.
1353 (ipa_merge_fn_summary_after_inlining): Use frequency function.
1354 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
1355 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
1356 (update_noncloned_counts): ... ths one; scale counts only.
1357 (clone_inlined_nodes): Do not scale frequency.
1358 (inline_call): Do not pass frequency.
1359 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
1360 (compute_inlined_call_time): Use IPA profile.
1361 (want_inline_small_function_p): Use IPA profile.
1362 (want_inline_self_recursive_call_p): Use IPA profile.
1363 (edge_badness): Use IPA profile.
1364 (lookup_recursive_calls): Use IPA profile.
1365 (recursive_inlining): Do not pass frequency.
1366 (resolve_noninline_speculation): Do not update frequency.
1367 (inline_small_functions): Collect max of IPA profile.
1368 (dump_overall_stats): Dump IPA porfile.
1369 (dump_inline_stats): Dump IPA porfile.
1370 (ipa_inline): Collect IPA stats.
1371 * ipa-inline.h (clone_inlined_nodes): Update prototype.
1372 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
1373 (ipa_propagate_frequency): Use frequency function.
1374 (ipa_profile): Cleanup.
1375 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
1376 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
1377 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
1378 (input_node): Do not stream frequency.
1379 (input_edge): Do not stream frequency.
1380 (merge_profile_summaries): Scale only IPA profiles.
1381 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
1382 * predict.c (drop_profile): Do not recompute frequency.
1383 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
1384 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
1385 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
1386 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
1387 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
1388 * tree-inline.c (copy_bb): Do not scale frequency.
1389 (expand_call_inline): Do not scale frequency.
1390 (tree_function_versioning): Do not scale frequency.
1391 * ubsan.c (ubsan_create_edge): Do not pass frequency.
1393 2017-11-10 Julia Koval <julia.koval@intel.com>
1395 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
1396 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
1397 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
1398 (_mm256_maskz_gf2p8affine_epi64_epi8)
1399 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
1400 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
1401 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
1402 (__builtin_ia32_vgf2p8affineqb_v32qi)
1403 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
1404 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
1406 2017-11-10 Tamar Christina <tamar.christina@arm.com>
1410 (arm_option_override): Refactor.
1411 (arm_option_reconfigure_globals): New.
1412 (arm_options_perform_arch_sanity_checks): New.
1413 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
1415 (arm_options_perform_arch_sanity_checks): Likewise
1417 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
1419 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1420 (power9-qpmul): New.
1421 * rs6000/rs6000.md ("type" attr): Add qmul.
1422 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1423 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1424 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1426 2017-11-10 Martin Sebor <msebor@redhat.com>
1429 * builtins.c (compute_objsize): Handle arrays that
1430 compute_builtin_object_size likes to fail for. Make extern.
1431 * builtins.h (compute_objsize): Declare.
1432 (check_strncpy_sizes): New function.
1433 (expand_builtin_strncpy): Call check_strncpy_sizes.
1434 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1435 -Wstringop-truncation.
1436 (gimple_fold_builtin_strncat): Same.
1437 * gimple.c (gimple_build_call_from_tree): Set call location.
1438 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1439 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1440 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1441 (handle_builtin_strlen): Use strlen_to_stridx.
1442 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1444 Use strlen_to_stridx.
1445 (pass_strlen::execute): Release strlen_to_stridx.
1446 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1447 (-Wstringop-truncation): Document new option.
1449 2017-11-10 Martin Liska <mliska@suse.cz>
1451 PR gcov-profile/82702
1452 * gcov.c (main): Handle intermediate files in a different
1454 (get_gcov_intermediate_filename): New function.
1455 (output_gcov_file): Remove support of intermediate files.
1456 (generate_results): Allocate intermediate file.
1457 (release_structures): Clean-up properly fn_end.
1458 (output_intermediate_file): Start iterating with line 1.
1460 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1462 PR tree-optimization/82929
1463 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1464 ops_swapped_p non-static data member.
1465 (store_immediate_info::store_immediate_info): Clear it.
1466 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1467 ops set ops_swapped_p.
1468 (count_multiple_uses): Handle ops_swapped_p.
1470 2017-11-10 Martin Liska <mliska@suse.cz>
1472 * coverage.c (coverage_init): Stream information about
1473 support of has_unexecuted_blocks.
1474 * doc/gcov.texi: Document that.
1475 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1476 * gcov.c (read_graph_file): Likewise.
1477 (output_line_beginning): Fix a small issue with
1480 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1482 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1483 reference of trivial component.
1485 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1488 * gimple-ssa-store-merging.c
1489 (pass_store_merging::terminate_all_aliasing_chains): For
1490 gimple_store_p stmts also call refs_output_dependent_p.
1492 PR rtl-optimization/82913
1493 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1496 2017-11-09 Jeff Law <law@redhat.com>
1498 * vr-values.h: New file with vr_values class.
1499 * tree-vrp.c: Include vr-values.h
1500 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1501 data objects into the vr_values class.
1502 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1503 (get_value_range): Make it a member function within vr_values class.
1504 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1505 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1506 (op_with_constant_singleton_value_range): Likewise.
1507 (extract_range_for_var_from_comparison_expr): Likewise.
1508 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1509 (extract_range_from_binary_expr): Likewise.
1510 (extract_range_from_unary_expr): Likewise.
1511 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1512 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1513 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1514 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1515 (compare_name_with_value, compare_names): Likewise.
1516 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1517 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1518 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1519 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1520 (extract_range_from_phi_node): Likewise.
1521 (simplify_truth_ops_using_ranges): Likewise.
1522 (simplify_div_or_mod_using_ranges): Likewise.
1523 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1524 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1525 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1526 (simplify_float_conversion_using_ranges): Likewise.
1527 (simplify_internal_call_using_ranges): Likewise.
1528 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1529 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1530 poitner via x_vr_values for calls into gimple folder.
1531 (vrp_initialize_lattice): Make this the vr_values ctor.
1532 (vrp_free_lattice): Make this the vr_values dtor.
1533 (set_vr_value): New function.
1534 (class vrp_prop): Add vr_values data member. Add various member
1535 functions as well as member functions that delegate to vr_values.
1536 (check_array_ref): Make a member function within vrp_prop class.
1537 (search_for_addr_array, vrp_initialize): Likewise.
1538 (vrp_finalize): Likewise. Revamp to avoid direct access to
1539 vr_value, values_propagated, etc.
1540 (check_array_bounds): Extract vrp_prop class instance pointer from
1541 walk info structure. Use it to call member functions.
1542 (check_all_array_refs): Make a member function within vrp_prop class.
1543 Smuggle class instance pointer via walk info structure.
1544 (x_vr_values): New local static.
1545 (vrp_valueize): Use x_vr_values to get class instance.
1546 (vr_valueize_1): Likewise.
1547 (class vrp_folder): Add vr_values data member. Add various member
1548 functions as well as member functions that delegate to vr_values.
1549 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1550 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1551 class instance from vr_values. Use it to call member functions.
1552 (vrp_dom_walker): Add vr_values data member.
1553 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1554 instance via x_vr_values.
1555 (identify_jump_threads): Accept vr_values as argument. Store
1556 it into the walker structure.
1557 (evrp_dom_walker): Add vr_values class data member. Add various
1559 (evrp_dom_walker::try_find_new_range): Use vr_values data
1560 member to access the memory allocator.
1561 (evrp_dom_walker::before_dom_children): Store vr_values class
1562 instance into the vrp_folder class.
1563 (evrp_dom_walker::push_value_range): Rework to avoid direct
1564 access to num_vr_values and vr_value.
1565 (evrp_dom_walker::pop_value_range): Likewise.
1566 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1567 Use vr_values to get to dump_all_value_ranges member function.
1568 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1569 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1570 Pass vr_values class instance down to identify_jump_threads.
1571 Remove call to vrp_free_lattice.
1572 (debug_all_value_ranges): Remove.
1574 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1575 (vrp_folder): Likewise.
1577 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1578 Get it from the existing bitmap instead.
1579 (vrp_intersect_ranges_1): Likewise.
1581 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1583 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1585 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1586 set bit_not_p to it.
1587 (imm_store_chain_info::coalesce_immediate_stores): Break group
1588 if bit_not_p is different.
1589 (count_multiple_uses, split_group,
1590 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1591 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1592 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1593 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1594 to store_immediate_info ctor.
1596 2017-11-09 Jim Wilson <jimw@sifive.com>
1598 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1599 (scan_prog_file): Likewise.
1601 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1603 * bb-reorder.c (max_entry_frequency): Remove.
1604 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1605 connect_traces, push_to_next_round_p): Remove prototypes.
1606 (find_traces_1_round): Use counts only.
1607 (push_to_next_round_p): Likewise.
1608 (find_traces): Likewise.
1609 (rotate_loop): Likewise.
1610 (find_traces_1_round): Likewise.
1611 (connect_traces): Likewise.
1612 (edge_order): Likewise.
1614 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1616 * config/arm/arm.c (output_return_instruction): Add comments to
1617 indicate requirement for cmse_nonsecure_entry return to account
1618 for the size of clearing instruction output here.
1619 (thumb_exit): Likewise.
1620 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1621 return in hardfloat mode.
1623 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1625 * config/rs6000/rs6000.c (machine_function): Add a bool,
1626 "toc_is_wrapped_separately".
1627 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1628 if it wasn't explicitly set or unset, we are optimizing for speed, and
1629 doing separate shrink-wrapping.
1630 (rs6000_get_separate_components): Enable the TOC component if
1631 saving the TOC register in the prologue.
1632 (rs6000_components_for_bb): Handle the TOC component.
1633 (rs6000_emit_prologue_components): Store the TOC register where needed.
1634 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1635 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1637 2017-11-09 Martin Jambor <mjambor@suse.cz>
1639 * ipa-param-manipulation.c: New file.
1640 * ipa-param-manipulation.h: Likewise.
1641 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1642 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1643 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1644 (ipa_parm_adjustment): Likewise.
1645 (ipa_parm_adjustment_vec): Likewise.
1646 (ipa_get_vector_of_formal_parms): Moved declaration to
1647 ipa-param-manipulation.h.
1648 (ipa_get_vector_of_formal_parm_types): Likewise.
1649 (ipa_modify_formal_parameters): Likewise.
1650 (ipa_modify_call_arguments): Likewise.
1651 (ipa_combine_adjustments): Likewise.
1652 (ipa_dump_param_adjustments): Likewise.
1653 (ipa_modify_expr): Likewise.
1654 (ipa_get_adjustment_candidate): Likewise.
1655 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1656 ipa-param-manipulation.c.
1657 (ipa_get_vector_of_formal_parm_types): Likewise.
1658 (ipa_modify_formal_parameters): Likewise.
1659 (ipa_modify_call_arguments): Likewise.
1660 (ipa_modify_expr): Likewise.
1661 (get_ssa_base_param): Likewise.
1662 (ipa_get_adjustment_candidate): Likewise.
1663 (index_in_adjustments_multiple_times_p): Likewise.
1664 (ipa_combine_adjustments): Likewise.
1665 (ipa_dump_param_adjustments): Likewise.
1666 * tree-sra.c: Also include ipa-param-manipulation.h
1667 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1670 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1671 Alan Hayward <alan.hayward@arm.com>
1672 David Sherwood <david.sherwood@arm.com>
1674 * doc/sourcebuild.texi (vect_masked_store): Document.
1676 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1677 Alan Hayward <alan.hayward@arm.com>
1678 David Sherwood <david.sherwood@arm.com>
1680 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1682 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1683 Alan Hayward <alan.hayward@arm.com>
1684 David Sherwood <david.sherwood@arm.com>
1686 * doc/sourcebuild.texi (vect_variable_length): Document.
1688 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1689 Alan Hayward <alan.hayward@arm.com>
1690 David Sherwood <david.sherwood@arm.com>
1692 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1694 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1695 Alan Hayward <alan.hayward@arm.com>
1696 David Sherwood <david.sherwood@arm.com>
1698 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1700 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1701 Alan Hayward <alan.hayward@arm.com>
1702 David Sherwood <david.sherwood@arm.com>
1704 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1705 previously undocumented selectors.
1706 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1708 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1710 * doc/rtl.texi (const_vector): Say that elements can be
1711 const_wide_ints too.
1712 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1713 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1714 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1715 * optabs.c (expand_vector_broadcast): Likewise.
1717 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1718 Alan Hayward <alan.hayward@arm.com>
1719 David Sherwood <david.sherwood@arm.com>
1721 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1722 scaled index even if the unscaled address was invalid.
1723 Don't increase the complexity of using a scale in that case.
1725 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1726 Alan Hayward <alan.hayward@arm.com>
1727 David Sherwood <david.sherwood@arm.com>
1729 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1730 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1731 UNITS_PER_WORD bytes.
1732 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1733 based on REGMODE_NATURAL_SIZE of the inner mode.
1734 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1735 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1736 * expmed.c (lowpart_bit_field_p): Divide the value up into
1737 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1738 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1739 whether something is likely to occupy more than one register.
1741 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1744 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1745 profile_count::adjust_for_ipa_scaling.
1746 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1747 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1749 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1751 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1753 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1754 (split_group): Add total_orig and total_new arguments, estimate the
1755 number of statements related to the store group without store merging
1756 and with store merging.
1757 (imm_store_chain_info::output_merged_store): Adjust split_group
1758 callers, punt if estimated number of statements with store merging
1759 is not smaller than estimated number of statements without it.
1761 (handled_load): Remove has_single_use checks.
1762 (pass_store_merging::process_store): Likewise.
1764 2017-11-09 Richard Biener <rguenther@suse.de>
1766 PR tree-optimization/82902
1767 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1769 2017-11-09 Martin Liska <mliska@suse.cz>
1772 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1775 2017-11-09 Martin Liska <mliska@suse.cz>
1777 PR tree-optimization/82669
1778 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1780 2017-11-09 Martin Liska <mliska@suse.cz>
1782 PR gcov-profile/48463
1783 * coverage.c (coverage_begin_function): Output also end locus
1784 of a function and information whether the function is
1786 * gcov-dump.c (tag_function): Parse and print the information.
1787 * gcov.c (INCLUDE_MAP): Add include.
1788 (INCLUDE_SET): Likewise.
1789 (struct line_info): Move earlier in the source file because
1790 of vector<line_info> in function_info structure.
1791 (line_info::line_info): Likewise.
1792 (line_info::has_block): Likewise.
1793 (struct source_info): Add new member index.
1794 (source_info::get_functions_at_location): New function.
1795 (function_info::group_line_p): New function.
1796 (output_intermediate_line): New function.
1797 (output_intermediate_file): Use the mentioned function.
1798 (struct function_start): New.
1799 (struct function_start_pair_hash): Likewise.
1800 (process_file): Add code that identifies group functions.
1801 Assign lines either to global or function scope.
1802 (generate_results): Skip artificial functions.
1803 (find_source): Assign index for each source file.
1804 (read_graph_file): Read new flag artificial and end_line.
1805 (add_line_counts): Assign it either to global of function scope.
1806 (accumulate_line_counts): Isolate core of the function to
1807 accumulate_line_info and call it for both function and global
1809 (accumulate_line_info): New function.
1810 (output_line_beginning): Fix GNU coding style.
1811 (print_source_line): New function.
1812 (output_line_details): Likewise.
1813 (output_function_details): Likewise.
1814 (output_lines): Iterate both source (global) scope and function
1816 (struct function_line_start_cmp): New class.
1817 * doc/gcov.texi: Reflect changes in documentation.
1819 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1822 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1823 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1824 and similarly for not instead of neg.
1826 2017-11-08 Andi Kleen <ak@linux.intel.com>
1828 * config/i386/i386.opt: Add -mforce-indirect-call.
1829 * config/i386/predicates.md: Check for flag_force_indirect_call.
1830 * doc/invoke.texi: Document -mforce-indirect-call
1832 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1834 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1836 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1837 config/riscv/riscv.c (predict.h): New include.
1838 (riscv_slow_unaligned_access_p): No longer static.
1839 (riscv_block_move_straight): Add require.
1840 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1842 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1845 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1846 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1847 nonimmediate_operand predicate for operand 1 instead of
1850 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1852 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1854 * config/aarch64/constraints.md (Uml): New constraint.
1855 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1858 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1860 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1861 of two vec_duplicates into a vec_concat.
1863 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1865 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1866 Simplify vec_merge of vec_duplicate and vec_concat.
1867 * config/aarch64/constraints.md (Utq): New constraint.
1868 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1871 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1873 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1874 Simplify vec_merge of vec_duplicate and const_vector.
1875 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1877 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1878 mode iterator. Update predicate on operand 1 to
1879 handle non-const_vec constants. Delete constraints.
1880 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1882 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1884 PR tree-optimization/78821
1885 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1887 (store_operand_info::store_operand_info): Initialize it to false.
1888 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1889 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1890 store in the group, and if chain_info is non-NULL, to ignore altogether
1892 (compatible_load_p): Fail if bit_not_p does not match.
1893 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1894 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1895 (pass_store_merging::process_store): Adjust
1896 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1897 call terminate_all_aliasing_chains newly when adding a store into
1898 a chain with non-NULL chain_info.
1900 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1902 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1904 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1906 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1908 (aarch64_layout_frame): Initialise emit_frame_chain.
1909 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1910 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1912 2017-11-08 Martin Liska <mliska@suse.cz>
1914 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1915 of gimple_call_internal_p.
1917 2017-11-07 Tom de Vries <tom@codesourcery.com>
1919 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1921 2017-11-07 Tom de Vries <tom@codesourcery.com>
1923 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1926 2017-11-08 Martin Liska <mliska@suse.cz>
1929 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1931 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1933 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1935 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1938 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1939 (*cmp<mode>_ccz_1): New insn with $k alternative.
1942 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1943 RTX_COMM_COMPARE as commutative as well.
1944 (ix86_binary_operator_ok): Formatting fix.
1945 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1946 *<code><mode>3<mask_name><round_saeonly_name>,
1947 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1948 *<s>mul<mode>3_highpart<mask_name>,
1949 *vec_widen_umult_even_v16si<mask_name>,
1950 *vec_widen_umult_even_v8si<mask_name>,
1951 *vec_widen_umult_even_v4si<mask_name>,
1952 *vec_widen_smult_even_v16si<mask_name>,
1953 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1954 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1955 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1956 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1957 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1958 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1959 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1960 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1961 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1962 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1963 ix86_binary_operator_ok. Formatting fixes.
1964 (*<plusminus_insn><mode>3<mask_name><round_name>,
1965 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1968 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1970 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1971 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1972 eq and ne if TARGET_ISEL.
1973 (cmp): New code_iterator.
1974 (UNS, UNSU_, UNSIK): New code_attrs.
1975 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1976 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1978 ("eq<mode>3"): ... this.
1979 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1981 ("ne<mode>3"): ... this.
1983 2017-11-07 Julia Koval <julia.koval@intel.com>
1986 * common/config/i386/i386-common.c
1987 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1988 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1989 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1990 * config/i386/i386.opt: Ditto.
1991 * config/i386/i386.c (ix86_target_string): Ditto.
1992 (ix86_option_override_internal): Ditto.
1993 (ix86_init_mpx_builtins): Move MPX to args2.
1994 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1995 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1996 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1997 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1998 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1999 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2000 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
2002 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
2005 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
2006 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
2007 (zero-extendsidi peephole2): Remove peephole.
2009 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
2012 * stor-layout.c: Include attribs.h.
2013 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
2014 explicit lookup of "aligned" attribute.
2016 2017-11-07 Andrew Waterman <andrew@sifive.com>
2018 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
2019 (riscv_expand_block_move): Likewise.
2020 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
2022 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
2023 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
2024 * config/riscv/riscv.c (riscv_block_move_straight): New function.
2025 (riscv_adjust_block_mem): Likewise.
2026 (riscv_block_move_loop): Likewise.
2027 (riscv_expand_block_move): Likewise.
2028 gcc/config/riscv/riscv.md (movmemsi): New pattern.
2030 2017-11-07 Michael Clark <michaeljclark@mac.com>
2032 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
2033 (MUSL_DYNAMIC_LINKER): Likewise.
2035 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
2037 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
2038 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
2041 2017-11-07 Richard Biener <rguenther@suse.de>
2043 * match.pd: Fix build.
2045 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
2046 Jackson Woodruff <jackson.woodruff@arm.com>
2048 PR tree-optimization/71026
2049 * match.pd: Canonicalize negate in division.
2051 2017-11-07 Sudakshina Das <sudi.das@arm.com>
2054 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
2056 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2058 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
2059 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
2061 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
2063 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
2064 non-scalar integral types.
2065 * match.pd (negate_expr_p): Handle MINUS_EXPR.
2066 (-(A-B), -(~A)): New transformations.
2068 2017-11-07 Tom de Vries <tom@codesourcery.com>
2070 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
2071 semicolon after "do {} while (0)".
2072 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2073 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2074 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2075 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2076 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2078 2017-11-07 Tom de Vries <tom@codesourcery.com>
2080 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
2081 after "do {} while (0)".
2082 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2083 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2084 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2085 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2086 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
2088 2017-11-07 Tom de Vries <tom@codesourcery.com>
2091 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
2093 (arm_rtx_costs_internal): Add missing semicolon after
2094 HANDLE_NARROW_SHIFT_ARITH call.
2096 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2098 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
2099 disable isel if it was not set explicitly.
2101 2017-11-06 James Bowman <james.bowman@ftdichip.com>
2103 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
2105 (add_type_attribute) likewise.
2107 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
2109 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
2110 of struct ix86_frame.
2111 (ix86_initial_elimination_offset): Likewise.
2112 (ix86_expand_split_stack_prologue): Likewise.
2114 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
2116 * tree-vrp.h (enum value_range_type): Update stale comment.
2118 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2119 Alan Hayward <alan.hayward@arm.com>
2120 David Sherwood <david.sherwood@arm.com>
2122 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
2123 (aarch64_expand_vec_perm_const): Take the number of units too.
2124 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
2125 (aarch64_expand_vec_perm_const): Likewise.
2126 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
2127 (vec_perm<mode>): Update accordingly.
2129 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2130 Alan Hayward <alan.hayward@arm.com>
2131 David Sherwood <david.sherwood@arm.com>
2133 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
2134 Take the number of units too.
2135 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
2136 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
2137 but check for a vector mode before rather than after the call.
2138 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
2139 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
2140 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
2141 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
2142 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
2143 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
2144 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
2145 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
2146 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
2147 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
2148 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
2149 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
2150 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
2151 (aarch64_sqdmull2_n<mode>): Update accordingly.
2153 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2154 Alan Hayward <alan.hayward@arm.com>
2155 David Sherwood <david.sherwood@arm.com>
2157 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
2158 the number of units too.
2159 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
2160 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
2161 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
2162 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
2163 (vec_store_lanesxi<mode>): Update accordingly.
2165 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2166 Alan Hayward <alan.hayward@arm.com>
2167 David Sherwood <david.sherwood@arm.com>
2169 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
2170 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
2171 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
2172 of units rather than the mode.
2173 * config/aarch64/iterators.md (nunits): New mode attribute.
2174 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
2175 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
2176 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
2177 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
2178 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
2179 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
2180 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
2181 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
2182 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
2183 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
2184 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
2185 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
2186 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
2187 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
2188 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
2189 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
2190 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
2191 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
2192 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
2193 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
2194 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
2195 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
2196 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
2197 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
2198 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
2199 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
2200 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
2201 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
2202 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
2203 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
2204 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
2205 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
2206 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
2207 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
2208 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
2209 (aarch64_simd_vec_setv2di): Likewise.
2211 2017-11-06 Carl Love <cel@us.ibm.com>
2213 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
2215 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
2216 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
2218 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
2220 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
2221 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
2222 Add power 8 macro expansions.
2223 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
2224 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
2225 power 8 instructions. (VSX_XXBR): Add iterator.
2227 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
2229 * config/arm/arm.md (predicable_short_it): Change default to "no",
2230 improve documentation, remove uses that are identical to the default.
2231 (enabled_for_depr_it): Rename to enabled_for_short_it.
2232 * gcc/config/arm/arm-fixed.md (predicable_short_it):
2233 Remove default uses.
2234 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
2235 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
2236 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
2237 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
2239 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2242 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
2243 float128 helper macros, which are no longer used after deleting
2244 the old 'q' built-in functions, and moving the round to odd
2245 built-in functions to being special built-in functions.
2246 (BU_FLOAT128_2): Likewise.
2247 (BU_FLOAT128_1_HW): Likewise.
2248 (BU_FLOAT128_2_HW): Likewise.
2249 (BU_FLOAT128_3_HW): Likewise.
2250 (FABSQ): Delete old 'q' built-in functions.
2251 (COPYSIGNQ): Likewise.
2252 (SQRTF128_ODD): Move round to odd built-in functions to be
2253 special built-in functions, so that we can handle
2254 -mabi=ieeelongdouble.
2255 (TRUNCF128_ODD): Likewise.
2256 (ADDF128_ODD): Likewise.
2257 (SUBF128_ODD): Likewise.
2258 (MULF128_ODD): Likewise.
2259 (DIVF128_ODD): Likewise.
2260 (FMAF128_ODD): Likewise.
2261 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
2262 built-in names to 'f128'.
2263 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
2264 old 'q' built-in functions, as the machine independent code for
2265 'f128' built-in functions handles this.
2266 (rs6000_expand_builtin): Add expansion for float128 round to odd
2267 functions, keying off on -mabi=ieeelongdouble of whether to use
2268 the KFmode or TFmode variant.
2269 (rs6000_init_builtins): Initialize the _Float128 round to odd
2271 * doc/extend.texi (PowerPC Built-in Functions): Document the old
2272 _Float128 'q' built-in functions are now mapped into the new
2273 'f128' built-in functions.
2275 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
2277 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
2278 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
2280 2017-11-06 David Malcolm <dmalcolm@redhat.com>
2283 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
2284 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
2285 * toplev.c: Include "ipa-fnsummary.h".
2286 (toplev::finalize): Call ipa_fnsummary_c_finalize.
2288 2017-11-06 Jakub Jelinek <jakub@redhat.com>
2290 PR tree-optimization/82838
2291 * gimple-ssa-store-merging.c
2292 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
2293 on a separate gimple_seq which is then appended to seq.
2295 2017-11-06 Jeff Law <law@redhat.com>
2298 * config/i386/i386.c (PROBE_INTERVAL): Remove.
2299 (get_probe_interval): New functions.
2300 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
2301 (ix86_adjust_stack_and_probe): Likewise.
2302 (output_adjust_stack_and_probe): Likewise.
2303 (ix86_emit_probe_stack_range): Likewise.
2304 (ix86_expand_prologue): Likewise.
2306 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2308 PR tree-optimization/82816
2309 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
2310 if the modes of the two types are the same.
2311 (convert_plusminus_to_widen): Likewise.
2313 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2315 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
2317 (usadv16qi): New define_expand.
2318 (usadv8hi): New define_expand.
2320 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
2323 * ipa-inline-transform.c (update_noncloned_frequencies): Always
2325 (inline_transform): Likewise.
2326 * predict.c (counts_to_freqs): Remove useless conditional.
2327 * profile-count.h (profile_count::apply_scale): Move sanity check.
2328 * tree-inline.c (copy_bb): Always scale.
2329 (copy_cfg_body): Likewise.
2331 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
2334 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
2336 (*cmp_ite0): Add enabled_for_depr_it attribute.
2337 (*cmp_ite1): Likewise.
2339 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
2341 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
2344 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
2346 * tree-vrp.c (vrp_int_const_binop): Return true on success and
2347 return the value by pointer.
2348 (extract_range_from_multiplicative_op_1): Update accordingly.
2349 Return as soon as an operation fails.
2351 2017-11-05 Tom de Vries <tom@codesourcery.com>
2354 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
2355 (DEF_SANITIZER_BUILTIN): ... here.
2356 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
2357 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
2359 2017-11-05 Tom de Vries <tom@codesourcery.com>
2362 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
2364 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
2365 ASM_OUTPUT_BEFORE_CASE_LABEL call.
2366 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
2368 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2369 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2370 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
2372 2017-11-05 Tom de Vries <tom@codesourcery.com>
2375 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
2378 2017-11-04 Michael Clark <michaeljclark@mac.com>
2380 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
2381 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
2383 (*addsi3_extended): Likewise.
2384 (*addsi3_extended2): Likewise.
2385 (<optab>si3): Likewise.
2386 (<optab>di3): Likewise.
2387 (<optab><mode>3): Likewise.
2388 (<*optabe>si3_internal): Likewise.
2389 (zero_extendqi<SUPERQI:mode>2): Likewise.
2390 (*add<mode>hi3): Likewise.
2391 (*xor<mode>hi3): Likewise.
2392 (<optab>di3): Likewise.
2393 (*<optab>si3_extend): Likewise.
2394 (*sge<u>_<X:mode><GPR:mode>): Likewise.
2395 (*slt<u>_<X:mode><GPR:mode>): Likewise.
2396 (*sle<u>_<X:mode><GPR:mode>): Likewise.
2398 2017-11-04 Andrew Waterman <andrew@sifive.com>
2400 * config/riscv/riscv.c (riscv_option_override): Conditionally set
2401 TARGET_STRICT_ALIGN based upon -mtune argument.
2403 2017-11-04 Andrew Waterman <andrew@sifive.com>
2405 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
2407 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
2409 config/i386/i386.c (choose_basereg): Use optional scratch
2410 register and add assertion.
2411 (x86_emit_outlined_ms2sysv_save): Use scratch register when
2412 needed, and don't allocate stack.
2413 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
2414 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
2415 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
2417 2017-11-03 Jeff Law <law@redhat.com>
2419 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2420 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2421 to probe at the start of a noreturn function.
2423 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2425 PR tree-optimization/78821
2426 * gimple-ssa-store-merging.c: Update the file comment.
2427 (MAX_STORE_ALIAS_CHECKS): Define.
2428 (struct store_operand_info): New type.
2429 (store_operand_info::store_operand_info): New constructor.
2430 (struct store_immediate_info): Add rhs_code and ops data members.
2431 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2432 and op1r arguments to the ctor, initialize corresponding data members.
2433 (struct merged_store_group): Add load_align_base and load_align
2435 (merged_store_group::merged_store_group): Initialize them.
2436 (merged_store_group::do_merge): Update them.
2437 (merged_store_group::apply_stores): Pick the constant for
2438 encode_tree_to_bitpos from one of the two operands, or skip
2439 encode_tree_to_bitpos if neither operand is a constant.
2440 (class pass_store_merging): Add process_store method decl. Remove
2441 bool argument from terminate_all_aliasing_chains method decl.
2442 (pass_store_merging::terminate_all_aliasing_chains): Remove
2443 var_offset_p argument and corresponding handling.
2444 (stmts_may_clobber_ref_p): New function.
2445 (compatible_load_p): New function.
2446 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2447 if there is overlap and rhs_code is not INTEGER_CST. For
2448 non-overlapping stores terminate group if rhs is not mergeable.
2449 (get_alias_type_for_stmts): Change first argument from
2450 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2451 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2452 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2454 (get_location_for_stmts): Change first argument from
2455 auto_vec<gimple *> & to vec<gimple *> &.
2456 (struct split_store): Remove orig_stmts data member, add orig_stores.
2457 (split_store::split_store): Create orig_stores rather than orig_stmts.
2458 (find_constituent_stmts): Renamed to ...
2459 (find_constituent_stores): ... this. Change second argument from
2460 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2461 to info structures rather than the statements.
2462 (split_group): Rename ALLOW_UNALIGNED argument to
2463 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2464 it. Adjust find_constituent_stores caller.
2465 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2466 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2467 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2468 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2469 (mem_valid_for_store_merging): New function.
2470 (handled_load): New function.
2471 (pass_store_merging::process_store): New method.
2472 (pass_store_merging::execute): Use process_store method. Adjust
2473 terminate_all_aliasing_chains caller.
2475 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2477 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2478 Return true for more constants, symbols and label references.
2479 (aarch64_valid_floating_const): Remove unused function.
2481 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2484 * config/aarch64/aarch64.c (aarch64_layout_frame):
2485 Undo forcing of LR at bottom of frame.
2487 2017-11-03 Jeff Law <law@redhat.com>
2490 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2491 for int_registers_saved.
2493 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2494 extracted from tree-ssa-dom.c.
2495 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2496 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2497 (record_equivalences_from_incoming_edge): Add additional argument
2498 to single_pred_edge_ignoring_loop_edges call.
2499 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2500 (uncprop_dom_walker::before_dom_children): Add additional argument
2501 to single_pred_edge_ignoring_loop_edges call.
2502 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2503 single_pred_edge_ignoring_loop_edges rather than open coding.
2504 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2506 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2508 * match.pd (-(-A)): Rewrite.
2510 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2512 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2513 (rs6000_emit_int_cmove): New declaration.
2514 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2515 (rs6000_emit_sISEL): Delete.
2516 (rs6000_emit_int_cmove): Make non-static.
2517 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2518 instead of rs6000_emit_sISEL.
2520 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2522 * asan.c (create_cond_insert_point): Maintain profile.
2523 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2525 * basic-block.h (struct basic_block_def): Remove frequency.
2526 (EDGE_FREQUENCY): Use to_frequency
2527 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2529 (find_traces): Update to use to_frequency.
2530 (find_traces_1_round): Likewise; use only IPA counts.
2531 (bb_to_key): Likewise.
2532 (connect_traces): Use IPA counts only.
2533 (copy_bb_p): Update to use to_frequency.
2534 (fix_up_crossing_landing_pad): Likewise.
2535 (sanitize_hot_paths): Likewise.
2536 * bt-load.c (basic_block_freq): Likewise.
2537 * cfg.c (init_flow): Set count_max to uninitialized.
2538 (check_bb_profile): Remove frequencies; check counts.
2539 (dump_bb_info): Do not dump frequencies.
2540 (update_bb_profile_for_threading): Update counts only.
2541 (scale_bbs_frequencies_int): Likewise.
2542 (MAX_SAFE_MULTIPLIER): Remove.
2543 (scale_bbs_frequencies_gcov_type): Update counts only.
2544 (scale_bbs_frequencies_profile_count): Update counts only.
2545 (scale_bbs_frequencies): Update counts only.
2546 * cfg.h (struct control_flow_graph): Add count-max.
2547 (update_bb_profile_for_threading): Update prototype.
2548 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2549 (find_many_sub_basic_blocks): Likewise.
2550 * cfgcleanup.c (try_forward_edges): Likewise.
2551 (try_crossjump_to_edge): Likewise.
2552 * cfgexpand.c (expand_gimple_cond): Likewise.
2553 (expand_gimple_tailcall): Likewise.
2554 (construct_init_block): Likewise.
2555 (construct_exit_block): Likewise.
2556 * cfghooks.c (verify_flow_info): Check consistency of counts.
2557 (dump_bb_for_graph): Do not dump frequencies.
2558 (split_block_1): Do not update frequencies.
2559 (split_edge): Do not update frequencies.
2560 (make_forwarder_block): Do not update frequencies.
2561 (duplicate_block): Do not update frequencies.
2562 (account_profile_record): Do not update frequencies.
2563 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2564 for global heuristics.
2565 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2566 (expected_loop_iterations_unbounded): Use counts only.
2567 * cfgloopmanip.c (scale_loop_profile): Simplify.
2568 (create_empty_loop_on_edge): Simplify
2570 (duplicate_loop_to_header_edge): Simplify
2571 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2572 (update_br_prob_note): Take care of removing note when profile
2574 (relink_block_chain): Do not dump frequency.
2575 (rtl_account_profile_record): Use to_frequency.
2576 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2577 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2578 (cgraph_update_edges_for_call_stmt_node): Likewise.
2579 (cgraph_edge::verify_count_and_frequency): Update.
2580 (cgraph_node::verify_node): Temporarily disable frequency verification.
2581 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2582 to_cgraph_frequency.
2583 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2584 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2586 (cgraph_node::expand_thunk): Update profile.
2587 * except.c (dw2_build_landing_pads): Do not update frequency.
2588 * final.c (compute_alignments): Use to_frequency.
2589 (dump_basic_block_info): Do not dump frequency.
2590 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2591 (dump_gimple_bb_header): Do not dump frequency.
2592 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2594 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2595 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2596 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2597 (init_before_recovery): Do not update frequency.
2598 (sched_create_recovery_edges): Do not update frequency.
2599 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2600 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2601 (ipa_cp_c_finalize): Set max_count to uninitialized.
2602 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2603 (param_change_prob): Use counts.
2604 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2606 * ipa-split.c (consider_split): Use to_frequency.
2607 (split_function): Use to_frequency.
2608 * ira-build.c (loop_compare_func): Likewise.
2609 (mark_loops_for_removal): Likewise.
2610 (mark_all_loops_for_removal): Likewise.
2611 * loop-doloop.c (doloop_modify): Do not update frequency.
2612 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2614 * lto-streamer-in.c (input_function): Update count_max.
2615 * omp-expand.c (expand_omp_taskreg): Update count_max.
2616 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2617 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2618 (maybe_hot_count_p): Use ipa counts only.
2619 (maybe_hot_bb_p): Simplify.
2620 (maybe_hot_edge_p): Simplify.
2621 (probably_never_executed): Do not take frequency argument.
2622 (probably_never_executed_bb_p): Do not pass frequency.
2623 (probably_never_executed_edge_p): Likewise.
2624 (combine_predictions_for_bb): Check that profile is nonzero.
2625 (propagate_freq): Do not set frequency.
2626 (drop_profile): Simplify.
2627 (counts_to_freqs): Simplify.
2628 (expensive_function_p): Use to_frequency.
2629 (propagate_unlikely_bbs_forward): Simplify.
2630 (determine_unlikely_bbs): Simplify.
2631 (estimate_bb_frequencies): Add hack to silence graphite issues.
2632 (compute_function_frequency): Use ipa counts.
2633 (pass_profile::execute): Update.
2634 (rebuild_frequencies): Use counts only.
2635 (force_edge_cold): Use counts only.
2636 * profile-count.c (profile_count::dump): Dump new count types.
2637 (profile_count::differs_from_p): Check compatiblity.
2638 (profile_count::to_frequency): New function.
2639 (profile_count::to_cgraph_frequency): New function.
2640 * profile-count.h (struct function): Declare.
2641 (enum profile_quality): Add profile_guessed_local and
2642 profile_guessed_global0.
2643 (class profile_proability): Decrease number of bits to 29;
2644 update from_reg_br_prob_note and to_reg_br_prob_note.
2645 (class profile_count: Update comment; decrease number of bits
2646 to 61. Check compatibility.
2647 (profile_count::compatible_p): New private member function.
2648 (profile_count::ipa_p): New member function.
2649 (profile_count::operator<): Handle global zero correctly.
2650 (profile_count::operator>): Handle global zero correctly.
2651 (profile_count::operator<=): Handle global zero correctly.
2652 (profile_count::operator>=): Handle global zero correctly.
2653 (profile_count::nonzero_p): New member function.
2654 (profile_count::force_nonzero): New member function.
2655 (profile_count::max): New member function.
2656 (profile_count::apply_scale): Handle IPA scalling.
2657 (profile_count::guessed_local): New member function.
2658 (profile_count::global0): New member function.
2659 (profile_count::ipa): New member function.
2660 (profile_count::to_frequency): Declare.
2661 (profile_count::to_cgraph_frequency): Declare.
2662 * profile.c (OVERLAP_BASE): Delete.
2663 (compute_frequency_overlap): Delete.
2664 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2665 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2666 * sched-ebb.c (rank): Use counts only.
2667 * shrink-wrap.c (handle_simple_exit): Use counts only.
2668 (try_shrink_wrapping): Use counts only.
2669 (place_prologue_for_one_component): Use counts only.
2670 * tracer.c (find_best_predecessor): Use to_frequency.
2671 (find_trace): Use to_frequency.
2672 (tail_duplicate): Use to_frequency.
2673 * trans-mem.c (expand_transaction): Do not update frequency.
2674 * tree-call-cdce.c: Do not update frequency.
2675 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2676 (gimple_merge_blocks): Likewise.
2677 (gimple_split_edge): Likewise.
2678 (gimple_duplicate_sese_region): Likewise.
2679 (gimple_duplicate_sese_tail): Likewise.
2680 (move_sese_region_to_fn): Likewise.
2681 (gimple_account_profile_record): Likewise.
2682 (insert_cond_bb): Likewise.
2683 * tree-complex.c (expand_complex_div_wide): Likewise.
2684 * tree-eh.c (lower_resx): Update profile.
2685 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2687 (initialize_cfun): Do not initialize frequencies
2688 (freqs_to_counts): Delete.
2689 (copy_cfg_body): Ignore count parameter.
2690 (copy_body): Update.
2691 (expand_call_inline): Update count_max.
2692 (optimize_inline_calls): Update count_max.
2693 (tree_function_versioning): Update count_max.
2694 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2695 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2697 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2698 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2699 (try_peel_loop): Likewise.
2700 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2702 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2703 (tree_transform_and_unroll_loop): Do not use frequencies
2704 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2705 Use reliable prediction only.
2706 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2707 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2708 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2709 probability scaling.
2710 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2711 not update frequency
2712 (any_remaining_duplicated_blocks): Likewise.
2713 (update_profile): Likewise.
2714 (estimated_freqs_path): Delete.
2715 (freqs_to_counts_path): Delete.
2716 (clear_counts_path): Delete.
2717 (ssa_fix_duplicate_block_edges): Likewise.
2718 (duplicate_thread_path): Likewise.
2719 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2720 * tree-tailcall.c (decrease_profile): Do not update frequency.
2721 (eliminate_tail_call): Likewise.
2722 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2723 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2724 (optimize_mask_stores): Likewise.
2725 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2726 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2727 (ubsan_expand_ptr_ifn): Update profile.
2728 * value-prof.c (gimple_ic): Simplify.
2729 * value-prof.h (gimple_ic): Update prototype.
2730 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2731 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2733 (want_inline_self_recursive_call_p): Likewise.
2734 (resolve_noninline_speculation): Only cummulate defined counts.
2735 (inline_small_functions): Use nonzero_p.
2736 (ipa_inline): Do not access freed node.
2738 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2740 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2741 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2743 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2745 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2746 non-legitimate address.
2748 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2750 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2751 (*lt0_<mode>di, *lt0_<mode>si): New.
2753 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2755 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2756 TARGET_PAIRED_FLOAT.
2758 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2759 Jim Wilson <jim.wilson@linaro.org>
2761 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2762 * config/aarch64/aarch64-tune.md: Regenerated.
2763 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2764 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2766 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2768 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2769 (arc_expand_prologue): Restore blink for millicode.
2770 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2772 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2775 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2776 gen_vec_duplicate after forcing the scalar into a register.
2778 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2780 * combine (try_combine): Print the insns input to try_combine to the
2783 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2786 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2787 Remove second argument from aarch64_process_target_attr call.
2788 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2790 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2792 (aarch64_handle_attr_arch): Remove second argument.
2793 (aarch64_handle_attr_cpu): Ditto.
2794 (aarch64_handle_attr_tune): Ditto.
2795 (aarch64_handle_attr_isa_flags): Ditto.
2796 (aarch64_process_one_target_attr): Ditto.
2797 (aarch64_process_target_attr): Ditto.
2798 (aarch64_option_valid_attribute_p): Remove second argument.
2799 on aarch64_process_target_attr call.
2801 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2803 * diagnostic.c: Include "selftest-diagnostic.h".
2804 (selftest::assert_location_text): New function.
2805 (selftest::test_diagnostic_get_location_text): New function.
2806 (selftest::diagnostic_c_tests): Call it.
2808 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2810 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2811 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2812 (class selftest::test_diagnostic_context): Move to...
2813 * selftest-diagnostic.c: New file.
2814 * selftest-diagnostic.h: New file.
2816 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2818 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2819 offset range for FT32B.
2820 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2821 * config/ft32/ft32.md: Add TARGET_NOPM.
2822 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2823 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2825 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2827 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2829 2017-11-02 Jeff Law <law@redhat.com>
2831 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2832 virtual keyword on FINAL OVERRIDE members.
2834 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2835 virtuals together. Add virtual destructor.
2836 (substitute_and_fold_engine): Similarly.
2838 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2840 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2842 2017-11-02 Richard Biener <rguenther@suse.de>
2844 PR tree-optimization/82795
2845 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2847 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2849 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2851 (gcc_GAS_CHECK_FEATURE): Remove.
2852 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2853 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2854 * configure: Regenerate.
2856 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2858 * config/arc/arc.c (hwloop_optimize): Account for empty
2861 2017-11-02 Richard Biener <rguenther@suse.de>
2864 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2865 Truncate ARRAY_REF index and element size.
2867 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2869 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2871 2017-11-01 Jeff Law <law@redhat.com>
2873 * tree-ssa-ccp.c (ccp_folder): New class derived from
2874 substitute_and_fold_engine.
2875 (ccp_folder::get_value): New member function.
2876 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2877 (ccp_fold_stmt): Remove prototype.
2878 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2879 * tree-ssa-copy.c (copy_folder): New class derived from
2880 substitute_and_fold_engine.
2881 (copy_folder::get_value): Renamed from get_value.
2882 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2883 * tree-vrp.c (vrp_folder): New class derived from
2884 substitute_and_fold_engine.
2885 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2886 (vrp_folder::get_value): New member function.
2887 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2888 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2889 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2890 provide a class interface to folder/substitute routines.
2891 (ssa_prop_fold_stmt_fn): Remove typedef.
2892 (ssa_prop_get_value_fn): Likewise.
2893 (subsitute_and_fold): Remove prototype.
2894 (replace_uses_in): Likewise.
2895 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2896 Renamed from replace_uses_in. Call the virtual member function
2897 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2898 (substitute_and_fold_dom_walker): Remove initialization of
2899 data member entries for calbacks. Add substitute_and_fold_engine
2900 member and initialize it.
2901 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2902 member functions for get_value, replace_phi_args_in c
2903 replace_uses_in, and fold_stmt calls.
2904 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2905 substitute_and_fold. Remove assert. Update ctor call.
2907 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2908 (ssa_prop_visit_phi_fn): Likewise.
2909 (class ssa_propagation_engine): New class to provide an interface
2911 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2913 (ssa_prop_visit_phi): Likewise.
2914 (ssa_propagation_engine::simulate_stmt): Moved into class.
2915 Call visit_phi/visit_stmt from the class rather than via
2916 file scoped static variables.
2917 (ssa_propagation_engine::simulate_block): Moved into class.
2918 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2919 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2920 set file scoped statics for the visit_stmt/visit_phi callbacks.
2921 * tree-complex.c (complex_propagate): New class derived from
2922 ssa_propagation_engine.
2923 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2924 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2925 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2927 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2928 ssa_propagation_engine.
2929 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2930 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2931 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2932 * tree-ssa-copy.c (copy_prop): New class derived from
2933 ssa_propagation_engine.
2934 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2935 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2936 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2937 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2938 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2939 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2940 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2942 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2944 PR rtl-optimization/82778
2945 PR rtl-optimization/82597
2946 * compare-elim.c (struct comparison): Add in_a_setter field.
2947 (find_comparison_dom_walker::before_dom_children): Remove killed
2948 bitmap and df_simulate_find_defs call, instead walk the defs.
2949 Compute last_setter and initialize in_a_setter. Merge definitions
2950 with first initialization for a few variables.
2951 (try_validate_parallel): Use insn_invalid_p instead of
2952 recog_memoized. Return insn rather than just the pattern.
2953 (try_merge_compare): Fix up comment. Don't uselessly test if
2954 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2956 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2959 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2960 Alan Hayward <alan.hayward@arm.com>
2961 David Sherwood <david.sherwood@arm.com>
2963 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2964 aarch64_hard_regno_nregs to get the number of registers
2967 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2968 Alan Hayward <alan.hayward@arm.com>
2969 David Sherwood <david.sherwood@arm.com>
2971 * config/aarch64/constraints.md (Upl): Rename to...
2973 * config/aarch64/aarch64.md
2974 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2977 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2978 Alan Hayward <alan.hayward@arm.com>
2979 David Sherwood <david.sherwood@arm.com>
2981 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2982 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2985 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2986 Alan Hayward <alan.hayward@arm.com>
2987 David Sherwood <david.sherwood@arm.com>
2989 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2990 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2991 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2993 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2994 of a single element.
2995 * config/aarch64/iterators.md: Add a comment above the permute
2996 unspecs to say that they are generated directly by
2997 aarch64_expand_vec_perm_const.
2998 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
3000 2017-11-01 Nathan Sidwell <nathan@acm.org>
3002 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
3004 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
3006 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
3007 and medany code models, and describe what they do.
3009 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3011 Revert accidental duplicate:
3013 * combine.c (can_change_dest_mode): Reject changes in
3014 REGMODE_NATURAL_SIZE.
3016 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
3018 PR rtl-optimization/64682
3019 PR rtl-optimization/69567
3020 PR rtl-optimization/69737
3021 PR rtl-optimization/82683
3022 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
3023 register mentioned in the note, drop the note, unless it came from I3,
3024 in which case it should go to I3 again.
3026 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3028 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
3029 and return false if not.
3030 (clear_bytes_written_by, live_bytes_read): Update accordingly.
3032 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3034 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
3035 range is known to be empty.
3037 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3038 Alan Hayward <alan.hayward@arm.com>
3039 David Sherwood <david.sherwood@arm.com>
3041 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
3042 and CONST_VECTOR_NUNITS instead of computing the number of units from
3043 the byte sizes of the vector and element.
3044 (simplify_binary_operation_1): Likewise.
3045 (simplify_const_binary_operation): Likewise.
3046 (simplify_ternary_operation): Likewise.
3048 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3049 Alan Hayward <alan.hayward@arm.com>
3050 David Sherwood <david.sherwood@arm.com>
3052 * var-tracking.c (INT_MEM_OFFSET): Replace with...
3053 (int_mem_offset): ...this new function.
3054 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
3055 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
3056 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
3059 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3060 Alan Hayward <alan.hayward@arm.com>
3061 David Sherwood <david.sherwood@arm.com>
3063 * lower-subreg.c (interesting_mode_p): New function.
3064 (compute_costs, find_decomposable_subregs, decompose_register)
3065 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
3066 (resolve_clobber, dump_choices): Use it.
3068 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3069 Alan Hayward <alan.hayward@arm.com>
3070 David Sherwood <david.sherwood@arm.com>
3072 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
3074 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3075 Alan Hayward <alan.hayward@arm.com>
3076 David Sherwood <david.sherwood@arm.com>
3078 * alias.c (find_base_value, find_base_term): Only process integer
3079 truncations. Check the precision rather than the size.
3081 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3082 Alan Hayward <alan.hayward@arm.com>
3083 David Sherwood <david.sherwood@arm.com>
3085 * machmode.h (is_narrower_int_mode): New function
3086 * optabs.c (expand_float, expand_fix): Use it.
3087 * dwarf2out.c (rotate_loc_descriptor): Likewise.
3089 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3090 Alan Hayward <alan.hayward@arm.com>
3091 David Sherwood <david.sherwood@arm.com>
3093 * rtl.h (narrower_subreg_mode): New function.
3094 * ira-color.c (update_costs_from_allocno): Use it.
3096 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3097 Alan Hayward <alan.hayward@arm.com>
3098 David Sherwood <david.sherwood@arm.com>
3100 * optabs-query.h (convert_optab_p): New function, split out from...
3101 (convert_optab_handler): ...here.
3102 (widening_optab_handler): Delete.
3103 (find_widening_optab_handler): Remove permit_non_widening parameter.
3104 (find_widening_optab_handler_and_mode): Likewise. Provide an
3105 override that operates on mode class wrappers.
3106 * optabs-query.c (widening_optab_handler): Delete.
3107 (find_widening_optab_handler_and_mode): Remove permit_non_widening
3108 parameter. Assert that the two modes are the same class and that
3109 the "from" mode is narrower than the "to" mode. Use
3110 convert_optab_handler instead of widening_optab_handler.
3111 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
3112 instead of widening_optab_handler.
3113 * expr.c (expand_expr_real_2): Update calls to
3114 find_widening_optab_handler.
3115 * optabs.c (expand_widen_pattern_expr): Likewise.
3116 (expand_binop_directly): Take the insn_code as a parameter.
3117 (expand_binop): Only call find_widening_optab_handler for
3118 conversion optabs; use optab_handler otherwise. Update calls
3119 to find_widening_optab_handler and expand_binop_directly.
3120 Use convert_optab_handler instead of widening_optab_handler.
3121 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
3122 find_widening_optab_handler and use scalar_mode rather than
3124 (convert_plusminus_to_widen): Likewise.
3126 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3127 Alan Hayward <alan.hayward@arm.com>
3128 David Sherwood <david.sherwood@arm.com>
3130 * machmode.h (fixed_size_mode): New class.
3131 * rtl.h (get_pool_mode): Return fixed_size_mode.
3132 * gengtype.c (main): Add fixed_size_mode.
3133 * target.def (get_raw_result_mode): Return a fixed_size_mode.
3134 (get_raw_arg_mode): Likewise.
3135 * doc/tm.texi: Regenerate.
3136 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
3137 * targhooks.c (default_get_reg_raw_mode): Likewise.
3138 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
3139 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
3140 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
3141 (msp430_get_raw_result_mode): Likewise.
3142 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
3143 * dbxout.c (dbxout_parms): Require fixed-size modes.
3144 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
3145 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
3146 * omp-low.c (lower_oacc_reductions): Likewise.
3147 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
3148 (simplify_subreg): Update accordingly.
3149 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
3150 (force_const_mem): Update accordingly. Return NULL_RTX for modes
3151 that aren't fixed-size.
3152 (get_pool_mode): Return a fixed_size_mode.
3153 (output_constant_pool_2): Take a fixed_size_mode.
3155 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3156 Alan Hayward <alan.hayward@arm.com>
3157 David Sherwood <david.sherwood@arm.com>
3159 * doc/rtl.texi (vec_series): Document.
3160 (const): Say that the operand can be a vec_series.
3161 * rtl.def (VEC_SERIES): New rtx code.
3162 * rtl.h (const_vec_series_p_1): Declare.
3163 (const_vec_series_p): New function.
3164 * emit-rtl.h (gen_const_vec_series): Declare.
3165 (gen_vec_series): Likewise.
3166 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
3167 (gen_vec_series): Likewise.
3168 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
3169 * simplify-rtx.c (simplify_unary_operation): Handle negations
3171 (simplify_binary_operation_series): New function.
3172 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
3173 (test_vector_ops_series): New function.
3174 (test_vector_ops): Call it.
3175 * config/powerpcspe/altivec.md (altivec_lvsl): Use
3176 gen_const_vec_series.
3177 (altivec_lvsr): Likewise.
3178 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
3180 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3181 Alan Hayward <alan.hayward@arm.com>
3182 David Sherwood <david.sherwood@arm.com>
3184 * doc/rtl.texi (const): Update description of address constants.
3185 Say that vector constants are allowed too.
3186 * common.md (E, F): Use CONSTANT_P instead of checking for
3188 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
3189 checking for CONST_VECTOR.
3190 * expmed.c (make_tree): Use build_vector_from_val for a CONST
3192 * expr.c (expand_expr_real_2): Check for vector modes instead
3193 of checking for CONST_VECTOR.
3194 * rtl.h (const_vec_p): New function.
3195 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
3196 (unwrap_const_vec_duplicate): Handle them here too.
3198 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3199 David Malcolm <dmalcolm@redhat.com>
3200 Alan Hayward <alan.hayward@arm.com>
3201 David Sherwood <david.sherwood@arm.com>
3203 * rtl.h (vec_duplicate_p): New function.
3204 * selftest-rtl.c (assert_rtx_eq_at): New function.
3205 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
3206 (assert_rtx_eq_at): Declare.
3207 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
3208 * selftest-run-tests.c (selftest::run_tests): Call it.
3209 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
3210 (simplify_unary_operation_1): Recursively handle vector duplicates.
3211 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
3213 (simplify_subreg): Handle subregs of vector duplicates.
3214 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
3215 (selftest::simplify_rtx_c_tests): New functions.
3217 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3218 Alan Hayward <alan.hayward@arm.com>
3219 David Sherwood <david.sherwood@arm.com>
3221 * emit-rtl.h (gen_const_vec_duplicate): Declare.
3222 (gen_vec_duplicate): Likewise.
3223 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
3225 (gen_const_vector): ...here.
3226 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
3227 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
3228 whose elements are all equal.
3229 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
3230 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
3231 (simplify_relational_operation): Likewise.
3232 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
3234 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
3235 (aarch64_expand_vector_init): Likewise.
3236 * config/arm/arm.c (neon_vdup_constant): Likewise.
3237 (neon_expand_vector_init): Likewise.
3238 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
3239 (arm_block_set_unaligned_vect): Likewise.
3240 (arm_block_set_aligned_vect): Likewise.
3241 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
3242 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
3243 (expand_vec_perm_even_odd_pack): Likewise.
3244 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
3245 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
3246 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
3247 gen_const_vec_duplicate.
3248 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
3249 * config/mips/mips.c (mips_gen_const_int_vector): Use
3250 gen_const_vec_duplicate.
3251 (mips_expand_vector_init): Use CONST0_RTX.
3252 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
3253 (define_split): Use gen_const_vec_duplicate.
3254 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
3255 (define_split): Use gen_const_vec_duplicate.
3256 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
3257 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
3258 * config/spu/spu.c (spu_const): Likewise.
3260 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
3261 Alan Hayward <alan.hayward@arm.com>
3262 David Sherwood <david.sherwood@arm.com>
3264 * combine.c (can_change_dest_mode): Reject changes in
3265 REGMODE_NATURAL_SIZE.
3267 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
3269 * configure.ac (--enable-libssp): New.
3270 (gcc_cv_libc_provides_ssp): Check for explicit setting before
3271 trying to determine target-specific default. Adjust indentation.
3272 * configure: Regenerated.
3273 * doc/install.texi (Configuration): Expand --disable-libssp
3276 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
3278 config/i386/i386.c (ix86_expand_epilogue): Correct stack
3281 2017-10-31 Martin Jambor <mjambor@suse.cz>
3284 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
3286 2017-10-31 David Malcolm <dmalcolm@redhat.com>
3288 * auto-profile.c (autofdo_source_profile::read): Use
3289 UNKNOWN_LOCATION rather than 0.
3290 * diagnostic-core.h (warning_at_rich_loc): Rename to...
3291 (warning_at): ...this overload.
3292 (warning_at_rich_loc_n): Rename to...
3293 (warning_n): ...this overload.
3294 (error_at_rich_loc): Rename to...
3295 (error_at): ...this overload.
3296 (pedwarn_at_rich_loc): Rename to...
3297 (pedwarn): ...this overload.
3298 (permerror_at_rich_loc): Rename to...
3299 (permerror): ...this overload.
3300 (inform_at_rich_loc): Rename to...
3301 (inform): ...this overload.
3302 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
3303 (diagnostic_n_impl_richloc): Rename to...
3304 (diagnostic_n_impl): ...this rich_location *-based decl.
3305 (inform_at_rich_loc): Rename to...
3306 (inform): ...this, and add an assertion.
3307 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
3308 (warning_at_rich_loc): Rename to...
3309 (warning_at): ...this, and add an assertion.
3310 (warning_at_rich_loc_n): Rename to...
3311 (warning_n): ...this, and add an assertion.
3312 (warning_n): Update location_t-based implementation for removal of
3313 location_t-based diagnostic_n_impl.
3314 (pedwarn_at_rich_loc): Rename to...
3315 (pedwarn): ...this, and add an assertion.
3316 (permerror_at_rich_loc): Rename to...
3317 (permerror): ...this, and add an assertion.
3318 (error_n): Update for removal of location_t-based diagnostic_n_impl.
3319 (error_at_rich_loc): Rename to...
3320 (error_at): ...this, and add an assertion.
3321 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
3322 (driver::do_spec_on_infiles): Likewise.
3323 * substring-locations.c (format_warning_va): Update for renaming
3324 of inform_at_rich_loc.
3326 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3328 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
3329 _Float<N>X built-in functions so that the variant without the
3330 "__builtin_" prefix is only enabled for the GNU C and Objective C
3331 languages when they are in non-strict ANSI/ISO mode.
3332 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
3333 * target.def (floatn_builtin_p): Add a target hook to control
3334 whether _Float<N> and _Float<N>X built-in functions without the
3335 "__builtin_" prefix are enabled, and return true for C and
3336 Objective C in the default hook. Include langhooks.h in
3338 * targhooks.h (default_floatn_builtin_p): Likewise.
3339 * targhooks.c (default_floatn_builtin_p): Likewise.
3340 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
3341 floatn_builtin_p target hook.
3342 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
3344 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
3345 Eric Botcazou <ebotcazou@adacore.com>
3347 PR rtl-optimization/81803
3348 * lra-constraints.c (curr_insn_transform): Also reload the whole
3349 register for a strict subreg no wider than a word if this is for
3350 a WORD_REGISTER_OPERATIONS target.
3352 2017-10-31 Jason Merrill <jason@redhat.com>
3354 * gdbinit.in: Skip over inlines from timevar.h.
3356 2017-10-31 Martin Liska <mliska@suse.cz>
3358 * doc/gcov.texi: Document new option.
3359 * gcov.c (print_usage): Likewise print it.
3360 (process_args): Support the argument.
3361 (format_count): New function.
3362 (format_gcov): Use the function.
3364 2017-10-31 Martin Liska <mliska@suse.cz>
3366 * gcov.c (struct name_map): do not use typedef.
3367 Define operator== and operator<.
3368 (name_search): Remove.
3369 (name_sort): Remove.
3370 (main): Do not allocate names.
3371 (process_file): Add vertical space.
3372 (generate_results): Use std::find.
3373 (release_structures): Do not release memory.
3374 (find_source): Use std::find.
3376 2017-10-31 Martin Liska <mliska@suse.cz>
3378 * gcov.c (struct line_info): Remove it's typedef.
3379 (line_info::line_info): Add proper ctor.
3380 (line_info::has_block): Do not use a typedef.
3381 (struct source_info): Do not use typedef.
3382 (circuit): Likewise.
3383 (get_cycles_count): Likewise.
3384 (output_intermediate_file): Iterate via vector iterator.
3385 (add_line_counts): Use std::vector methods.
3386 (accumulate_line_counts): Likewise.
3387 (output_lines): Likewise.
3389 2017-10-31 Martin Liska <mliska@suse.cz>
3391 * gcov.c (struct source_info): Remove typedef.
3392 (source_info::source_info): Add proper ctor.
3393 (accumulate_line_counts): Use struct, not it's typedef.
3394 (output_gcov_file): Likewise.
3395 (output_lines): Likewise.
3396 (main): Do not allocate an array.
3397 (output_intermediate_file): Use size of vector container.
3398 (process_file): Resize the vector.
3399 (generate_results): Do not preallocate, use newly added vector
3401 (release_structures): Do not release sources.
3402 (find_source): Use vector methods.
3403 (add_line_counts): Do not use typedef.
3405 2017-10-31 Martin Liska <mliska@suse.cz>
3407 * doc/gcov.texi: Document that.
3408 * gcov.c (add_line_counts): Mark lines with a non-executed
3410 (output_line_beginning): Handle such lines.
3411 (output_lines): Pass new argument.
3412 (output_intermediate_file): Print it in intermediate format.
3414 2017-10-31 Martin Liska <mliska@suse.cz>
3416 * color-macros.h: New file.
3417 * diagnostic-color.c: Factor out color related to macros to
3419 * doc/gcov.texi: Document -k option.
3420 * gcov.c (INCLUDE_STRING): Include string.h.
3421 (print_usage): Add -k option.
3422 (process_args): Parse it.
3423 (pad_count_string): New function.
3424 (output_line_beginning): Likewise.
3425 (DEFAULT_LINE_START): New macro.
3426 (output_lines): Support color output.
3428 2017-10-31 Martin Liska <mliska@suse.cz>
3430 PR gcov-profile/82633
3431 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3432 their interaction with GCOV infrastructure.
3433 * configure.ac: Add -fkeep-{inline,static}-functions to
3435 * configure: Regenerate.
3437 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3440 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3442 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3445 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3446 into a register if it does not fit into an immediate offset field.
3448 2017-10-31 Olivier Hainque <hainque@adacore.com>
3450 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3452 2017-10-31 Julia Koval <julia.koval@intel.com>
3454 * config.gcc: Add gfniintrin.h.
3455 * config/i386/gfniintrin.h: New.
3456 * config/i386/i386-builtin-types.def
3457 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3458 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3459 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3460 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3461 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3462 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3463 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3464 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3465 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3466 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3467 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3468 * config/i386/immintrin.h: Include gfniintrin.h.
3469 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3471 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3473 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3475 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3477 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3478 (arm_ashldi3_1bit): Remove pattern.
3479 (ashrdi3): Remove shift by 1 expansion.
3480 (arm_ashrdi3_1bit): Remove pattern.
3481 (lshrdi3): Remove shift by 1 expansion.
3482 (arm_lshrdi3_1bit): Remove pattern.
3483 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3484 cost of ashldi3 by 1.
3485 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3486 (<shift>di3_neon): Likewise.
3488 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3490 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3491 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3492 and (*aarch64_simd_mov<VQ:mode>).
3493 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3494 pattern alternative.
3495 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3496 attributes to match pattern alternative.
3498 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3500 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3501 * config/rs6000/emmintrin.h: New file.
3502 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3504 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3506 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3507 * (movdi_vfp_cortexa8): Remove pattern.
3509 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3511 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3513 (Specific, alpha*-dec-osf5.1): Remove.
3514 (Specific, mips-sgi-irix5): Remove.
3515 (Specific, mips-sgi-irix6): Remove.
3517 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3520 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3521 arguments to clear_bit_region_be.
3523 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3525 * gimplify.c: Include memmodel.h.
3527 2017-10-30 Martin Jambor <mjambor@suse.cz>
3529 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3530 condition whether loop should be executed at all.
3532 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3534 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3535 gimple folding of vec_madd() intrinsics.
3536 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3537 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3538 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3540 2017-10-30 Richard Biener <rguenther@suse.de>
3542 PR tree-optimization/82762
3544 2017-10-23 Richard Biener <rguenther@suse.de>
3546 PR tree-optimization/82129
3548 2017-08-01 Richard Biener <rguenther@suse.de>
3550 PR tree-optimization/81181
3551 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3552 (compute_antic): ... end of iteration here.
3554 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3556 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3557 -std=iso9899:2017 and -std=gnu17.
3558 * doc/standards.texi (C Language): Document C17 support.
3559 * doc/cpp.texi (Overview): Mention -std=c17.
3560 (Standard Predefined Macros): Document C11 and C17 values of
3561 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3562 * doc/extend.texi (Inline): Do not list individual options for
3563 standards newer than C99.
3564 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3566 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3569 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3571 * asan.c (asan_finish_file): Align asan globals array by shadow
3574 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3577 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3578 (struct store_immediate_info): Add bitregion_start and bitregion_end
3580 (store_immediate_info::store_immediate_info): Add brs and bre
3581 arguments and initialize bitregion_{start,end} from those.
3582 (struct merged_store_group): Add bitregion_start, bitregion_end,
3583 align_base and mask fields. Drop unnecessary struct keyword from
3584 struct store_immediate_info. Add do_merge method.
3585 (clear_bit_region_be): Use memset instead of loop storing zeros.
3586 (merged_store_group::do_merge): New method.
3587 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3588 stores as long as the surrounding bitregions have no gaps.
3589 (merged_store_group::merge_overlapping): Use do_merge.
3590 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3591 is byte aligned, rather than requiring that start and width are
3592 byte aligned. Drop unnecessary struct keyword from
3593 struct store_immediate_info. Allocate and populate also mask array.
3594 Make start of the arrays relative to bitregion_start rather than
3595 start and size them according to bitregion_{end,start} difference.
3596 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3597 struct store_immediate_info.
3598 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3599 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3600 struct keyword from struct store_immediate_info.
3601 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3602 between stores as long as the surrounding bitregions have no gaps.
3604 (struct split_store): Add orig non-static data member.
3605 (split_store::split_store): Initialize orig to false.
3606 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3607 if there is exactly a single original stmt. Change stmts argument
3608 to pointer from reference, if NULL, don't push anything to it. Add
3609 first argument, use it to optimize skipping over orig stmts that
3610 are known to be before bitpos already. Simplify.
3611 (split_group): Return unsigned int count how many stores are or
3612 would be needed rather than a bool. Add allow_unaligned argument.
3613 Change split_stores argument from reference to pointer, if NULL,
3614 only do a dry run computing how many stores would be produced.
3615 Rewritten algorithm to use both alignment and misalign if
3616 !allow_unaligned and handle bitfield stores with gaps.
3617 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3618 from bitregion_start instead of start. Compute allow_unaligned
3619 here, if true, do 2 split_group dry runs to compute which one
3620 produces fewer stores and prefer aligned if equal. Punt if
3621 new count is bigger or equal than original before emitting any
3622 statements, rather than during that. Remove no longer needed
3623 new_ssa_names tracking. Replace num_stmts with
3624 split_stores.length (). Use 32-bit stack allocated entries
3625 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3626 if possible. Handle bitfields with gaps.
3627 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3628 Compute bitregion_{start,end} for the stores and construct
3629 store_immediate_info with that. Formatting fixes.
3631 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3634 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3635 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3637 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3639 * gimplify.c: Include tm_p.h.
3641 * common.opt (gcoff): Re-add as ignored option.
3642 (gcoff1, gcoff2, gcoff3): Likewise.
3644 * Makefile.in (OBJS): Delete sdbout.o.
3645 (GTFILES): Delete $(srcdir)/sdbout.c.
3646 * debug.h: Delete sdb_debug_hooks.
3647 * final.c: Delete sdbout.h include.
3648 (final_scan_insn): Delete SDB_DEBUG check.
3649 (rest_of_clean_state): Likewise.
3650 * output.h: Delete sdb_begin_function_line.
3653 * toplev.c: Delete sdbout.h include.
3654 (process_options): Delete SDB_DEBUG check.
3655 * tree-core.h (tree_type_common): Delete pointer field of
3657 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3658 TYPE_SYMTAB_POINTER.
3659 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3660 (TYPE_SYMTAB_IS_POINTER): Delete.
3661 (TYPE_SYMTAB_IS_DIE): Renumber.
3662 * xcoffout.c: Refer to former sdbout.c file.
3663 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3665 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3666 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3667 (Debugging Options): Delete -gcoff.
3668 (-gstabs): Delete SDB reference.
3670 (-gcoff@var{level}): Delete.
3671 * doc/passes.texi (Debugging information output): Delete SDB and
3672 sdbout.c references.
3673 * doc/tm.texi: Regenerate.
3674 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3675 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3677 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3678 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3679 (SDB_DEBUGGING_INFO): Delete.
3680 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3681 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3682 * target.def (output_source_filename): Delete COFF reference.
3684 * common.opt (gcoff): Delete.
3685 (gxcoff+): Update Negative chain.
3686 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3688 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3689 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3690 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3692 (expand_function_start): Change sdb reference to past tense.
3693 (expand_function_end): Change sdb reference to past tense.
3694 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3695 * opts.c (debug_type_names): Delete coff entry.
3696 (common_handle_option): Delete OPT_gcoff case.
3697 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3699 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3700 * config/cris/cris.h: Delete SDB reference in comment.
3701 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3702 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3703 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3704 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3706 (ix86_expand_prologue): Likewise.
3707 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3708 * config/ia64/ia64.h: Likewise.
3709 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3710 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3712 * config/mmix/mmix.h: Likewise.
3713 * config/nds32/nds32.c: Likewise.
3714 * config/stormy/storym16.h: Likewise.
3715 * config/visium/visium.h: Likewise.
3716 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3718 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3720 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3721 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3722 FRAME_POINTER_REGNUM point at high end of local var area.
3724 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3726 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3727 Move comment around. Do not reset best_edge for a copiable
3728 destination if the copy would cause a partition change.
3729 (better_edge_p): Remove redundant check.
3731 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3733 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3735 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3737 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3738 for math functions that have _Float<N> and _Float<N>X variants.
3739 (mathfn_built_in_2): Add support for math functions that have
3740 _Float<N> and _Float<N>X variants.
3741 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3742 (expand_builtin_mathfn_ternary): Add support for fma with
3743 _Float<N> and _Float<N>X variants.
3744 (expand_builtin): Likewise.
3745 (fold_builtin_3): Likewise.
3746 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3747 create math function _Float<N> and _Float<N>X variants as external
3749 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3750 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3751 the __builtin_ prefix and if not strict ansi, without the prefix.
3752 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3753 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3754 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3755 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3756 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3757 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3758 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3759 function signatures for fma _Float<N> and _Float<N>X variants.
3760 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3761 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3762 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3763 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3764 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3765 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3766 * gencfn-macros.c (print_case_cfn): Add support for math functions
3767 that have _Float<N> and _Float<N>X variants.
3768 (print_define_operator_list): Likewise.
3769 (fltfn_suffixes): Likewise.
3771 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3772 for math functions that have _Float<N> and _Float<N>X variants.
3773 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3774 and _Float<N>X variants.
3775 (COPYSIGN): Likewise.
3778 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3779 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3781 (integer_valued_read_call_p): Likewise.
3782 * fold-const-call.c (fold_const_call_ss): Likewise.
3783 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3784 _Float<N> and _Float<N>X variants.
3785 (fold_const_call_ssss): Add support for fma _Float<N> and
3786 _Float<N>X variants.
3787 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3788 support for copysign and fma _Float<N> and _Float<N>X variants.
3789 (backprop::process_builtin_call_use): Likewise.
3790 * tree-call-cdce.c (can_test_argument_range); Add support for
3791 sqrt _Float<N> and _Float<N>X variants.
3792 (edom_only_function): Likewise.
3793 (get_no_error_domain): Likewise.
3794 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3795 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3796 copysign _Float<N> and _Float<N>X variants.
3797 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3798 handled by machine independent code.
3799 (FMAF128): Likewise.
3800 * doc/cpp.texi (Common Predefined Macros): Document defining
3801 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3802 fma _Float<N> and _Float<N>X variants.
3804 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3807 * config/i386/i386-modes.def (CCFPU): Remove definition.
3808 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3809 (ix86_cc_modes_compatible): Ditto.
3810 (ix86_expand_carry_flag_compare): Ditto.
3811 (ix86_expand_int_movcc): Ditto.
3812 (ix86_expand_int_addcc): Ditto.
3813 (ix86_reverse_condition): Ditto.
3814 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3815 Return true/false for unordered/ordered fp comparisons.
3816 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3817 (ix86_prepare_fp_compare_args): Update for rename.
3818 (ix86_expand_fp_compare): Update for rename. Generate unordered
3819 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3820 (ix86_expand_sse_compare_and_jump): Ditto.
3821 * config/i386/predicates.md (fcmov_comparison_operator):
3822 Remove CCFPU mode handling.
3823 (ix86_comparison_operator): Ditto.
3824 (ix86_carry_flag_operator): Ditto.
3825 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3826 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3827 (*cmpu<mode>_cc_i387): Ditto.
3828 (FPCMP): Remove mode iterator.
3829 (unord): Remove mode attribute.
3830 (unord_subst): New define_subst transformation
3831 (unord): New define_subst attribute.
3833 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3834 (*cmpi<unord>xf_i387): Ditto.
3835 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3836 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3837 using unord_subst transformation.
3838 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3839 (round_saeonly): Also handle CCFP mode.
3840 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3841 Remove UNSPEC_SAHF unspec handling.
3843 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3845 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3847 2017-10-27 Jeff Law <law@redhat.com>
3849 * gimple-ssa-sprintf.c: Include domwalk.h.
3850 (class sprintf_dom_walker): New class, derived from dom_walker.
3851 (sprintf_dom_walker::before_dom_children): New function.
3852 (struct call_info): Moved into sprintf_dom_walker class
3853 (compute_formath_length, handle_gimple_call): Likewise.
3854 (sprintf_length::execute): Call the dominator walker rather
3855 than walking the statements.
3857 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3858 gimple statement locations.
3859 (check_array_bounds): Corresponding changes. Get the statement's
3860 location directly from wi->stmt.
3862 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3865 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3867 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3869 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3870 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3872 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3875 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3876 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3877 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3878 maybe_get_pool_constant.
3879 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3882 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3884 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3885 2.26 caveat. Update gas and gld versions.
3886 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3889 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3891 * cgraph.h (set_malloc_flag): Declare.
3892 * cgraph.c (set_malloc_flag_1): New function.
3893 (set_malloc_flag): Likewise.
3894 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3895 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3897 (read_ipa_call_summary): Add support for reading is_return_callee.
3898 (write_ipa_call_summary): Stream is_return_callee.
3899 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3900 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3901 ipa-prop.h, ipa-fnsummary.h.
3902 (pure_const_names): Change to static.
3903 (malloc_state_e): Define.
3904 (malloc_state_names): Define.
3905 (funct_state_d): Add field malloc_state.
3906 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3907 (check_retval_uses): New function.
3908 (malloc_candidate_p): Likewise.
3909 (analyze_function): Add support for malloc attribute.
3910 (pure_const_write_summary): Stream malloc_state.
3911 (pure_const_read_summary): Add support for reading malloc_state.
3912 (dump_malloc_lattice): New function.
3913 (propagate_malloc): New function.
3914 (warn_function_malloc): New function.
3915 (ipa_pure_const::execute): Call propagate_malloc and
3916 ipa_free_fn_summary.
3917 (pass_local_pure_const::execute): Add support for malloc attribute.
3918 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3919 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3921 2017-10-27 Martin Liska <mliska@suse.cz>
3923 PR gcov-profile/82457
3924 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3925 for fork-like functions to be properly instrumented.
3927 2017-10-27 Richard Biener <rguenther@suse.de>
3930 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3931 info when we redirected EH.
3933 2017-10-26 Michael Collison <michael.collison@arm.com>
3935 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3937 (<optab>_trunchf<GPI:mode>2: New pattern.
3938 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3939 * config/aarch64/iterators.md (wv): New mode attribute.
3940 (vf, VF): New mode attributes.
3941 (vgp, VGP): New mode attributes.
3942 (s): Update attribute with SImode and DImode prefixes.
3944 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3946 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3947 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3948 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3949 (nios2_option_overide): Initialize it. Don't allow R0-relative
3950 addressing with PIC.
3951 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3952 (nios2_symbolic_constant_p): Likewise.
3953 (nios2_legitimate_address_p): Likewise.
3954 (nios2_r0rel_section_name_p): New.
3955 (nios2_symbol_ref_in_r0rel_data_p): New.
3956 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3957 (r0rel_constant_p): New.
3958 (nios2_print_operand_address): Handle r0rel_constant_p.
3959 (nios2_cdx_narrow_form_p): Likewise.
3960 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3961 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3962 (Nios II Options): Document -mr0rel-sec.
3964 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3966 * config/nios2/nios2.c: Include xregex.h.
3967 (nios2_gprel_sec_regex): New.
3968 (nios2_option_overide): Initialize it. Don't allow GP-relative
3969 addressing with PIC.
3970 (nios2_small_section_name_p): Check for regex match.
3971 * config/nios2/nios2.opt (mgprel-sec=): New option.
3972 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3973 (Nios II Options): Document -mgprel-sec.
3975 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3977 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3979 2017-10-26 Tom de Vries <tom@codesourcery.com>
3981 PR tree-optimization/82707
3982 * gimple.c (gimple_copy): Fix unsharing of
3983 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3985 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3987 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3988 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3989 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3990 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3991 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3992 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3993 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3994 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3995 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3996 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3997 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3998 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3999 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
4000 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
4001 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
4002 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
4003 _mm512_mask_cmpunord_ps_mask): New intrinsics.
4005 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4007 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
4009 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
4010 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
4011 warning to rs6000.c. Remove the Undocumented flag, since it has
4013 (-mabi=ibmlongdouble): Likewise.
4014 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
4015 already set, set the default format for long double.
4016 (rs6000_debug_reg_global): Print whether long double is IBM or
4018 (rs6000_option_override_internal): Rework setting long double
4019 format. Only warn if the user is changing the long double default
4020 and they did not use -Wno-psabi.
4021 * doc/invoke.texi (PowerPC options): Update the documentation for
4022 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
4024 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4025 Alan Hayward <alan.hayward@arm.com>
4026 David Sherwood <david.sherwood@arm.com>
4028 * rtl.h (wider_subreg_mode): New function.
4029 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
4030 rather than an unsigned int *.
4031 * ira-color.c (regno_max_ref_width): Replace with...
4032 (regno_max_ref_mode): ...this new variable.
4033 (coalesced_pseudo_reg_slot_compare): Update accordingly.
4034 Use wider_subreg_mode.
4035 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
4036 rather than an unsigned int *.
4037 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
4038 (process_alt_operands): Likewise.
4039 (invariant_p): Likewise.
4040 * lra-spills.c (assign_mem_slot): Likewise.
4041 (add_pseudo_to_slot): Likewise.
4042 * lra.c (collect_non_operand_hard_regs): Likewise.
4043 (add_regs_to_insn_regno_info): Likewise.
4044 * reload1.c (regno_max_ref_width): Replace with...
4045 (regno_max_ref_mode): ...this new variable.
4046 (reload): Update accordingly. Update call to
4047 ira_sort_regnos_for_alter_reg.
4048 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
4049 (init_eliminable_invariants): Update to use regno_max_ref_mode.
4050 (scan_paradoxical_subregs): Likewise.
4052 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4054 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
4055 (aarch64_frame): Add emit_frame_chain boolean.
4056 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4057 Move eh_return case to aarch64_layout_frame.
4058 (aarch64_layout_frame): Initialize emit_frame_chain.
4059 (aarch64_expand_prologue): Use emit_frame_chain.
4061 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4063 * config/aarch64/aarch64.c (aarch64_layout_frame):
4064 Ensure LR is always stored at the bottom of the callee-saves.
4065 Remove rarely used frame layout which saves callee-saves at top of
4066 frame, so the store of LR can be used as a valid probe in all cases.
4068 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
4070 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
4071 Improve unaligned TImode/TFmode base/offset split.
4073 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4074 Alan Hayward <alan.hayward@arm.com>
4075 David Sherwood <david.sherwood@arm.com>
4077 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
4078 * combine.c (find_single_use_1): Likewise.
4079 (expand_field_assignment): Likewise.
4080 (move_deaths): Likewise.
4081 * lra-constraints.c (simplify_operand_subreg): Likewise.
4082 (curr_insn_transform): Likewise.
4083 * lra.c (collect_non_operand_hard_regs): Likewise.
4084 (add_regs_to_insn_regno_info): Likewise.
4085 * rtlanal.c (reg_referenced_p): Likewise.
4086 (covers_regno_no_parallel_p): Likewise.
4088 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4090 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
4091 Don't print any bits outside the precision of the value.
4092 * wide-int.cc (test_printing): Add some new tests.
4094 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4096 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
4097 supports -xbrace_comment option.
4098 * configure: Regenerate.
4099 * config.in: Regenerate.
4100 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
4101 (ASM_CPU_SPEC): Use it.
4103 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
4105 * target.def (static_rtx_alignment): New hook.
4106 * targhooks.h (default_static_rtx_alignment): Declare.
4107 * targhooks.c (default_static_rtx_alignment): New function.
4108 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
4109 * doc/tm.texi: Regenerate.
4110 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
4111 instead of targetm.constant_alignment. Remove call to
4113 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4114 (cris_preferred_mininum_alignment): New function, split out from...
4115 (cris_constant_alignment): ...here.
4116 (cris_static_rtx_alignment): New function.
4117 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
4119 (ix86_constant_alignment): ...here.
4120 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4121 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4122 (mmix_static_rtx_alignment): New function.
4123 * config/spu/spu.c (spu_static_rtx_alignment): New function.
4124 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
4126 2017-10-26 Tamar Christina <tamar.christina@arm.com>
4129 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
4130 Add flag_trapping_math and flag_fp_int_builtin_inexact.
4132 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
4134 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
4135 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
4136 mark as a sign-extending load.
4137 (local_pic_load_u): Define.
4139 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
4142 * fold-const.c (operand_equal_for_comparison_p): Also return true
4143 if ARG0 is a simple variant of ARG1 with narrower precision.
4144 (fold_ternary_loc): Always pass unstripped operands to the predicate.
4146 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
4148 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
4150 * i386.h (processor_costs): Add gather_static, gather_per_elt,
4151 scatter_static, scatter_per_elt.
4152 * x86-tune-costs.h: Add new cost entries.
4154 2017-10-25 Richard Biener <rguenther@suse.de>
4156 * tree-ssa-sccvn.h (vn_eliminate): Declare.
4157 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
4158 class pass_fre): Move to ...
4159 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
4160 class pass_fre): ... here and adjust for statistics.
4162 2017-10-25 Jakub Jelinek <jakub@redhat.com>
4165 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
4166 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
4167 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
4168 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
4171 2017-10-25 Richard Biener <rguenther@suse.de>
4173 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
4174 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
4175 eliminate_push_avail, eliminate_insert): Move inside...
4176 (class eliminate_dom_walker): ... this class in preparation
4178 (fini_eliminate): Remove by merging with ...
4179 (eliminate): ... this function. Adjust for class changes.
4180 (pass_pre::execute): Remove fini_eliminate call.
4181 (pass_fre::execute): Likewise.
4183 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4186 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
4187 (VPERMI2, VPERMI2I): New mode iterators.
4188 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
4189 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
4191 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
4192 mode iterator. Remove 3 old define_insn patterns.
4193 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
4194 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
4195 VPERMI2 mode iterator, remove the other two expanders.
4196 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
4197 to use VPERMI2 mode iterator, add another alternative for vpermi2*
4198 instructions, remove the other two patterns.
4199 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
4200 mode iterator, remove the other two patterns.
4201 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
4202 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
4203 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
4204 and adjust argument order accordingly.
4205 (ix86_expand_vec_perm): Adjust caller.
4206 (expand_vec_perm_1): Likewise.
4207 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
4208 (expand_vec_perm_vpermt2_vpshub2): ... this.
4209 (ix86_expand_vec_perm_const_1): Adjust caller.
4210 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
4213 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
4214 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
4215 (vec_shl_<mode>): Remove unused expander.
4216 (avx512bw_<shift_insn><mode>3): New define_insn.
4217 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
4218 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
4220 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
4223 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
4226 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4228 PR rtl-optimization/82396
4229 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
4230 (autopref_multipass_init): Simplify initialization.
4231 (autopref_rank_data): Simplify sort order.
4232 * gcc/sched-int.h (autopref_multipass_data_): Remove
4233 multi_mem_insn_p, min_offset and max_offset.
4235 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
4238 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
4239 Check special value of flag_omit_frame_pointer.
4240 (aarch64_can_eliminate): Likewise.
4241 (aarch64_override_options_after_change_1): Simplify handling of
4242 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
4244 2017-10-24 Richard Biener <rguenther@suse.de>
4246 PR tree-optimization/82697
4247 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
4248 zero for conditional load and unconditional store.
4250 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4252 * doc/install.texi: Document bootstrap-cet.
4254 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
4257 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
4258 ENDBR instruction at function entrance if function is only
4261 2017-10-24 Jakub Jelinek <jakub@redhat.com>
4264 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
4265 patterns to better describe from which operation the CF is computed.
4266 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
4267 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
4268 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
4269 is 0, use _0 suffixed expanders instead of emitting a comparison
4272 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
4274 * config/i386/i386.md(*movsf_internal, *movdf_internal):
4275 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
4277 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
4280 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
4281 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
4282 * loop-iv.c (iv_get_reaching_def): Likewise.
4283 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
4284 variable is promoted and the partition contains undefined values.
4286 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4288 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
4290 (nios2_address_cost): Define.
4291 (nios2_legitimize_address): Recognize (exp + constant) directly.
4292 (TARGET_ADDRESS_COST): Define.
4294 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4296 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
4297 (nios2_symbolic_memory_operand_p): Declare.
4298 (nios2_split_large_constant): Declare.
4299 (nios2_split_symbolic_memory_operand): Declare.
4300 * config/nios2/nios2.c: Adjust includes.
4301 (nios2_symbolic_constant_allowed): New.
4302 (nios2_symbolic_constant_p): New.
4303 (nios2_plus_symbolic_constant_p): New.
4304 (nios2_valid_addr_expr_p): Recognize addresses involving
4306 (nios2_legitimate_address_p): Likewise, also LO_SUM.
4307 (nios2_symbolic_memory_operand_p): New.
4308 (nios2_large_constant_p): New.
4309 (nios2_split_large_constant): New.
4310 (nios2_split_plus_large_constant): New.
4311 (nios2_split_symbolic_memory_operand): New.
4312 (nios2_legitimize_address): Code refactoring. Handle addresses
4313 involving symbolic constants.
4314 (nios2_emit_move_sequence): Likewise.
4315 (nios2_print_operand): Improve error output.
4316 (nios2_print_operand_address): Handle LO_SUM.
4317 (nios2_cdx_narrow_form_p): Likewise.
4318 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
4319 operands involving symbolic constants.
4320 (movhi_internal, movsi_internal): Likewise.
4321 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
4322 (extendhisi2, extendqi<mode>2): Likewise.
4324 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4326 * tree-pass.h (PROP_rtl_split_insns): Define.
4327 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
4329 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
4331 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
4333 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4336 * target.def (const_not_ok_for_debug_p): Default to
4337 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
4338 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
4339 * targhooks.c (default_const_not_ok_for_debug_p): New function.
4340 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
4341 which targetm.const_not_ok_for_debug_p returned true.
4342 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
4344 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
4346 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
4347 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
4348 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
4350 (ix86_const_not_ok_for_debug_p): New function.
4351 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
4352 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
4354 2017-10-23 David Malcolm <dmalcolm@redhat.com>
4357 * system.h: Conditionally include "unique-ptr.h" if
4358 INCLUDE_UNIQUE_PTR is defined.
4359 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
4360 of defining INCLUDE_UNIQUE_PTR before including "system.h".
4362 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
4364 * config/rl78/rl78.md: New define_expand "subdi3".
4366 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
4369 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
4370 DF_REF_INSN if DF_REF_INSN_INFO is false.
4372 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
4374 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
4375 xmm_move instead of sse_move.
4376 (sse_store_index): New function.
4377 (ix86_register_move_cost): Be more sensible about mismatch stall;
4378 model AVX moves correctly; make difference between sse->integer and
4380 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
4381 moves; make difference between SSE and AVX.
4382 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
4383 and zmm_move. Increase size of sse load and store tables;
4384 add unaligned load and store tables; add ssemmx_to_integer.
4385 * x86-tune-costs.h: Update all entries according to real
4386 move latencies from Agner Fog's manual and chip documentation.
4388 2017-10-23 Jakub Jelinek <jakub@redhat.com>
4391 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
4392 * config/i386/constraints.md (Wf): New constraint.
4393 * config/i386/i386.md (UNSPEC_SBB): New unspec.
4394 (cmp<dwi>_doubleword): Removed.
4395 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
4396 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
4397 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
4398 expand with cmp<dwi>_doubleword. For LTU and GEU use
4399 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
4401 * common.opt (gcolumn-info): Enable by default.
4402 * doc/invoke.texi (gcolumn-info): Document new default.
4404 2017-10-23 Richard Biener <rguenther@suse.de>
4406 PR tree-optimization/82672
4407 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
4408 Fold the stmt if we propagated into it.
4410 2017-10-23 Richard Biener <rguenther@suse.de>
4412 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
4413 (bitmap_remove_expr_from_set): ... this. All callers call this
4414 for non-constant values.
4415 (bitmap_set_subtract): Rename to...
4416 (bitmap_set_subtract_expressions): ... this. Adjust and
4418 (bitmap_set_contains_value): Remove superfluous check.
4419 (bitmap_set_replace_value): Inline into single caller ...
4420 (bitmap_value_replace_in_set): ... here and simplify.
4421 (dependent_clean): Merge into ...
4422 (clean): ... this using an overload. Adjust.
4423 (prune_clobbered_mems): Adjust.
4424 (compute_antic_aux): Likewise.
4425 (compute_partial_antic_aux): Likewise.
4427 2017-10-23 Richard Biener <rguenther@suse.de>
4429 PR tree-optimization/82129
4431 2017-08-01 Richard Biener <rguenther@suse.de>
4433 PR tree-optimization/81181
4434 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4435 (compute_antic): ... end of iteration here.
4437 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4439 * target.def (starting_frame_offset): New hook.
4440 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4441 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4442 * doc/tm.texi.in: Regenerate.
4443 * hooks.h (hook_hwi_void_0): Declare.
4444 * hooks.c (hook_hwi_void_0): New function.
4445 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4446 STARTING_FRAME_OFFSET.
4447 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4448 * reload1.c (reload): Likewise.
4449 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4450 instead of STARTING_FRAME_OFFSET.
4451 * function.c (try_fit_stack_local): Likewise.
4452 (assign_stack_local_1): Likewise
4453 (instantiate_virtual_regs): Likewise.
4454 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4455 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4456 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4457 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4458 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4459 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4460 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4461 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4462 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4463 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4464 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4465 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4466 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4467 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4468 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4469 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4470 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4471 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4472 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4473 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4474 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4475 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4476 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4477 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4478 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4479 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4480 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4481 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4482 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4483 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4484 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4485 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4486 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4487 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4488 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4489 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4490 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4491 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4492 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4493 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4494 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4495 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4496 return a HOST_WIDE_INT.
4497 (avr_builtin_setjmp_frame_value): Use it instead of
4498 STARTING_FRAME_OFFSET.
4499 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4500 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4501 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4503 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4504 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4505 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4506 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4507 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4508 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4509 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4510 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4511 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4512 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4513 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4514 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4516 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4517 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4518 * config/mips/mips.c (mips_compute_frame_info): Refer to
4519 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4520 (mips_starting_frame_offset): New function.
4521 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4522 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4523 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4524 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4525 and return a HOST_WIDE_INT.
4526 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4527 (mmix_initial_elimination_offset): Refer to
4528 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4529 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4530 * config/pa/pa.c (pa_starting_frame_offset): New function.
4531 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4532 (pa_expand_prologue): Likewise.
4533 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4534 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4535 !FRAME_GROWS_DOWNWARD handling to...
4536 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4537 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4538 !FRAME_GROWS_DOWNWARD handling to...
4539 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4540 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4541 !FRAME_GROWS_DOWNWARD handling to...
4542 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4543 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4545 (rs6000_starting_frame_offset): New function.
4546 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4547 !FRAME_GROWS_DOWNWARD handling to...
4548 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4549 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4550 !FRAME_GROWS_DOWNWARD handling to...
4551 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4552 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4553 !FRAME_GROWS_DOWNWARD handling to...
4554 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4555 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4556 (rs6000_starting_frame_offset): New function.
4557 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4558 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4559 * config/vax/vax.c (vax_starting_frame_offset): New function.
4560 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4561 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4562 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4563 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4564 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4565 * system.h (STARTING_FRAME_OFFSET): Poison.
4567 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4569 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4570 SCALAR_TYPE_MODE instead of TYPE_MODE.
4572 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4573 Alan Hayward <alan.hayward@arm.com>
4574 David Sherwood <david.sherwood@arm.com>
4576 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4578 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4579 Alan Hayward <alan.hayward@arm.com>
4580 David Sherwood <david.sherwood@arm.com>
4582 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4584 2017-10-23 Richard Biener <rguenther@suse.de>
4586 PR tree-optimization/82129
4587 * tree-ssa-pre.c (bitmap_set_and): Remove.
4588 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4589 canonicalizing expressions in the set to those with lowest
4590 ID rather than taking that from the first edge.
4592 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4594 * combine.c (rtx_equal_for_field_assignment_p): Use
4595 byte_lowpart_offset.
4597 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4598 Alan Hayward <alan.hayward@arm.com>
4599 David Sherwood <david.sherwood@arm.com>
4601 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4602 to a SUBREG_PROMOTED_VAR.
4604 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4605 Alan Hayward <alan.hayward@arm.com>
4606 David Sherwood <david.sherwood@arm.com>
4608 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4609 (expand_debug_source_expr): Likewise.
4610 * combine.c (combine_simplify_rtx): Likewise.
4611 * cse.c (fold_rtx): Likewise.
4612 * optabs.c (expand_float): Likewise.
4613 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4614 (simplify_binary_operation_1): Likewise.
4616 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4617 Alan Hayward <alan.hayward@arm.com>
4618 David Sherwood <david.sherwood@arm.com>
4620 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4621 (record_promoted_value): Likewise.
4622 * expr.c (expand_expr_real_2): Likewise.
4623 * ree.c (update_reg_equal_equiv_notes): Likewise.
4624 (combine_set_extension): Likewise.
4625 * rtlanal.c (low_bitmask_len): Likewise.
4626 * simplify-rtx.c (neg_const_int): Likewise.
4627 (simplify_binary_operation_1): Likewise.
4629 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4630 Alan Hayward <alan.hayward@arm.com>
4631 David Sherwood <david.sherwood@arm.com>
4633 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4634 * regcprop.c (maybe_mode_change): Likewise.
4635 * reload1.c (alter_reg): Likewise.
4637 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4639 * inchash.h (inchash::hash::add_wide_int): New function.
4640 * lto-streamer-out.c (hash_tree): Use it.
4642 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4644 * inchash.h (inchash::hash::add_wide_int): Rename to...
4645 (inchash::hash::add_hwi): ...this.
4646 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4647 (polymorphic_call_target_hasher::hash): Likewise.
4648 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4649 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4650 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4651 * lto-streamer-out.c (hash_tree): Likewise.
4652 * optc-save-gen.awk: Likewise.
4653 * tree.c (add_expr): Likewise.
4655 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4658 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4659 for ordered inequality comparisons even with TARGET_IEEE_FP.
4661 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4664 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4665 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4666 Expand with cmp<dwi>_doubleword.
4668 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4670 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4671 List CET intrinsics.
4672 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4673 specific to -fcf-protection option.
4675 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4677 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4678 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4679 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4680 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4681 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4682 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4683 (extra_objs): Add cet.o for Linux/x86 targets.
4684 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4685 * config/i386/cet.c: New file.
4686 * config/i386/cetintrin.h: Likewise.
4687 * config/i386/t-cet: Likewise.
4688 * config/i386/cpuid.h (bit_SHSTK): New.
4689 (bit_IBT): Likewise.
4690 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4691 pass IBT and SHSTK bits.
4692 * config/i386/i386-builtin-types.def
4693 (VOID_FTYPE_UNSIGNED_PVOID): New.
4694 (VOID_FTYPE_UINT64_PVOID): Likewise.
4695 * config/i386/i386-builtin.def: Add CET intrinsics.
4696 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4697 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4698 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4699 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4701 * config/i386/i386.c (rest_of_insert_endbranch): New.
4702 (pass_data_insert_endbranch): Likewise.
4703 (pass_insert_endbranch): Likewise.
4704 (make_pass_insert_endbranch): Likewise.
4705 (ix86_notrack_prefixed_insn_p): Likewise.
4706 (ix86_target_string): Add -mibt, -mshstk flags.
4707 (ix86_option_override_internal): Add flag_cf_protection
4709 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4710 (ix86_print_operand): Add 'notrack' prefix output.
4711 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4712 (ix86_expand_builtin): Expand CET intrinsics.
4713 (x86_output_mi_thunk): Add 'endbranch' instruction.
4714 * config/i386/i386.h (TARGET_IBT): New.
4715 (TARGET_IBT_P): Likewise.
4716 (TARGET_SHSTK): Likewise.
4717 (TARGET_SHSTK_P): Likewise.
4718 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4719 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4720 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4721 (builtin_setjmp_setup): New pattern.
4722 (builtin_longjmp): Likewise.
4723 (rdssp<mode>): Likewise.
4724 (incssp<mode>): Likewise.
4725 (saveprevssp): Likewise.
4726 (rstorssp): Likewise.
4727 (wrss<mode>): Likewise.
4728 (wruss<mode>): Likewise.
4729 (setssbsy): Likewise.
4730 (clrssbsy): Likewise.
4731 (nop_endbr): Likewise.
4732 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4734 * config/i386/immintrin.h: Include <cetintrin.h>.
4735 * config/i386/linux-common.h
4736 (file_end_indicate_exec_stack_and_cet): New prototype.
4737 (TARGET_ASM_FILE_END): New.
4739 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4741 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4742 latencies instead of having separate table; make difference between
4743 integer and float costs.
4744 * i386.h (processor_costs): Remove scalar_stmt_cost,
4745 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4746 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4748 * x86-tune-costs.h: Remove entries which has been removed in
4749 procesor_costs from all tables; make cond_taken_branch_cost
4750 and cond_not_taken_branch_cost COST_N_INSNS based.
4752 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4754 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4756 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4758 * config/i386/i386.md (isa): Remove fma_avx512f.
4759 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4760 <avx512>_fmadd_<mode>_mask3<round_name>,
4761 <avx512>_fmsub_<mode>_mask<round_name>,
4762 <avx512>_fmsub_<mode>_mask3<round_name>,
4763 <avx512>_fnmadd_<mode>_mask<round_name>,
4764 <avx512>_fnmadd_<mode>_mask3<round_name>,
4765 <avx512>_fnmsub_<mode>_mask<round_name>,
4766 <avx512>_fnmsub_<mode>_mask3<round_name>,
4767 <avx512>_fmaddsub_<mode>_mask<round_name>,
4768 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4769 <avx512>_fmsubadd_<mode>_mask<round_name>,
4770 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4771 (*vec_widen_umult_even_v16si<mask_name>,
4772 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4773 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4775 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4777 * extend.texi: Add 'nocf_check' documentation.
4778 * gimple.texi: Add second parameter to
4779 gimple_build_call_from_tree.
4780 * invoke.texi: Add -fcf-protection documentation.
4781 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4783 2017-10-20 Richard Biener <rguenther@suse.de>
4785 PR tree-optimization/82473
4786 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4787 the largest input type.
4789 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4791 * c-attribs.c (handle_nocf_check_attribute): New function.
4792 (c_common_attribute_table): Add 'nocf_check' handling.
4793 * gimple-parser.c: Add second argument NULL to
4794 gimple_build_call_from_tree.
4795 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4796 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4798 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4799 * common.opt: Add fcf-protection flag.
4800 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4801 * flag-types.h: Add enum cf_protection_level.
4802 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4803 Add 'nocf_check' attribute propagation to gimple call.
4804 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4805 (gimple_build_call_from_tree): Update prototype.
4806 (gimple_call_nocf_check_p): New function.
4807 (gimple_call_set_nocf_check): Likewise.
4808 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4809 * ipa-icf.c: Add nocf_check attribute in statement hash.
4810 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4811 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4812 * toplev.c (process_options): Add flag_cf_protection handling.
4814 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4816 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4818 2017-10-20 Richard Biener <rguenther@suse.de>
4820 PR tree-optimization/82603
4821 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4822 remove false predicated stores.
4824 2017-10-20 Richard Biener <rguenther@suse.de>
4826 * graphite-isl-ast-to-gimple.c
4827 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4828 Remove return value and simplify, dump copied stmt after lhs
4830 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4831 Reduce dump verbosity.
4832 (gsi_insert_earliest): Likewise.
4833 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4834 * graphite.c (print_global_statistics): Adjust dumping.
4835 (print_graphite_scop_statistics): Likewise.
4836 (print_graphite_statistics): Do not dump loops here.
4837 (graphite_transform_loops): But here.
4839 2017-10-20 Nicolas Roche <roche@adacore.com>
4841 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4842 * configure: Regenerate.
4844 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4847 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4848 functions when optimizing replace GIMPLE_RETURN stmts with
4849 calls to __builtin_unreachable ().
4852 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4853 for -fsanitize=thread link of executables.
4854 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4855 link of executables.
4858 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4860 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4861 define_insns for logical vector shifts to use VI248_AVX512BW
4862 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4863 condition, useless isa and prefix attributes. Change the first
4864 2 of these define_insns to ...
4865 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4866 define_insn for avx512vl.
4867 (<shift_insn><mode>3): ... and this, new define_insn without
4868 masking for non-avx512vl.
4871 * config/i386/sse.md (*andnot<mode>3,
4872 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4873 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4874 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4875 not applied use empty suffix even for TARGET_AVX512VL.
4876 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4877 is applied, supply evex,evex or evex,evex,evex instead of just
4880 2017-10-20 Julia Koval <julia.koval@intel.com>
4882 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4883 (OPTION_MASK_ISA_GFNI_UNSET): New.
4884 (ix86_handle_option): Handle OPT_mgfni.
4885 * config/i386/cpuid.h (bit_GFNI): New.
4886 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4887 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4888 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4889 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4890 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4891 * config/i386/i386.opt: Add mgfni.
4893 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4895 * config/msp430/msp430.c (msp430_option_override): Disable
4896 -fdelete-null-pointer-checks.
4897 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4899 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4901 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4902 of x87 and SSE instructions.
4904 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4906 * asan.c (create_cond_insert_point): Do not update edge count.
4907 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4908 (afdo_propagate_circuit): Likewise.
4909 (afdo_calculate_branch_prob): Likewise.
4910 (afdo_annotate_cfg): Likewise.
4911 * basic-block.h (struct edge_def): Remove count.
4912 (edge_def::count): New accessor.
4913 * bb-reorder.c (rotate_loop): Update.
4914 (find_traces_1_round): Update.
4915 (connect_traces): Update.
4916 (sanitize_hot_paths): Update.
4917 * cfg.c (unchecked_make_edge): Update.
4918 (make_single_succ_edge): Update.
4919 (check_bb_profile): Update.
4920 (dump_edge_info): Update.
4921 (update_bb_profile_for_threading): Update.
4922 (scale_bbs_frequencies_int): Update.
4923 (scale_bbs_frequencies_gcov_type): Update.
4924 (scale_bbs_frequencies_profile_count): Update.
4925 (scale_bbs_frequencies): Update.
4926 * cfganal.c (connect_infinite_loops_to_exit): Update.
4927 * cfgbuild.c (compute_outgoing_frequencies): Update.
4928 (find_many_sub_basic_blocks): Update.
4929 * cfgcleanup.c (try_forward_edges): Update.
4930 (try_crossjump_to_edge): Update
4931 * cfgexpand.c (expand_gimple_cond): Update
4932 (expand_gimple_tailcall): Update
4933 (construct_exit_block): Update
4934 * cfghooks.c (verify_flow_info): Update
4935 (redirect_edge_succ_nodup): Update
4936 (split_edge): Update
4937 (make_forwarder_block): Update
4938 (duplicate_block): Update
4939 (account_profile_record): Update
4940 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4941 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4942 * cfgloopmanip.c (scale_loop_profile): Update.
4944 (lv_adjust_loop_entry_edge): Update.
4945 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4946 (force_nonfallthru_and_redirect): Update.
4947 (purge_dead_edges): Update.
4948 (rtl_flow_call_edges_add): Update.
4949 * cgraphunit.c (init_lowered_empty_function): Update.
4950 (cgraph_node::expand_thunk): Update.
4951 * gimple-pretty-print.c (dump_probability): Update.
4952 (dump_edge_probability): Update.
4953 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4954 * haifa-sched.c (sched_create_recovery_edges): Update.
4955 * hsa-gen.c (convert_switch_statements): Update.
4956 * ifcvt.c (dead_or_predicable): Update.
4957 * ipa-inline-transform.c (inline_transform): Update.
4958 * ipa-split.c (split_function): Update.
4959 * ipa-utils.c (ipa_merge_profiles): Update.
4960 * loop-doloop.c (add_test): Update.
4961 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4962 * lto-streamer-in.c (input_cfg): Update.
4963 (input_function): Update.
4964 * lto-streamer-out.c (output_cfg): Update.
4965 * modulo-sched.c (sms_schedule): Update.
4966 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4967 * predict.c (maybe_hot_edge_p): Update.
4968 (unlikely_executed_edge_p): Update.
4969 (probably_never_executed_edge_p): Update.
4970 (dump_prediction): Update.
4971 (drop_profile): Update.
4972 (propagate_unlikely_bbs_forward): Update.
4973 (determine_unlikely_bbs): Update.
4974 (force_edge_cold): Update.
4975 * profile.c (compute_branch_probabilities): Update.
4976 * reg-stack.c (better_edge): Update.
4977 * shrink-wrap.c (handle_simple_exit): Update.
4978 * tracer.c (better_p): Update.
4979 * trans-mem.c (expand_transaction): Update.
4980 (split_bb_make_tm_edge): Update.
4981 * tree-call-cdce.c: Update.
4982 * tree-cfg.c (gimple_find_sub_bbs): Update.
4983 (gimple_split_edge): Update.
4984 (gimple_duplicate_sese_region): Update.
4985 (gimple_duplicate_sese_tail): Update.
4986 (gimple_flow_call_edges_add): Update.
4987 (insert_cond_bb): Update.
4988 (execute_fixup_cfg): Update.
4989 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4990 * tree-complex.c (expand_complex_div_wide): Update.
4991 * tree-eh.c (lower_resx): Update.
4992 (unsplit_eh): Update.
4993 (cleanup_empty_eh_move_lp): Update.
4994 * tree-inline.c (copy_edges_for_bb): Update.
4995 (freqs_to_counts): Update.
4996 (copy_cfg_body): Update.
4997 * tree-ssa-dce.c (remove_dead_stmt): Update.
4998 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4999 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
5000 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
5001 (unloop_loops): Update.
5002 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
5003 * tree-ssa-loop-split.c (connect_loops): Update.
5004 (split_loop): Update.
5005 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
5006 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
5007 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
5008 * tree-ssa-reassoc.c (branch_fixup): Update.
5009 * tree-ssa-tail-merge.c (replace_block_by): Update.
5010 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
5011 (compute_path_counts): Update.
5012 (update_profile): Update.
5013 (recompute_probabilities): Update.
5014 (update_joiner_offpath_counts): Update.
5015 (estimated_freqs_path): Update.
5016 (freqs_to_counts_path): Update.
5017 (clear_counts_path): Update.
5018 (ssa_fix_duplicate_block_edges): Update.
5019 (duplicate_thread_path): Update.
5020 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
5021 (case_bit_test_cmp): Update.
5022 (collect_switch_conv_info): Update.
5023 (gen_inbound_check): Update.
5024 (do_jump_if_equal): Update.
5025 (emit_cmp_and_jump_insns): Update.
5026 * tree-tailcall.c (decrease_profile): Update.
5027 (eliminate_tail_call): Update.
5028 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
5029 (vect_do_peeling): Update.
5030 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
5031 * ubsan.c (ubsan_expand_null_ifn): Update.
5032 (ubsan_expand_ptr_ifn): Update.
5033 * value-prof.c (gimple_divmod_fixed_value): Update.
5034 (gimple_mod_pow2): Update.
5035 (gimple_mod_subtract): Update.
5036 (gimple_ic): Update.
5037 (gimple_stringop_fixed_value): Update.
5039 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5042 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
5044 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
5046 PR rtl-optimization/82395
5047 * ira-color.c (allocno_priority_compare_func): Fix comparison step
5048 based on non_spilled_static_chain_regno_p.
5050 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
5052 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
5053 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
5054 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
5056 2017-10-19 Martin Sebor <msebor@redhat.com>
5058 PR tree-optimization/82596
5059 * tree.c (array_at_struct_end_p): Handle STRING_CST.
5061 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5063 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
5064 (get_mem_refs_of_builtin_call): Likewise.
5065 * builtins.c (expand_builtin_apply): Adjust call to
5066 allocate_dynamic_stack_space.
5067 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
5068 the third argument to allocate_dynamic_stack_space, otherwise -1.
5069 (expand_builtin): Deal with all alloca variants.
5070 (is_inexpensive_builtin): Likewise.
5071 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
5072 * calls.c (special_function_p): Deal with all alloca variants.
5073 (initialize_argument_information): Adjust call to
5074 allocate_dynamic_stack_space.
5075 (expand_call): Likewise.
5076 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
5077 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
5078 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
5079 use it for the stack usage computation.
5080 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
5081 * function.c (gimplify_parameters): Call build_alloca_call_expr.
5082 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
5083 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
5084 (in_loop_p): Remove first argument and useless check.
5085 (pass_walloca::execute): Remove useless test and adjust call to above.
5086 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
5087 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
5088 (gimplify_call_expr): Deal with all alloca variants.
5089 * hsa-gen.c (gen_hsa_alloca): Likewise.
5090 (gen_hsa_insns_for_call): Likewise.
5091 * ipa-pure-const.c (special_builtin_state): Likewise.
5092 * tree-chkp.c (chkp_build_returned_bound): Likewise.
5093 * tree-object-size.c (alloc_object_size): Likewise.
5094 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
5095 (call_may_clobber_ref_p_1): Likewise.
5096 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
5097 (ccp_fold_stmt): Likewise.
5098 (optimize_stack_restore): Likewise.
5099 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5100 (mark_all_reaching_defs_necessary_1): Likewise.
5101 (propagate_necessity): Likewise.
5102 (eliminate_unnecessary_stmts): Likewise.
5103 * tree.c (build_common_builtin_nodes): Build
5104 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
5105 (build_alloca_call_expr): New function.
5106 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
5107 (CASE_BUILT_IN_ALLOCA): Likewise.
5108 (build_alloca_call_expr): Declare.
5109 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
5111 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
5114 * dwarf2out.c (new_die_raw): New static inline function.
5115 (new_die): Use it to create the DIE.
5116 (add_AT_external_die_ref): Likewise.
5117 (clone_die): Likewise.
5118 (clone_as_declaration): Likewise.
5119 (dwarf2out_vms_debug_main_pointer): Likewise.
5120 (base_type_die): Likewise. Remove early return for corner cases.
5121 Do not call add_pubtype on the DIE here.
5122 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
5123 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
5124 typedefs for base types with DW_AT_endianity. Make sure a DIE with
5125 native order exists for base types, attach the DIE manually and call
5126 add_pubtype on it. Do not equate a reverse order DIE to the type.
5128 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
5130 * config/arm/arm.c (align_ok_ldrd_strd): New function.
5131 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
5133 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
5135 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5137 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
5138 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
5139 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
5140 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
5141 * opts.c (sanitizer_opts): Add builtin.
5142 * ubsan.c (instrument_builtin): New function.
5143 (pass_ubsan::execute): Call it.
5144 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
5145 * doc/invoke.texi: Document -fsanitize=builtin.
5147 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
5148 builtins, store max (log2 (align), 0) into uchar field instead of
5149 align into uptr field.
5150 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
5151 store uchar 0 field instead of uptr 0 field.
5152 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
5153 instead of passing one address of struct with 2 locations pass
5154 two addresses of structs with 1 location each.
5155 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
5156 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
5157 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
5158 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
5159 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
5160 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
5161 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
5162 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
5164 2017-10-19 Martin Liska <mliska@suse.cz>
5167 * file-find.c (remove_prefix): Remove.
5168 * file-find.h (remove_prefix): Likewise.
5169 * gcc-ar.c: Remove smartness of lookup.
5171 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5173 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
5174 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
5175 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
5177 2017-10-19 Jakub Jelinek <jakub@redhat.com>
5180 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
5181 (setcc + and to xor + setcc): New peephole2.
5183 2017-10-19 Tom de Vries <tom@codesourcery.com>
5185 * doc/sourcebuild.texi (Test Directives, Variants of
5186 dg-require-support): Add dg-require-stack-size.
5188 2017-10-19 Martin Liska <mliska@suse.cz>
5191 * gimplify.c (gimplify_decl_expr): Do not instrument variables
5192 that have a large alignment.
5193 (gimplify_target_expr): Likewise.
5195 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
5197 PR rtl-optimization/82602
5198 * ira.c (rtx_moveable_p): Return false for volatile asm.
5200 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
5203 * config/i386/i386-modes.def (CCGZ): New CC mode.
5204 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
5205 * config/i386/predicates.md (ix86_comparison_operator):
5207 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
5208 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
5209 with double-word subtraction.
5210 (put_condition_code): Handle CCGZmode.
5212 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
5214 * wide-int.cc (debug (const wide_int &)): New.
5215 (debug (const wide_int *)): New.
5216 (debug (const widest_int &)): New.
5217 (debug (const widest_int *)): New.
5219 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
5222 * lra-constraints.c (curr_insn_transform): Use non-input operand
5223 instead of output one for matched reload.
5225 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5227 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
5228 (tree-ssa-loop-ivopts.h): New header file.
5229 (struct builtin_info): New fields.
5230 (classify_builtin_1): Compute and record base and offset parts for
5231 memset builtin partition by calling strip_offset.
5232 (offset_cmp, fuse_memset_builtins): New functions.
5233 (finalize_partitions): Fuse adjacent memset partitions by calling
5235 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
5236 Expose the interface.
5237 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
5239 2017-10-18 Bin Cheng <bin.cheng@arm.com>
5241 PR tree-optimization/82574
5242 * tree-loop-distribution.c (find_single_drs): New parameter. Check
5243 that data reference must be executed exactly once per iteration
5244 against the outermost loop in nest.
5245 (classify_partition): Update call to above function.
5247 2017-10-18 Richard Biener <rguenther@suse.de>
5249 PR tree-optimization/82591
5250 * graphite.c (graphite_transform_loops): Move code gen message
5252 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5253 Here. Handle scop_to_isl_ast failing.
5254 (scop_to_isl_ast): Limit the number of ISL operations.
5256 2017-10-18 Richard Biener <rguenther@suse.de>
5258 * graphite-isl-ast-to-gimple.c
5259 (translate_isl_ast_to_gimple::set_rename): Simplify.
5260 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
5261 (graphite_copy_stmts_from_block): ... here.
5262 (copy_bb_and_scalar_dependences): Simplify.
5263 (add_parameters_to_ivs_params): Canonicalize.
5264 (generate_entry_out_of_ssa_copies): Simplify.
5265 * graphite-sese-to-poly.c (extract_affine_name): Simplify
5266 by passing in ISL dimension.
5267 (parameter_index_in_region_1): Rename to ...
5268 (parameter_index_in_region): ... this.
5269 (extract_affine): Adjust assert, pass down parameter index.
5270 (add_param_constraints): Use range-info when available.
5271 (build_scop_context): Adjust.
5272 * sese.c (new_sese_info): Adjust.
5273 (free_sese_info): Likewise.
5274 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
5275 Remove unused typedefs.
5276 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
5278 2017-10-18 Martin Liska <mliska@suse.cz>
5280 * combine.c (simplify_compare_const): Add gcc_fallthrough.
5282 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5284 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
5285 (s390_sched_init): Do not reset s390_sched_state if we entered the
5286 current basic block via a fallthru edge and all others are unlikely.
5288 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5290 * config/s390/s390.c (NUM_SIDES): New variable.
5291 (LONGRUNNING_THRESHOLD): New variable.
5292 (LATENCY_FACTOR): New variable.
5293 (s390_sched_score): Decrease score for long-running instructions on
5295 (s390_sched_variable_issue): Perform bookkeeping for long-running
5298 2017-10-18 Richard Biener <rguenther@suse.de>
5300 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
5301 Simplify with removal of the parameter rename map.
5302 (set_rename): Likewise.
5303 (should_copy_to_new_region): Likewise.
5304 (graphite_copy_stmts_from_block): Likewise.
5305 (copy_bb_and_scalar_dependences): Remove initialization of
5306 unused copied_bb_map.
5308 (copy_internal_parameters): Likewise.
5309 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
5310 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
5311 Use INTEGRAL_TYPE_P.
5312 (parameter_index_in_region_1): Rename to ...
5313 (assign_parameter_index_in_region): ... this. Assert we have
5314 a parameter we handle.
5315 (scan_tree_for_params): Adjust.
5316 * sese.h (parameter_rename_map_t): Remove.
5317 (struct sese_info_t): Remove unused parameter_rename_map and
5318 copied_bb_map members.
5319 * sese.c (new_sese_info): Adjust.
5320 (free_sese_info): Likewise.
5322 2017-10-18 Martin Liska <mliska@suse.cz>
5325 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
5326 on an abnormal edge.
5328 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
5330 * doc/invoke.texi (ffunction-sections and fdata-sections):
5333 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5335 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
5336 the use statement can throw internally.
5338 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
5340 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
5341 any RTX present on the RHS of a SET.
5342 * compare-elim.c (try_eliminate_compare): Restore comment.
5344 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5346 * langhooks.h (struct lang_hooks): Document that tree_size langhook
5347 may be also called on tcc_type nodes.
5348 * langhooks.c (lhd_tree_size): Likewise.
5350 2017-10-17 David Malcolm <dmalcolm@redhat.com>
5352 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
5353 format_warning_at_substring.
5354 (maybe_warn): Convert source_range * param to a location_t. Pass
5355 UNKNOWN_LOCATION rather than NULL to fmtwarn.
5356 (format_directive): Remove code to extract source_ranges and
5357 source_range * in favor of just a location_t.
5358 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
5360 * substring-locations.c (format_warning_va): Convert
5361 source_range * param to a location_t.
5362 (format_warning_at_substring): Likewise.
5363 * substring-locations.h (format_warning_va): Likewise.
5364 (format_warning_at_substring): Likewise.
5366 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
5368 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
5370 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
5371 and scatter/gather ops.
5373 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
5374 vec_gather_load and vec_scatter_store.
5375 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
5376 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
5377 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
5378 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
5379 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
5380 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
5382 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
5384 * reg-stack.c (compare_for_stack_reg): Add bool argument.
5385 Detect FTST instruction and handle its register pops. Only pop
5386 second operand if can_pop_second_op is true.
5387 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
5388 set can_pop_second_op to false in the compare_for_stack_reg call.
5390 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
5391 output_fp_compare for stack register operands.
5392 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
5393 instructions here. Do not emit stack register pops here. Assert
5394 that FCOMPP pops next to top stack register. Rewrite function.
5396 2017-10-17 Nathan Sidwell <nathan@acm.org>
5399 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
5400 use DECL_ASSEMBLER_NAME_RAW.
5403 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
5406 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
5407 Wilco Dijkstra <wilco.dijkstra@arm.com>
5409 * builtins.c (expand_builtin_update_setjmp_buf): Add a
5410 converstion to Pmode from the buf_addr.
5412 2017-10-17 Richard Biener <rguenther@suse.de>
5414 * graphite-dependences.c (scop_get_reads_and_writes): Change
5415 output parameters to references.
5417 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
5419 PR 71026/tree-optimization
5420 * fold-const.c (distribute_real_division): Removed.
5421 (fold_binary_loc): Remove calls to distribute_real_divison.
5423 2017-10-17 Richard Biener <rguenther@suse.de>
5425 * graphite-scop-detection.c
5426 (scop_detection::stmt_has_simple_data_refs_p): Always use
5427 the full nest as region.
5428 (try_generate_gimple_bb): Likewise.
5429 * sese.c (scalar_evolution_in_region): Simplify now that
5430 SCEV can handle instantiation in regions.
5431 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5432 in the non-loop part of a function if requested.
5434 2017-10-17 Richard Biener <rguenther@suse.de>
5436 PR tree-optimization/82563
5437 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5439 (graphite_regenerate_ast_isl): Call it.
5440 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5442 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5444 PR tree-optimization/82549
5445 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5446 Formatting fixes. Instead of calling make_bit_field_ref with negative
5449 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5451 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5452 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5453 _mm_maskz_reduce_ss): New.
5454 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5455 __builtin_ia32_reducess_mask): Ditto..
5456 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5457 * config/i386/sse.md (reduces<mode>): Renamed to ...
5458 (reduces<mode><mask_scalar_name>): ... this.
5459 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5461 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5462 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5464 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5466 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5467 * selftest-run-tests.c (selftest::run_tests): Call
5468 selftest::unique_ptr_tests_cc_tests.
5469 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5470 * unique-ptr-tests.cc: New file.
5472 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5475 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5477 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5478 (make_hard_regno_born, make_hard_regno_dead): Update
5479 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5481 2017-10-16 Jeff Law <law@redhat.com>
5483 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5485 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5487 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5489 2017-10-16 Olivier Hainque <hainque@adacore.com>
5491 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5492 with_cpu if we were configured for an e500v2 target cpu name.
5494 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5496 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5497 * doc/invoke.texi: Document +nodsp as a valid extension for
5500 2017-10-16 Martin Liska <mliska@suse.cz>
5502 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5503 (test_set_range): Likewise.
5504 (test_range_functions): Rename to ...
5505 (test_bit_in_range): ... this.
5506 (sbitmap_c_tests): Add new test.
5508 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5510 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5512 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5513 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5515 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5517 * config/aarch64/aarch64-builtins.c
5518 (aarch64_types_quadopu_lane_qualifiers): New.
5519 (TYPES_QUADOPU_LANE): New.
5520 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5521 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5522 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5523 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5524 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5525 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5526 (Vdottype, DOTPROD): New.
5527 (sur): Add SDOT and UDOT.
5529 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5531 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5532 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5533 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5535 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5536 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5537 Enable TARGET_DOTPROD.
5538 (cortex-a75.cortex-a55): Likewise.
5539 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5541 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5543 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5544 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5546 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5548 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5549 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5550 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5551 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5552 * config/arm/types.md (neon_dot, neon_dot_q): New.
5553 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5555 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5557 * config/arm/arm.h (TARGET_DOTPROD): New.
5558 * config/arm/arm.c (arm_arch_dotprod): New.
5559 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5560 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5561 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5562 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5563 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5564 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5565 * doc/invoke.texi (armv8.2-a): Document dotprod
5567 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5569 * i386.c (ix86_vec_cost): New function.
5570 (ix86_rtx_costs): Handle vector operations better.
5571 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5572 * x86-tune-costs.h: Add new costs to all tables.
5574 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5576 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5578 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5579 divsd, sqrtss and sqrtsd
5580 * x86-tune-costs.h: Add new entries to all costs.
5581 (znver1_cost): Fix to match real instruction latencies.
5583 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5584 Michael Collison <michael.collison@arm.com>
5586 * compare-elim.c: Include emit-rtl.h.
5587 (can_merge_compare_into_arith): New function.
5588 (try_validate_parallel): Likewise.
5589 (try_merge_compare): Likewise.
5590 (try_eliminate_compare): Call the above when no previous clobber
5592 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5595 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5599 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5600 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5604 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5605 to be any operand_equal_p operands. For & (B - 1) require
5606 B to be power of 2. Recognize
5607 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5609 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5612 * optabs.c (expand_memory_blockage): Fix call of
5613 targetm.have_memory_blockage.
5615 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5618 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5619 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5620 objects to extra_objs instead of overwriting it.
5622 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5624 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5625 Use any_fp_register_operand as operand[3] predicate. Simplify
5626 equality test for operands[2] and operands[4] memory location.
5627 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5628 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5629 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5630 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5631 any_fp_register_operand as operand[1] predicate. Simplify
5632 equality test for operands[0] and operands[3] memory location.
5633 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5634 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5635 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5637 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5639 * target-insns.def: Add memory_blockage.
5640 * optabs.c (expand_memory_blockage): New function.
5641 (expand_asm_memory_barrier): Rename ...
5642 (expand_asm_memory_blockage): ... to this.
5643 (expand_mem_thread_fence): Call expand_memory_blockage
5644 instead of expand_asm_memory_barrier.
5645 (expand_mem_singnal_fence): Ditto.
5646 (expand_atomic_load): Ditto.
5647 (expand_atomic_store): Ditto.
5648 * doc/md.texi (Standard Pattern Names For Generation):
5649 Document memory_blockage instruction pattern.
5651 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5653 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5654 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5655 * config/rl78/rl78.md: New define_expand "adddi3".
5657 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5659 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5662 2017-10-13 Jeff Law <law@redhat.com>
5664 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5666 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5669 * internal-fn.c (expand_mul_overflow): If both operands have
5670 the same highpart of -1 or 0 and the topmost bit of lowpart
5671 is different, overflow is if res <= 0 rather than res < 0.
5673 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5675 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5676 TARGET_P9_VECTOR code for unaligned_load case.
5678 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5680 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5682 2017-10-13 Nathan Sidwell <nathan@acm.org>
5684 * tree-core.h (tree_contains_struct): Make bool.
5685 * tree.c (tree_contains_struct): Likewise.
5686 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5687 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5688 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5689 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5690 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5692 2017-10-13 Richard Biener <rguenther@suse.de>
5694 * graphite-isl-ast-to-gimple.c
5695 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5696 parameters and dominance check.
5697 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5698 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5699 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5700 Do not update SSA form here or do intermediate IL verification.
5701 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5702 (graphite_initialize): Remove check on the number of loops in
5703 the function and inline into graphite_transform_loops.
5704 (graphite_finalize): Inline into graphite_transform_loops.
5705 (graphite_transform_loops): Perform SSA update and IL verification
5707 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5709 2017-10-13 Richard Biener <rguenther@suse.de>
5711 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5712 graphite_expression_type_precision): Avoid global constructor
5714 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5715 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5716 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5717 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5718 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5719 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5721 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5724 * config/i386/i386.h (ix86_red_zone_size): New.
5725 * config/i386/i386.md (push peephole2s): Replace
5726 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5728 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5729 Alan Hayward <alan.hayward@arm.com>
5730 David Sherwood <david.sherwood@arm.com>
5732 * combine.c (can_change_dest_mode): Reject changes in
5733 REGMODE_NATURAL_SIZE.
5735 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5736 Alan Hayward <alan.hayward@arm.com>
5737 David Sherwood <david.sherwood@arm.com>
5739 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5740 (expand_debug_source_expr): Likewise.
5741 * combine.c (combine_simplify_rtx): Likewise.
5742 * cse.c (fold_rtx): Likewise.
5743 * fwprop.c (canonicalize_address): Likewise.
5744 * targhooks.c (default_shift_truncation_mask): Likewise.
5746 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5747 Alan Hayward <alan.hayward@arm.com>
5748 David Sherwood <david.sherwood@arm.com>
5750 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5751 (widened_mode): Likewise.
5752 (expand_unop): Likewise.
5753 * ree.c (transform_ifelse): Likewise.
5754 (merge_def_and_ext): Likewise.
5755 (combine_reaching_defs): Likewise.
5756 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5758 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5759 Alan Hayward <alan.hayward@arm.com>
5760 David Sherwood <david.sherwood@arm.com>
5762 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5763 * combine.c (gen_lowpart_for_combine): Likewise.
5764 * dwarf2out.c (rtl_for_decl_location): Likewise.
5765 * final.c (alter_subreg): Likewise.
5766 * rtlhooks.c (gen_lowpart_general): Likewise.
5767 (gen_lowpart_if_possible): Likewise.
5769 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5770 Alan Hayward <alan.hayward@arm.com>
5771 David Sherwood <david.sherwood@arm.com>
5773 * calls.c (expand_call): Use subreg_lowpart_offset.
5774 * cse.c (cse_insn): Likewise.
5775 * regcprop.c (copy_value): Likewise.
5776 (copyprop_hardreg_forward_1): Likewise.
5778 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5781 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5782 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5783 =Q constraints to +Q and into insn condition add check
5784 that operands[0] and operands[1] are equal.
5785 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5786 =Q constraints to +Q and into insn condition add check
5787 that operands[0] is equal to either operands[1] or operands[2].
5790 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5791 instead of handling MINUS_EXPR twice (once for each argument),
5792 canonicalize operand order and handle just once, use rtype where
5793 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5796 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5797 any values of __C while still being pattern recognizable as a simple
5800 2017-10-13 Richard Biener <rguenther@suse.de>
5802 PR tree-optimization/82451
5804 2017-10-02 Richard Biener <rguenther@suse.de>
5806 PR tree-optimization/82355
5807 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5808 a mapping for the enclosing loop but avoid generating one for
5810 (copy_bb_and_scalar_dependences): Remove premature codegen
5811 error on PHIs in blocks duplicated into multiple places.
5812 * graphite-scop-detection.c
5813 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5814 in the region use it as loop and nest to analyze the DR in.
5815 (try_generate_gimple_bb): Likewise.
5816 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5817 (add_loop_constraints): For blocks in a loop not in the region
5818 create a dimension with a single iteration.
5819 * sese.h (gbb_loop_at_index): Remove assert.
5821 * cfgloop.c (loop_preheader_edge): For the loop tree root
5822 return the single successor of the entry block.
5823 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5824 Reset the SCEV hashtable and niters.
5825 * graphite-scop-detection.c
5826 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5827 assert that we only have POLYNOMIAL_CHREC that vary in loops
5828 contained in the region.
5829 (scop_detection::graphite_can_represent_expr): Adjust.
5830 (scop_detection::stmt_has_simple_data_refs_p): For loops
5831 not in the region set loop to NULL. The nest is now the
5832 entry edge to the region.
5833 (try_generate_gimple_bb): Likewise.
5834 * sese.c (scalar_evolution_in_region): Adjust for
5835 instantiate_scev change.
5836 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5837 Make nest parameter the edge into the region.
5838 (create_data_ref): Likewise.
5839 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5840 entry edge into a region and adjust instantiate_scev calls.
5841 (create_data_ref): Likewise.
5842 (graphite_find_data_references_in_stmt): Likewise.
5843 (find_data_references_in_stmt): Pass the loop preheader edge
5844 from the nest argument.
5845 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5846 parameter the edge into the region.
5847 (instantiate_parameters): Use the loop preheader edge as entry.
5848 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5850 (get_instantiated_value_entry): Make instantiate_below parameter
5851 the edge into the region.
5852 (instantiate_scev_name): Likewise. Adjust dominance checks,
5853 when we cannot use loop-based instantiation instantiate by
5854 walking use-def chains.
5855 (instantiate_scev_poly): Adjust.
5856 (instantiate_scev_binary): Likewise.
5857 (instantiate_scev_convert): Likewise.
5858 (instantiate_scev_not): Likewise.
5859 (instantiate_array_ref): Remove.
5860 (instantiate_scev_3): Likewise.
5861 (instantiate_scev_2): Likewise.
5862 (instantiate_scev_1): Likewise.
5863 (instantiate_scev_r): Do not blindly handle N-operand trees.
5864 Do not instantiate array-refs. Handle all constants and invariants.
5865 (instantiate_scev): Make instantiate_below parameter
5866 the edge into the region.
5867 (resolve_mixers): Use the loop preheader edge for the region
5868 parameter to instantiate_scev_r.
5869 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5871 2017-10-13 Richard Biener <rguenther@suse.de>
5873 PR tree-optimization/82525
5874 * graphite-isl-ast-to-gimple.c
5875 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5877 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5878 Fail code generation when we cannot represent the isl integer.
5879 (binary_op_to_tree): Elide modulo operations that are no-ops
5880 in the type we code generate. Remove now superfluous code
5883 2017-10-13 Richard Biener <rguenther@suse.de>
5885 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5886 (scop_detection::harmful_loop_in_region): Remove premature
5887 IV type restriction.
5888 (scop_detection::graphite_can_represent_scev): We can handle
5889 pointer IVs just fine.
5891 2017-10-13 Alan Modra <amodra@gmail.com>
5893 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5894 "Clobbers and Scratch Registers". Add paragraph on
5895 alternative to clobbers for scratch registers and OpenBLAS
5898 2017-10-13 Alan Modra <amodra@gmail.com>
5900 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5901 example of a memory input for a string of known length. Move
5902 commentary out of table. Add a number of new examples
5903 covering array memory inputs.
5905 2017-10-12 Martin Liska <mliska@suse.cz>
5907 PR tree-optimization/82493
5908 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5909 (test_range_functions): New function.
5910 (sbitmap_c_tests): Likewise.
5911 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5912 * selftest.h (sbitmap_c_tests): New function.
5914 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5916 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5918 * config/rs6000/amo.h: Fix spacing issue.
5920 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5923 * config/i386/i386.md (*ashl<mode>3_mask_1,
5924 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5925 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5928 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5930 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5931 (profile_probability): Set max_probability
5932 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5934 * profile-count.c (profile_probability::differs_from_p): Do not
5935 rely on max_probaiblity == 10000
5937 2017-10-12 Jeff Law <law@redhat.com>
5939 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5942 2017-10-12 Martin Sebor <msebor@redhat.com>
5946 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5947 (handle_alias_pairs): Call it.
5948 * common.opt (-Wattribute-alias): New option.
5949 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5950 * doc/invoke.texi (-Wattribute-alias): Document.
5952 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5955 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5957 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5959 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5960 (make_hard_regno_born, make_hard_regno_dead): Update
5961 bb_killed_pseudos and bb_gen_pseudos.
5963 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5965 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5967 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5969 * config/alpha/alpha.c (alpha_split_conditional_move):
5970 Use std::swap instead of manually swapping.
5971 (alpha_stdarg_optimize_hook): Ditto.
5972 (alpha_canonicalize_comparison): Ditto.
5974 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5976 * tree-loop-distribution.c (struct builtin_info): New struct.
5977 (struct partition): Refactor fields into struct builtin_info.
5978 (partition_free): Free struct builtin_info.
5979 (build_size_arg_loc, build_addr_arg_loc): Delete.
5980 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5981 information from struct builtin_info.
5982 (find_single_drs): New function refactored from classify_partition.
5983 Also moved builtin validity checks to this function.
5984 (compute_access_range, alloc_builtin): New functions.
5985 (classify_builtin_st, classify_builtin_ldst): New functions.
5986 (classify_partition): Refactor code into functions find_single_drs,
5987 classify_builtin_st and classify_builtin_ldst.
5988 (distribute_loop): Don't do runtime alias check when distributing
5990 (find_seed_stmts_for_distribution): New function.
5991 (pass_loop_distribution::execute): Refactor code finding seed
5992 stmts into above function. Support distribution for the innermost
5993 two-level loop nest. Adjust dump information.
5995 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5997 * tree-loop-distribution.c: Adjust the general comment.
5998 (NUM_PARTITION_THRESHOLD): New macro.
5999 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
6000 (classify_partition): Skip builtin pattern of loop nest's inner loop.
6001 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
6002 in call to build_partition_graph.
6003 (finalize_partitions): New parameter. Make loop distribution more
6004 conservative by fusing more partitions.
6005 (distribute_loop): Don't do runtime alias check in case of loop nest
6007 (find_seed_stmts_for_distribution): New function.
6008 (prepare_perfect_loop_nest): New function.
6009 (pass_loop_distribution::execute): Refactor code finding seed stmts
6010 and loop nest into above functions. Support loop nest distribution.
6011 Adjust dump information accordingly.
6013 2017-10-12 Bin Cheng <bin.cheng@arm.com>
6015 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
6016 and set PTYPE_SEQUENTIAL for merged partition.
6018 2017-10-12 Richard Biener <rguenther@suse.de>
6020 PR tree-optimization/69728
6022 2017-09-19 Richard Biener <rguenther@suse.de>
6024 PR tree-optimization/69728
6025 * graphite-sese-to-poly.c (schedule_error): New global.
6026 (add_loop_schedule): Handle empty domain by failing the
6028 (build_original_schedule): Handle schedule_error.
6030 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
6031 domain by returning an unchanged schedule.
6033 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6035 * genrecog.c (validate_pattern): For VEC_SELECT verify that
6036 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
6038 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
6040 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
6043 2017-10-12 Jakub Jelinek <jakub@redhat.com>
6046 * expr.c (store_field): Don't optimize away bitsize == 0 store
6047 from CALL_EXPR with addressable return type.
6049 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6051 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
6052 * config/rs6000/rs6000.md (sel): Delete mode attribute.
6053 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
6054 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
6055 TARGET_ISEL instead of TARGET_ISEL<sel>.
6057 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
6059 * config/rs6000/rs6000.c
6060 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
6062 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
6064 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
6066 (reg_or_cint_operand): Fix comment.
6067 (reg_or_zero_operand): New predicate.
6068 * config/rs6000/rs6000-protos.h (output_isel): Delete.
6069 * config/rs6000/rs6000.c (output_isel): Delete.
6070 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
6071 instead of reg_or_cint_operand. Output instruction directly (not via
6073 (isel_unsigned_<mode>): Ditto.
6074 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
6075 gpc_reg_operand. Add an instruction alternative for this. Output
6076 instruction directly.
6077 (*isel_reversed_unsigned_<mode>): Ditto.
6079 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6081 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
6082 (TARGET_CANONICALIZE_COMPARISON): Define.
6084 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
6087 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
6088 Check whether the dest is REG before adding REG_EQUIV note.
6090 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
6093 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
6095 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
6096 (make_hard_regno_born, make_hard_regno_dead): Update
6097 bb_killed_pseudos and bb_gen_pseudos.
6099 2017-10-11 Nathan Sidwell <nathan@acm.org>
6101 * incpath.h (enum incpath_kind): Name enum, prefix values.
6102 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
6103 * incpath.c (heads, tails): Use INC_MAX.
6104 (add_env_var_paths, add_standard_paths): Use incpath_kind.
6105 (merge_include_chains, split_quote_chain,
6106 register_include_chains): Update incpath_kind names.
6107 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
6108 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
6110 (add_framework_path, darwin_register_objc_includes): Likewise.
6111 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
6113 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6115 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6116 Do not use float_operator operator predicate.
6117 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6118 * config/i386/predicates.md (float_operator): Remove predicate.
6120 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
6122 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
6123 (*jccxf_i387): Ditto.
6124 (*jcc<mode>_i387): Ditto.
6125 (*jccu<mode>_i387): Ditto.
6126 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
6127 (*jcc_*_i387 splitters): Remove.
6128 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
6129 * config/i386/i386.c (ix86_split_fp_branch): Remove.
6130 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
6133 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6135 * profile-count.h (slow_safe_scale_64bit): New function.
6136 (safe_scale_64bit): New inline.
6137 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
6138 * profile-count.c: Include wide-int.h
6139 (slow_safe_scale_64bit): New.
6141 2017-10-11 Nathan Sidwell <nathan@acm.org>
6143 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
6144 HAS_DECL_ASSEMBLER_NAME_P.
6145 * gimple-expr.c (gimple_decl_printable_name: Check
6146 HAS_DECL_ASSEMBLER_NAME_P too.
6147 * ipa-utils.h (type_in_anonymous_namespace_p): Check
6148 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
6149 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
6150 * passes.c (rest_of_decl_compilation): Check
6151 HAS_DECL_ASSEMBLER_NAME_P too.
6152 * recog.c (verify_changes): Likewise.
6153 * tree-pretty-print.c (dump_decl_name): Likewise.
6154 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
6156 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
6157 (SET_DECL_ASSEMBLER_NAME): Use it.
6158 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
6159 (COPY_DECL_ASSEMBLER_NAME): Likewise.
6160 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
6162 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
6164 * config.gcc (i386, x86_64): Add extra objects.
6165 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
6166 (ix86_min_insn_size): Declare.
6167 (ix86_issue_rate): Declare.
6168 (ix86_adjust_cost): Declare.
6169 (ia32_multipass_dfa_lookahead): Declare.
6170 (ix86_macro_fusion_p): Declare.
6171 (ix86_macro_fusion_pair_p): Declare.
6172 (ix86_bd_has_dispatch): Declare.
6173 (ix86_bd_do_dispatch): Declare.
6174 (ix86_core2i7_init_hooks): Declare.
6175 (ix86_atom_sched_reorder): Declare.
6176 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
6177 (COSTS_N_BYTES): Move to x86-tune-costs.h.
6178 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
6179 (rip_relative_addr_p): Rename to ...
6180 (ix86_rip_relative_addr_p): ... this one; export.
6181 (memory_address_length): Update.
6182 (ix86_issue_rate): Move to x86-tune-sched.c.
6183 (ix86_flags_dependent): Move to x86-tune-sched.c.
6184 (ix86_agi_dependent): Move to x86-tune-sched.c.
6185 (exact_dependency_1): Move to x86-tune-sched.c.
6186 (exact_store_load_dependency): Move to x86-tune-sched.c.
6187 (ix86_adjust_cost): Move to x86-tune-sched.c.
6188 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
6189 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
6190 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
6191 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
6192 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
6193 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
6194 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
6195 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
6196 (min_insn_size): Rename to ...
6197 (ix86_min_insn_size): ... this one; export.
6198 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
6199 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
6200 (core2i7_first_cycle_multipass_backtrack): Move to
6201 x86-tune-sched-core.c.
6202 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
6203 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
6204 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
6205 (ix86_avoid_jump_mispredicts): Update.
6206 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
6207 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
6208 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
6209 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
6210 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
6211 (MAX_INSN): Move to ix86-tune-sched-bd.c.
6212 (MAX_IMM): Move to ix86-tune-sched-bd.c.
6213 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
6214 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
6215 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
6216 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
6217 (MAX_STORE): Move to ix86-tune-sched-bd.c.
6218 (BIG): Move to ix86-tune-sched-bd.c.
6219 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
6220 (enum insn_path): Move to ix86-tune-sched-bd.c.
6221 (get_mem_group): Move to ix86-tune-sched-bd.c.
6222 (is_cmp): Move to ix86-tune-sched-bd.c.
6223 (dispatch_violation): Move to ix86-tune-sched-bd.c.
6224 (is_branch): Move to ix86-tune-sched-bd.c.
6225 (is_prefetch): Move to ix86-tune-sched-bd.c.
6226 (init_window): Move to ix86-tune-sched-bd.c.
6227 (allocate_window): Move to ix86-tune-sched-bd.c.
6228 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
6229 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
6230 (process_end_window): Move to ix86-tune-sched-bd.c.
6231 (allocate_next_window): Move to ix86-tune-sched-bd.c.
6232 (find_constant): Move to ix86-tune-sched-bd.c.
6233 (get_num_immediates): Move to ix86-tune-sched-bd.c.
6234 (has_immediate): Move to ix86-tune-sched-bd.c.
6235 (get_insn_path): Move to ix86-tune-sched-bd.c.
6236 (get_insn_group): Move to ix86-tune-sched-bd.c.
6237 (count_num_restricted): Move to ix86-tune-sched-bd.c.
6238 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
6239 (add_insn_window): Move to ix86-tune-sched-bd.c.
6240 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
6241 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
6242 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
6243 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
6244 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
6245 (do_dispatch): Move to ix86-tune-sched-bd.c.
6246 (has_dispatch): Move to ix86-tune-sched-bd.c.
6247 * config/i386/t-i386: Add new object files.
6248 * config/i386/x86-tune-costs.h: New file.
6249 * config/i386/x86-tune-sched-atom.c: New file.
6250 * config/i386/x86-tune-sched-bd.c: New file.
6251 * config/i386/x86-tune-sched-core.c: New file.
6252 * config/i386/x86-tune-sched.c: New file.
6254 2017-10-11 Liu Hao <lh_mouse@126.com>
6256 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
6257 the generic version below instead.
6258 (should_colorize): Recognize Windows consoles as terminals
6260 * pretty-print.c [__MINGW32__] (write_all): New function.
6261 [__MINGW32__] (find_esc_head): Likewise.
6262 [__MINGW32__] (find_esc_terminator): Likewise.
6263 [__MINGW32__] (eat_esc_sequence): Likewise.
6264 [__MINGW32__] (mingw_ansi_fputs): New function that handles
6266 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
6269 2017-10-11 Richard Biener <rguenther@suse.de>
6271 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
6272 Properly call analyze_scalar_evolution with the loop of the stmt.
6274 2017-10-11 Richard Biener <rguenther@suse.de>
6276 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
6277 * tree-core.h (tree_base): Add chrec_var union member.
6278 * tree.h (CHREC_VAR): Remove.
6279 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
6280 * tree-chrec.h (build_polynomial_chrec): Adjust.
6281 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
6282 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
6284 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
6286 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
6287 * match.pd: ... here.
6288 ((T) X == (T) Y): Relax condition.
6290 2017-10-11 Bin Cheng <bin.cheng@arm.com>
6292 PR tree-optimization/82472
6293 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
6295 (break_alias_scc_partitions): Update postorder number.
6297 2017-10-11 Martin Liska <mliska@suse.cz>
6300 * opts.c (parse_no_sanitize_attribute): Do not use error_value
6302 * opts.h (parse_no_sanitize_attribute): Remove last argument.
6304 2017-10-11 Martin Liska <mliska@suse.cz>
6306 * print-rtl.c (print_insn): Move declaration of idbuf
6307 to same scope as name.
6309 2017-10-11 Martin Liska <mliska@suse.cz>
6314 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6317 2017-10-11 Martin Liska <mliska@suse.cz>
6320 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
6323 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
6325 * config.gcc (default_gnu_indirect_function): Default to yes for
6326 arm*-*-linux* with glibc.
6328 2017-10-11 Richard Biener <rguenther@suse.de>
6330 * tree-scalar-evolution.c (get_scalar_evolution): Handle
6331 default-defs and types we do not want to analyze.
6332 (interpret_loop_phi): Replace unreachable code with an assert.
6333 (compute_scalar_evolution_in_loop): Remove and inline ...
6334 (analyze_scalar_evolution_1): ... here, replacing condition with
6335 what makes the intent clearer. Remove handling of cases
6336 get_scalar_evolution now handles.
6338 2017-10-10 Jim Wilson <wilson@tuliptree.org>
6340 PR rtl-optimization/81434
6341 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
6342 comment for main loop. In sched_group_found if, also add checks for
6343 pass and min_cost_group.
6345 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
6347 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
6348 (rs6000_insn_cost): New function.
6349 * config/rs6000/rs6000.md (cost): New attribute.
6351 2017-10-10 Jakub Jelinek <jakub@redhat.com>
6352 H.J. Lu <hongjiu.lu@intel.com>
6356 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
6357 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
6358 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
6359 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
6360 to OPTION_MASK_ISA_AVX512VL - builtins that have both
6361 OPTION_MASK_ISA_MMX and some other bit set require both
6362 mmx and the ISAs without the mmx bit.
6363 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
6364 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
6365 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
6366 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
6367 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
6368 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
6369 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
6370 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
6371 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
6372 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
6373 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
6374 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
6375 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
6376 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
6377 Add OPTION_MASK_ISA_MMX.
6379 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
6381 * config.gcc (armv7*-*-freebsd*): New target.
6382 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
6384 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
6386 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
6387 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
6390 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
6392 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
6393 a template parameter.
6394 (WIDE_INT_REF_FOR): Update accordingly.
6395 * tree.h (wi::int_traits <const_tree>): Delete.
6396 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
6397 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
6398 (wi::tree_to_wide_ref): New typedef.
6399 (wi::to_wide): New function.
6400 * calls.c (get_size_range): Use wi::to_wide when operating on
6402 * cgraph.c (cgraph_node::create_thunk): Likewise.
6403 * config/i386/i386.c (ix86_data_alignment): Likewise.
6404 (ix86_local_alignment): Likewise.
6405 * dbxout.c (stabstr_O): Likewise.
6406 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
6407 * expr.c (const_vector_from_tree): Likewise.
6408 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
6409 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
6410 (fold_negate_expr_1, int_const_binop_1, const_binop)
6411 (fold_convert_const_int_from_real, optimize_bit_field_compare)
6412 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
6413 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
6414 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
6415 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
6416 (fold_not_const, round_up_loc): Likewise.
6417 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
6418 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
6419 (alloca_call_type): Likewise.
6420 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6421 * godump.c (go_output_typedef): Likewise.
6422 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6423 * internal-fn.c (get_min_precision): Likewise.
6424 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6425 * ipa-polymorphic-call.c
6426 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6427 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6428 (ipa_modify_call_arguments): Likewise.
6429 * match.pd: Likewise.
6430 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6431 * print-tree.c (print_node_brief, print_node): Likewise.
6432 * stmt.c (expand_case): Likewise.
6433 * stor-layout.c (layout_type): Likewise.
6434 * tree-affine.c (tree_to_aff_combination): Likewise.
6435 * tree-cfg.c (group_case_labels_stmt): Likewise.
6436 * tree-data-ref.c (dr_analyze_indices): Likewise.
6437 (prune_runtime_alias_test_list): Likewise.
6438 * tree-dump.c (dequeue_and_dump): Likewise.
6439 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6440 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6441 * tree-pretty-print.c (dump_generic_node): Likewise.
6442 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6443 (simple_iv_with_niters): Likewise.
6444 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6445 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6446 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6447 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6448 (refine_value_range_using_guard, number_of_iterations_ne_max)
6449 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6450 (get_cst_init_from_scev, record_nonwrapping_iv)
6451 (scev_var_range_cant_overflow): Likewise.
6452 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6453 * tree-ssa-pre.c (compute_avail): Likewise.
6454 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6455 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6456 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6457 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6458 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6459 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6460 (dump_case_nodes, try_switch_expansion): Likewise.
6461 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6462 (vect_do_peeling): Likewise.
6463 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6464 * tree-vect-stmts.c (vectorizable_load): Likewise.
6465 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6466 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6467 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6468 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6469 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6470 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6471 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6472 (evrp_dom_walker::before_dom_children): Likewise.
6473 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6474 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6475 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6476 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6477 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6478 (get_range_pos_neg): Likewise.
6479 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6480 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6481 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6482 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6483 * config/avr/avr.c (avr_fold_builtin): Likewise.
6484 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6485 * config/msp430/msp430.c (msp430_attr): Likewise.
6486 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6487 * config/powerpcspe/powerpcspe-c.c
6488 (altivec_resolve_overloaded_builtin): Likewise.
6489 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6490 (rs6000_expand_ternop_builtin): Likewise.
6491 * config/rs6000/rs6000-c.c
6492 (altivec_resolve_overloaded_builtin): Likewise.
6493 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6494 (rs6000_expand_ternop_builtin): Likewise.
6495 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6497 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6499 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6500 when copying loop nest with only one inner loop.
6502 2017-10-10 Richard Biener <rguenther@suse.de>
6504 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6505 blocks if SCEV is active.
6506 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6508 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6509 (scev_initialize): Assert we are not yet initialized.
6511 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6513 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6514 inner loop's exit stmt by making it always exit the loop, otherwise
6515 we would generate an infinite empty loop.
6517 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6519 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6520 renaming variables in new preheader if it's deleted.
6522 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6524 * tree-loop-distribution.c (struct partition): Remove unused field
6525 loops of the structure.
6526 (partition_alloc, partition_free): Ditto.
6527 (build_rdg_partition_for_vertex): Ditto.
6529 2017-10-09 Jeff Law <law@redhat.com>
6531 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6532 return type to match prototype and documentation.
6534 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6536 * config/rs6000/rs6000.c (processor_costs): Move to ...
6537 * config/rs6000/rs6000.h: ... here.
6538 (rs6000_cost): Declare.
6540 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6542 * except.c (setjmp_fn): New global variable.
6543 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6544 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6545 if DONT_USE_BUILTIN_SETJMP is defined.
6547 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6549 * target.def (insn_cost): New hook.
6550 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6551 * doc/tm.texi: Regenerate.
6552 * rtlanal.c (insn_cost): Use the new hook.
6554 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6556 * combine.c (combine_validate_cost): Compute the new insn_cost,
6557 not just pattern_cost.
6558 (try_combine): Adjust comment.
6560 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6562 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6564 * combine.c (uid_insn_cost): Adjust comment.
6565 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6567 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6568 * dse.c (find_shift_sequence): Ditto.
6569 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6570 (bb_valid_for_noce_process_p): Use pattern_cost.
6571 * rtl.h (insn_rtx_cost): Delete.
6572 (pattern_cost): New prototype.
6573 (insn_cost): New prototype.
6574 * rtlanal.c (insn_rtx_cost): Rename to...
6575 (pattern_cost): ... this.
6578 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6580 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6581 (*jcc<mode>_0_r_i387): Ditto.
6582 (*jccxf_r_i387): Ditto.
6583 (*jcc<mode>_r_i387): Ditto.
6584 (*jccu<mode>_r_i387): Ditto.
6585 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6586 (*jcc): Rename from *jcc_1.
6588 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6590 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6591 deferred rescans after the lvx/stvx recombination pre-pass.
6593 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6595 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6596 memory operation instruction support.
6597 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6598 (rs6000-ibm-aix[789]*): Likewise.
6599 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6600 Document new functions.
6602 2017-10-09 Richard Biener <rguenther@suse.de>
6604 PR tree-optimization/82397
6605 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6606 equality only for semantically equal trees.
6608 2017-10-09 Richard Biener <rguenther@suse.de>
6610 PR tree-optimization/82449
6611 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6612 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6613 allow constant addresses.
6614 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6617 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6619 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6622 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6625 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6628 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6631 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6633 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6636 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6637 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6639 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6641 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6642 (WI_BINARY_PREDICATE_RESULT): Likewise.
6643 (wi::binary_traits::operator_result): New type.
6644 (wi::binary_traits::predicate_result): Likewise.
6645 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6646 (generic_wide_int::operator==, generic_wide_int::operator!=)
6647 (generic_wide_int::operator&, generic_wide_int::and_not)
6648 (generic_wide_int::operator|, generic_wide_int::or_not)
6649 (generic_wide_int::operator^, generic_wide_int::operator+
6650 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6652 (operator~, unary operator-, operator==, operator!=, operator&)
6653 (operator|, operator^, operator+, binary operator-, operator*): New
6655 * expr.c (get_inner_reference): Use wi::bit_and_not.
6656 * fold-const.c (fold_binary_loc): Likewise.
6657 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6658 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6659 (bit_value_binop): Likewise.
6660 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6661 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6662 (extract_range_from_binary_expr_1): Likewise.
6663 (masked_increment): Likewise.
6664 (simplify_bit_ops_using_ranges): Likewise.
6666 2017-10-09 Martin Jambor <mjambor@suse.cz>
6669 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6670 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6671 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6673 (hsa_fixup_mov_insn_type): New function.
6674 (hsa_op_with_type::get_in_type): Use it.
6675 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6676 immediates in an assert.
6677 (hsa_op_with_type::extend_int_to_32bit): New method.
6678 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6679 types. Convert to dest type if necessary.
6680 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6681 (reg_for_gimple_ssa): Pass false as min32int to
6682 hsa_type_for_scalar_tree_type.
6683 (gen_hsa_addr): Fixup type when creating addresable temporary.
6684 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6685 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6686 necessary. Call hsa_fixup_mov_insn_type.
6687 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6688 extend operands and convert to dest type if necessary.
6689 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6690 to dest type if necessary.
6691 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6692 if conversion nt necessary and size matches.
6693 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6694 to dest type if necessary.
6695 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6696 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6698 (gen_hsa_clrsb): Likewise.
6699 (gen_hsa_ffs): Likewise.
6700 (gen_hsa_divmod): Extend operands and convert to dest type if
6702 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6704 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6706 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6707 Remove empty default arguments. Use a brace block as output
6709 (conditional return): Ditto.
6711 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6712 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6714 (group_ending_nop): Ditto.
6715 (doloop_end): Ditto.
6716 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6717 (splitters for those): Ditto.
6719 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6721 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6722 a conditional jump (and the compare for it) so that pc_rtx is the
6724 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6725 for the deleted and renamed ctr<mode>_internal[234] patterns.
6726 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6727 Delete second conditional return pattern.
6728 (ctr<mode>_internal2): Delete this second bdnz pattern.
6729 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6730 (ctr<mode>_internal4): Delete this second bdz pattern.
6732 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6734 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6735 (always_initialized_rtx_for_ssa_name_p): New predicate.
6736 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6737 (finish_out_of_ssa): Free new field of SA.
6738 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6739 * tree-ssa-coalesce.c: Include tree-ssa.h.
6740 (get_parm_default_def_partitions): Remove extern keyword.
6741 (get_undefined_value_partitions): New function.
6742 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6743 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6745 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6747 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6749 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6751 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6753 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6754 for targets that preffer 128bit.
6756 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6758 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6760 2017-10-08 Olivier Hainque <hainque@adacore.com>
6762 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6763 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6765 (thumb_set_return_address): Likewise.
6767 2017-10-08 Olivier Hainque <hainque@adacore.com>
6769 * common/config/arm/arm-common.c (arm_except_unwind_info):
6770 Handle DWARF2_UNWIND_INFO.
6772 2017-10-07 Michael Collison <michael.collison@arm.com>
6774 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6777 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6779 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6780 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6781 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6782 defined, force the creation of a new block for a dispatch label.
6784 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6786 * invoke.texi (Wsuggest-attribute=cold): Document.
6787 * common.opt (Wsuggest-attribute=cold): New
6788 * ipa-pure-const.c (warn_function_cold): New function.
6789 * predict.c (compute_function_frequency): Use it.
6790 * predict.h (warn_function_cold): Declare.
6792 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6794 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6797 2017-10-06 Martin Liska <mliska@suse.cz>
6799 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6800 keyword for member functions.
6801 (struct sanopt_tree_couple): New struct.
6802 (struct sanopt_tree_couple_hash): New function.
6803 (struct sanopt_ctx): Add new hash_map.
6804 (has_dominating_ubsan_ptr_check): New function.
6805 (record_ubsan_ptr_check_stmt): Likewise.
6806 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6807 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6808 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6810 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6813 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6814 aarch64_simd_valid_immediate on CONST_VECTORs.
6815 (aarch64_reg_or_bic_imm): Likewise.
6817 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6819 PR rtl-optimization/82396
6820 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6822 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6824 * graphite-dependences.c (scop_get_reads): Move code to...
6825 (scop_get_must_writes): Move code to...
6826 (scop_get_may_writes): Move code to...
6827 (scop_get_reads_and_writes): ... here.
6828 (scop_get_dependences): Call scop_get_reads_and_writes.
6830 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6832 PR tree-optimization/82434
6833 * fold-const.h (can_native_encode_type_p,
6834 can_native_encode_string_p): Remove.
6835 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6836 don't encode anything, just return what would be otherwise returned.
6837 (native_encode_fixed, native_encode_complex, native_encode_vector):
6839 (native_encode_string): Likewise. Inline by hand
6840 can_native_encode_string_p.
6841 (can_native_encode_type_p): Remove.
6842 (can_native_encode_string_p): Remove.
6843 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6844 STRING_CSTs using can_native_encode_string_p, test all
6845 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6846 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6847 argument from native_encode_expr.
6848 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6849 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6852 2017-10-06 Richard Biener <rguenther@suse.de>
6854 PR tree-optimization/82397
6855 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6856 operand_equal_p but rely on data_ref_compare_tree for detecting
6858 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6859 to match up with dr_group_sort_cmp.
6861 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6864 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6866 * config/s390/s390-builtin-types.def: Regenerate.
6868 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6871 * config/s390/s390-builtin-types.def: Regenerate.
6872 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6873 Change flag from B_VXE to B_VX.
6874 (s390_vec_min_dbl): Remove B_VXE flag.
6876 2017-10-06 Richard Biener <rguenther@suse.de>
6878 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6879 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6880 translate_isl_ast_to_gimple::is_valid_rename,
6881 translate_isl_ast_to_gimple::get_rename,
6882 translate_isl_ast_to_gimple::get_def_bb_for_const,
6883 translate_isl_ast_to_gimple::get_new_name,
6884 translate_isl_ast_to_gimple::collect_all_ssa_names,
6885 translate_isl_ast_to_gimple::copy_loop_phi_args,
6886 translate_isl_ast_to_gimple::collect_all_ssa_names,
6887 translate_isl_ast_to_gimple::copy_loop_phi_args,
6888 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6889 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6890 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6891 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6892 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6893 translate_isl_ast_to_gimple::copy_cond_phi_args,
6894 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6895 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6896 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6897 translate_isl_ast_to_gimple::rename_uses,
6898 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6899 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6900 (set_rename_for_each_def): Likewise.
6901 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6902 here. Handle rewriting SCEV analyzable uses here.
6903 (copy_bb_and_scalar_dependences): Generate code for PHI
6905 (graphite_regenerate_ast_isl): Adjust.
6906 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6907 (add_write, add_read): New functions.
6908 (build_cross_bb_scalars_def): Use it and simplify.
6909 (build_cross_bb_scalars_use): Likewise.
6910 (graphite_find_cross_bb_scalar_vars): Inline into...
6911 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6912 simulating out-of-SSA. Compute liveout and add dependencies.
6913 (build_scops): Force an empty entry block.
6914 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6916 (sese_build_liveouts): Declare.
6917 (sese_trivially_empty_bb_p): Likewise.
6918 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6919 compute liveout and debug_liveout.
6920 (sese_bad_liveouts_use): Remove.
6921 (sese_reset_debug_liveouts_bb): Likewise.
6922 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6923 (sese_build_liveouts): Build liveout and debug_liveout and store
6925 (new_sese_info): Adjust.
6926 (free_sese_info): Likewise.
6927 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6928 do not build liveout here.
6929 (move_sese_in_condition): Adjust region entry.
6930 (scev_analyzable_p): Match up with chrec_apply requirements.
6931 (sese_trivially_empty_bb_p): New.
6932 * tree-into-ssa.c (get_reaching_def): Properly support generating
6933 default-defs for incremental rewrite of anonymous names.
6935 2017-10-06 Richard Biener <rguenther@suse.de>
6937 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6938 precision do not perform modulo reduction.
6940 2017-10-06 Richard Biener <rguenther@suse.de>
6942 PR tree-optimization/82436
6943 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6944 conservatively choose the vectorization factor when checking
6945 whether we can perform the required load permutation.
6946 (vect_transform_slp_perm_load): Assert when we may not fail.
6948 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6950 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6951 message for incompatible -msdata=* and -mcall-* options.
6953 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6955 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6956 rate for post-reload scheduling.
6958 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6960 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6962 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6964 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6965 to improve monte carlo in scimark.
6967 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6969 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6970 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6971 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6972 pentium4_cost, nocona_cost): Set reassociation width to 1.
6973 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6974 width to 2 for fp operations and 1 otherwise.
6975 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6977 (atom_cost): Set reassociation width to 2.
6978 (slm_cost, generic_cost): Set fp reassociation width
6979 to 2 and 1 otherwise.
6980 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6981 (core_cost): Set fp reassociation width to 4 and vector to 2.
6982 (ix86_reassociation_width): Rewrite using cost table; special case
6983 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6984 and TARGET_AVX128_OPTIMAL.
6985 * config/i386/i386.h (processor_costs): Add
6986 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6987 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6988 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6989 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6990 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6991 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6993 2017-10-05 Nathan Sidwell <nathan@acm.org>
6995 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6997 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6999 * config/arm/arm.c (arm_test_fpu_data): New.
7000 (arm_run_selftests): Call arm_test_fpu_data.
7002 2017-10-04 Nathan Sidwell <nathan@acm.org>
7004 * toplev.c (toplev::main): Remove excess parens on pretty_printer
7006 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
7008 2017-10-04 Sudakshina Das <sudi.das@arm.com>
7010 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
7011 check type for aarch64_simd_valid_immediate.
7012 (aarch64_output_simd_mov_immediate): Update prototype.
7013 (aarch64_simd_valid_immediate): Update prototype.
7014 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
7015 support for ORR-immediate.
7016 (and<mode>3): modified pattern to add support for BIC-immediate.
7017 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
7018 now checks for valid immediate for BIC and ORR based on new enum
7020 (aarch64_output_simd_mov_immediate): Function now used to output
7021 BIC/ORR imm as well based on new enum argument.
7022 * config/aarch64/constraints.md (Do): New vector immediate constraint.
7024 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
7025 (aarch64_reg_or_bic_imm): Likewise.
7027 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7029 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
7030 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
7032 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7036 PR rtl-optimization/82396
7037 * haifa-sched.c (autopref_multipass_init): Simplify
7039 (autopref_rank_data): Simplify sort order.
7040 * sched-int.h (autopref_multipass_data_): Remove
7041 multi_mem_insn_p, min_offset and max_offset.
7043 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7045 * doc/sourcebuild.texi: Document vect_peeling_profitable.
7047 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7049 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
7052 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7054 * doc/sourcebuild.texi: Document vect_long_mult.
7056 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
7058 PR tree-optimization/82413
7059 * fold-const.c (build_range_check): Use widest_int when comparing
7060 the maximum ETYPE value with HIGH.
7062 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
7064 PR rtl-optimization/82396
7065 * haifa-sched.c (autopref_multipass_init): Simplify
7067 (autopref_rank_data): Simplify sort order.
7068 * sched-int.h (autopref_multipass_data_): Remove
7069 multi_mem_insn_p, min_offset and max_offset.
7071 2017-10-04 Jakub Jelinek <jakub@redhat.com>
7073 PR tree-optimization/82381
7074 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
7075 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
7078 PR tree-optimization/82374
7079 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
7080 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
7081 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
7082 current_function_decl to the new decl.
7084 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7086 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
7087 helper macro for IEEE float128 hardware built-in functions.
7088 (SQRTF128_ODD): Add built-in functions with the round-to-odd
7090 (TRUNCF128_ODD): Likewise.
7091 (ADDF128_ODD): Likewise.
7092 (SUBF128_ODD): Likewise.
7093 (MULF128_ODD): Likewise.
7094 (DIVF128_ODD): Likewise.
7095 (FMAF128_ODD): Likewise.
7096 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
7097 UNSPEC_TRUNC_ROUND_TO_ODD.
7098 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
7099 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
7100 floating point round to odd instructions.
7101 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
7102 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
7103 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
7104 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
7105 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
7106 (trunc<mode>sf2_hw): Change the truncate with round to odd
7107 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
7108 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
7109 to odd hardware instructions.
7110 (sub<mode>3_odd): Likewise.
7111 (mul<mode>3_odd): Likewise.
7112 (div<mode>3_odd): Likewise.
7113 (sqrt<mode>2_odd): Likewise.
7114 (fma<mode>4_odd): Likewise.
7115 (fms<mode>4_odd): Likewise.
7116 (nfma<mode>4_odd): Likewise.
7117 (nfms<mode>4_odd): Likewise.
7118 (trunc<mode>df2_odd): Change the truncate with round to odd
7119 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
7121 * doc/extend.texi (PowerPC built-in functions): Update documentation
7122 for existing IEEE float128-bit built-in functions. Add built-in
7123 functions that generate the IEEE 128-bit floating point round to
7126 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
7128 PR rtl-optimization/77729
7129 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
7130 to (X&(C1&~C2))|C2 transformations.
7132 2017-10-03 Martin Jambor <mjambor@suse.cz>
7134 PR tree-optimization/82363
7135 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
7136 mismatch, mark lacc written regardless of racc.
7138 2017-10-03 Jakub Jelinek <jakub@redhat.com>
7140 PR tree-optimization/82381
7141 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
7142 stmt_to_insert nor wheather SSA_NAMEs are default defs.
7143 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
7144 fallthrough into reassoc_stmt_dominates_stmt_p.
7147 * combine.c (combine_instructions): Don't combine in unreachable
7150 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
7153 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
7154 function to not use the have_cpu variable. Do not set cpu_index,
7155 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
7157 (rs6000_valid_attribute_p): Remove duplicate initializations of
7158 old_optimize and func_optimize.
7159 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
7160 (rs6000_activate_target_options): Make global.
7161 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
7164 2017-10-02 Jakub Jelinek <jakub@redhat.com>
7166 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
7167 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
7168 Set *poffset to 0 and *psize and *pmax_size to -1 if
7169 *poffset + *psize overflows in HOST_WIDE_INT.
7171 PR tree-optimization/82387
7172 PR tree-optimization/82388
7173 PR tree-optimization/82389
7174 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
7175 instead of live_bytes non-NULL.
7177 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
7180 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
7183 2017-10-02 Richard Biener <rguenther@suse.de>
7185 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
7186 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
7187 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
7189 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7191 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
7192 requested precision matches the type's.
7193 * calls.c (alloc_max_size): Calculate the new candidate size as
7194 a widest_int and use wi::to_widest when comparing it with the
7195 current candidate size.
7196 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
7197 zero rather than integer_zero_node.
7198 * match.pd: Check for a no-op conversion before using wi::add
7199 rather than after. Use tree_to_uhwi when summing small shift
7200 counts into an unsigned int.
7202 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
7203 Alan Hayward <alan.hayward@arm.com>
7204 David Sherwood <david.sherwood@arm.com>
7207 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
7208 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
7209 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
7210 POINTER_AND_FP_REGS.
7212 2017-10-02 Richard Biener <rguenther@suse.de>
7214 PR tree-optimization/82355
7215 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
7216 a mapping for the enclosing loop but avoid generating one for
7218 (copy_bb_and_scalar_dependences): Remove premature codegen
7219 error on PHIs in blocks duplicated into multiple places.
7220 * graphite-scop-detection.c
7221 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
7222 in the region use it as loop and nest to analyze the DR in.
7223 (try_generate_gimple_bb): Likewise.
7224 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
7225 (add_loop_constraints): For blocks in a loop not in the region
7226 create a dimension with a single iteration.
7227 * sese.h (gbb_loop_at_index): Remove assert.
7229 2017-10-01 Kevin Buettner <kevinb@redhat.com>
7231 * omp-expand.c (adjust_context_scope): New function.
7232 (expand_parallel_call): Call adjust_context_scope.
7234 2017-10-01 Jeff Law <law@redhat.com>
7236 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
7237 dom_opt_dom_walker class with direct access to private members.
7238 Add comments. Call test_for_singularity.
7239 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
7240 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
7241 m_dummy_cond anymore.
7242 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
7244 (pass_dominator:execute): Build the dummy_cond here and pass it
7245 to the dom_opt_dom_walker ctor.
7246 (test_for_singularity): New function.
7248 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
7249 Maya Rashish <coypu@sdf.org>
7251 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
7252 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
7253 (alpha*-*-netbsd*) Use nbsd_tm_file.
7254 (arm*-*-netbsdelf*) Likewise.
7255 (i[34567]86-*-netbsdelf*) Likewise.
7256 (x86_64-*-netbsd*) Likewise.
7257 (mips*-*-netbsd*) Likewise.
7258 (powerpc-*-netbsd*) Likewise.
7259 (sh*-*-netbsd*) Likewise.
7260 (sparc-*-netbsdelf*) Likewise.
7261 (sparc64-*-netbsd*) Likewise.
7262 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
7264 (vax-*-netbsdelf*) Likewise.
7265 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
7266 (UINT_FAST8_TYPE) Likewise.
7267 (INT_FAST16_TYPE) Check CHAR_FAST16.
7268 (UINT_FAST16_TYPE) Likewise.
7270 2017-09-30 Jakub Jelinek <jakub@redhat.com>
7273 * config/i386/i386.md
7274 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
7275 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
7276 *divmodsi4_zext_2): New define_insn_and_split.
7277 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
7278 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
7279 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
7280 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
7281 New define_insn_and_split.
7282 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
7283 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
7284 operands[1] having DImode when mode is SImode.
7286 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
7287 always SImode for DIV and MOD in REG_EQUAL notes.
7289 2017-09-29 Yury Gribov <tetra2005@gmail.com>
7292 * match.pd: Fix handling of NaNs in pattern.
7294 2017-09-29 Jeff Law <law@redhat.com>
7296 * sbitmap.c (bitmap_bit_in_range_p): New function.
7297 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
7298 * tree-ssa-dse.c (live_bytes_read): New function.
7299 (dse_classify_store): Ignore reads of dead bytes.
7301 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
7302 typos and whitespace errors.
7303 * config/i386/predicates.md (address_no_seg_operand): Likewise.
7304 * config/s390/s390.c (s390_emit_prologue): Likewise.
7306 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7309 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
7310 with a symbol for LRA.
7312 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
7314 PR rtl-optimization/82338
7315 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
7317 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
7319 * genmodes.c (calc_wider_mode): Suppress qsort macro.
7320 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
7321 (qsort_chk): Declare.
7322 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
7323 (qsort_chk): New function.
7325 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7327 PR tree-optimization/82337
7328 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
7329 phi definition if the PHI result appears in an abnormal PHI.
7330 (find_basis_for_base_expr): Don't record a basis if the LHS of the
7331 basis appears in an abnormal PHI.
7333 2017-09-29 Richard Biener <rguenther@suse.de>
7335 * graphite-isl-ast-to-gimple.c
7336 (translate_isl_ast_to_gimple::set_codegen_error): New function.
7337 (binary_op_to_tree): Use it.
7338 (get_rename_from_scev): Likewise.
7339 (copy_loop_phi_nodes): Likewise.
7340 (copy_bb_and_scalar_dependences): Likewise.
7341 (translate_pending_phi_nodes): Likewise.
7343 2017-09-29 Jakub Jelinek <jakub@redhat.com>
7346 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
7347 for movabsq $(i32 << shift), r64.
7349 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
7351 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
7352 index when encoding %esp as %rsp to avoid 0x67 prefix.
7354 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
7356 * config/i386/i386.md (*movsf_internal, *movdf_internal):
7357 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
7359 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
7361 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
7362 Extensions with more than 16 double VFP registers.
7363 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
7364 to_clear_mask and all code related to it. Replace the remaining
7365 entry by a sbitmap and adapt code accordingly.
7367 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
7369 * brig-builtins.def: Change pure attributes to const.
7371 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7373 * config.gcc (default_gnu_indirect_function): Default to yes for
7374 sparc*-*-linux* with glibc.
7376 2017-09-28 Joseph Myers <joseph@codesourcery.com>
7378 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
7379 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
7380 when creating .init_array and .fini_array sections with priority
7383 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
7386 * config/aarch64/aarch64.c
7387 (aarch64_builtin_support_vector_misalignment): Always return false
7388 when misalignment is unknown.
7390 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
7392 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
7393 this function to return false if the definition used by the swap
7394 instruction is artificial, or if the memory address from which the
7395 constant value is loaded is not represented by a base address held
7396 in a register or if the base address register is a frame or stack
7397 pointer. Additionally, return false if the base address of the
7398 loaded constant is a SYMBOL_REF but is not considered to be a
7400 (replace_swapped_load_constant): New function.
7401 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
7402 loaded constant vector with a load of a swapped constant vector.
7404 2017-09-27 Carl Love <cel@us.ibm.com>
7406 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
7407 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
7408 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
7411 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
7413 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
7414 first, always call autopref_rank_data otherwise.
7416 2017-09-27 Richard Biener <rguenther@suse.de>
7418 * graphite-scop-detection.c (find_scop_parameters): Move
7419 loop bound handling ...
7420 (gather_bbs::before_dom_children): ... here, avoiding the need
7421 to build scop_info->loop_nest.
7422 (record_loop_in_sese): Remove.
7423 * sese.h (sese_info_t::loop_nest): Remove.
7424 * sese.c (new_sese_info): Do not allocate loop_nest.
7425 (free_sese_info): Do not free loop_nest.
7427 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7430 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7431 lhs from calls if the lhs has addressable type.
7433 2017-09-27 Richard Biener <rguenther@suse.de>
7435 * graphite.h (scop::max_alias_set): New member.
7436 * graphite-scop-detection.c: Remove references to non-existing
7437 --param in comments.
7438 (build_alias_sets): Record the maximum alias set used for drs.
7439 (build_scops): Support zero as unlimited for
7440 --param graphite-max-arrays-per-scop.
7441 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7443 (build_poly_sr_1): ... here. Compute alias set based on the
7444 maximum alias set used for drs rather than
7445 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7447 2017-09-27 Richard Biener <rguenther@suse.de>
7449 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7450 --param loop-block-tile-size=0 to disable tiling.
7452 2017-09-27 Richard Biener <rguenther@suse.de>
7454 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7455 (graphite-max-nb-scop-params): Document special value zero.
7456 * domwalk.h (dom_walker::STOP): New symbolical constant.
7457 (dom_walker::dom_walker): Add optional parameter for bb to
7459 (dom_walker::~dom_walker): Declare.
7460 (dom_walker::before_dom_children): Document STOP return value.
7461 (dom_walker::m_user_bb_to_rpo): New member.
7462 (dom_walker::m_bb_to_rpo): Likewise.
7463 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7464 mapping here if not provided by the user.
7465 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7466 provided by the user.
7467 (dom_walker::STOP): Define.
7468 (dom_walker::walk): Do not compute bb to RPO mapping here.
7469 Support STOP return value from before_dom_children to stop
7471 * graphite-optimize-isl.c (optimize_isl): If the schedule
7472 is the same still generate code if -fgraphite-identity
7473 or -floop-parallelize-all are given.
7474 * graphite-scop-detection.c: Include cfganal.h.
7475 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7477 (gather_bbs::before_dom_children): Return STOP for BBs
7479 (build_scops): Compute bb to RPO mapping and pass it to
7480 the domwalk. Treat --param graphite-max-nb-scop-params=0
7481 as not limiting the number of params.
7482 * graphite.c (graphite_initialize): Remove limit on the
7483 number of basic-blocks in a function.
7484 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7485 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7486 default value of 10.
7488 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7490 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7491 Adjust code to eliminate needing to do the shift right 32-bits
7492 operation after XSCVDPSPN.
7494 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7496 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7497 ((X / Y) != 0 -> X >= Y): Likewise.
7499 2017-09-26 Carl Love <cel@us.ibm.com>
7501 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7502 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7503 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7504 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7505 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7506 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7507 definitions and overloading.
7508 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7509 statement for P9V_BUILTIN_XST_LEN_R.
7510 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7511 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7512 define_expand and define_insn for the instructions and builtins.
7513 * doc/extend.texi: Update the built-in documentation file for the new
7515 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7516 define_insn for the instructions
7518 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7521 * gcc/config/netbsd-protos.h: New file.
7522 * gcc/config/netbsd.c: New file.
7523 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7524 * gcc/config/t-netbsd: New file.
7525 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7526 (tmake_file) Add t-netbsd.
7527 (extra_objs) Add netbsd.o.
7529 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7533 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7535 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7537 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7538 sign extension from a vector register to a GPR by doing a 32-bit
7539 direct move and then an EXTSW.
7540 (extendsi<mode>2 splitter): Likewise.
7541 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7542 right or vector extract after doing XSCVDPSPN. Use
7543 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7545 (movdi_from_sf_zero_ext): Likewise.
7546 (reload_gpr_from_vsxsf): Likewise.
7547 (p8_mfvsrd_4_disf): Delete, no longer used.
7548 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7549 then needing to move the SFmode to a GPR to use the XSCVDPSP
7550 instruction instead of FRSP and XSCVDPSPN.
7551 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7552 it is adjacent to the other XSCVSPDP insns.
7553 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7554 SFmode to be in traditional Altivec registers.
7555 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7556 (vsx_xscvspdpn): Likewise.
7557 (vsx_xscvspdpn_scalar): Likewise.
7559 2017-09-26 Martin Jambor <mjambor@suse.cz>
7561 * tree-sra.c (compare_access_positions): Put integral types first,
7562 stabilize sorting of integral types, remove conditions putting
7563 non-full-precision integers last.
7564 (sort_and_splice_var_accesses): Disable scalarization if a
7565 non-integert would be represented by a non-full-precision integer.
7567 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7569 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7570 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7571 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7572 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7573 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7574 conditionals inside the function instead of around it. Call
7575 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7576 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7578 2017-09-26 Richard Biener <rguenther@suse.de>
7580 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7582 (scop_detection::build_scop_breadth): ... this. Removed.
7583 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7584 (scop_detection::harmful_stmt_in_bb): Likewise.
7585 (scop_detection::graphite_can_represent_stmt): Likewise.
7586 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7587 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7588 (scop_detection::can_represent_loop_1): ... this. Removed.
7589 (scop_detection::harmful_loop_in_region): Simplify after inlining
7590 the above and remove more quadraticness.
7591 (build_scops): Adjust.
7592 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7595 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7598 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7599 REGNO (base) == SP_REG if base is a REG.
7602 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7603 if it is different SSA_NAME.
7604 (optimize_range_tests_cmp_bitwise): New function.
7605 (optimize_range_tests): Call it.
7607 2017-09-26 Richard Biener <rguenther@suse.de>
7609 PR tree-optimization/82321
7610 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7611 for the def being inside the loop.
7613 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7615 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7617 * config/s390/s390-builtins.def: Fix constraint on op4.
7619 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7621 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7622 independent expanders.
7623 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7624 ("vec_ordered", "vec_unordered"): New expanders.
7626 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7628 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7631 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7633 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7634 vec_unpacks_lo_v16qi.
7635 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7637 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7639 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7640 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7641 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7643 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7645 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7647 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7649 ("*vec_slb<mode>"): New insn pattern.
7650 ("vec_shr_<mode>"): New expander.
7651 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7652 and force the shift count operand to V16QImode.
7653 ("vec_srb<mode>"): Set shift count mode to V16QI.
7655 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7657 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7658 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7659 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7661 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7664 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7666 2017-09-26 Richard Biener <rguenther@suse.de>
7668 PR tree-optimization/82320
7669 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7672 2017-09-25 Jeff Law <law@redhat.com>
7674 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7675 prototype for new argument.
7676 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7677 mostly extracted from rs6000_emit_allocate_stack.
7678 (rs6000_emit_probe_stack_range_stack_clash): New function.
7679 (rs6000_emit_allocate_stack): Call
7680 rs6000_emit_probe_stack_range_stack_clash as needed.
7681 (rs6000_emit_probe_stack_range): Add additional argument
7682 to call to gen_probe_stack_range{si,di}.
7683 (output_probe_stack_range): New.
7684 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7685 (output_probe_stack_range_stack_clash): New.
7686 (rs6000_emit_prologue): Emit notes into dump file as requested.
7687 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7688 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7689 Add additional operand and pass it to output_probe_stack_range.
7691 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7693 PR tree-optimization/82163
7694 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7695 (checking_verify_loop_closed_ssa): New parameter.
7696 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7697 (check_loop_closed_ssa_stmt): Delete.
7698 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7699 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7700 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7703 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7705 * brig-builtins.def: Treat HSAIL barrier builtins as
7706 setjmp/longjump style functions.
7708 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7710 * target.def (constant_alignment): New hook.
7711 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7712 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7713 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7714 * doc/tm.texi: Regenerate.
7715 * targhooks.h (default_constant_alignment): Declare.
7716 (constant_alignment_word_strings): Likewise.
7717 * targhooks.c (default_constant_alignment): New function.
7718 (constant_alignment_word_strings): Likewise.
7719 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7720 instead of CONSTANT_ALIGNMENT.
7721 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7722 (force_const_mem): Likewise.
7723 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7724 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7725 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7726 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7727 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7729 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7730 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7731 constant_alignment_word_strings.
7732 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7733 (CONSTANT_ALIGNMENT): Likewise.
7734 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7735 (arm_constant_alignment): New function.
7736 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7737 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7738 constant_alignment_word_strings.
7739 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7740 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7741 constant_alignment_word_strings.
7742 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7743 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7744 (cris_constant_alignment): New function.
7745 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7746 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7747 (epiphany_constant_alignment): New function.
7748 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7749 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7750 constant_alignment_word_strings.
7751 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7752 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7753 constant_alignment_word_strings.
7754 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7755 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7756 constant_alignment_word_strings.
7757 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7758 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7759 * config/i386/i386.c (ix86_constant_alignment): Make static.
7760 Use the same interface as the target hook.
7761 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7762 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7763 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7764 constant_alignment_word_strings.
7765 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7766 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7767 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7768 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7769 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7770 constant_alignment_word_strings.
7771 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7772 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7773 constant_alignment_word_strings.
7774 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7775 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7776 constant_alignment_word_strings.
7777 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7778 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7780 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7781 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7782 * config/mips/mips.c (mips_constant_alignment): New function.
7783 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7784 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7785 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7786 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7787 (mmix_constant_alignment): Make static. Use the same interface
7789 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7790 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7791 constant_alignment_word_strings.
7792 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7793 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7794 constant_alignment_word_strings.
7795 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7796 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7797 constant_alignment_word_strings.
7798 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7799 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7800 (rs6000_constant_alignment): New function.
7801 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7802 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7803 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7804 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7805 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7806 (rs6000_constant_alignment): New function.
7807 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7808 * config/s390/s390.c (s390_constant_alignment): New function.
7809 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7810 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7811 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7812 constant_alignment_word_strings.
7813 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7814 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7815 (sparc_constant_alignment): New function.
7816 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7817 * config/spu/spu.c (spu_constant_alignment): New function.
7818 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7819 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7820 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7821 constant_alignment_word_strings.
7822 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7823 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7824 constant_alignment_word_strings.
7825 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7826 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7827 constant_alignment_word_strings.
7828 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7829 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7830 (visium_constant_alignment): New function.
7831 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7832 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7833 (xtensa_constant_alignment): New function.
7834 * system.h (CONSTANT_ALIGNMENT): Poison.
7836 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7838 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7839 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7840 (rs6000_builtin_valid_without_lhs): New helper function.
7841 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7842 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7844 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7846 * target.h (vec_perm_indices): Use unsigned short rather than
7848 (auto_vec_perm_indices): Likewise.
7849 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7850 Use unsigned int rather than unsigned char.
7851 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7853 2017-09-25 Richard Biener <rguenther@suse.de>
7855 * cfgloop.h (sort_sibling_loops): Declare.
7856 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7857 (sort_sibling_loops): New function sorting the sibling loop list
7859 * graphite.c (graphite_transform_loops): Sort sibling loops.
7861 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7863 * target.def (vec_perm_const_ok): Change sel parameter to
7865 * optabs-query.c (can_vec_perm_p): Update accordingly.
7866 * doc/tm.texi: Regenerate.
7867 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7868 auto_vec_perm_indices and remove separate nelt field.
7869 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7870 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7871 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7872 (aarch64_expand_vec_perm_const): Update accordingly.
7873 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7874 to vec_perm_indices.
7875 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7876 auto_vec_perm_indices and remove separate nelt field.
7877 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7878 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7879 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7881 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7882 to vec_perm_indices.
7883 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7884 sel to vec_perm_indices.
7885 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7886 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7887 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7889 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7892 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7895 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7896 on the FUNCTION_DECL function context if it has a DIE that is a
7899 2017-09-25 Richard Biener <rguenther@suse.de>
7901 PR tree-optimization/82285
7902 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7905 2017-09-25 Tom de Vries <tom@codesourcery.com>
7909 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7912 2017-09-25 Richard Biener <rguenther@suse.de>
7914 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7915 ISL errors other than isl_error_quota happen. Dump if the
7916 schedule is the same.
7917 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7918 errors instead of aborting inside ISL.
7920 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7923 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7924 of libgcc_eh for m64.
7925 * config/i386/darwin64.h: Likewise.
7927 2017-09-25 Richard Biener <rguenther@suse.de>
7930 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7931 attribute for incomplete types nor twice for complete ones.
7933 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7936 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7937 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7940 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7943 * config/i386/i386.opt (mprefer-avx256): Use
7944 ix86_target_flags variable.
7945 * config/i386/i386.c (ix86_target_string): Move
7946 -mprefer-avx256 to flag2_opts.
7948 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7951 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7952 and x != -1 | y != -1 into (x & y) != -1.
7954 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7956 * config.gcc: Add new case statement to set
7957 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7958 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7959 s390x-*-linux* case statements. Added aarch64 to the list of
7960 supported architectures.
7962 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7964 PR tree-optimization/82289
7965 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7966 STMT_VINFO_RELEVANT_P.
7968 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7969 Alan Hayward <alan.hayward@arm.com>
7970 David Sherwood <david.sherwood@arm.com>
7972 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7973 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7974 (extract_range_from_binary_expr_1): Don't call
7975 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7977 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7978 Alan Hayward <alan.hayward@arm.com>
7979 David Sherwood <david.sherwood@arm.com>
7981 * target.def (preferred_vector_alignment): New hook.
7982 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7984 * doc/tm.texi: Regenerate.
7985 * targhooks.h (default_preferred_vector_alignment): Declare.
7986 * targhooks.c (default_preferred_vector_alignment): New function.
7987 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7989 (DR_TARGET_ALIGNMENT): New macro.
7990 (aligned_access_p): Update commentary.
7991 (vect_known_alignment_in_bytes): New function.
7992 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7994 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7995 Calculate the misalignment based on the target alignment rather than
7997 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7998 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7999 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
8000 the target alignment, rather than masking the element misalignment
8001 with the number of elements in a vector. Also use the target
8002 alignment when calculating the maximum number of peels.
8003 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
8004 instead of TYPE_ALIGN_UNIT.
8005 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
8006 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
8007 (vect_create_addr_base_for_vector_ref): Update call accordingly.
8008 (vect_create_data_ref_ptr): Likewise.
8009 (vect_setup_realignment): Realign by ANDing with
8010 -DR_TARGET_MISALIGNMENT.
8011 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
8012 the number of peels based on DR_TARGET_ALIGNMENT.
8013 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
8014 with the guaranteed alignment boundary when deciding whether
8016 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
8017 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
8018 (ensure_base_align): Remove stmt_info parameter. Get the
8019 target base alignment from DR_TARGET_ALIGNMENT.
8020 (vectorizable_store): Update call accordingly. Interpret
8021 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
8023 (vectorizable_load): Likewise.
8025 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
8026 Alan Hayward <alan.hayward@arm.com>
8027 David Sherwood <david.sherwood@arm.com>
8029 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
8030 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
8031 (vect_enhance_data_refs_alignment): Likewise.
8033 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8035 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8036 error. Only quit immediately if parsing is complete.
8037 (BEGIN): Initialize fatal_err and parse_done.
8038 (begin fpu, end fpu): Check number of arguments.
8039 (begin arch, end arch): Likewise.
8040 (begin cpu, end cpu): Likewise.
8041 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8042 (optalias): Likewise.
8044 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
8046 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8047 * config/arm/arm-isa.h: Delete. Move definitions to ...
8048 * arm-cpus.in: ... here. Use new feature and fgroup values.
8049 * config/arm/arm.c (arm_option_override): Use lower case for feature
8051 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8052 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8053 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8055 (print_isa_bits_for): New function.
8056 (gen_isa): New function.
8057 (gen_comm_data): Use print_isa_bits_for.
8058 (define feature): New keyword.
8059 (define fgroup): New keyword.
8060 * config/arm/t-arm (TM_H): Remove.
8061 (GTM_H): Add arm-isa.h.
8062 (arm-isa.h): Add rule to generate file.
8063 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8064 case for feature bit names.
8066 2017-09-22 Richard Biener <rguenther@suse.de>
8068 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
8070 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
8071 print of no dependency loops ...
8072 * graphite.c (graphite_transform_loops): ... here.
8073 (canonicalize_loop_closed_ssa_form): Work from inner to outer
8075 (same_close_phi_node, remove_duplicate_close_phi,
8076 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
8077 (canonicalize_loop_closed_ssa): ... here and simplify.
8078 * graphite-optimize-isl.c: Include tree-vectorizer.h.
8079 (optimize_isl): Use dump_printf_loc to tell when we stopped
8080 optimizing because of an ISL timeout.
8082 2017-09-22 Richard Biener <rguenther@suse.de>
8084 PR tree-optimization/82291
8085 * tree-if-conv.c (predicate_mem_writes): Make sure to
8086 remove writes in blocks predicated with false.
8088 2017-09-22 Richard Biener <rguenther@suse.de>
8090 * sese.c: Include cfganal.h.
8091 (if_region_set_false_region): Remove.
8092 (create_if_region_on_edge): Likewise.
8093 (move_sese_in_condition): Re-implement without destroying
8096 2017-09-22 Richard Biener <rguenther@suse.de>
8098 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
8099 Verify both BBs contain loop PHI nodes before dispatching to
8101 (graphite_regenerate_ast_isl): Do not recompute dominators,
8102 do not verify three times. Restructure for clarity.
8103 * graphite-scop-detection.c (same_close_phi_node,
8104 remove_duplicate_close_phi, make_close_phi_nodes_unique,
8105 defined_in_loop_p, canonicalize_loop_closed_ssa,
8106 canonicalize_loop_closed_ssa_form): Simplify, remove excess
8107 checking and SSA rewrite, move to ...
8108 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
8109 (graphite_initialize): Do not pass in ctx, do not reset the
8110 SCEV cache, compute only dominators.
8111 (graphite_transform_loops): Allocate ISL ctx after
8112 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
8113 Maintain post-dominators only around build_scops.
8114 * sese.c (if_region_set_false_region): Make static. Free
8115 and recompute dominators.
8116 (move_sese_in_condition): Assert we don't get called with
8117 post-dominators computed.
8118 * sese.h (if_region_set_false_region): Remove.
8120 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
8122 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
8123 mode attribute for TARGET_AVX512VL.
8125 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
8127 * config/i386/i386.opt (mprefer-avx256): New option.
8128 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
8130 (ix86_preferred_simd_mode): Return 256-bit AVX modes
8131 for TARGET_PREFER_AVX256.
8132 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
8134 2017-09-21 Jeff Law <law@redhat.com>
8136 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
8137 Fix dump output if the only stack space is for pushed registers.
8139 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8141 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
8144 2017-09-21 Martin Sebor <msebor@redhat.com>
8147 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
8148 code (in C++) or code that triggers warnings.
8150 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
8152 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
8154 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
8156 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
8157 * sched-rgn.c: Ditto.
8158 * sel-sched-ir.c: Ditto.
8160 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
8162 * toplev.h (set_random_seed): Adjust return type.
8163 * toplev.c (init_local_tick): Move eager initialization of random_seed
8164 to get_random_seed. Adjust comment.
8165 (init_random_seed): Inline to get_random_seed, delete.
8166 (get_random_seed): Initialize random_seed lazily.
8167 (set_random_seed): Do not return previous value.
8168 (print_switch_value): Do not call get_random_seed.
8170 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
8172 * cgraph.c (delete_function_version): New, broken out from...
8173 (cgraph_node::delete_function_version): ...here. Rename to
8174 cgraph_node::delete_function_version_by_decl. Update all uses.
8175 (cgraph_node::remove): Call delete_function_version.
8177 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8180 * tree-inline.c (expand_call_inline): Emit clobber stmts for
8181 VAR_DECLs to which addressable non-volatile parameters are mapped
8182 and for id->retvar after the return value assignment. Clear
8183 id->retval and id->retbnd after inlining.
8185 2017-09-21 Richard Biener <rguenther@suse.de>
8187 PR tree-optimization/82276
8188 PR tree-optimization/82244
8189 * tree-vrp.c (build_assert_expr_for): Set
8190 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
8192 (remove_range_assertions): Revert earlier change.
8194 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
8197 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
8199 2017-09-21 Richard Biener <rguenther@suse.de>
8201 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
8202 Restore valid IL after code generation errors.
8203 * graphite.c (graphite_transform_loops): Diagnose code
8204 generation issues as MSG_MISSED_OPTIMIZATION and continue
8205 with processing SCOPs.
8207 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8208 Alan Hayward <alan.hayward@arm.com>
8209 David Sherwood <david.sherwood@arm.com>
8211 * calls.c (compute_argument_addresses): Use simplify_gen_binary
8212 rather than choosing between plus_constant and gen_rtx_<CODE>.
8213 * expr.c (emit_push_insn): Likewise.
8214 (expand_expr_real_2): Likewise.
8216 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8217 Alan Hayward <alan.hayward@arm.com>
8218 David Sherwood <david.sherwood@arm.com>
8220 * loop-unroll.c (split_iv): Call copy_rtx on the step.
8222 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8223 Alan Hayward <alan.hayward@arm.com>
8224 David Sherwood <david.sherwood@arm.com>
8226 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
8227 calling tree_to_uhwi.
8229 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8230 Alan Hayward <alan.hayward@arm.com>
8231 David Sherwood <david.sherwood@arm.com>
8233 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
8234 INTEGER_CST rather than a negative test for ADDR_EXPR.
8236 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
8237 Alan Hayward <alan.hayward@arm.com>
8238 David Sherwood <david.sherwood@arm.com>
8240 * tree-vrp.c (extract_range_from_binary_expr_1): Check
8241 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
8243 2017-09-21 Richard Biener <rguenther@suse.de>
8245 PR tree-optimization/71351
8246 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
8247 graphite_create_new_loop_guard): Remove, fold remaining parts
8249 (translate_isl_ast_node_for): ... here and simplify.
8251 2017-09-21 Jakub Jelinek <jakub@redhat.com>
8254 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
8255 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
8256 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
8257 alternative always use QI mode, for -Os imov (=R,R) alternative
8258 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
8261 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8262 Jeff Law <law@redhat.com>
8264 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
8265 (allocate_stack_space): New function, partially extracted from
8267 (s390_emit_prologue): Track offset to most recent stack probe.
8268 Code to allocate space moved into allocate_stack_space.
8269 Dump actions when no stack is allocated.
8270 (s390_prologue_plus_offset): New function.
8271 (s390_emit_stack_probe): Likewise.
8273 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
8275 * common.opt (Wa, Wl, Wp, g, gz=): Add
8277 (gno-column-info): Remove.
8278 (gcolumn-info): Drop RejectNegative.
8280 (gno-record-gcc-switches): Remove.
8281 (grecord-gcc-switches): Drop RejectNegative.
8282 (gno-split-dwarf): Remove.
8283 (gsplit-dwarf): Drop RejectNegative.
8284 (gno-strict-dwarf): Remove.
8285 (gstrict-dwarf): Drop RejectNegative.
8286 * config/darwin.opt (gfull, gused): Add RejectNegative.
8287 * dwarf2out.c (gen_producer_string): Drop
8288 gno-record-gcc-switches handler.
8289 * optc-gen.awk: Add g to prefixes with negative forms.
8290 * opts-common.c (remapping_prefix_p): New.
8291 (find_opt): Check it.
8292 (generate_canonical_option): Test g prefix.
8293 (option_map): Add -gno- mapping.
8294 (add_misspelling_candidates): Check remapping_prefix_p.
8296 2017-09-20 Jeff Law <law@redhat.com>
8298 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
8299 thinko in stack clash protection support.
8301 * explow.c (compute_stack_clash_protection_loop_data): Use
8302 CONST_INT_P instead of explicit test. Verify object is a
8303 CONST_INT_P before looking at INTVAL.
8304 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
8305 instead of explicit test.
8307 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8310 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
8311 address instead of to r1 and r11.
8313 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
8315 * config.gcc: Support "knm".
8316 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
8317 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
8319 * config/i386/i386.c (m_KNM): Define.
8320 (processor_target_table): Add "knm".
8322 (ix86_option_override_internal): Add "knm".
8323 (ix86_issue_rate): Add PROCESSOR_KNM.
8324 (ix86_adjust_cost): Ditto.
8325 (ia32_multipass_dfa_lookahead): Ditto.
8326 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
8327 (fold_builtin_cpu): Add M_INTEL_KNM.
8328 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
8329 (processor_type): Add PROCESSOR_KNM.
8330 * config/i386/x86-tune.def: Add m_KNM.
8331 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
8333 2017-09-20 Richard Biener <rguenther@suse.de>
8335 PR tree-optimization/80213
8336 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
8337 are allowed in empty BBs as well.
8338 (canonicalize_loop_closed_ssa): Also look for other complex
8340 (scop_detection::get_sese): Include the loop-closed PHI block
8342 (scop_detection::merge_sese): Remove code adding extra blocks.
8343 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
8344 (build_scops): Assert the final returned scop is invalid.
8346 2017-09-20 Richard Biener <rguenther@suse.de>
8348 PR tree-optimization/82264
8349 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
8351 (vn_phi_lookup): Likewise.
8352 (vn_phi_insert): Likewise.
8354 2017-09-20 Jakub Jelinek <jakub@redhat.com>
8356 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
8357 that fits into uhwi or shwi, add DW_AT_const_value regardless
8358 of early_dwarf without going through RTL, using add_AT_unsigned
8361 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
8362 (DEBUG_LTO_ABBREV_SECTION): Likewise.
8363 (DEBUG_LTO_MACINFO_SECTION): Likewise.
8364 (DEBUG_MACRO_SECTION): Likewise.
8365 (DEBUG_LTO_MACRO_SECTION): Likewise.
8366 (DEBUG_STR_DWO_SECTION): Likewise.
8367 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
8368 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
8369 (DEBUG_LTO_DWO_LINE_SECTION): Define.
8370 (DEBUG_LTO_LINE_STR_SECTION): Define.
8371 (init_sections_and_labels): Initialize debug_line_str_section
8372 variable. Initialize debug_loc_section for -gdwarf-5 to
8373 DEBUG_LOCLISTS_SECTION. Formatting fixes.
8375 2017-09-20 Richard Biener <rguenther@suse.de>
8377 * graphite-sese-to-poly.c (extract_affine): Properly handle
8378 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
8380 2017-09-20 Richard Biener <rguenther@suse.de>
8382 PR tree-optimization/81373
8383 * graphite-scop-detection.c (build_cross_bb_scalars_def):
8384 Force SESE live-out defs to be handled even if they are
8387 2017-09-19 Jeff Law <law@redhat.com>
8389 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
8390 nothing for stack adjustments with REG_STACK_CHECK.
8391 * sched-deps.c (parse_add_or_inc): Reject insns with
8392 REG_STACK_CHECK from dependency breaking.
8393 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
8394 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
8395 * reg-notes.def (STACK_CHECK): New note.
8397 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
8398 (ix86_expand_prologue): Dump stack clash info as needed.
8399 Call ix86_adjust_stack_and_probe_stack_clash as needed.
8401 * function.c (dump_stack_clash_frame_info): New function.
8402 * function.h (dump_stack_clash_frame_info): Prototype.
8403 (enum stack_clash_probes): New enum.
8405 * config/alpha/alpha.c (alpha_expand_prologue): Also check
8406 flag_stack_clash_protection.
8407 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
8408 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
8409 (arm_frame_pointer_required): Likewise.
8410 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
8411 (ia64_expand_prologue): Likewise.
8412 * config/mips/mips.c (mips_expand_prologue): Likewise.
8413 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
8414 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8415 (sparc_flat_expand_prologue): Likewise.
8416 * config/spu/spu.c (spu_expand_prologue): Likewise.
8418 * explow.c: Include "params.h".
8419 (anti_adjust_stack_and_probe_stack_clash): New function.
8420 (get_stack_check_protect): Likewise.
8421 (compute_stack_clash_protection_loop_data): Likewise.
8422 (emit_stack_clash_protection_loop_start): Likewise.
8423 (emit_stack_clash_protection_loop_end): Likewise.
8424 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8425 Use anti_adjust_stack_and_probe_stack_clash.
8426 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8427 (emit_stack_clash_protection_loop_start): Likewise.
8428 (emit_stack_clash_protection_loop_end): Likewise.
8429 * rtl.h (get_stack_check_protect): Prototype.
8430 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8431 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8432 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8434 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8436 * doc/tm.texi: Rebuilt.
8437 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8438 get_stack_check_protect.
8439 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8440 * config/arm/arm.c (arm_expand_prologue): Likewise.
8441 (arm_frame_pointer_required): Likewise.
8442 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8443 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8444 * config/mips/mips.c (mips_expand_prologue): Likewise.
8445 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8446 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8447 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8448 (sparc_flat_expand_prologue): Likewise.
8450 * common.opt (-fstack-clash-protection): New option.
8451 * flag-types.h (enum stack_check_type): Note difference between
8452 -fstack-check= and -fstack-clash-protection.
8453 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8454 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8455 * toplev.c (process_options): Issue warnings/errors for cases
8456 not handled with -fstack-clash-protection.
8457 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8458 (-fstack-check): Note additional problem with -fstack-check=generic.
8459 Note that -fstack-check is primarily for Ada and refer users
8460 to -fstack-clash-protection for stack-clash-protection.
8461 Document new params for stack clash protection.
8463 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8465 * config/i386/i386.c (ix86_split_long_move): Do not handle
8466 address used for LEA in a special way.
8468 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8470 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8472 2017-09-19 Martin Sebor <msebor@redhat.com>
8475 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8476 of incompatible types.
8478 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8480 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8481 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8482 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8483 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8485 2017-09-19 Richard Biener <rguenther@suse.de>
8487 PR tree-optimization/82244
8488 * tree-vrp.c (remove_range_assertions): Do not propagate
8489 a constant to abnormals but replace the assert with a copy.
8491 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8493 PR rtl-optimization/57878
8494 PR rtl-optimization/68988
8495 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8496 avoidance test involving non_reload_pseudos. Move frequency test
8497 below the general fragmentation avoidance test.
8499 2017-09-19 Richard Biener <rguenther@suse.de>
8501 PR tree-optimization/69728
8502 * graphite-sese-to-poly.c (schedule_error): New global.
8503 (add_loop_schedule): Handle empty domain by failing the
8505 (build_original_schedule): Handle schedule_error.
8507 2017-09-19 Richard Biener <rguenther@suse.de>
8509 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8510 Do not iterate to sibling loops but only to siblings of inner
8513 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8516 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8517 that the registers are different.
8519 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8521 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8522 to processor_model and "amdfam17h" to arch_names_table.
8523 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8525 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8528 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8530 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8531 Alan Hayward <alan.hayward@arm.com>
8532 David Sherwood <david.sherwood@arm.com>
8534 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8536 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8538 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8539 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8540 vect_schedule_slp_instance.
8541 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8542 Update call to vect_slp_analyze_node_operations. Simplify return
8544 (vect_slp_analyze_bb_1): Update call accordingly.
8545 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8546 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8547 (vect_schedule_slp): Update call accordingly.
8549 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8550 Alan Hayward <alan.hayward@arm.com>
8551 David Sherwood <david.sherwood@arm.com>
8553 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8554 with types that aren't in fact scalar.
8556 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8558 * tree-vect-slp.c (vect_record_max_nunits): New function,
8560 (vect_build_slp_tree_1): ...here.
8561 (vect_build_slp_tree_2): Call it for phis too.
8563 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8565 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8566 to vect_get_vec_def_for_operand when getting the mask operand.
8568 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8569 Alan Hayward <alan.hayward@arm.com>
8570 David Sherwood <david.sherwood@arm.com>
8572 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8575 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8576 Alan Hayward <alan.hayward@arm.com>
8577 David Sherwood <david.sherwood@arm.com>
8579 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8580 calculation for vector booleans.
8582 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8583 Alan Hayward <alan.hayward@arm.com>
8584 David Sherwood <david.sherwood@arm.com>
8586 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8588 (vect_transform_stmt): ...here.
8589 (vect_analyze_stmt): Use it instead of calling
8590 vectorizable_live_operation directly.
8592 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8594 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8595 non-SIMT targets in acc vector loops.
8597 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8599 * configure.ac: Add arc and check if assembler supports gdwarf2.
8600 * configure: Regenerate.
8602 2017-09-18 Richard Biener <rguenther@suse.de>
8604 PR tree-optimization/82220
8605 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8606 epilogue niters from the min_profitable_iters compute.
8608 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8611 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8612 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8613 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8615 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8618 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8619 switching to a new text section.
8621 2017-09-18 Richard Biener <rguenther@suse.de>
8623 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8625 (build_alias_set): Reject aliases with no access function.
8627 2017-09-18 Richard Biener <rguenther@suse.de>
8629 PR tree-optimization/79622
8630 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8632 (build_cross_bb_scalars_use): Likewise.
8634 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8636 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8638 2017-09-18 Alan Modra <amodra@gmail.com>
8641 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8642 stack_pointer_rtx for count 0. Update comments. Break up
8643 large rtl expression.
8645 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8647 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8648 Increase to 20 bytes.
8649 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8650 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8651 or avx version of the stub.
8653 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8656 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8657 compute the minimum stack alignment. Also update preferred stack
8658 boundary for leaf functions.
8660 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8662 PR tree-optimization/82228
8663 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8666 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8668 * common/config/nds32/nds32-common.c
8669 (nds32_option_optimization_table): Refine formatting.
8670 (nds32_option_optimization_table): Use -fsched-pressure and
8671 -fomit-frame-pointer for specific optimization level.
8673 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8675 * config/nds32/nds32.c: Refine formatting and comments.
8676 * config/nds32/nds32.h: Likewise.
8677 * config/nds32/nds32.md: Likewise.
8678 * config/nds32/nds32-cost.c: Likewise.
8679 * config/nds32/nds32-isr.c: Likewise.
8680 * config/nds32/nds32-md-auxiliary.c: Likewise.
8681 * config/nds32/nds32-multiple.md: Likewise.
8682 * config/nds32/nds32-predicates.c: Likewise.
8684 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8685 Jakub Jelinek <jakub@redhat.com>
8687 Add support for -std=c++2a.
8688 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8690 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8692 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8695 * doc/extend.texi (Common Function Attributes): Add
8696 references to ARM, AArch64, and S/390 specific attributes.
8697 (Function Specific Option Pragmas): Add AArch64 and S/390
8698 to list of back ends that support the target pragma.
8700 2017-09-15 Nathan Sidwell <nathan@acm.org>
8702 * doc/standards.texi: Fix C++17 description. Update URLs for
8705 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8707 * common.opt (Wcast-align=strict): New warning option.
8708 * doc/invoke.texi: Document -Wcast-align=strict.
8710 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8712 * cgraph.h (cgraph_thunk_info): Add comments.
8713 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8714 assert for VIRTUAL_* arguments stricter.
8716 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8718 PR tree-optimization/71026
8719 * match.pd: Move RDIV patterns from fold-const.c
8720 * fold-const.c (distribute_real_division): Removed.
8721 (fold_binary_loc): Remove calls to distribute_real_divison.
8723 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8725 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8726 c++1z and gnu++1z as deprecated. Change other references to
8727 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8728 Change -Wc++1z-compat to -Wc++17-compat.
8729 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8730 * dwarf2out.c (highest_c_language): Handle C++17.
8731 (gen_compile_unit_die): Likewise.
8733 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8735 PR rtl-optimization/82192
8736 * combine.c (make_extraction): Don't look through non-paradoxical
8737 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8740 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8741 Alan Hayward <alan.hayward@arm.com>
8742 David Sherwood <david.sherwood@arm.com>
8744 * target.def (function_arg_offset): New hook.
8745 * targhooks.h (default_function_arg_offset): Declare.
8746 * targhooks.c (default_function_arg_offset): New function.
8747 * function.c (locate_and_pad_parm): Use
8748 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8749 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8750 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8751 * doc/tm.texi: Regenerate.
8752 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8753 * config/spu/spu.c (spu_function_arg_offset): New function.
8754 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8755 * system.h (FUNCTION_ARG_OFFSET): Poison.
8757 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8758 Alan Hayard <alan.hayward@arm.com>
8759 David Sherwood <david.sherwood@arm.com>
8761 * target.def (truly_noop_truncation): New hook.
8762 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8763 than TRULY_NOOP_TRUNCATION.
8764 * hooks.h (hook_bool_uint_uint_true): Declare.
8765 * hooks.c (hook_bool_uint_uint_true): New function.
8766 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8767 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8768 * doc/tm.texi: Regenerate.
8769 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8770 rather than TRULY_NOOP_TRUNCATION in comments.
8771 (simplify_comparison): Likewise.
8772 (record_truncated_value): Likewise.
8773 * expmed.c (extract_bit_field_1): Likewise.
8774 (extract_split_bit_field): Likewise.
8775 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8776 instead of TRULY_NOOP_TRUNCATION.
8777 * function.c (assign_parm_setup_block): Likewise.
8778 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8779 * rtlhooks.c: Include target.h.
8780 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8781 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8782 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8783 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8784 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8785 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8786 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8787 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8788 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8789 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8790 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8791 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8792 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8793 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8794 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8795 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8796 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8797 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8798 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8799 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8800 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8801 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8802 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8803 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8804 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8805 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8806 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8807 rather than TRULY_NOOP_TRUNCATION in comments.
8808 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8809 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8810 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8811 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8812 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8813 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8814 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8815 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8816 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8817 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8818 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8819 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8820 rather than TRULY_NOOP_TRUNCATION in comments.
8821 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8822 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8823 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8824 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8825 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8826 TRULY_NOOP_TRUNCATION condition.
8827 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8828 (TRULY_NOOP_TRUNCATION): Delete.
8829 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8830 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8831 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8832 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8833 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8834 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8835 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8836 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8837 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8838 rather than TRULY_NOOP_TRUNCATION in comments.
8839 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8840 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8841 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8842 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8843 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8844 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8846 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8849 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8850 (*cmp_ior): Likewise.
8851 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8852 (*ior_scc_scc_cmp): Likewise.
8853 (*and_scc_scc): Likewise.
8854 (*and_scc_scc_cmp): Likewise.
8856 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8857 Alan Hayard <alan.hayward@arm.com>
8858 David Sherwood <david.sherwood@arm.com>
8860 * target.def (can_change_mode_class): New hook.
8861 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8862 (hard_regno_nregs): Likewise.
8863 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8864 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8865 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8866 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8867 (LOAD_EXTEND_OP): Update accordingly.
8868 * doc/tm.texi: Regenerate.
8869 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8870 CANNOT_CHANGE_MODE_CLASS.
8871 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8872 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8873 * combine.c (simplify_set): Update accordingly.
8874 * emit-rtl.c (validate_subreg): Likewise.
8875 * recog.c (general_operand): Likewise.
8876 * regcprop.c (mode_change_ok): Likewise.
8877 * reload1.c (choose_reload_regs): Likewise.
8878 (inherit_piecemeal_p): Likewise.
8879 * rtlanal.c (simplify_subreg_regno): Likewise.
8880 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8881 instead of CANNOT_CHANGE_MODE_CLASS.
8882 (reload_cse_simplify_operands): Likewise.
8883 * reload.c (push_reload): Use targetm.can_change_mode_class
8884 instead of CANNOT_CHANGE_MODE_CLASS.
8885 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8886 REG_CANNOT_CHANGE_MODE_P.
8887 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8888 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8889 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8890 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8891 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8892 (arm_can_change_mode_class): New function.
8893 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8894 than CANNOT_CHANGE_MODE_CLASS in comments.
8895 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8896 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8897 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8898 (ix86_can_change_mode_class): ...this new function, inverting the
8899 sense of the return value.
8900 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8901 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8902 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8903 (ia64_can_change_mode_class): New function.
8904 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8905 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8906 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8907 (m32c_can_change_mode_class): ...this new function, inverting the
8908 sense of the return value.
8909 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8910 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8911 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8912 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8913 (mips_can_change_mode_class): ...this new function, inverting the
8914 sense of the return value.
8915 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8916 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8917 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8918 (msp430_can_change_mode_class): New function.
8919 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8920 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8921 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8922 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8923 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8924 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8925 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8926 (pa_cannot_change_mode_class): Replace with...
8927 (pa_can_change_mode_class): ...this new function, inverting the
8928 sense of the return value.
8929 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8930 than CANNOT_CHANGE_MODE_CLASS in comments.
8931 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8932 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8933 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8934 (pdp11_cannot_change_mode_class): Replace with...
8935 (pdp11_can_change_mode_class): ...this new function, inverting the
8936 sense of the return value.
8937 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8938 * config/powerpcspe/powerpcspe-protos.h
8939 (rs6000_cannot_change_mode_class_ptr): Delete.
8940 * config/powerpcspe/powerpcspe.c
8941 (rs6000_cannot_change_mode_class_ptr): Delete.
8942 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8943 (rs6000_option_override_internal): Assign to
8944 targetm.can_change_mode_class instead of
8945 rs6000_cannot_change_mode_class_ptr.
8946 (rs6000_cannot_change_mode_class): Replace with...
8947 (rs6000_can_change_mode_class): ...this new function, inverting the
8948 sense of the return value.
8949 (rs6000_debug_cannot_change_mode_class): Replace with...
8950 (rs6000_debug_can_change_mode_class): ...this new function.
8951 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8952 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8953 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8954 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8955 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8957 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8958 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8959 (rs6000_option_override_internal): Assign to
8960 targetm.can_change_mode_class instead of
8961 rs6000_cannot_change_mode_class_ptr.
8962 (rs6000_cannot_change_mode_class): Replace with...
8963 (rs6000_can_change_mode_class): ...this new function, inverting the
8964 sense of the return value.
8965 (rs6000_debug_cannot_change_mode_class): Replace with...
8966 (rs6000_debug_can_change_mode_class): ...this new function.
8967 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8968 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8969 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8970 (s390_can_change_mode_class): ...this new function, inverting the
8971 sense of the return value.
8972 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8973 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8974 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8975 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8976 (sh_cannot_change_mode_class): Replace with...
8977 (sh_can_change_mode_class): ...this new function, inverting the
8978 sense of the return value.
8979 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8980 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8981 (sparc_can_change_mode_class): New function.
8982 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8983 * config/spu/spu.c (spu_can_change_mode_class): New function.
8984 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8985 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8986 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8987 (visium_can_change_mode_class): New function.
8988 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8990 2017-09-15 Richard Biener <rguenther@suse.de>
8992 PR tree-optimization/82217
8993 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8994 but not undefined case.
8996 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8999 * postreload.c (reload_cse_simplify_operands): Skip
9000 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
9002 2017-09-15 Richard Biener <rguenther@suse.de>
9004 PR tree-optimization/68823
9005 * graphite-scop-detection.c (build_alias_set): If we have a
9006 possible dependence check whether we can handle them by just
9007 looking at the DRs DR_ACCESS_FNs.
9008 (build_scops): If build_alias_set fails, fail the SCOP.
9010 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9012 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
9013 to support float128 built-in functions that require the ISA 3.0
9015 (BU_FLOAT128_3_HW): Likewise.
9016 (SQRTF128): Add support for the IEEE 128-bit square root and fma
9018 (FMAF128): Likewise.
9020 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
9021 support for built-in functions that need the ISA 3.0 IEEE 128-bit
9022 floating point instructions.
9023 (rs6000_invalid_builtin): Likewise.
9024 (rs6000_builtin_mask_names): Likewise.
9025 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
9026 (RS6000_BTM_FLOAT128_HW): Likewise.
9027 (RS6000_BTM_COMMON): Likewise.
9028 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
9030 * doc/extend.texi (RS/6000 built-in functions): Document the
9031 IEEE 128-bit floating point square root and fused multiply-add
9034 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
9036 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
9037 reg (r2) isn't in the set of registers defined in the prologue.
9039 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9040 Alan Hayward <alan.hayward@arm.com>
9041 David Sherwood <david.sherwood@arm.com>
9043 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
9044 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
9045 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
9046 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
9047 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
9049 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
9050 max_vectorization_factor.
9051 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
9053 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9054 Alan Hayward <alan.hayward@arm.com>
9055 David Sherwood <david.sherwood@arm.com>
9057 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
9058 (vect_worthwhile_without_simd_p): Declare.
9059 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
9060 (vectorizable_reduction): Use it.
9061 * tree-vect-stmts.c (vectorizable_shift): Likewise.
9062 (vectorizable_operation): Likewise.
9064 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9065 Alan Hayward <alan.hayward@arm.com>
9066 David Sherwood <david.sherwood@arm.com>
9068 * tree-vectorizer.h (vect_get_num_copies): New function.
9069 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
9070 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9071 (vectorizable_induction): Likewise.
9072 (vectorizable_live_operation): Likewise.
9073 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
9074 (vectorizable_bswap): Likewise.
9075 (vectorizable_call): Likewise.
9076 (vectorizable_conversion): Likewise.
9077 (vectorizable_assignment): Likewise.
9078 (vectorizable_shift): Likewise.
9079 (vectorizable_operation): Likewise.
9080 (vectorizable_store): Likewise.
9081 (vectorizable_load): Likewise.
9082 (vectorizable_condition): Likewise.
9083 (vectorizable_comparison): Likewise.
9084 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
9086 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9087 Alan Hayward <alan.hayward@arm.com>
9088 David Sherwood <david.sherwood@arm.com>
9090 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
9091 of vect_init_vector.
9093 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9094 Alan Hayward <alan.hayward@arm.com>
9095 David Sherwood <david.sherwood@arm.com>
9097 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
9098 an inline wrapper that provides a location.
9099 (gimple_build_vector): Likewise.
9100 * gimple-fold.c (gimple_build_vector_from_val): New function.
9101 (gimple_build_vector): Likewise.
9102 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
9103 functions to build the initial value. Always return a gimple value.
9104 (get_initial_defs_for_reduction): Likewise. Only compute
9106 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
9107 vect_init_vector on the results from get_initial_def(s)_for_reduction.
9108 (vectorizable_induction): Use gimple_build_vector rather than
9111 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9112 Alan Hayward <alan.hayward@arm.com>
9113 David Sherwood <david.sherwood@arm.com>
9115 * target.h (vec_perm_indices): New typedef.
9116 (auto_vec_perm_indices): Likewise.
9117 * optabs-query.h: Include target.h
9118 (can_vec_perm_p): Take a vec_perm_indices *.
9119 * optabs-query.c (can_vec_perm_p): Likewise.
9120 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
9121 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9122 * tree-vect-generic.c (lower_vec_perm): Likewise.
9123 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
9124 (vect_grouped_load_supported): Likewise.
9125 (vect_shift_permute_load_chain): Likewise.
9126 (vect_permute_store_chain): Use auto_vec_perm_indices.
9127 (vect_permute_load_chain): Likewise.
9128 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
9129 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
9130 Update uses of can_vec_perm_p.
9131 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
9132 mode with a number of elements. Take a vec_perm_indices *.
9133 (vect_create_epilog_for_reduction): Update accordingly.
9134 Use auto_vec_perm_indices.
9135 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
9136 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
9137 (vect_transform_slp_perm_load): Likewise.
9138 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
9139 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
9140 (vect_gen_perm_mask_checked): Likewise.
9141 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
9142 (vect_gen_perm_mask_checked): Likewise.
9143 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
9144 (vectorizable_store): Likewise.
9145 (vectorizable_load): Likewise.
9146 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
9147 (vectorizable_bswap): Likewise.
9149 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9150 Alan Hayward <alan.hayward@arm.com>
9151 David Sherwood <david.sherwood@arm.com>
9153 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
9154 * tree.c (build_vector): Likewise.
9155 (build_vector_from_ctor): Update accordingly.
9156 (build_vector_from_val): Likewise.
9157 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9158 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
9159 * tree-vect-generic.c (add_rshift): Likewise.
9160 (expand_vector_divmod): Likewise.
9161 (optimize_vector_constructor): Likewise.
9162 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
9163 (vect_transform_slp_perm_load): Likewise.
9164 (vect_schedule_slp_instance): Likewise.
9165 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
9166 (vectorizable_call): Likewise.
9167 (vect_gen_perm_mask_any): Likewise. Add elements in order.
9168 * expmed.c (make_tree): Likewise.
9169 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
9170 a vector passed to build_vector.
9171 (fold_convert_const): Likewise.
9172 (exact_inverse): Likewise.
9173 (fold_ternary_loc): Likewise.
9174 (fold_relational_const): Likewise.
9175 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
9176 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
9177 (const_unop): Likewise. Store the reduction accumulator in a
9178 variable rather than an array.
9179 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
9180 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
9181 the new vector, rather than constructing it after the input arrays.
9182 (native_interpret_vector): Use auto_vec<tree> when building
9183 a vector passed to build_vector. Add elements in order.
9184 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
9185 auto_vec<tree> when building a vector passed to build_vector.
9186 (vect_create_epilog_for_reduction): Likewise.
9187 (vectorizable_induction): Likewise.
9188 (get_initial_def_for_reduction): Likewise. Fix indentation of
9190 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
9192 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
9193 passed to build_vector.
9195 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
9196 Alan Hayward <alan.hayward@arm.com>
9197 David Sherwood <david.sherwood@arm.com>
9199 * tree-core.h (tree_base::u): Add an "nelts" field.
9200 (tree_vector): Use VECTOR_CST_NELTS as the length.
9201 * tree.c (tree_size): Likewise.
9202 (make_vector): Initialize VECTOR_CST_NELTS.
9203 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
9204 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
9205 TYPE_VECTOR_SUBPARTS.
9206 * expr.c (const_vector_mask_from_tree): Consistently use "units"
9207 as the number of units, setting it from VECTOR_CST_NELTS.
9208 (const_vector_from_tree): Likewise.
9209 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
9210 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
9211 (fold_negate_expr_1): Likewise.
9212 (fold_convert_const): Likewise.
9213 (const_binop): Likewise. Differentiate the number of output and
9215 (const_unop): Likewise.
9216 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
9217 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
9218 in cases that did the opposite.
9220 2017-09-14 Richard Biener <rguenther@suse.de>
9222 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
9225 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
9227 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
9229 2017-09-14 Jakub Jelinek <jakub@redhat.com>
9232 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
9233 if and where to split a bb, except for splitting before debug insn
9234 sequences followed by non-label real insn. Delete debug insns
9235 in between basic blocks.
9237 * combine.c (make_compound_operation_int): Formatting fixes.
9239 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
9240 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9241 * config/netbsd.h (LINK_EH_SPEC): Likewise.
9242 * config/sol2.h (LINK_EH_SPEC): Likewise.
9243 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9244 * config/s390/linux.h (LINK_SPEC): Likewise.
9245 * config/freebsd.h (LINK_EH_SPEC): Likewise.
9246 * config/openbsd.h (LINK_EH_SPEC): Likewise.
9247 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9248 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
9249 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
9250 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9251 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
9252 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
9254 2017-09-13 Jakub Jelinek <jakub@redhat.com>
9256 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
9258 (ENDFILE_LINUX_SPEC): Likewise.
9259 (LINK_EH_SPEC): Likewise.
9260 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
9261 (LINK_OS_LINUX_SPEC32): Likewise.
9262 (LINK_OS_LINUX_SPEC64): Likewise.
9263 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
9264 (LINK_OS_LINUX_SPEC): Likewise.
9266 2017-09-13 Martin Liska <mliska@suse.cz>
9269 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
9270 CASE_HIGH is NULL_TREE.
9272 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9273 Alan Hayward <alan.hayward@arm.com>
9274 David Sherwood <david.sherwood@arm.com>
9276 * target.def (secondary_memory_needed): New hook.
9277 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
9278 instead of SECONDARY_MEMORY_NEEDED.
9279 (secondary_memory_needed_mode): Likewise.
9280 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
9281 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
9282 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
9283 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
9284 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
9285 * doc/tm.texi: Regenerate.
9286 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
9287 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
9288 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9289 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
9290 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
9291 * config/i386/i386.c (inline_secondary_memory_needed): Put the
9292 mode argument first and change the reg_class arguments to reg_class_t.
9293 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
9294 Make static. Update the call to inline_secondary_memory_needed.
9295 (ix86_register_move_cost): Update the call to
9296 inline_secondary_memory_needed.
9297 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9298 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
9300 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
9301 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9303 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
9304 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
9305 * config/mips/mips.c (mips_secondary_memory_needed): Make static
9306 and match hook interface. Add comment from mips.h.
9307 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9308 * config/mmix/mmix.md (truncdfsf2): Refer to
9309 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9311 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
9312 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
9313 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9314 (pa_secondary_memory_needed): New function.
9315 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
9316 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
9317 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9318 (pdp11_secondary_memory_needed): Make static and match hook interface.
9319 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
9320 * config/powerpcspe/powerpcspe-protos.h
9321 (rs6000_secondary_memory_needed_ptr): Delete.
9322 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
9324 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9325 (rs6000_option_override_internal): Assign to
9326 targetm.secondary_memory_needed rather than
9327 rs6000_secondary_memory_needed_ptr.
9328 (rs6000_secondary_memory_needed): Match hook interface.
9329 (rs6000_debug_secondary_memory_needed): Likewise.
9330 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
9331 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
9332 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
9333 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9334 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
9335 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
9337 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
9338 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9339 (rs6000_option_override_internal): Assign to
9340 targetm.secondary_memory_needed rather than
9341 rs6000_secondary_memory_needed_ptr.
9342 (rs6000_secondary_memory_needed): Match hook interface.
9343 (rs6000_debug_secondary_memory_needed): Likewise.
9344 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
9345 * config/s390/s390.c (s390_secondary_memory_needed): New function.
9346 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9347 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
9348 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
9349 (sparc_secondary_memory_needed): New function.
9350 * lra-constraints.c (check_and_process_move): Refer to
9351 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
9353 (curr_insn_transform): Likewise.
9354 (process_alt_operands): Use targetm.secondary_memory_needed
9355 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9356 (check_secondary_memory_needed_p): Likewise.
9357 (choose_split_class): Likewise.
9358 * reload.c: Unconditionally include code that was previously
9359 conditional on SECONDARY_MEMORY_NEEDED.
9360 (push_secondary_reload): Use targetm.secondary_memory_needed
9361 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9362 (push_reload): Likewise.
9363 * reload1.c: Unconditionally include code that was previously
9364 conditional on SECONDARY_MEMORY_NEEDED.
9365 (choose_reload_regs): Use targetm.secondary_memory_needed
9366 instead of TARGET_SECONDARY_MEMORY_NEEDED.
9367 (gen_reload): Likewise.
9368 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
9370 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
9371 Alan Hayward <alan.hayward@arm.com>
9372 David Sherwood <david.sherwood@arm.com>
9374 * target.def (secondary_memory_needed_mode): New hook:
9375 * targhooks.c (default_secondary_memory_needed_mode): Declare.
9376 * targhooks.h (default_secondary_memory_needed_mode): New function.
9377 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
9378 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
9379 * doc/tm.texi: Regenerate.
9380 * lra-constraints.c (check_and_process_move): Use
9381 targetm.secondary_memory_needed_mode instead of
9382 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
9383 (curr_insn_transform): Likewise.
9384 * reload.c (get_secondary_mem): Likewise.
9385 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9386 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
9388 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9389 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9390 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
9391 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9392 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
9394 * config/powerpcspe/powerpcspe-protos.h
9395 (rs6000_secondary_memory_needed_mode): Delete.
9396 * config/powerpcspe/powerpcspe.c
9397 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9398 (rs6000_secondary_memory_needed_mode): Make static.
9399 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9400 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
9402 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9404 (rs6000_secondary_memory_needed_mode): Make static.
9405 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9406 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
9407 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
9408 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
9409 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
9411 (sparc_secondary_memory_needed_mode): New function.
9412 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
9414 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
9416 * config/aarch64/constraints.md (Umq): New constraint.
9417 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
9419 (mov<mode>): Update condition.
9421 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9423 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9424 when bitposition is the same.
9426 2017-09-13 Richard Biener <rguenther@suse.de>
9428 * dwarf2out.c (output_die_symbol): Remove.
9429 (output_die): Do not output a DIEs symbol.
9431 2017-09-13 Richard Biener <rguenther@suse.de>
9434 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9435 default-def to avoid breaking iterator update with the weird
9436 interaction with cgraph_update_edges_for_call_stmt_node.
9438 2017-09-13 Richard Biener <rguenther@suse.de>
9440 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9441 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9442 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9443 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9445 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9447 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9448 Disable pc relative literal load irrespective of
9449 TARGET_FIX_ERR_A53_84341 for default.
9451 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9453 * config/sparc/sparc.c (output_return): Output the source location of
9454 the insn in the delay slot, if any.
9455 (output_sibcall): Likewise.
9457 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9460 * common.opt (-static-pie): New alias.
9461 (shared): Negate static-pie.
9462 (-no-pie): Update help text.
9464 (static-pie): New option.
9465 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9466 -static-pie support.
9467 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9468 (LINK_EH_SPEC): Likewise.
9469 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9470 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9471 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9472 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9473 (init_gcc_specs): Likewise.
9474 (init_spec): Likewise.
9475 (display_help): Update help message for -pie.
9476 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9479 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9481 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9482 (movdi_aarch64): Likewise.
9483 (movti_aarch64): Likewise.
9485 2017-09-12 Simon Wright <simon@pushface.org>
9488 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9489 calculation of the minor version, always output as 0.
9491 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9494 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9495 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9496 on it early, rather than manual conversion late. For
9497 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9498 instead of performing manual conversion.
9500 2017-09-12 Carl Love <cel@us.ibm.com>
9502 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9503 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9505 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9506 VMULOSW): Add definitions.
9507 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9508 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9509 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9510 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9511 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9513 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9515 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9517 (movti_aarch64): Likewise.
9518 (movdf_aarch64): Likewise.
9519 (movtf_aarch64): Likewise.
9520 (load_pairdi): Likewise.
9521 (store_pairdi): Likewise.
9522 (load_pairdf): Likewise.
9523 (store_pairdf): Likewise.
9524 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9525 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9526 (ldr_got_small_<mode>): Likewise.
9527 (ldr_got_small_28k_<mode>): Likewise.
9528 (ldr_got_tiny): Likewise.
9529 * config/aarch64/iterators.md (ldst_sz): New.
9530 (ldpstp_sz): Likewise.
9531 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9533 (thunderx_load): Split load_8 to load_16.
9534 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9536 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9537 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9538 (xgene1_store_pair): Split store_8 to store_16.
9539 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9540 (falkor_st_0_st_sd): Split store_8 to store_16.
9542 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9544 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9545 and store1/2/3/4 to store_4/8/12/16.
9546 * config/aarch64/aarch64.md: Update for rename.
9547 * config/arm/arm.md: Likewise.: Likewise.
9548 * config/arm/arm.c: Likewise.
9549 * config/arm/thumb1.md: Likewise.
9550 * config/arm/thumb2.md: Likewise.
9551 * config/arm/vfp.md: Likewise.
9552 * config/arm/arm-generic.md: Likewise.
9553 * config/arm/arm1020e.md: Likewise.
9554 * config/arm/arm1026ejs.md: Likewise.
9555 * config/arm/arm1136jfs.md: Likewise.
9556 * config/arm/arm926ejs.md: Likewise.
9557 * config/arm/cortex-a15.md: Likewise.
9558 * config/arm/cortex-a17.md: Likewise.
9559 * config/arm/cortex-a5.md: Likewise.
9560 * config/arm/cortex-a53.md: Likewise.
9561 * config/arm/cortex-a57.md: Likewise.
9562 * config/arm/cortex-a7.md: Likewise.
9563 * config/arm/cortex-a8.md: Likewise.
9564 * config/arm/cortex-a9.md: Likewise.
9565 * config/arm/cortex-m4.md: Likewise.
9566 * config/arm/cortex-m7.md: Likewise.
9567 * config/arm/cortex-r4.md: Likewise.
9568 * config/arm/exynos-m1.md: Likewise.
9569 * config/arm/fa526.md: Likewise.
9570 * config/arm/fa606te.md: Likewise.
9571 * config/arm/fa626te.md: Likewise.
9572 * config/arm/fa726te.md: Likewise.
9573 * config/arm/fmp626.md: Likewise.
9574 * config/arm/iwmmxt.md: Likewise.
9575 * config/arm/ldmstm.md: Likewise.
9576 * config/arm/marvell-pj4.md: Likewise.
9577 * config/arm/xgene1.md: Likewise.
9578 * config/aarch64/thunderx.md: Likewise.
9579 * config/aarch64/thunderx2t99.md: Likewise.
9580 * config/aarch64/falkor.md: Likewise.
9582 2017-09-12 Martin Liska <mliska@suse.cz>
9584 * attribs.c (private_lookup_attribute): New function.
9585 * attribs.h (private_lookup_attribute): Declared here.
9586 (lookup_attribute): Called from this place.
9588 2017-09-12 Richard Biener <rguenther@suse.de>
9590 PR tree-optimization/82157
9591 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9592 stmts with side-effects.
9594 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9595 Alan Hayward <alan.hayward@arm.com>
9596 David Sherwood <david.sherwood@arm.com>
9598 * target.def (hard_regno_nregs): New hook.
9599 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9600 * targhooks.h (default_hard_regno_nregs): Declare.
9601 * targhooks.c (default_hard_regno_nregs): New function.
9602 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9603 (TARGET_HARD_REGNO_NREGS): ...this hook.
9604 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9605 (CLASS_MAX_NREGS): Likewise.
9606 * doc/tm.texi: Regenerate.
9607 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9608 instead of HARD_REGNO_NREGS.
9609 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9610 HARD_REGNO_NREGS in the comment.
9611 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9612 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9613 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9614 Return an unsigned int.
9615 (TARGET_HARD_REGNO_NREGS): Redefine.
9616 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9617 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9618 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9619 (arc_hard_regno_nregs): New function.
9620 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9621 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9622 (arm_hard_regno_nregs): New function.
9623 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9624 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9625 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9626 (TARGET_HARD_REGNO_NREGS): Redefine.
9627 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9628 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9629 (HARD_REGNO_NREGS): Delete.
9630 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9631 (cr16_hard_regno_nregs): New function.
9632 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9633 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9634 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9635 (cris_hard_regno_nregs): New function.
9636 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9637 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9638 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9639 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9640 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9641 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9642 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9643 (frv_hard_regno_nregs): Make static. Take and return an
9645 (frv_class_max_nregs): Remove outdated copy of documentation.
9646 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9647 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9648 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9649 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9650 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9651 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9652 (TARGET_HARD_REGNO_NREGS): Redefine.
9653 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9654 (CLASS_MAX_NREGS): Update comment.
9655 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9656 (ia64_hard_regno_nregs): New function.
9657 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9658 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9659 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9660 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9661 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9663 (m32c_hard_regno_nregs): Likewise. Make static.
9664 (TARGET_HARD_REGNO_NREGS): Redefine.
9665 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9666 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9667 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9668 (m68k_hard_regno_nregs): New function.
9669 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9670 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9671 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9672 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9673 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9674 Take and return an unsigned int.
9675 (TARGET_HARD_REGNO_NREGS): Redefine.
9676 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9677 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9678 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9679 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9680 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9681 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9682 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9683 (msp430_hard_regno_nregs): Make static. Take and return an
9685 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9686 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9687 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9688 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9689 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9690 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9691 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9692 (TARGET_HARD_REGNO_NREGS): Redefine.
9693 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9694 (PA_HARD_REGNO_NREGS): ...this.
9695 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9696 (PA_HARD_REGNO_NREGS): ...this.
9697 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9698 (pa_hard_regno_nregs): New function.
9699 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9700 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9701 (pdp11_hard_regno_nregs): New function.
9702 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9703 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9704 (rs6000_hard_regno_nregs_hook): New function.
9705 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9706 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9707 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9708 Take and return an unsigned int. Move earlier in file.
9709 (TARGET_HARD_REGNO_NREGS): Redefine.
9710 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9711 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9712 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9713 (rl78_hard_regno_nregs): Make static. Take and return an
9715 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9716 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9717 (rs6000_hard_regno_nregs_hook): New function.
9718 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9719 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9720 (TARGET_HARD_REGNO_NREGS): Redefine.
9721 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9722 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9723 instead of HARD_REGNO_NREGS.
9724 (s390_hard_regno_nregs): New function.
9725 (s390_hard_regno_mode_ok): Add comment from s390.h.
9726 (TARGET_HARD_REGNO_NREGS): Redefine.
9727 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9728 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9729 (sh_hard_regno_nregs): New function.
9730 (sh_pass_in_reg_p): Use it.
9731 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9732 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9733 (sparc_hard_regno_nregs): New function.
9734 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9735 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9736 (spu_function_arg_advance): Use it, supplying a valid register number.
9737 (TARGET_HARD_REGNO_NREGS): Redefine.
9738 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9739 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9740 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9741 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9742 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9743 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9744 (CLASS_MAX_NREGS): Remove copy of old documentation.
9745 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9746 (visium_hard_regno_nregs): New function.
9747 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9748 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9749 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9750 xtensa_hard_regno_nregs): New function.
9751 * system.h (HARD_REGNO_NREGS): Poison.
9753 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9755 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9756 hard_regno_nregs instead of HARD_REGNO_NREGS.
9757 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9758 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9759 (c6x_expand_epilogue): Likewise.
9760 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9761 (frv_read_iacc_argument): Likewise.
9762 * config/sh/sh.c: Include regs.h.
9763 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9764 (regs_used): Likewise.
9765 (output_stack_adjust): Likewise.
9766 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9767 * expmed.c: Include regs.h.
9768 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9769 * ree.c: Include regs.h.
9770 (combine_reaching_defs): Use hard_regno_nregs instead of
9772 (add_removable_extension): Likewise.
9774 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9776 * regs.h (hard_regno_nregs): Turn into a function.
9777 (end_hard_regno): Update accordingly.
9778 * caller-save.c (setup_save_areas): Likewise.
9779 (save_call_clobbered_regs): Likewise.
9780 (replace_reg_with_saved_mem): Likewise.
9781 (insert_restore): Likewise.
9782 (insert_save): Likewise.
9783 * combine.c (can_change_dest_mode): Likewise.
9784 (move_deaths): Likewise.
9785 (distribute_notes): Likewise.
9786 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9787 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9788 (rs6000_split_multireg_move): Likewise.
9789 (rs6000_register_move_cost): Likewise.
9790 (rs6000_memory_move_cost): Likewise.
9791 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9792 (rs6000_split_multireg_move): Likewise.
9793 (rs6000_register_move_cost): Likewise.
9794 (rs6000_memory_move_cost): Likewise.
9795 * cselib.c (cselib_reset_table): Likewise.
9796 (cselib_lookup_1): Likewise.
9797 * emit-rtl.c (set_mode_and_regno): Likewise.
9798 * function.c (aggregate_value_p): Likewise.
9799 * ira-color.c (setup_profitable_hard_regs): Likewise.
9800 (check_hard_reg_p): Likewise.
9801 (calculate_saved_nregs): Likewise.
9802 (assign_hard_reg): Likewise.
9803 (improve_allocation): Likewise.
9804 (calculate_spill_cost): Likewise.
9805 * ira-emit.c (modify_move_list): Likewise.
9806 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9807 (ira_hard_reg_in_set_p): Likewise.
9808 * ira.c (setup_reg_mode_hard_regset): Likewise.
9809 (clarify_prohibited_class_mode_regs): Likewise.
9810 (check_allocation): Likewise.
9811 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9812 (lra_setup_reg_renumber): Likewise.
9813 (setup_try_hard_regno_pseudos): Likewise.
9814 (spill_for): Likewise.
9815 (assign_hard_regno): Likewise.
9816 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9817 * lra-constraints.c (in_class_p): Likewise.
9818 (lra_constraint_offset): Likewise.
9819 (simplify_operand_subreg): Likewise.
9820 (lra_constraints): Likewise.
9821 (split_reg): Likewise.
9822 (split_if_necessary): Likewise.
9823 (invariant_p): Likewise.
9824 (inherit_in_ebb): Likewise.
9825 * lra-lives.c (process_bb_lives): Likewise.
9826 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9827 (get_hard_regs): Likewise.
9828 (do_remat): Likewise.
9829 * lra-spills.c (assign_spill_hard_regs): Likewise.
9830 * mode-switching.c (create_pre_exit): Likewise.
9831 * postreload.c (reload_combine_recognize_pattern): Likewise.
9832 * recog.c (peep2_find_free_register): Likewise.
9833 * regcprop.c (kill_value_regno): Likewise.
9834 (set_value_regno): Likewise.
9835 (copy_value): Likewise.
9836 (maybe_mode_change): Likewise.
9837 (find_oldest_value_reg): Likewise.
9838 (copyprop_hardreg_forward_1): Likewise.
9839 * regrename.c (check_new_reg_p): Likewise.
9840 (regrename_do_replace): Likewise.
9841 * reload.c (push_reload): Likewise.
9842 (combine_reloads): Likewise.
9843 (find_dummy_reload): Likewise.
9844 (operands_match_p): Likewise.
9845 (find_reloads): Likewise.
9846 (find_equiv_reg): Likewise.
9847 (reload_adjust_reg_for_mode): Likewise.
9848 * reload1.c (count_pseudo): Likewise.
9849 (count_spilled_pseudo): Likewise.
9850 (find_reg): Likewise.
9851 (clear_reload_reg_in_use): Likewise.
9852 (free_for_value_p): Likewise.
9853 (allocate_reload_reg): Likewise.
9854 (choose_reload_regs): Likewise.
9855 (reload_adjust_reg_for_temp): Likewise.
9856 (emit_reload_insns): Likewise.
9857 (delete_output_reload): Likewise.
9858 * rtlanal.c (subreg_get_info): Likewise.
9859 * sched-deps.c (sched_analyze_reg): Likewise.
9860 * sel-sched.c (init_regs_for_mode): Likewise.
9861 (mark_unavailable_hard_regs): Likewise.
9862 (choose_best_reg_1): Likewise.
9863 (verify_target_availability): Likewise.
9864 * valtrack.c (dead_debug_insert_temp): Likewise.
9865 * var-tracking.c (track_loc_p): Likewise.
9866 (emit_note_insn_var_location): Likewise.
9867 * varasm.c (make_decl_rtl): Likewise.
9868 * reginfo.c (choose_hard_reg_mode): Likewise.
9869 (init_reg_modes_target): Refer directly to
9870 this_target_regs->x_hard_regno_nregs.
9872 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9874 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9875 instead of hard_regno_nregs.
9877 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9879 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9880 end_hard_regno instead of hard_regno_nregs.
9881 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9882 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9883 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9884 * ira-color.c (improve_allocation): Likewise.
9885 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9886 * lra-lives.c (mark_regno_live): Likewise.
9887 (mark_regno_dead): Likewise.
9888 * lra-remat.c (operand_to_remat): Likewise.
9889 * lra.c (collect_non_operand_hard_regs): Likewise.
9890 * postreload.c (reload_combine_note_store): Likewise.
9891 (move2add_valid_value_p): Likewise.
9892 * reload.c (regno_clobbered_p): Likewise.
9894 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9896 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9898 * config/v850/v850.c (v850_reorg): Likewise.
9899 * reload.c (refers_to_regno_for_reload_p): Likewise.
9900 (find_equiv_reg): Likewise.
9901 * reload1.c (reload_reg_reaches_end_p): Likewise.
9903 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9905 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9907 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9908 * config/arm/arm.c (output_move_neon): Likewise.
9909 (arm_attr_length_move_neon): Likewise.
9910 (neon_split_vcombine): Likewise.
9911 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9912 (c6x_mark_reg_written): Likewise.
9913 (c6x_dwarf_register_span): Likewise.
9914 * config/i386/i386.c (ix86_save_reg): Likewise.
9915 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9916 (rws_access_reg): Likewise.
9917 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9918 * mode-switching.c (create_pre_exit): Likewise.
9919 * ree.c (combine_reaching_defs): Likewise.
9920 (add_removable_extension): Likewise.
9921 * regcprop.c (find_oldest_value_reg): Likewise.
9922 (copyprop_hardreg_forward_1): Likewise.
9923 * reload.c (reload_inner_reg_of_subreg): Likewise.
9924 (push_reload): Likewise.
9925 (combine_reloads): Likewise.
9926 (find_dummy_reload): Likewise.
9927 (reload_adjust_reg_for_mode): Likewise.
9928 * reload1.c (find_reload_regs): Likewise.
9929 (forget_old_reloads_1): Likewise.
9930 (reload_reg_free_for_value_p): Likewise.
9931 (reload_adjust_reg_for_temp): Likewise.
9932 (emit_reload_insns): Likewise.
9933 (delete_output_reload): Likewise.
9934 * sel-sched.c (choose_best_reg_1): Likewise.
9935 (choose_best_pseudo_reg): Likewise.
9937 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9938 Alan Hayward <alan.hayward@arm.com>
9939 David Sherwood <david.sherwood@arm.com>
9941 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9942 * target.def (slow_unaligned_access): New hook.
9943 * targhooks.h (default_slow_unaligned_access): Declare.
9944 * targhooks.c (default_slow_unaligned_access): New function.
9945 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9946 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9947 * doc/tm.texi: Regenerate.
9948 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9949 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9950 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9952 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9953 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9955 (rs6000_slow_unaligned_access): New function.
9956 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9957 (expand_block_compare): Likewise.
9958 (expand_strn_compare): Likewise.
9959 (rs6000_rtx_costs): Likewise.
9960 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9961 (riscv_slow_unaligned_access): Likewise.
9962 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9963 (riscv_slow_unaligned_access_p): ...this and make static.
9964 (riscv_option_override): Update accordingly.
9965 (riscv_slow_unaligned_access): New function.
9966 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9967 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9968 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9969 (rs6000_slow_unaligned_access): New function.
9970 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9971 (rs6000_rtx_costs): Likewise.
9972 * config/rs6000/rs6000-string.c (expand_block_compare)
9973 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9974 of SLOW_UNALIGNED_ACCESS.
9975 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9976 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9977 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9978 of SLOW_UNALIGNED_ACCESS.
9979 * expmed.c (simple_mem_bitfield_p): Likewise.
9980 * expr.c (alignment_for_piecewise_move): Likewise.
9981 (emit_group_load_1): Likewise.
9982 (emit_group_store): Likewise.
9983 (copy_blkmode_from_reg): Likewise.
9984 (emit_push_insn): Likewise.
9985 (expand_assignment): Likewise.
9986 (store_field): Likewise.
9987 (expand_expr_real_1): Likewise.
9988 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9989 * lra-constraints.c (simplify_operand_subreg): Likewise.
9990 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9991 * gimple-ssa-store-merging.c: Likewise in block comment at start
9993 * tree-ssa-strlen.c: Include target.h.
9994 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9995 of SLOW_UNALIGNED_ACCESS.
9996 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9998 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
10000 PR rtl-optimization/82185
10001 * expmed.c (emit_store_flag_int): Only test tem if it has been
10004 2017-09-12 Richard Biener <rguenther@suse.de>
10006 PR middle-end/82149
10007 * match.pd ((FTYPE) N CMP CST): Fix typo.
10009 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
10011 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
10013 (mips_near_type_p): Add 'short_call' attribute as a synonym
10015 * doc/extend.texi (short_call): Document new function attribute.
10017 2017-09-12 Jakub Jelinek <jakub@redhat.com>
10020 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
10021 assertion check that in the condition.
10022 (get_atomic_generic_size): Likewise. Before testing if parameter
10023 has pointer type, if it has array type, call for C++
10024 default_conversion to perform array-to-pointer conversion.
10026 2017-09-12 Richard Biener <rguenther@suse.de>
10028 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
10029 for operations we cannot scalarize.
10031 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
10033 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
10034 vectors heap vectors. Clean up comments.
10035 Make visited_bbs a reference.
10036 (profitable_jump_thread_path): Make GC
10037 vectors heap vectors. Clean up comments.
10039 (convert_and_register_jump_thread_path): Make GC vectors heap
10041 (check_subpath_and_update_thread_path): Same. Clean up comments.
10042 Make visited_bbs a reference.
10043 (handle_phi): Abstract common code to to
10044 register_jump_thread_path_if_profitable.
10045 Rename VAR_BB to DEF_BB.
10047 Make GC vectors heap vectors.
10048 Make visited_bbs a reference.
10049 (handle_assignment): Same.
10050 (register_jump_thread_path_if_profitable): New.
10051 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
10053 Make GC vectors heap vectors. Clean up comments.
10054 Make visited_bbs a reference.
10055 (find_jump_threads_backwards): Make visited_bbs live in the stack.
10056 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
10059 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
10062 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
10063 words of E_DImode object are reachable by xtensa_uimm8x4 access.
10065 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
10067 Revert r251800 and r251799.
10069 2017-09-11 Martin Jambor <mjambor@suse.cz>
10072 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
10073 arguments in advance.
10074 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
10075 use it to find predecessor edges.
10076 (naive_outof_ssa): Collect vector of predecessors.
10078 2017-09-08 Jason Merrill <jason@redhat.com>
10080 PR c++/70029 - ICE with ref-qualifier and -flto
10081 * langhooks.h (struct lang_hooks_for_types): Add
10082 copy_lang_qualifiers.
10083 * attribs.c (build_type_attribute_qual_variant): Use it.
10084 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
10086 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
10087 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
10089 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
10092 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
10093 (*mulsi3_sp64): New instruction.
10094 (mulsi3): New expander.
10096 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
10098 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
10100 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10102 * sancov.c: Include memmodel.h.
10104 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
10107 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
10110 2017-09-07 Carl Love <cel@us.ibm.com>
10112 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
10113 the sldi instruction.
10115 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
10117 * sancov.c: Include tm_p.h.
10119 2017-09-07 Jakub Jelinek <jakub@redhat.com>
10122 * output.h (switch_to_other_text_partition): New declaration.
10123 * varasm.c (switch_to_other_text_partition): New function.
10124 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
10125 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
10126 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
10127 to the other text partition before emitting LCL label and switch back
10128 after emitting the word after it.
10130 2017-09-07 Richard Biener <rguenther@suse.de>
10132 * passes.def (pass_split_crit_edges): Remove instance before PRE.
10133 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
10134 critical edges here, after loop init.
10135 (pass_data_pre): Remove PROP_no_crit_edges flags.
10136 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
10137 for valueization of call args to avoid leaking VN_TOP.
10138 (visit_use): Assert we do not visit default defs.
10139 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
10140 Use error_mark_node to more easily detect leaking VN_TOP.
10141 All default-defs are varying, not VN_TOP. Mark them visited.
10142 (run_scc_vn): Make code match comment.
10144 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
10146 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
10147 OPTION_MASK_FLOAT128_KEYWORD.
10148 (POWERPC_MASKS): Likewise.
10149 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
10150 support for the -mfloat128-type option, and make -mfloat128
10151 default on PowerPC Linux systems. Define or undefine
10152 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
10153 Define __float128 to be __ieee128 if IEEE 128-bit support is
10154 enabled, or undefine it.
10155 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
10156 Delete defining __FLOAT128_TYPE__.
10157 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
10158 -mfloat128-type option and make -mfloat128 default on PowerPC
10160 (TARGET_FLOAT128_TYPE): Likewise.
10161 (-mfloat128-type): Likewise.
10162 * config/rs6000/rs6000.c (rs6000_option_override_internal):
10163 Delete the -mfloat128-type option and make -mfloat128 default on
10164 PowerPC Linux systems. Always use __ieee128 to be the keyword for
10165 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
10166 128-bit floating point is enabled. Change tests from using
10167 -mfloat128-type to -mfloat128.
10168 (rs6000_mangle_type): Use the correct mangling for the __float128
10169 type even if normal long double is restricted to 64-bits.
10170 (floatn_mode): Enable the _Float128 type by default on VSX Linux
10172 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
10173 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
10174 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
10176 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
10177 documentation for -mfloat128.
10179 2017-09-06 Olivier Hainque <hainque@adacore.com>
10181 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
10183 2017-09-06 Wish Wu <wishwu007@gmail.com>
10184 Jakub Jelinek <jakub@redhat.com>
10186 * asan.c (initialize_sanitizer_builtins): Add
10187 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
10188 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
10189 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
10190 BT_FN_VOID_UINT64_PTR variables.
10191 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
10192 (BT_FN_VOID_UINT16_UINT16): Likewise.
10193 (BT_FN_VOID_UINT32_UINT32): Likewise.
10194 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
10195 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
10196 (BT_FN_VOID_UINT64_PTR): Likewise.
10197 * common.opt (flag_sanitize_coverage): New variable.
10198 (fsanitize-coverage=trace-pc): Remove.
10199 (fsanitize-coverage=): Add.
10200 * flag-types.h (enum sanitize_coverage_code): New enum.
10201 * fold-const.c (fold_range_test): Disable non-short-circuit
10202 optimization if flag_sanitize_coverage.
10203 (fold_truth_andor): Likewise.
10204 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
10205 * opts.c (COVERAGE_SANITIZER_OPT): Define.
10206 (coverage_sanitizer_opts): New array.
10207 (get_closest_sanitizer_option): Add OPTS argument, handle also
10208 OPT_fsanitize_coverage_.
10209 (parse_sanitizer_options): Adjusted to also handle
10210 OPT_fsanitize_coverage_.
10211 (common_handle_option): Add OPT_fsanitize_coverage_.
10212 * sancov.c (instrument_comparison, instrument_switch): New function.
10213 (sancov_pass): Add trace-cmp support.
10214 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
10215 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
10216 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
10217 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
10218 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
10219 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
10220 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
10221 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
10222 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
10223 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
10225 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10227 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
10228 error. Only quit immediately if parsing is complete.
10229 (BEGIN): Initialize fatal_err and parse_done.
10230 (begin fpu, end fpu): Check number of arguments.
10231 (begin arch, end arch): Likewise.
10232 (begin cpu, end cpu): Likewise.
10233 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
10234 (optalias): Likewise.
10236 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
10238 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
10239 * config/arm/arm-isa.h: Delete. Move definitions to ...
10240 * arm-cpus.in: ... here. Use new feature and fgroup values.
10241 * config/arm/arm.c (arm_option_override): Use lower case for feature
10243 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
10244 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
10245 * config/arm/parsecpu.awk (END): Add new command 'isa'.
10247 (print_isa_bits_for): New function.
10248 (gen_isa): New function.
10249 (gen_comm_data): Use print_isa_bits_for.
10250 (define feature): New keyword.
10251 (define fgroup): New keyword.
10252 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
10253 (arm-isa.h): Add rule to generate file.
10254 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
10255 case for feature bit names.
10257 2017-09-06 Richard Biener <rguenther@suse.de>
10259 * tree-ssa-pre.c (NECESSARY): Remove.
10260 (create_expression_by_pieces): Do not touch pass-local flags.
10261 (insert_into_preds_of_block): Likewise.
10262 (do_pre_regular_insertion): Likewise.
10263 (eliminate_insert): Likewise.
10264 (eliminate_dom_walker::before_dom_children): Likewise.
10265 (fini_eliminate): Do not look at inserted_exprs.
10266 (mark_operand_necessary): Remove.
10267 (remove_dead_inserted_code): Replace with simple work-list
10268 algorithm based on inserted_exprs and SSA uses.
10269 (pass_pre::execute): Re-order fini_eliminate and
10270 remove_dead_inserted_code.
10272 2017-09-06 Olivier Hainque <hainque@adacore.com>
10274 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
10275 for VxWorks 7. Adjust surrounding comments.
10277 2017-09-06 Richard Biener <rguenther@suse.de>
10279 * gimple-ssa-strength-reduction.c
10280 (find_candidates_dom_walker::before_dom_children): Also allow
10283 2017-09-06 Richard Biener <rguenther@suse.de>
10285 PR tree-optimization/82108
10286 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
10287 for gap in the non-permutation SLP case.
10289 2017-09-06 Martin Jambor <mjambor@suse.cz>
10291 PR tree-optimization/82078
10292 * tree-sra.c (sort_and_splice_var_accesses): Move call to
10293 add_access_to_work_queue...
10294 (build_accesses_from_assign): ...here.
10295 (propagate_all_subaccesses): Make sure racc is the group
10296 representative, if there is one.
10298 2017-09-06 Jakub Jelinek <jakub@redhat.com>
10300 PR middle-end/82095
10301 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
10304 2017-09-06 Richard Biener <rguenther@suse.de>
10306 * gimple-ssa-strength-reduction.c
10307 (find_candidates_dom_walker::before_doom_children): Use a
10308 type and not a mode check.
10310 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
10313 * config/arm/predicates.md (arm_general_adddi_operand): Create new
10315 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
10317 2017-09-05 Jeff Law <law@redhat.com>
10319 PR tree-optimization/64910
10320 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
10321 cases where we have 3 or more operands.
10323 2017-09-05 Jakub Jelinek <jakub@redhat.com>
10325 PR middle-end/81768
10326 * omp-low.c (lower_omp_for): Recompute tree invariant if
10327 gimple_omp_for_initial/final is ADDR_EXPR.
10329 PR middle-end/81768
10330 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
10331 into gimple val before gimplification fo the COND_EXPR.
10333 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
10335 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
10336 REGION_COPY argument.
10337 (thread_through_all_blocks): Remove unused argument to
10338 duplicate_thread_path.
10340 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10341 Alan Hayward <alan.hayward@arm.com>
10342 David Sherwood <david.sherwood@arm.com>
10344 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
10345 Take a scalar_mode rather than a machine_mode.
10346 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10347 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
10348 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
10349 (aarch64_gen_adjusted_ldpstp): Likewise.
10350 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
10352 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10353 Alan Hayward <alan.hayward@arm.com>
10354 David Sherwood <david.sherwood@arm.com>
10356 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
10357 Take a scalar_int_mode instead of a machine_mode.
10358 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10359 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10360 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10361 (aarch64_simd_attr_length_rglist): Delete.
10362 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
10363 a scalar_int_mode instead of a machine_mode.
10364 (aarch64_add_offset): Likewise.
10365 (aarch64_internal_mov_immediate): Likewise
10366 (aarch64_add_constant_internal): Likewise.
10367 (aarch64_add_constant): Likewise.
10368 (aarch64_movw_imm): Likewise.
10369 (aarch64_rtx_arith_op_extract_p): Likewise.
10370 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
10371 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
10372 Remove assert that the mode isn't a vector.
10373 (aarch64_output_scalar_simd_mov_immediate): Likewise.
10374 (aarch64_expand_mov_immediate): Update calls after above changes.
10375 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
10376 (aarch64_and_bitmask_imm): Check for scalar integer modes.
10377 (aarch64_move_imm): Likewise.
10378 (aarch64_can_const_movi_rtx_p): Likewise.
10379 (aarch64_strip_extend): Likewise.
10380 (aarch64_extr_rtx_p): Likewise.
10381 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
10382 a CONST_INT when the mode parameter is VOIDmode.
10383 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
10385 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10387 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
10388 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10389 (bitwise_type_for_mode): Update accordingly.
10391 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10393 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
10394 * stor-layout.c (mode_for_size_tree): Likewise.
10395 (mode_for_array): Update accordingly.
10396 (layout_decl): Likewise.
10397 (compute_record_mode): Likewise. Only set the mode once.
10399 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10401 * target.def (get_mask_mode): Change return type to opt_mode.
10403 * doc/tm.texi: Regenerate.
10404 * targhooks.h (default_get_mask_mode): Return an opt_mode.
10405 * targhooks.c (default_get_mask_mode): Likewise.
10406 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10407 * optabs-query.c (can_vec_mask_load_store_p): Update use of
10408 targetm.get_mask_mode.
10409 * tree.c (build_truth_vector_type): Likewise.
10411 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10413 * machmode.h (mode_for_vector): Return an opt_mode.
10414 * stor-layout.c (mode_for_vector): Likewise.
10415 (mode_for_int_vector): Update accordingly.
10416 (layout_type): Likewise.
10417 * config/i386/i386.c (emit_memmov): Likewise.
10418 (ix86_expand_set_or_movmem): Likewise.
10419 (ix86_expand_vector_init): Likewise.
10420 (ix86_get_mask_mode): Likewise.
10421 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10423 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10424 * expmed.c (extract_bit_field_1): Likewise.
10425 * expr.c (expand_expr_real_2): Likewise.
10426 * optabs-query.c (can_vec_perm_p): Likewise.
10427 (can_vec_mask_load_store_p): Likewise.
10428 * optabs.c (expand_vec_perm): Likewise.
10429 * targhooks.c (default_get_mask_mode): Likewise.
10430 * tree-vect-stmts.c (vectorizable_store): Likewise.
10431 (vectorizable_load): Likewise.
10432 (get_vectype_for_scalar_type_and_size): Likewise.
10434 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10436 * machmode.h (mode_for_int_vector): New function.
10437 * stor-layout.c (mode_for_int_vector): Likewise.
10438 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10439 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10440 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10441 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10442 (s390_expand_vcond): Likewise.
10444 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10446 * machmode.h (opt_machine_mode): New type.
10447 (opt_mode<T>): Allow construction from anything that can be
10449 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10450 (mode_for_size): Return an opt_machine_mode.
10451 * stor-layout.c (mode_for_size): Likewise.
10452 (mode_for_size_tree): Update call accordingly.
10453 (bitwise_mode_for_mode): Likewise.
10454 (make_fract_type): Likewise.
10455 (make_accum_type): Likewise.
10456 * caller-save.c (replace_reg_with_saved_mem): Update call
10458 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10459 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10460 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10461 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10462 * expmed.c (extract_bit_field_1): Likewise.
10463 * reload.c (get_secondary_mem): Likewise.
10464 * varasm.c (assemble_integer): Likewise.
10465 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10468 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10470 * machmode.h (decimal_float_mode_for_size): New function.
10471 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10472 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10473 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10474 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10475 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10476 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10478 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10480 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10481 (get_builtin_sync_mode): Likewise.
10482 (expand_ifn_atomic_compare_exchange): Likewise.
10483 (expand_builtin_atomic_clear): Likewise.
10484 (expand_builtin_atomic_test_and_set): Likewise.
10485 (fold_builtin_atomic_always_lock_free): Likewise.
10486 * calls.c (compute_argument_addresses): Likewise.
10487 (emit_library_call_value_1): Likewise.
10488 (store_one_arg): Likewise.
10489 * combine.c (combine_instructions): Likewise.
10490 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10491 * config/arm/arm.c (arm_function_value): Likewise.
10492 (aapcs_allocate_return_reg): Likewise.
10493 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10494 * config/i386/i386.c (construct_container): Likewise.
10495 (ix86_gimplify_va_arg): Likewise.
10496 (ix86_expand_sse_cmp): Likewise.
10497 (emit_memmov): Likewise.
10498 (emit_memset): Likewise.
10499 (expand_small_movmem_or_setmem): Likewise.
10500 (ix86_expand_pextr): Likewise.
10501 (ix86_expand_pinsr): Likewise.
10502 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10503 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10505 * config/mips/mips.c (mips_function_value_1) Likewise.
10506 (mips_block_move_straight): Likewise.
10507 (mips_expand_ins_as_unaligned_store): Likewise.
10508 * config/powerpcspe/powerpcspe.c
10509 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10510 (rs6000_darwin64_record_arg_flush): Likewise.
10511 * config/rs6000/rs6000.c
10512 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10513 (rs6000_darwin64_record_arg_flush): Likewise.
10514 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10515 (sparc_function_value_1): Likewise.
10516 * config/spu/spu.c (adjust_operand): Likewise.
10517 (spu_emit_branch_or_set): Likewise.
10518 (arith_immediate_p): Likewise.
10519 * emit-rtl.c (gen_lowpart_common): Likewise.
10520 * expr.c (expand_expr_real_1): Likewise.
10521 * function.c (assign_parm_setup_block): Likewise.
10522 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10523 * reload1.c (alter_reg): Likewise.
10524 * stor-layout.c (mode_for_vector): Likewise.
10525 (layout_type): Likewise.
10527 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10529 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10530 (spu_convert_move): Likewise.
10531 * lower-subreg.c (resolve_simple_move): Likewise.
10533 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10536 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10537 define_insn to a define_expand.
10538 (altivec_vsum2sws_direct): New define_insn.
10539 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10541 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10543 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10546 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10551 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10552 realign stack if stack alignment needed is less than incoming
10555 2017-09-05 Marek Polacek <polacek@redhat.com>
10558 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10561 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10563 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10565 2017-09-05 Richard Biener <rguenther@suse.de>
10567 PR tree-optimization/82084
10568 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10570 2017-09-05 Richard Biener <rguenther@suse.de>
10572 PR tree-optimization/82102
10573 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10575 2017-09-05 Martin Liska <mliska@suse.cz>
10577 PR tree-optimization/82032
10578 * tree-cfg.c (generate_range_test): New function.
10579 * tree-cfg.h (generate_range_test): Declared here.
10580 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10581 (cleanup_control_expr_graph): Use it.
10582 * tree-switch-conversion.c (try_switch_expansion): Remove
10584 (emit_case_nodes): Use generate_range_test.
10586 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10589 * config/i386/i386.md (*<btsc><mode>_mask): Add
10590 TARGET_USE_BT to insn constraint.
10591 (*btr<mode>_mask): Ditto.
10593 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10595 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10596 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10598 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10601 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10602 TARGET_NEON and TARGET_IWMMXT.
10603 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10604 TARGET_NEON and TARGET_IWMMXT.
10605 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10607 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10609 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10610 (ix86_rewrite_tls_address): Ditto.
10611 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10612 (ix86_rewrite_tls_address_1): Ditto.
10613 (ix86_rewrite_tls_address): Ditto.
10614 * config/i386/predicates.md (tls_address_pattern): New predicate.
10615 * config/i386/i386.md (TLS address splitter): New splitter.
10617 2017-09-04 Richard Biener <rguenther@suse.de>
10619 PR tree-optimization/82084
10620 * fold-const.h (can_native_encode_string_p): Declare.
10621 * fold-const.c (can_native_encode_string_p): Factor out from ...
10622 (native_encode_string): ... here.
10623 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10624 vectorizing stores from constants we later cannot handle.
10626 2017-09-04 Marek Polacek <polacek@redhat.com>
10629 * doc/invoke.texi: Update -Wtautological-compare documentation.
10631 2017-09-04 Jeff Law <law@redhat.com>
10633 PR tree-optimization/64910
10634 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10635 swap the first and last operand if the last is a constant.
10637 2017-09-04 Marek Polacek <polacek@redhat.com>
10640 * convert.c (do_narrow): When sanitizing signed integer overflows,
10641 bail out for signed types.
10642 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10644 2017-09-04 Richard Biener <rguenther@suse.de>
10646 PR tree-optimization/82060
10647 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10648 Move devirtualization after stmt folding and before EH/AB/noreturn
10649 cleanup to get the stmt refs canonicalized. Use a bool instead
10650 of gimple_modified_p since that doesn't work for NOPs. Schedule
10651 NOPs generated by folding for removal.
10653 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10654 Alan Hayward <alan.hayward@arm.com>
10655 David Sherwood <david.sherwood@arm.com>
10657 * coretypes.h (pad_direction): New enum.
10658 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10659 (FUNCTION_ARG_PADDING): Likewise.
10660 * target.def (function_arg_padding): New hook.
10661 * targhooks.h (default_function_arg_padding): Declare.
10662 * targhooks.c (default_function_arg_padding): New function.
10663 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10664 (TARGET_FUNCTION_ARG_PADDING): ...this.
10665 * doc/tm.texi: Regenerate.
10666 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10667 instead of direction.
10668 (compute_argument_addresses): Likewise.
10669 (load_register_parameters): Likewise.
10670 (emit_library_call_value_1): Likewise.
10671 (store_one_arg): Use targetm.calls.function_arg_padding instead
10672 of FUNCTION_ARG_PADDING.
10673 (must_pass_in_stack_var_size_or_pad): Likewise.
10674 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10675 (emit_group_store): Likewise.
10676 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10677 instead of FUNCTION_ARG_PADDING.
10678 (emit_push_insn): Likewise, and propagate enum change throughout
10680 * function.h (direction): Delete.
10681 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10683 * function.c (assign_parm_find_stack_rtl): Likewise.
10684 (assign_parm_setup_block_p): Likewise.
10685 (assign_parm_setup_block): Likewise.
10686 (gimplify_parameters): Likewise.
10687 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10688 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10690 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10691 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10692 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10693 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10694 (aarch64_function_arg_padding): ...this new function.
10695 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10696 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10697 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10698 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10699 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10700 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10701 (arm_pad_arg_upward): Replace with...
10702 (arm_function_arg_padding): ...this new function.
10703 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10705 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10706 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10707 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10708 (ia64_hpux_function_arg_padding): Replace with...
10709 (ia64_function_arg_padding): ...this new function. Use pad_direction
10710 instead of direction. Check for TARGET_HPUX.
10711 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10712 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10713 (iq2000_function_arg_padding): New function.
10714 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10715 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10716 (mips_function_arg_padding): ...this new function.
10717 (mips_pad_reg_upward): Update accordingly.
10718 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10719 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10720 targetm.calls.function_arg_padding.
10721 (FUNCTION_ARG_PADDING): Delete.
10722 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10723 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10724 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10725 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10726 (nios2_block_reg_padding): Return pad_direction instead of direction.
10727 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10728 instead of direction.
10729 (nios2_function_arg_padding): Likewise. Make static.
10730 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10731 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10732 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10733 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10734 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10735 (pa_function_arg_padding): Make static. Return pad_direction instead
10737 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10738 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10739 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10740 instead of direction. Use targetm.calls.function_arg_padding.
10741 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10742 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10743 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10744 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10745 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10747 (function_arg_padding): Rename to...
10748 (rs6000_function_arg_padding): ...this. Make static. Return
10749 pad_direction instead of direction.
10750 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10751 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10752 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10753 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10754 instead of direction. Use targetm.calls.function_arg_padding.
10755 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10756 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10757 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10758 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10759 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10760 (function_arg_padding): Rename to...
10761 (rs6000_function_arg_padding): ...this. Make static. Return
10762 pad_direction instead of direction.
10763 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10764 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10765 * config/s390/s390.c (s390_function_arg_padding): New function.
10766 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10767 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10768 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10769 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10770 (function_arg_padding): Rename to...
10771 (sparc_function_arg_padding): ...this. Make static. Return
10772 pad_direction instead of direction.
10773 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10774 * config/spu/spu.c (spu_function_arg_padding): New function.
10775 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10776 * system.h (FUNCTION_ARG_PADDING): Poison.
10778 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10779 Alan Hayward <alan.hayward@arm.com>
10780 David Sherwood <david.sherwood@arm.com>
10782 * target.def (modes_tieable_p): New hook.
10783 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10784 (TARGET_MODES_TIEABLE_P): ...this.
10785 * doc/tm.texi.in: Regenerate.
10786 * hooks.h (hook_bool_mode_mode_true): Declare.
10787 * hooks.c (hook_bool_mode_mode_true): New function.
10788 * combine.c (subst): Use targetm.modes_tieable_p instead of
10790 * dse.c (find_shift_sequence): Likewise.
10791 * expmed.c (extract_low_bits): Likewise.
10792 * lower-subreg.c: Include target.h.
10793 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10795 * rtlanal.c (rtx_cost): Likewise.
10796 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10797 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10798 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10799 (TARGET_MODES_TIEABLE_P): Redefine.
10800 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10801 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10802 (TARGET_MODES_TIEABLE_P): Redefine.
10803 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10804 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10805 (arc_modes_tieable_p): New function.
10806 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10807 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10808 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10809 (arm_modes_tieable_p): Make static.
10810 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10811 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10812 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10813 (TARGET_MODES_TIEABLE_P): Redefine.
10814 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10815 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10816 (TARGET_MODES_TIEABLE_P): Redefine.
10817 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10818 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10819 (cr16_modes_tieable_p): New function.
10820 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10821 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10822 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10823 (TRULY_NOOP_TRUNCATION): Update comment.
10824 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10825 (TRULY_NOOP_TRUNCATION): Update comment.
10826 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10827 (frv_modes_tieable_p): New function.
10828 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10829 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10830 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10831 (TARGET_MODES_TIEABLE_P): Redefine.
10832 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10833 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10834 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10835 (TARGET_MODES_TIEABLE_P): Redefine.
10836 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10837 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10838 (ia64_modes_tieable_p): New function.
10839 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10840 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10841 (iq2000_modes_tieable_p): New function.
10842 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10843 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10844 (lm32_modes_tieable_p): New function.
10845 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10846 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10847 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10848 (TARGET_MODES_TIEABLE_P): Redefine.
10849 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10850 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10851 (m32r_modes_tieable_p): New function.
10852 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10853 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10854 (m68k_modes_tieable_p): New function.
10855 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10856 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10857 (mcore_modes_tieable_p): New function.
10858 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10859 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10861 (TARGET_MODES_TIEABLE_P): Redefine.
10862 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10863 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10864 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10865 (TARGET_MODES_TIEABLE_P): Redefine.
10866 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10867 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10868 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10869 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10870 (mn10300_modes_tieable_p): ...this and make static.
10871 (TARGET_MODES_TIEABLE_P): Redefine.
10872 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10873 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10874 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10875 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10876 (msp430_modes_tieable_p): Make static.
10877 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10878 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10879 (TARGET_MODES_TIEABLE_P): Redefine.
10880 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10881 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10882 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10883 (TARGET_MODES_TIEABLE_P): Redefine.
10884 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10885 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10886 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10887 (TARGET_MODES_TIEABLE_P): Redefine.
10888 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10889 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10890 (pdp11_modes_tieable_p): New function.
10891 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10892 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10893 (rs6000_modes_tieable_p): New function.
10894 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10895 * config/powerpcspe/powerpcspe.md: Update comment.
10896 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10897 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10898 (TARGET_MODES_TIEABLE_P): Redefine.
10899 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10900 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10901 (rl78_modes_tieable_p): New function.
10902 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10903 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10904 (rs6000_modes_tieable_p): New function.
10905 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10906 * config/rs6000/rs6000.md: Update comment.
10907 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10908 * config/rx/rx.c (rx_modes_tieable_p): New function.
10909 (TARGET_MODES_TIEABLE_P): Redefine.
10910 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10911 * config/s390/s390.c (s390_modes_tieable_p): New function.
10912 (TARGET_MODES_TIEABLE_P): Redefine.
10913 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10914 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10915 (sh_modes_tieable_p): New function.
10916 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10917 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10918 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10919 (sparc_modes_tieable_p): Make static.
10920 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10921 * config/spu/spu.c (spu_modes_tieable_p): New function.
10922 (TARGET_MODES_TIEABLE_P): Redefine.
10923 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10924 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10925 (TARGET_MODES_TIEABLE_P): Redefine.
10926 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10927 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10928 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10929 * config/v850/v850.c (v850_modes_tieable_p): New function.
10930 (TARGET_MODES_TIEABLE_P): Redefine.
10931 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10932 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10933 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10934 (visium_modes_tieable_p): New function.
10935 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10936 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10937 (xtensa_modes_tieable_p): New function.
10938 * system.h (MODES_TIEABLE_P): Poison.
10940 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10941 Alan Hayward <alan.hayward@arm.com>
10942 David Sherwood <david.sherwood@arm.com>
10944 * target.def (hard_regno_mode_ok): New hook.
10945 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10946 (TARGET_HARD_REGNO_MODE_OK): ...this.
10947 * doc/tm.texi.in: Regenerate.
10948 * hooks.h (hook_bool_uint_mode_true): Declare.
10949 * hooks.c (hook_bool_uint_mode_true): New function.
10950 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10951 HARD_REGNO_MODE_OK.
10952 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10953 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10954 instead of HARD_REGNO_MODE_OK.
10955 * caller-save.c: Include target.h.
10956 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10957 HARD_REGNO_MODE_OK.
10958 * combine.c (can_combine_p): Likewise.
10959 (combinable_i3pat): Likewise.
10960 (can_change_dest_mode): Likewise.
10961 * expr.c (init_expr_target): Likewise.
10962 (convert_move): Likewise.
10963 (convert_modes): Likewise.
10964 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10965 (setup_prohibited_mode_move_regs): Likewise.
10966 * ira.h (target_ira): Likewise.
10967 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10968 * lra-constraints.c (process_alt_operands): Likewise.
10969 (split_reg): Likewise.
10970 * recog.c (peep2_find_free_register): Likewise.
10971 * ree.c (combine_reaching_defs): Likewise.
10972 * regcprop.c (maybe_mode_change): Likewise.
10973 * reginfo.c (init_reg_sets_1): Likewise.
10974 (choose_hard_reg_mode): Likewise.
10975 (simplifiable_subregs): Likewise.
10976 * regrename.c (check_new_reg_p): Likewise.
10977 * reload.c (find_valid_class): Likewise.
10978 (find_valid_class_1): Likewise.
10979 (reload_inner_reg_of_subreg): Likewise.
10980 (push_reload): Likewise.
10981 (combine_reloads): Likewise.
10982 (find_dummy_reload): Likewise.
10983 (find_reloads): Likewise.
10984 * reload1.c (find_reg): Likewise.
10985 (set_reload_reg): Likewise.
10986 (allocate_reload_reg): Likewise.
10987 (choose_reload_regs): Likewise.
10988 (reload_adjust_reg_for_temp): Likewise.
10989 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10990 (simplify_subreg_regno): Likewise.
10991 * sel-sched.c (init_regs_for_mode): Likewise.
10992 * varasm.c (make_decl_rtl): Likewise.
10993 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10994 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10995 HARD_REGNO_MODE_OK.
10996 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10997 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10998 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10999 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
11000 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
11001 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11002 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
11003 (arc_mode_class): Delete.
11004 (HARD_REGNO_MODE_OK): Delete.
11005 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11006 (arc_hard_regno_mode_ok): Rename old array to...
11007 (arc_hard_regno_mode_ok_modes): ...this.
11008 (arc_conditional_register_usage): Update accordingly.
11009 (arc_mode_class): Make static.
11010 (arc_hard_regno_mode_ok): New function.
11011 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
11012 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
11013 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11014 (arm_hard_regno_mode_ok): Make static.
11015 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
11016 HARD_REGNO_MODE_OK.
11017 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
11018 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
11019 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
11021 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11022 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
11023 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
11024 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
11025 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
11026 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11027 * config/bfin/predicates.md (valid_reg_operand): Use
11028 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
11029 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
11030 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
11031 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11032 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
11033 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
11034 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11035 (cr16_hard_regno_mode_ok): Make static and return a bool.
11036 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
11037 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11038 (cris_hard_regno_mode_ok): New function.
11039 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
11040 (epiphany_mode_class): Delete.
11041 (HARD_REGNO_MODE_OK): Delete.
11042 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
11043 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11044 (hard_regno_mode_ok): Rename to...
11045 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
11046 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
11047 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
11048 HARD_REGNO_MODE_OK.
11049 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
11050 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
11051 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11052 (frv_hard_regno_mode_ok): Make static and return a bool.
11053 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
11054 HARD_REGNO_MODE_OK.
11055 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
11056 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
11057 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
11058 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
11060 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11061 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
11062 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
11063 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
11065 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11066 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
11067 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11068 (ia64_hard_regno_mode_ok): New function.
11069 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
11070 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11071 (iq2000_hard_regno_mode_ok): New function.
11072 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
11073 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11074 (lm32_hard_regno_mode_ok): New function.
11075 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
11076 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
11077 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
11078 instead of HARD_REGNO_MODE_OK.
11079 (m32c_hard_regno_ok): Rename to...
11080 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
11081 (m32c_cannot_change_mode_class): Update accordingly.
11082 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11083 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
11084 (m32r_mode_class): Delete.
11085 (HARD_REGNO_MODE_OK): Delete.
11086 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11087 (m32r_hard_regno_mode_ok): Rename to...
11088 (m32r_hard_regno_modes): ...this.
11089 (m32r_mode_class): Make static.
11090 (m32r_hard_regno_mode_ok): New function.
11091 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
11092 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
11093 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11094 (m68k_hard_regno_mode_ok): Make static.
11095 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
11096 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11097 (mcore_hard_regno_mode_ok): New function.
11098 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
11099 (HARD_REGNO_MODE_OK): Delete.
11100 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
11102 (microblaze_hard_regno_mode_ok_p): ...this and make static.
11103 (microblaze_hard_regno_mode_ok): New function.
11104 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11105 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
11106 (mips_hard_regno_mode_ok): Delete.
11107 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
11108 (mips_hard_regno_mode_ok_p): ...this and make static.
11109 (mips_hard_regno_mode_ok_p): Rename to...
11110 (mips_hard_regno_mode_ok_uncached): ...this.
11111 (mips_hard_regno_mode_ok): New function.
11112 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
11113 of HARD_REGNO_MODE_OK.
11114 (mips_option_override): Update after above name changes.
11115 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11116 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
11117 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
11118 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
11119 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
11120 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11121 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
11122 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
11123 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
11124 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11125 (msp430_hard_regno_mode_ok): Make static and return a bool.
11126 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
11127 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
11128 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
11130 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11131 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
11132 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
11133 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
11134 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
11135 (PA_HARD_REGNO_MODE_OK): ...this
11136 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
11137 (PA_HARD_REGNO_MODE_OK): ...this.
11138 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11139 (pa_hard_regno_mode_ok): New function.
11140 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
11141 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11142 (pdp11_hard_regno_mode_ok): New function.
11143 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
11144 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
11146 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
11148 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11149 (rs6000_hard_regno_mode_ok): Rename to...
11150 (rs6000_hard_regno_mode_ok_uncached): ...this.
11151 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11152 (rs6000_hard_regno_mode_ok): New function.
11153 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
11154 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
11155 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
11156 (riscv_hard_regno_mode_ok): ...this and make static.
11157 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11158 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
11159 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
11160 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11161 (rl78_hard_regno_mode_ok): Make static and return bool.
11162 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
11163 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
11165 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
11166 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11167 (rs6000_hard_regno_mode_ok): Rename to...
11168 (rs6000_hard_regno_mode_ok_uncached): ...this.
11169 (rs6000_init_hard_regno_mode_ok): Update accordingly.
11170 (rs6000_hard_regno_mode_ok): New function.
11171 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
11172 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
11173 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11174 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
11175 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
11176 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
11177 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11178 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
11179 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
11180 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11181 (sh_hard_regno_mode_ok): Make static.
11182 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
11183 instead of HARD_REGNO_MODE_OK.
11184 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
11185 (sparc_mode_class): Delete.
11186 (HARD_REGNO_MODE_OK): Delete.
11187 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11188 (hard_regno_mode_classes): Make static.
11189 (sparc_mode_class): Likewise.
11190 (sparc_hard_regno_mode_ok): New function.
11191 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
11192 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
11193 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
11195 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11196 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
11197 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
11198 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
11199 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
11200 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11201 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
11202 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
11203 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
11204 (visium_hard_regno_mode_ok): New function.
11205 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
11206 instead of HARD_REGNO_MODE_OK.
11207 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
11208 (HARD_REGNO_MODE_OK): Delete.
11209 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
11210 (xtensa_hard_regno_mode_ok_p): ...this and make static.
11211 (xtensa_option_override): Update accordingly.
11212 (TARGET_HARD_REGNO_MODE_OK): Redefine.
11213 (xtensa_hard_regno_mode_ok): New function.
11214 * system.h (HARD_REGNO_MODE_OK): Poison.
11216 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11217 Alan Hayward <alan.hayward@arm.com>
11218 David Sherwood <david.sherwood@arm.com>
11220 * target.def (hard_regno_call_part_clobbered): New hook.
11221 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
11222 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
11223 * doc/tm.texi: Regenerate.
11224 * hooks.h (hook_bool_uint_mode_false): Declare.
11225 * hooks.c (hook_bool_uint_mode_false): New function.
11226 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11227 * cselib.c (cselib_process_insn): Use
11228 targetm.hard_regno_call_part_clobbered instead of
11229 HARD_REGNO_CALL_PART_CLOBBERED.
11230 * ira-conflicts.c (ira_build_conflicts): Likewise.
11231 * ira-costs.c (ira_tune_allocno_costs): Likewise.
11232 * lra-constraints.c (need_for_call_save_p): Likewise.
11233 * lra-lives.c: Include target.h.
11234 (check_pseudos_live_through_calls): Use
11235 targetm.hard_regno_call_part_clobbered instead of
11236 HARD_REGNO_CALL_PART_CLOBBERED.
11237 * regcprop.c: Include target.h.
11238 (copyprop_hardreg_forward_1): Use
11239 targetm.hard_regno_call_part_clobbered instead of
11240 HARD_REGNO_CALL_PART_CLOBBERED.
11241 * reginfo.c (choose_hard_reg_mode): Likewise.
11242 * regrename.c (check_new_reg_p): Likewise.
11243 * reload.c (find_equiv_reg): Likewise.
11244 * reload1.c (emit_reload_insns): Likewise.
11245 * sched-deps.c (deps_analyze_insn): Likewise.
11246 * sel-sched.c (init_regs_for_mode): Likewise.
11247 (mark_unavailable_hard_regs): Likewise.
11248 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
11249 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11250 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
11252 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11253 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11254 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
11256 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
11258 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11259 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11260 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
11262 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11263 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11264 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
11266 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11267 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
11269 * config/powerpcspe/powerpcspe.c
11270 (rs6000_hard_regno_call_part_clobbered): New function.
11271 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11272 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11273 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
11275 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11276 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11277 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
11279 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
11280 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
11281 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
11283 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11284 Alan Hayward <alan.hayward@arm.com>
11285 David Sherwood <david.sherwood@arm.com>
11287 * rtl.h (subreg_memory_offset): Declare.
11288 * emit-rtl.c (subreg_memory_offset): New function.
11289 * expmed.c (store_bit_field_1): Use it.
11290 * expr.c (undefined_operand_subword_p): Likewise.
11291 * simplify-rtx.c (simplify_subreg): Likewise.
11293 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
11295 PR rtl-optimization/57448
11298 * optabs.c (expand_atomic_load): Place compiler memory barriers if
11299 using atomic_load pattern.
11300 (expand_atomic_store): Likewise.
11302 2017-09-04 Jakub Jelinek <jakub@redhat.com>
11305 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
11306 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
11307 handling. Use replace_call_with_value with NULL instead of
11308 gsi_replace, unlink_stmt_vdef and release_defs.
11310 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
11313 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
11315 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
11318 * rtl.h (emit_library_call_value_1): Declare.
11319 (emit_library_call): Replace declaration with a series of overloads.
11320 Remove the parameter count argument.
11321 (emit_library_call_value): Likewise.
11322 * calls.c (emit_library_call_value_1): Make global. Replace varargs
11323 with an "rtx_mode_t *".
11324 (emit_library_call_value): Delete.
11325 (emit_library_call): Likewise.
11326 * asan.c (asan_emit_stack_protection): Update calls accordingly.
11327 (asan_emit_allocas_unpoison): Likewise.
11328 * builtins.c (expand_builtin_powi): Likewise.
11329 (expand_asan_emit_allocas_unpoison): Likewise.
11330 * cfgexpand.c (expand_main_function): Likewise.
11331 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
11332 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
11333 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
11334 * config/arm/arm.c (arm_trampoline_init): Likewise.
11335 (arm_call_tls_get_addr): Likewise.
11336 (arm_expand_divmod_libfunc): Likewise.
11337 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
11338 (smulsi3_highpart): Likewise.
11339 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
11340 (c6x_expand_compare): Likewise.
11341 (c6x_expand_movmem): Likewise.
11342 * config/frv/frv.c (frv_trampoline_init): Likewise.
11343 * config/i386/i386.c (ix86_trampoline_init): Likewise.
11344 (ix86_expand_divmod_libfunc): Likewise.
11345 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
11346 (ia64_expand_compare): Likewise.
11347 (ia64_profile_hook): Likewise.
11348 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
11349 (nonlocal_goto): Likewise.
11350 (restore_stack_nonlocal): Likewise.
11351 * config/m32r/m32r.c (block_move_call): Likewise.
11352 (m32r_trampoline_init): Likewise.
11353 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
11354 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
11355 (m68k_call_m68k_read_tp): Likewise.
11356 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
11357 (microblaze_expand_divide): Likewise.
11358 * config/mips/mips.h (mips_args): Likewise.
11359 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
11360 (MIPS_ICACHE_SYNC): Likewise.
11361 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
11362 (nios2_trampoline_init): Likewise.
11363 * config/pa/pa.c (hppa_tls_call): Likewise.
11364 (pa_trampoline_init): Likewise.
11365 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
11366 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
11367 (expand_strn_compare): Likewise.
11368 (rs6000_generate_compare): Likewise.
11369 (rs6000_expand_float128_convert): Likewise.
11370 (output_profile_hook): Likewise.
11371 (rs6000_trampoline_init): Likewise.
11372 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
11373 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
11374 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
11375 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
11376 (rs6000_generate_compare): Likewise.
11377 (rs6000_expand_float128_convert): Likewise.
11378 (output_profile_hook): Likewise.
11379 (rs6000_trampoline_init): Likewise.
11380 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
11381 * config/sh/sh.c (sh_trampoline_init): Likewise.
11382 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
11383 (sparc_emit_float_lib_cmp): Likewise.
11384 (sparc32_initialize_trampoline): Likewise.
11385 (sparc64_initialize_trampoline): Likewise.
11386 (sparc_profile_hook): Likewise.
11387 * config/spu/spu.c (ea_load_store): Likewise.
11388 * config/spu/spu.md (floatunssidf2): Likewise.
11389 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
11390 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
11391 * config/visium/visium.c (expand_block_move_4): Likewise.
11392 (expand_block_move_2): Likewise.
11393 (expand_block_move_1): Likewise.
11394 (expand_block_set_4): Likewise.
11395 (expand_block_set_2): Likewise.
11396 (expand_block_set_1): Likewise.
11397 (visium_trampoline_init): Likewise.
11398 (visium_profile_hook): Likewise.
11399 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
11400 (xtensa_setup_frame_addresses): Likewise.
11401 (xtensa_trampoline_init): Likewise.
11402 * except.c (sjlj_emit_function_enter): Likewise.
11403 (sjlj_emit_function_exit): Likewise.
11404 * explow.c (allocate_dynamic_stack_space): Likewise.
11405 (probe_stack_range): Likewise.
11406 * expr.c (convert_mode_scalar): Likewise.
11407 * optabs.c (expand_binop): Likewise.
11408 (expand_twoval_binop_libfunc): Likewise.
11409 (expand_unop): Likewise.
11410 (prepare_cmp_insn): Likewise.
11411 (prepare_float_lib_cmp): Likewise.
11412 (expand_float): Likewise.
11413 (expand_fix): Likewise.
11414 (expand_fixed_convert): Likewise.
11415 (maybe_emit_sync_lock_test_and_set): Likewise.
11416 (expand_atomic_compare_and_swap): Likewise.
11417 (expand_mem_thread_fence): Likewise.
11418 (expand_atomic_fetch_op): Likewise.
11420 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11422 * doc/generic.texi (OpenACC): Adjust URL.
11423 * doc/invoke.texi (C Dialect Options): Ditto.
11425 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11427 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11428 predicate for operand 1. Add (m,<S>) constraint.
11429 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11430 Prevent memory operand 1 with register operand 2.
11432 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11434 PR rtl-optimization/82024
11435 * combine.c (try_combine): If the combination result is a PARALLEL,
11436 and we only need to retain the SET in there that would be placed
11437 at I2, check that we can place that at I3 instead, before doing so.
11439 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11442 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11444 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11445 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11448 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11449 Jeff Law <law@redhat.com>
11451 * varasm.c (bss_initializer_p): Do not put constants into .bss
11452 (categorize_decl_for_section): Handle bss_initializer_p returning
11453 false when DECL_INITIAL is NULL.
11455 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11458 * config/s390/s390.c (s390_can_inline_p): New function.
11460 2017-09-01 Jeff Law <law@redhat.com>
11462 PR tree-optimization/82052
11463 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11464 Always initialize the returned slot after a hash table miss
11465 when INSERT is true.
11467 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11469 * config/s390/s390.md (mem_signal_fence): Remove.
11470 * doc/md.texi (mem_signal_fence): Remove.
11471 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11473 * target-insns.def (mem_signal_fence): Remove.
11475 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11478 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11481 * asan.c (create_odr_indicator): Strip name encoding from assembler
11482 name before appending it after __odr_asan_.
11484 2017-09-01 Martin Liska <mliska@suse.cz>
11486 PR tree-optimization/82059
11487 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11488 frequency only when an edge is redirected.
11490 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11492 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11493 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11494 (arc_conditional_register_usage): Remove ARC600 lp_count
11496 (arc_file_start): Emit Tag_ARC_CPU_variation.
11497 (arc_can_use_doloop_p): New conditions to use ZOLs.
11498 (hwloop_fail): New function.
11499 (hwloop_optimize): Likewise.
11500 (hwloop_pattern_reg): Likewise.
11501 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11502 (arc_reorg_loops): New function, calls reorg_loops.
11503 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11504 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11506 (arc_loop_hazard): Remove function, functionality moved into
11508 (arc_hazard): Remove arc_loop_hazard call.
11509 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11510 into hwloop_optimize.
11511 (arc_label_align): Remove ZOL handling.
11512 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11513 * config/arc/arc.md (doloop_begin): Remove pattern.
11514 (doloop_begin_i): Likewise.
11515 (doloop_end_i): Likewise.
11516 (doloop_fallback): Likewise.
11517 (doloop_fallback_m): Likewise.
11518 (doloop_end): Reimplement expand.
11519 (arc_lp): New pattern for LP instruction.
11520 (loop_end): New pattern.
11521 (loop_fail): Likewise.
11522 (decrement_and_branch_until_zero): Likewise.
11523 * config/arc/arc.opt (mlpc-width): New option.
11524 * doc/invoke.texi (mlpc-width): Document option.
11526 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11528 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11529 (arc_ccfsm_advance): Fix checking for delay slots.
11530 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11532 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11534 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11536 (movhi_insn): Update store instruction constraint which are saving
11537 6-bit short immediates.
11538 (movsi_insn): Consider also short scaled load operations.
11539 (zero_extendhisi2_i): Use Usd constraint instead of T.
11540 (extendhisi2_i): Add q constraint.
11541 (arc_clzsi2): Add type and length attributes.
11542 (arc_ctzsi2): Likewise.
11543 * config/arc/constraints.md (Usc): Update constraint, the
11544 assembler can parse two relocations for a single instruction.
11546 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11548 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11549 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11551 2017-08-31 Olivier Hainque <hainque@adacore.com>
11553 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11554 match as powerpc-wrs-vxworks*.
11556 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11558 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11559 register constraint for by-element operand.
11560 (aarch64_mls_elt_merge<mode>): Likewise.
11562 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11564 * config/arc/arc.c (arc_can_follow_jump): Check for short
11567 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11569 * config.gcc: Use g.opt for arc.
11570 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11571 functionality moved to ...
11572 (legitimate_scaled_address_p): New function, ...here.
11573 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11574 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11575 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11577 (arc_override_options): Handle G option.
11578 (arc_output_pic_addr_const): Correct function definition.
11579 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11580 (arc_decl_anon_ns_mem_p): Delete.
11581 (arc_in_small_data_p): Overhaul this function to take into
11582 consideration the value given via G option.
11583 (arc_rewrite_small_data_1): Renamed and corrected old
11584 arc_rewrite_small_data function.
11585 (arc_rewrite_small_data): New function.
11586 (small_data_pattern): Don't use pic_offset_table_rtx.
11587 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11588 * config/arc/simdext.md (movmisalignv2hi): Use
11589 prepare_move_operands function.
11591 (movmisalign*): Likewise.
11592 * doc/invoke.texi (ARC options): Document -G option.
11594 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11596 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11598 * config/arc/arc.c (arc_print_operand): Output scalled address for
11599 sdata whenever is possible.
11600 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11601 load/stores are available.
11602 (compact_sda_memory_operand): Check for the alignment required by
11603 code density instructions.
11604 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11606 * config/arc/constraints.md (Usd): Update constraint.
11607 (Us0): New constraint.
11608 (Usc): Update constraint.
11610 2017-08-31 Richard Biener <rguenther@suse.de>
11612 PR middle-end/82054
11613 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11614 function only once.
11616 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11618 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11619 Resize type_signature.
11621 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11622 Alan Hayward <alan.hayward@arm.com>
11623 David Sherwood <david.sherwood@arm.com>
11625 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11626 subregs whose inner modes can be stored in GPRs.
11627 (aarch64_classify_index): Likewise.
11629 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11630 Alan Hayward <alan.hayward@arm.com>
11631 David Sherwood <david.sherwood@arm.com>
11633 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11634 (V_INT_EQUIV): ...this.
11635 (v_cmp_result): Rename to...
11636 (v_int_equiv): ...this.
11637 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11638 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11639 (copysign<mode>3): Likewise.
11640 (aarch64_simd_bsl<mode>_internal): Likewise.
11641 (aarch64_simd_bsl<mode>): Likewise.
11642 (vec_cmp<mode><mode>): Likewise.
11643 (vcond<mode><mode>): Likewise.
11644 (vcond<v_cmp_mixed><mode>): Likewise.
11645 (vcondu<mode><v_cmp_mixed>): Likewise.
11646 (aarch64_cm<optab><mode>): Likewise.
11647 (aarch64_cmtst<mode>): Likewise.
11648 (aarch64_fac<optab><mode>): Likewise.
11649 (vec_perm_const<mode>): Likewise.
11650 (vcond_mask_<mode><v_cmp_result>): Rename to...
11651 (vcond_mask_<mode><v_int_equiv>): ...this.
11652 (vec_cmp<mode><v_cmp_result>): Rename to...
11653 (vec_cmp<mode><v_int_equiv>): ...this.
11655 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11656 Alan Hayward <alan.hayward@arm.com>
11657 David Sherwood <david.sherwood@arm.com>
11659 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11661 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11662 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11663 (UNSPEC_LD4_DREG): New unspecs.
11664 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11665 (aarch64_ld2<mode>_dreg_be): Replace with...
11666 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11668 (aarch64_ld3<mode>_dreg_le)
11669 (aarch64_ld3<mode>_dreg_be): Replace with...
11670 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11672 (aarch64_ld4<mode>_dreg_le)
11673 (aarch64_ld4<mode>_dreg_be): Replace with...
11674 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11677 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11679 PR tree-optimization/81987
11680 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11681 insert an initializer in a location not dominated by the stride
11684 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11686 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11687 on the entire header of the finally block in the fallthru case.
11689 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11691 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11693 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11695 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11696 rs6000_emit_move_from_cr and call renamed function.
11697 (rs6000_emit_prologue): Call renamed functions.
11698 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11699 movesi_from_cr, remove volatile CRs.
11701 2017-08-30 Jon Beniston <jon@beniston.com>
11702 Richard Biener <rguenther@suse.de>
11704 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11705 of VECTOR_MODE_P check.
11706 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11707 element vector types.
11709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11711 * df.h (df_read_modify_subreg_p): Remove in favor of...
11712 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11713 const_rtx instead of an rtx.
11714 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11715 * df-problems.c (df_word_lr_mark_ref): Likewise.
11716 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11717 (mark_pseudo_reg_dead): Likewise.
11718 (mark_ref_dead): Likewise.
11719 * reginfo.c (init_subregs_of_mode): Likewise.
11720 * sched-deps.c (sched_analyze_1): Likewise.
11721 * df-scan.c (df_def_record_1): Likewise.
11722 (df_uses_record): Likewise.
11723 (df_read_modify_subreg_p): Remove in favor of...
11724 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11725 const_rtx instead of an rtx.
11727 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11728 Alan Hayward <alan.hayward@arm.com>
11729 David Sherwood <david.sherwood@arm.com>
11731 * rtl.h (partial_subreg_p): New function.
11732 * caller-save.c (save_call_clobbered_regs): Use it.
11733 * calls.c (expand_call): Likewise.
11734 * combine.c (combinable_i3pat): Likewise.
11735 (simplify_set): Likewise.
11736 (make_extraction): Likewise.
11737 (make_compound_operation_int): Likewise.
11738 (gen_lowpart_or_truncate): Likewise.
11739 (force_to_mode): Likewise.
11740 (make_field_assignment): Likewise.
11741 (reg_truncated_to_mode): Likewise.
11742 (record_truncated_value): Likewise.
11743 (move_deaths): Likewise.
11744 * cse.c (record_jump_cond): Likewise.
11745 (cse_insn): Likewise.
11746 * cselib.c (cselib_lookup_1): Likewise.
11747 * expmed.c (extract_bit_field_using_extv): Likewise.
11748 * function.c (assign_parm_setup_reg): Likewise.
11749 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11750 * ira-build.c (create_insn_allocnos): Likewise.
11751 * lra-coalesce.c (merge_pseudos): Likewise.
11752 * lra-constraints.c (match_reload): Likewise.
11753 (simplify_operand_subreg): Likewise.
11754 (curr_insn_transform): Likewise.
11755 * lra-lives.c (process_bb_lives): Likewise.
11756 * lra.c (new_insn_reg): Likewise.
11757 (lra_substitute_pseudo): Likewise.
11758 * regcprop.c (mode_change_ok): Likewise.
11759 (maybe_mode_change): Likewise.
11760 (copyprop_hardreg_forward_1): Likewise.
11761 * reload.c (push_reload): Likewise.
11762 (find_reloads): Likewise.
11763 (find_reloads_subreg_address): Likewise.
11764 * reload1.c (alter_reg): Likewise.
11765 (eliminate_regs_1): Likewise.
11766 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11768 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11770 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11771 back to if statements, including unpack.
11773 2017-08-30 Martin Liska <mliska@suse.cz>
11775 PR inline-asm/82001
11776 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11778 (func_checker::compare_asm_inputs_outputs): ... this function.
11779 (func_checker::compare_gimple_asm): Use the function to compare
11780 also ASM constrains.
11781 * ipa-icf-gimple.h: Rename the function.
11783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11784 Alan Hayward <alan.hayward@arm.com>
11785 David Sherwood <david.sherwood@arm.com>
11787 * coretypes.h (complex_mode): New type.
11788 * gdbhooks.py (build_pretty_printer): Handle it.
11789 * machmode.h (complex_mode): New class.
11790 (complex_mode::includes_p): New function.
11791 (is_complex_int_mode): Likewise.
11792 (is_complex_float_mode): Likewise.
11793 * genmodes.c (get_mode_class): Handle complex mode classes.
11794 * function.c (expand_function_end): Use is_complex_int_mode.
11796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11797 Alan Hayward <alan.hayward@arm.com>
11798 David Sherwood <david.sherwood@arm.com>
11800 * coretypes.h (scalar_mode_pod): New typedef.
11801 * gdbhooks.py (build_pretty_printer): Handle it.
11802 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11803 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11804 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11805 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11806 as_a <scalar_mode>.
11808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11809 Alan Hayward <alan.hayward@arm.com>
11810 David Sherwood <david.sherwood@arm.com>
11812 * machmode.h (mode_for_vector): Take a scalar_mode instead
11814 * stor-layout.c (mode_for_vector): Likewise.
11815 * explow.c (promote_mode): Use as_a <scalar_mode>.
11816 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11819 Alan Hayward <alan.hayward@arm.com>
11820 David Sherwood <david.sherwood@arm.com>
11822 * target.def (preferred_simd_mode): Take a scalar_mode
11823 instead of a machine_mode.
11824 * targhooks.h (default_preferred_simd_mode): Likewise.
11825 * targhooks.c (default_preferred_simd_mode): Likewise.
11826 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11827 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11828 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11829 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11830 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11831 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11832 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11833 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11835 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11836 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11837 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11838 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11839 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11840 * doc/tm.texi: Regenerate.
11841 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11844 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11845 Alan Hayward <alan.hayward@arm.com>
11846 David Sherwood <david.sherwood@arm.com>
11848 * target.def (scalar_mode_supported_p): Take a scalar_mode
11849 instead of a machine_mode.
11850 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11851 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11852 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11853 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11854 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11855 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11856 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11857 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11858 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11859 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11860 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11861 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11862 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11863 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11865 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11866 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11867 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11868 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11869 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11871 * doc/tm.texi: Regenerate.
11873 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11874 Alan Hayward <alan.hayward@arm.com>
11875 David Sherwood <david.sherwood@arm.com>
11877 * coretypes.h (opt_scalar_mode): New typedef.
11878 * gdbhooks.py (build_pretty_printers): Handle it.
11879 * machmode.h (mode_iterator::get_2xwider): Add overload for
11881 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11883 * expr.c (convert_mode_scalar): Likewise.
11884 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11885 * optabs.c (expand_float): Likewise.
11886 (expand_fix): Likewise.
11887 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11890 Alan Hayward <alan.hayward@arm.com>
11891 David Sherwood <david.sherwood@arm.com>
11893 * optabs.c (expand_float): Explicitly check for scalars before
11894 using a branching expansion.
11895 (expand_fix): Likewise.
11897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11898 Alan Hayward <alan.hayward@arm.com>
11899 David Sherwood <david.sherwood@arm.com>
11901 * expr.c (convert_mode): Split scalar handling out into...
11902 (convert_mode_scalar): ...this new function. Treat the modes
11905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11906 Alan Hayward <alan.hayward@arm.com>
11907 David Sherwood <david.sherwood@arm.com>
11909 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11911 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11913 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11914 Alan Hayward <alan.hayward@arm.com>
11915 David Sherwood <david.sherwood@arm.com>
11917 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11918 rather than a machine_mode.
11919 (fixed_from_string): Likewise.
11920 (fixed_convert): Likewise.
11921 (fixed_convert_from_int): Likewise.
11922 (fixed_convert_from_real): Likewise.
11923 (real_convert_from_fixed): Likewise.
11924 * fixed-value.c (fixed_from_double_int): Likewise.
11925 (fixed_from_string): Likewise.
11926 (fixed_convert): Likewise.
11927 (fixed_convert_from_int): Likewise.
11928 (fixed_convert_from_real): Likewise.
11929 (real_convert_from_fixed): Likewise.
11930 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11933 Alan Hayward <alan.hayward@arm.com>
11934 David Sherwood <david.sherwood@arm.com>
11936 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11937 of separate mode class checks. Do not allow vector modes here.
11938 (immed_wide_int_const): Use as_a <scalar_mode>.
11939 * explow.c (trunc_int_for_mode): Likewise.
11940 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11941 (wi::shwi): Likewise.
11942 (wi::min_value): Likewise.
11943 (wi::max_value): Likewise.
11944 * dwarf2out.c (loc_descriptor): Likewise.
11945 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11946 for CONST_WIDE_INT.
11948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11949 Alan Hayward <alan.hayward@arm.com>
11950 David Sherwood <david.sherwood@arm.com>
11952 * tree.h (SCALAR_TYPE_MODE): New macro.
11953 * expr.c (expand_expr_addr_expr_1): Use it.
11954 (expand_expr_real_2): Likewise.
11955 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11956 (fold_convert_const_fixed_from_int): Likewise.
11957 (fold_convert_const_fixed_from_real): Likewise.
11958 (native_encode_fixed): Likewise
11959 (native_encode_complex): Likewise
11960 (native_encode_vector): Likewise.
11961 (native_interpret_fixed): Likewise.
11962 (native_interpret_real): Likewise.
11963 (native_interpret_complex): Likewise.
11964 (native_interpret_vector): Likewise.
11965 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11966 (simd_clone_adjust_argument_types): Likewise.
11967 (simd_clone_init_simd_arrays): Likewise.
11968 (simd_clone_adjust): Likewise.
11969 * stor-layout.c (layout_type): Likewise.
11970 * tree.c (build_minus_one_cst): Likewise.
11971 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11972 * tree-inline.c (estimate_move_cost): Likewise.
11973 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11974 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11975 (vectorizable_reduction): Likewise.
11976 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11977 (vect_recog_mixed_size_cond_pattern): Likewise.
11978 (check_bool_pattern): Likewise.
11979 (adjust_bool_pattern): Likewise.
11980 (search_type_for_mask_1): Likewise.
11981 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11982 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11983 (vectorizable_load): Likewise.
11984 (vectorizable_store): Likewise.
11985 * ubsan.c (ubsan_encode_value): Likewise.
11986 * varasm.c (output_constant): Likewise.
11988 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11989 Alan Hayward <alan.hayward@arm.com>
11990 David Sherwood <david.sherwood@arm.com>
11992 * coretypes.h (scalar_mode): New class.
11993 * machmode.h (scalar_mode): Likewise.
11994 (scalar_mode::includes_p): New function.
11995 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11996 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11997 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11998 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11999 * expmed.c (store_bit_field_1): Likewise.
12000 (extract_bit_field_1): Likewise.
12001 * expr.c (write_complex_part): Likewise.
12002 (read_complex_part): Likewise.
12003 (emit_move_complex_push): Likewise.
12004 (expand_expr_real_2): Likewise.
12005 * function.c (assign_parm_setup_reg): Likewise.
12006 (assign_parms_unsplit_complex): Likewise.
12007 * optabs.c (expand_binop): Likewise.
12008 * rtlanal.c (subreg_get_info): Likewise.
12009 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12010 * varasm.c (output_constant_pool_2): Likewise.
12012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12013 Alan Hayward <alan.hayward@arm.com>
12014 David Sherwood <david.sherwood@arm.com>
12016 * expmed.c (extract_high_half): Use scalar_int_mode and remove
12018 (expmed_mult_highpart_optab): Likewise.
12019 (expmed_mult_highpart): Likewise.
12021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12022 Alan Hayward <alan.hayward@arm.com>
12023 David Sherwood <david.sherwood@arm.com>
12025 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
12026 instead of a machine_mode.
12027 (builtin_memset_read_str): Likewise.
12028 * builtins.c (c_readstr): Likewise.
12029 (builtin_memcpy_read_str): Likewise.
12030 (builtin_strncpy_read_str): Likewise.
12031 (builtin_memset_read_str): Likewise.
12032 (builtin_memset_gen_str): Likewise.
12033 (expand_builtin_signbit): Use scalar_int_mode for local variables.
12034 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
12035 instead of a machine_mode.
12036 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
12038 (make_extraction): Likewise.
12039 (try_widen_shift_mode): Take and return scalar_int_modes instead
12041 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
12042 a scalar_int_mode instead of a machine_mode.
12043 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
12044 (avr_addr_space_pointer_mode): Likewise.
12045 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
12046 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
12047 (msp430_unwind_word_mode): Likewise.
12048 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
12049 (spu_addr_space_pointer_mode): Likewise.
12050 (spu_addr_space_address_mode): Likewise.
12051 (spu_libgcc_cmp_return_mode): Likewise.
12052 (spu_libgcc_shift_count_mode): Likewise.
12053 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
12054 (rl78_addr_space_pointer_mode): Likewise.
12055 (fl78_unwind_word_mode): Likewise.
12056 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
12058 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
12059 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
12060 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
12061 (mips_valid_pointer_mode): Likewise.
12062 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
12063 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
12064 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
12066 (ft32_addr_space_address_mode): Likewise.
12067 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
12068 scalar_int_mode instead of a machine_mode.
12069 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
12071 (m32c_addr_space_address_mode): Likewise.
12072 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
12073 (rs6000_eh_return_filter_mode): Likewise.
12074 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
12075 (rs6000_eh_return_filter_mode): Likewise.
12076 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
12077 (s390_libgcc_shift_count_mode): Likewise.
12078 (s390_unwind_word_mode): Likewise.
12079 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
12081 * target.def (mode_rep_extended): Likewise.
12082 (valid_pointer_mode): Likewise.
12083 (addr_space.valid_pointer_mode): Likewise.
12084 (eh_return_filter_mode): Return a scalar_int_mode rather than
12086 (libgcc_cmp_return_mode): Likewise.
12087 (libgcc_shift_count_mode): Likewise.
12088 (unwind_word_mode): Likewise.
12089 (addr_space.pointer_mode): Likewise.
12090 (addr_space.address_mode): Likewise.
12091 * doc/tm.texi: Regenerate.
12092 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
12094 (do_jump): Use scalar_int_mode for local variables.
12095 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
12096 rather than a machine_mode.
12097 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
12098 (scompare_loc_descriptor_wide): Likewise.
12099 (scompare_loc_descriptor_narrow): Likewise.
12100 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
12102 * except.c (sjlj_emit_dispatch_table): Likewise.
12103 (expand_builtin_eh_copy_values): Likewise.
12104 * explow.c (convert_memory_address_addr_space_1): Likewise.
12105 Take a scalar_int_mode rather than a machine_mode.
12106 (convert_memory_address_addr_space): Take a scalar_int_mode rather
12107 than a machine_mode.
12108 (memory_address_addr_space): Use scalar_int_mode for local variables.
12109 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
12110 rather than a machine_mode.
12111 * expmed.c (mask_rtx): Likewise.
12112 (init_expmed_one_conv): Likewise.
12113 (expand_mult_highpart_adjust): Likewise.
12114 (extract_high_half): Likewise.
12115 (expmed_mult_highpart_optab): Likewise.
12116 (expmed_mult_highpart): Likewise.
12117 (expand_smod_pow2): Likewise.
12118 (expand_sdiv_pow2): Likewise.
12119 (emit_store_flag_int): Likewise.
12120 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
12122 (extract_low_bits): Likewise.
12123 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
12125 * expr.c (pieces_addr::adjust): Likewise.
12126 (can_store_by_pieces): Likewise.
12127 (store_by_pieces): Likewise.
12128 (clear_by_pieces_1): Likewise.
12129 (expand_expr_addr_expr_1): Likewise.
12130 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
12131 (expand_expr_real_1): Likewise.
12132 (try_casesi): Likewise.
12133 * final.c (shorten_branches): Likewise.
12134 * fold-const.c (fold_convert_const_int_from_fixed): Change the
12135 type of "mode" to machine_mode.
12136 * internal-fn.c (expand_arith_overflow_result_store): Take a
12137 scalar_int_mode rather than a machine_mode.
12138 (expand_mul_overflow): Use scalar_int_mode for local variables.
12139 * loop-doloop.c (doloop_modify): Likewise.
12140 (doloop_optimize): Likewise.
12141 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
12142 than a machine_mode.
12143 (expand_doubleword_shift_condmove): Likewise.
12144 (expand_doubleword_shift): Likewise.
12145 (expand_doubleword_clz): Likewise.
12146 (expand_doubleword_popcount): Likewise.
12147 (expand_doubleword_parity): Likewise.
12148 (expand_absneg_bit): Use scalar_int_mode for local variables.
12149 (prepare_float_lib_cmp): Likewise.
12150 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
12151 rather than a machine_mode.
12152 (convert_memory_address_addr_space): Likewise.
12153 (get_mode_bounds): Likewise.
12154 (get_address_mode): Return a scalar_int_mode rather than a
12156 * rtlanal.c (get_address_mode): Likewise.
12157 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
12158 than a machine_mode.
12159 * targhooks.c (default_mode_rep_extended): Likewise.
12160 (default_valid_pointer_mode): Likewise.
12161 (default_addr_space_valid_pointer_mode): Likewise.
12162 (default_eh_return_filter_mode): Return a scalar_int_mode rather
12163 than a machine_mode.
12164 (default_libgcc_cmp_return_mode): Likewise.
12165 (default_libgcc_shift_count_mode): Likewise.
12166 (default_unwind_word_mode): Likewise.
12167 (default_addr_space_pointer_mode): Likewise.
12168 (default_addr_space_address_mode): Likewise.
12169 * targhooks.h (default_eh_return_filter_mode): Likewise.
12170 (default_libgcc_cmp_return_mode): Likewise.
12171 (default_libgcc_shift_count_mode): Likewise.
12172 (default_unwind_word_mode): Likewise.
12173 (default_addr_space_pointer_mode): Likewise.
12174 (default_addr_space_address_mode): Likewise.
12175 (default_mode_rep_extended): Take a scalar_int_mode rather than
12177 (default_valid_pointer_mode): Likewise.
12178 (default_addr_space_valid_pointer_mode): Likewise.
12179 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
12181 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
12182 rather than a machine_mode.
12183 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
12184 for local variables.
12185 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12186 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
12187 than a machine_mode.
12189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12190 Alan Hayward <alan.hayward@arm.com>
12191 David Sherwood <david.sherwood@arm.com>
12193 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
12194 the mode argument to scalar_int_mode.
12195 (do_jump_by_parts_zero_rtx): Likewise.
12196 (do_jump_by_parts_equality_rtx): Likewise.
12197 (do_jump_by_parts_greater): Take a mode argument.
12198 (do_jump_by_parts_equality): Likewise.
12199 (do_jump_1): Update calls accordingly.
12201 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12202 Alan Hayward <alan.hayward@arm.com>
12203 David Sherwood <david.sherwood@arm.com>
12205 * is-a.h (safe_dyn_cast): New function.
12206 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
12207 (jump_table_for_label): Likewise.
12208 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
12209 instead of an rtx_insn *.
12210 (shorten_branches): Use dyn_cast instead of LABEL_P and
12211 JUMP_TABLE_DATA_P. Use jump_table_for_label and
12212 rtx_jump_table_data::get_data_mode.
12213 (final_scan_insn): Likewise.
12215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12216 Alan Hayward <alan.hayward@arm.com>
12217 David Sherwood <david.sherwood@arm.com>
12219 * combine.c (try_combine): Use is_a <scalar_int_mode> when
12220 trying to combine a full-register integer set with a subreg
12223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12224 Alan Hayward <alan.hayward@arm.com>
12225 David Sherwood <david.sherwood@arm.com>
12227 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
12228 that is always either address_mode or pointer_mode.
12230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12231 Alan Hayward <alan.hayward@arm.com>
12232 David Sherwood <david.sherwood@arm.com>
12234 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
12235 when the two are known to be equal.
12237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12238 Alan Hayward <alan.hayward@arm.com>
12239 David Sherwood <david.sherwood@arm.com>
12241 * simplify-rtx.c (simplify_const_unary_operation): Use
12242 is_a <scalar_int_mode> instead of checking for a nonzero
12243 precision. Forcibly convert op_mode to a scalar_int_mode
12244 in that case. More clearly differentiate the operand and
12245 result modes and use the former when deciding what the value
12246 of a count-bits operation should be. Use is_int_mode instead
12247 of checking for a MODE_INT. Remove redundant check for whether
12248 this mode has a zero precision.
12250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12251 Alan Hayward <alan.hayward@arm.com>
12252 David Sherwood <david.sherwood@arm.com>
12254 * optabs.c (widen_leading): Change the type of the mode argument
12255 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
12256 (widen_bswap): Likewise.
12257 (expand_parity): Likewise.
12258 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
12259 (expand_ffs): Likewise.
12260 (epand_unop): Check for scalar integer modes before calling the
12263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12264 Alan Hayward <alan.hayward@arm.com>
12265 David Sherwood <david.sherwood@arm.com>
12267 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
12269 (expand_expr_real_1): Update call accordingly.
12271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12272 Alan Hayward <alan.hayward@arm.com>
12273 David Sherwood <david.sherwood@arm.com>
12275 * expmed.c (store_bit_field_using_insv): Add op0_mode and
12276 value_mode arguments. Use scalar_int_mode internally.
12277 (store_bit_field_1): Rename the new integer mode from imode
12278 to op0_mode and use it instead of GET_MODE (op0). Update calls
12279 to store_split_bit_field, store_bit_field_using_insv and
12280 store_fixed_bit_field.
12281 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
12282 Use scalar_int_mode internally. Use a bit count rather than a mode
12283 when calculating the largest bit size for get_best_mode.
12284 Update calls to store_split_bit_field and store_fixed_bit_field_1.
12285 (store_fixed_bit_field_1): Add mode and value_mode arguments.
12286 Remove assertion that OP0 has a scalar integer mode.
12287 (store_split_bit_field): Add op0_mode and value_mode arguments.
12288 Update calls to extract_fixed_bit_field.
12289 (extract_bit_field_using_extv): Add an op0_mode argument.
12290 Use scalar_int_mode internally.
12291 (extract_bit_field_1): Rename the new integer mode from imode to
12292 op0_mode and use it instead of GET_MODE (op0). Update calls to
12293 extract_split_bit_field, extract_bit_field_using_extv and
12294 extract_fixed_bit_field.
12295 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
12296 to extract_split_bit_field and extract_fixed_bit_field_1.
12297 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
12298 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
12299 on the target mode.
12300 (extract_split_bit_field): Add an op0_mode argument. Update call
12301 to extract_fixed_bit_field.
12303 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12304 Alan Hayward <alan.hayward@arm.com>
12305 David Sherwood <david.sherwood@arm.com>
12307 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
12308 * explow.c (hard_function_value): Likewise.
12309 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
12310 convert_to_mode call outside the loop.
12311 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
12312 for the mode iterator. Require the mode specified by max_pieces
12314 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
12316 (copy_blkmode_to_reg): Likewise.
12317 (set_storage_via_setmem): Likewise.
12318 * optabs.c (prepare_cmp_insn): Likewise.
12319 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12320 * stor-layout.c (finish_bitfield_representative): Likewise.
12322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12323 Alan Hayward <alan.hayward@arm.com>
12324 David Sherwood <david.sherwood@arm.com>
12326 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
12327 * expr.c (convert_move): Use them.
12328 (convert_modes): Likewise.
12329 (store_expr_with_bounds): Likewise.
12331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12332 Alan Hayward <alan.hayward@arm.com>
12333 David Sherwood <david.sherwood@arm.com>
12335 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
12336 parameter for the mode of "x". Remove the "known_x", "known_mode"
12337 and "known_ret" arguments. Change the type of the mode argument
12338 to scalar_int_mode.
12339 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
12340 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
12341 (reg_num_sign_bit_copies_for_combine): Likewise.
12342 * rtlanal.c (nonzero_bits1): Likewise.
12343 (num_sign_bit_copies1): Likewise.
12344 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
12345 (reg_num_sign_bit_copies_general): Likewise.
12346 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
12347 (reg_nonzero_bits_general): Likewise.
12349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12350 Alan Hayward <alan.hayward@arm.com>
12351 David Sherwood <david.sherwood@arm.com>
12353 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
12354 than in subroutines. Return 1 for non-integer modes.
12355 (cached_num_sign_bit_copies): Change the type of the mode parameter
12356 to scalar_int_mode.
12357 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
12358 classes. Handle CONST_INT_P first and then check whether X also
12359 has a scalar integer mode. Check the same thing for inner registers
12360 of a SUBREG and for values that are being extended or truncated.
12362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12363 Alan Hayward <alan.hayward@arm.com>
12364 David Sherwood <david.sherwood@arm.com>
12366 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
12367 in subroutines. Return the mode mask for non-integer modes.
12368 (cached_nonzero_bits): Change the type of the mode parameter
12369 to scalar_int_mode.
12370 (nonzero_bits1): Likewise. Remove early exit for other mode
12371 classes. Handle CONST_INT_P first and then check whether X
12372 also has a scalar integer mode.
12374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12375 Alan Hayward <alan.hayward@arm.com>
12376 David Sherwood <david.sherwood@arm.com>
12378 * expr.c (widest_int_mode_for_size): Make the comment match the code.
12379 Return a scalar_int_mode and assert that the size is greater than
12381 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
12382 (op_by_pieces_d::op_by_pieces_d): Likewise.
12383 (op_by_pieces_d::run): Likewise.
12384 (can_store_by_pieces): Likewise.
12386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12387 Alan Hayward <alan.hayward@arm.com>
12388 David Sherwood <david.sherwood@arm.com>
12390 * combine.c (extract_left_shift): Add a mode argument and update
12392 (make_compound_operation_int): Change the type of the mode parameter
12393 to scalar_int_mode and update the call to extract_left_shift.
12395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12396 Alan Hayward <alan.hayward@arm.com>
12397 David Sherwood <david.sherwood@arm.com>
12399 * combine.c (simplify_and_const_int): Change the type of the mode
12400 parameter to scalar_int_mode.
12401 (simplify_and_const_int_1): Likewise. Update recursive call.
12403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12404 Alan Hayward <alan.hayward@arm.com>
12405 David Sherwood <david.sherwood@arm.com>
12407 * combine.c (simplify_compare_const): Check that the mode is a
12408 scalar_int_mode (rather than VOIDmode) before testing its
12410 (simplify_comparison): Move COMPARISON_P handling out of the
12411 loop and restrict the latter part of the loop to scalar_int_modes.
12412 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
12413 and when considering SUBREG_REGs. Use is_int_mode instead of
12414 checking GET_MODE_CLASS against MODE_INT.
12416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12417 Alan Hayward <alan.hayward@arm.com>
12418 David Sherwood <david.sherwood@arm.com>
12420 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12421 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12422 shift_unit_mode and for modes involved in scalar shifts.
12424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12425 Alan Hayward <alan.hayward@arm.com>
12426 David Sherwood <david.sherwood@arm.com>
12428 * combine.c (force_int_to_mode): New function, split out from...
12429 (force_to_mode): ...here. Keep xmode up-to-date and use it
12430 instead of GET_MODE (x).
12432 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12433 Alan Hayward <alan.hayward@arm.com>
12434 David Sherwood <david.sherwood@arm.com>
12436 * optabs-query.h (extraction_insn::struct_mode): Change type to
12437 opt_scalar_int_mode and update comment.
12438 (extraction_insn::field_mode): Change type to scalar_int_mode.
12439 (extraction_insn::pos_mode): Likewise.
12440 * combine.c (make_extraction): Update accordingly.
12441 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12442 (get_optab_extraction_insn): Likewise.
12443 * recog.c (simplify_while_replacing): Likewise.
12444 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12445 parameter to opt_scalar_int_mode.
12447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12448 Alan Hayward <alan.hayward@arm.com>
12449 David Sherwood <david.sherwood@arm.com>
12451 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12452 to a scalar_int_mode instead of a machine_mode.
12453 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12454 (get_best_mode): Return a boolean and use a pointer argument to store
12455 the selected mode. Replace the limit mode parameter with a bit limit.
12456 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12457 for the values returned by bit_field_mode_iterator::next_mode.
12458 (store_bit_field): Update call to get_best_mode.
12459 (store_fixed_bit_field): Likewise.
12460 (extract_fixed_bit_field): Likewise.
12461 * expr.c (optimize_bitfield_assignment_op): Likewise.
12462 * fold-const.c (optimize_bit_field_compare): Likewise.
12463 (fold_truth_andor_1): Likewise.
12464 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12465 Update for new type of m_mode.
12466 (get_best_mode): As above.
12468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12469 Alan Hayward <alan.hayward@arm.com>
12470 David Sherwood <david.sherwood@arm.com>
12472 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12473 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12474 (store_bit_field): Check is_a <scalar_int_mode> before calling
12475 strict_volatile_bitfield_p.
12476 (extract_bit_field): Likewise.
12478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12479 Alan Hayward <alan.hayward@arm.com>
12480 David Sherwood <david.sherwood@arm.com>
12482 * target.def (cstore_mode): Return a scalar_int_mode.
12483 * doc/tm.texi: Regenerate.
12484 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12485 * targhooks.h (default_cstore_mode): Likewise.
12486 * targhooks.c (default_cstore_mode): Likewise, using a forced
12488 * expmed.c (emit_cstore): Expect the target of the cstore to be
12491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12492 Alan Hayward <alan.hayward@arm.com>
12493 David Sherwood <david.sherwood@arm.com>
12495 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12497 (niter_desc): Likewise mode.
12498 (iv_analyze): Add a mode parameter.
12500 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12501 and change its type to scalar_int_mode.
12502 * loop-iv.c: Update commentary at head of file.
12503 (iv_constant): Pass the mode paraeter before the rtx it describes
12504 and change its type to scalar_int_mode. Remove VOIDmode handling.
12505 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12506 (iv_extend): Likewise.
12507 (shorten_into_mode): Likewise.
12508 (iv_add): Use scalar_int_mode.
12509 (iv_mult): Likewise.
12510 (iv_shift): Likewise.
12511 (canonicalize_iv_subregs): Likewise.
12512 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12513 it describes and change its mode to scalar_int_mode. Also change
12514 the type of the returned inner_mode to scalar_int_mode.
12515 (get_biv_step): Likewise, turning outer_mode from a pointer
12516 into a direct parameter. Update call to get_biv_step_1.
12517 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12518 iv_constant and get_biv_step.
12519 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12520 and change its type to scalar_int_mode. Don't initialise iv->mode
12521 to VOIDmode and remove later checks for its still being VOIDmode.
12522 Update calls to iv_analyze_op and iv_analyze_expr. Check
12523 is_a <scalar_int_mode> when changing the mode under consideration.
12524 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12525 Update call to iv_analyze_expr.
12526 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12527 inner register is not also a scalar_int_mode. Update call to
12529 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12530 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12531 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12532 separate mode class checks. Update calls to iv_analyze. Remove
12533 fix-up of VOIDmodes after iv_analyze_biv.
12534 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12535 don't have a scalar_int_mode. Update call to biv_p.
12537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12538 Alan Hayward <alan.hayward@arm.com>
12539 David Sherwood <david.sherwood@arm.com>
12541 * cfgexpand.c (convert_debug_memory_address): Use
12542 as_a <scalar_int_mode>.
12543 * combine.c (expand_compound_operation): Likewise.
12544 (make_extraction): Likewise.
12545 (change_zero_ext): Likewise.
12546 (simplify_comparison): Likewise.
12547 * cse.c (cse_insn): Likewise.
12548 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12549 (mem_loc_descriptor): Likewise.
12550 (loc_descriptor): Likewise.
12551 * expmed.c (init_expmed_one_mode): Likewise.
12552 (synth_mult): Likewise.
12553 (emit_store_flag_1): Likewise.
12554 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12555 of a comparison with size.
12556 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12557 (reduce_to_bit_field_precision): Likewise.
12558 * function.c (expand_function_end): Likewise.
12559 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12560 * loop-doloop.c (doloop_modify): Likewise.
12561 * optabs.c (expand_binop): Likewise.
12562 (expand_unop): Likewise.
12563 (expand_copysign_absneg): Likewise.
12564 (prepare_cmp_insn): Likewise.
12565 (maybe_legitimize_operand): Likewise.
12566 * recog.c (const_scalar_int_operand): Likewise.
12567 * rtlanal.c (get_address_mode): Likewise.
12568 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12569 (simplify_cond_clz_ctz): Likewise.
12570 * tree-nested.c (get_nl_goto_field): Likewise.
12571 * tree.c (build_vector_type_for_mode): Likewise.
12572 * var-tracking.c (use_narrower_mode): Likewise.
12574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12575 Alan Hayward <alan.hayward@arm.com>
12576 David Sherwood <david.sherwood@arm.com>
12578 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12579 * builtins.c (expand_builtin_signbit): Use it.
12580 * cfgexpand.c (expand_debug_expr): Likewise.
12581 * dojump.c (do_jump): Likewise.
12582 (do_compare_and_jump): Likewise.
12583 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12584 * expmed.c (make_tree): Likewise.
12585 * expr.c (expand_expr_real_2): Likewise.
12586 (expand_expr_real_1): Likewise.
12587 (try_casesi): Likewise.
12588 * fold-const-call.c (fold_const_call_ss): Likewise.
12589 * fold-const.c (unextend): Likewise.
12590 (extract_muldiv_1): Likewise.
12591 (fold_single_bit_test): Likewise.
12592 (native_encode_int): Likewise.
12593 (native_encode_string): Likewise.
12594 (native_interpret_int): Likewise.
12595 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12596 * internal-fn.c (expand_addsub_overflow): Likewise.
12597 (expand_neg_overflow): Likewise.
12598 (expand_mul_overflow): Likewise.
12599 (expand_arith_overflow): Likewise.
12600 * match.pd: Likewise.
12601 * stor-layout.c (layout_type): Likewise.
12602 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12603 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12604 * tree-ssanames.c (get_range_info): Likewise.
12605 * tree-switch-conversion.c (array_value_type) Likewise.
12606 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12607 (vect_recog_divmod_pattern): Likewise.
12608 (vect_recog_mixed_size_cond_pattern): Likewise.
12609 * tree-vrp.c (extract_range_basic): Likewise.
12610 (simplify_float_conversion_using_ranges): Likewise.
12611 * tree.c (int_fits_type_p): Likewise.
12612 * ubsan.c (instrument_bool_enum_load): Likewise.
12613 * varasm.c (mergeable_string_section): Likewise.
12614 (narrowing_initializer_constant_valid_p): Likewise.
12615 (output_constant): Likewise.
12617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12618 Alan Hayward <alan.hayward@arm.com>
12619 David Sherwood <david.sherwood@arm.com>
12621 * machmode.h (NARROWEST_INT_MODE): New macro.
12622 * expr.c (alignment_for_piecewise_move): Use it instead of
12623 GET_CLASS_NARROWEST_MODE (MODE_INT).
12624 (push_block): Likewise.
12625 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12627 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12630 Alan Hayward <alan.hayward@arm.com>
12631 David Sherwood <david.sherwood@arm.com>
12633 * postreload.c (move2add_valid_value_p): Change the type of the
12634 mode parameter to scalar_int_mode.
12635 (move2add_use_add2_insn): Add a mode parameter and use it instead
12637 (move2add_use_add3_insn): Likewise.
12638 (reload_cse_move2add): Update accordingly.
12640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12641 Alan Hayward <alan.hayward@arm.com>
12642 David Sherwood <david.sherwood@arm.com>
12644 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12646 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12647 * optabs.c (expand_unop): Likewise.
12649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12650 Alan Hayward <alan.hayward@arm.com>
12651 David Sherwood <david.sherwood@arm.com>
12653 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12654 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12655 (popcount_loc_descriptor): Likewise.
12656 (bswap_loc_descriptor): Likewise.
12657 (rotate_loc_descriptor): Likewise.
12658 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12659 calling the functions above.
12661 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12662 Alan Hayward <alan.hayward@arm.com>
12663 David Sherwood <david.sherwood@arm.com>
12665 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12667 (try_combine): Likewise.
12668 (simplify_if_then_else): Likewise.
12669 * cse.c (cse_insn): Likewise.
12670 * dwarf2out.c (mem_loc_descriptor): Likewise.
12671 * emit-rtl.c (gen_lowpart_common): Likewise.
12672 * simplify-rtx.c (simplify_truncation): Likewise.
12673 (simplify_binary_operation_1): Likewise.
12674 (simplify_const_relational_operation): Likewise.
12675 (simplify_ternary_operation): Likewise.
12676 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12679 Alan Hayward <alan.hayward@arm.com>
12680 David Sherwood <david.sherwood@arm.com>
12682 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12683 * reload.c (push_reload): Likewise.
12684 (find_reloads): Likewise.
12686 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12687 Alan Hayward <alan.hayward@arm.com>
12688 David Sherwood <david.sherwood@arm.com>
12690 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12691 (make_compound_operation_int): Likewise.
12692 (change_zero_ext): Likewise.
12693 * expr.c (convert_move): Likewise.
12694 (convert_modes): Likewise.
12695 * fwprop.c (forward_propagate_subreg): Likewise.
12696 * loop-iv.c (get_biv_step_1): Likewise.
12697 * optabs.c (widen_operand): Likewise.
12698 * postreload.c (move2add_valid_value_p): Likewise.
12699 * recog.c (simplify_while_replacing): Likewise.
12700 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12701 (simplify_binary_operation_1): Likewise. Remove redundant
12702 mode equality check.
12704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12705 Alan Hayward <alan.hayward@arm.com>
12706 David Sherwood <david.sherwood@arm.com>
12708 * combine.c (combine_simplify_rtx): Add checks for
12709 is_a <scalar_int_mode>.
12710 (simplify_if_then_else): Likewise.
12711 (make_field_assignment): Likewise.
12712 (simplify_comparison): Likewise.
12713 * ifcvt.c (noce_try_bitop): Likewise.
12714 * loop-invariant.c (canonicalize_address_mult): Likewise.
12715 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12718 Alan Hayward <alan.hayward@arm.com>
12719 David Sherwood <david.sherwood@arm.com>
12721 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12722 is_a <scalar_int_mode> instead of != BLKmode.
12724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12725 Alan Hayward <alan.hayward@arm.com>
12726 David Sherwood <david.sherwood@arm.com>
12728 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12729 instead of != VOIDmode.
12730 * combine.c (if_then_else_cond): Likewise.
12731 (change_zero_ext): Likewise.
12732 * dwarf2out.c (mem_loc_descriptor): Likewise.
12733 (loc_descriptor): Likewise.
12734 * rtlanal.c (canonicalize_condition): Likewise.
12735 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12738 Alan Hayward <alan.hayward@arm.com>
12739 David Sherwood <david.sherwood@arm.com>
12741 * simplify-rtx.c (simplify_binary_operation_1): Use
12742 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12745 Alan Hayward <alan.hayward@arm.com>
12746 David Sherwood <david.sherwood@arm.com>
12748 * wide-int.h (int_traits<unsigned char>) New class.
12749 (int_traits<unsigned short>) Likewise.
12750 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12751 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12753 * combine.c (set_nonzero_bits_and_sign_copies): Use
12754 is_a <scalar_int_mode>.
12755 (find_split_point): Likewise.
12756 (combine_simplify_rtx): Likewise.
12757 (simplify_logical): Likewise.
12758 (expand_compound_operation): Likewise.
12759 (expand_field_assignment): Likewise.
12760 (make_compound_operation): Likewise.
12761 (extended_count): Likewise.
12762 (change_zero_ext): Likewise.
12763 (simplify_comparison): Likewise.
12764 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12765 (ucompare_loc_descriptor): Likewise.
12766 (minmax_loc_descriptor): Likewise.
12767 (mem_loc_descriptor): Likewise.
12768 (loc_descriptor): Likewise.
12769 * expmed.c (init_expmed_one_mode): Likewise.
12770 * lra-constraints.c (lra_constraint_offset): Likewise.
12771 * optabs.c (prepare_libcall_arg): Likewise.
12772 * postreload.c (move2add_note_store): Likewise.
12773 * reload.c (operands_match_p): Likewise.
12774 * rtl.h (load_extend_op): Likewise.
12775 * rtlhooks.c (gen_lowpart_general): Likewise.
12776 * simplify-rtx.c (simplify_truncation): Likewise.
12777 (simplify_unary_operation_1): Likewise.
12778 (simplify_binary_operation_1): Likewise.
12779 (simplify_const_binary_operation): Likewise.
12780 (simplify_const_relational_operation): Likewise.
12781 (simplify_subreg): Likewise.
12782 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12783 * var-tracking.c (adjust_mems): Likewise.
12784 (prepare_call_arguments): Likewise.
12786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12787 Alan Hayward <alan.hayward@arm.com>
12788 David Sherwood <david.sherwood@arm.com>
12790 * machmode.h (is_int_mode): New fuction.
12791 * combine.c (find_split_point): Use it.
12792 (combine_simplify_rtx): Likewise.
12793 (simplify_if_then_else): Likewise.
12794 (simplify_set): Likewise.
12795 (simplify_shift_const_1): Likewise.
12796 (simplify_comparison): Likewise.
12797 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12798 * cse.c (notreg_cost): Likewise.
12799 (cse_insn): Likewise.
12800 * cselib.c (cselib_lookup_1): Likewise.
12801 * dojump.c (do_jump_1): Likewise.
12802 (do_compare_rtx_and_jump): Likewise.
12803 * dse.c (get_call_args): Likewise.
12804 * dwarf2out.c (rtl_for_decl_init): Likewise.
12805 (native_encode_initializer): Likewise.
12806 * expmed.c (emit_store_flag_1): Likewise.
12807 (emit_store_flag): Likewise.
12808 * expr.c (convert_modes): Likewise.
12809 (store_field): Likewise.
12810 (expand_expr_real_1): Likewise.
12811 * fold-const.c (fold_read_from_constant_string): Likewise.
12812 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12813 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12814 * optabs.c (expand_binop): Likewise.
12815 (expand_unop): Likewise.
12816 (expand_abs_nojump): Likewise.
12817 (expand_one_cmpl_abs_nojump): Likewise.
12818 * simplify-rtx.c (mode_signbit_p): Likewise.
12819 (val_signbit_p): Likewise.
12820 (val_signbit_known_set_p): Likewise.
12821 (val_signbit_known_clear_p): Likewise.
12822 (simplify_relational_operation_1): Likewise.
12823 * tree.c (vector_type_mode): Likewise.
12825 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12826 Alan Hayward <alan.hayward@arm.com>
12827 David Sherwood <david.sherwood@arm.com>
12829 * machmode.h (smallest_mode_for_size): Fix formatting.
12830 (smallest_int_mode_for_size): New function.
12831 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12832 instead of smallest_mode_for_size.
12833 * combine.c (make_extraction): Likewise.
12834 * config/arc/arc.c (arc_expand_movmem): Likewise.
12835 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12836 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12837 * config/s390/s390.c (s390_expand_insv): Likewise.
12838 * config/sparc/sparc.c (assign_int_registers): Likewise.
12839 * config/spu/spu.c (spu_function_value): Likewise.
12840 (spu_function_arg): Likewise.
12841 * coverage.c (get_gcov_type): Likewise.
12842 (get_gcov_unsigned_t): Likewise.
12843 * dse.c (find_shift_sequence): Likewise.
12844 * expmed.c (store_bit_field_1): Likewise.
12845 * expr.c (convert_move): Likewise.
12846 (store_field): Likewise.
12847 * internal-fn.c (expand_arith_overflow): Likewise.
12848 * optabs-query.c (get_best_extraction_insn): Likewise.
12849 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12850 * stor-layout.c (layout_type): Likewise.
12851 (initialize_sizetypes): Likewise.
12852 * targhooks.c (default_get_mask_mode): Likewise.
12853 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12856 Alan Hayward <alan.hayward@arm.com>
12857 David Sherwood <david.sherwood@arm.com>
12859 * machmode.h (opt_mode::else_blk): New function.
12860 (int_mode_for_mode): Declare.
12861 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12862 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12864 * cfgexpand.c (expand_debug_expr): Likewise.
12865 * combine.c (gen_lowpart_or_truncate): Likewise.
12866 (gen_lowpart_for_combine): Likewise.
12867 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12868 * config/avr/avr.c (avr_to_int_mode): Likewise.
12869 (avr_out_plus_1): Likewise.
12870 (avr_out_plus): Likewise.
12871 (avr_out_round): Likewise.
12872 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12873 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12874 (s390_expand_vcond): Likewise.
12875 * config/spu/spu.c (spu_split_immediate): Likewise.
12876 (spu_expand_mov): Likewise.
12877 * dse.c (get_stored_val): Likewise.
12878 * expmed.c (store_bit_field_1): Likewise.
12879 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12881 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12882 (extract_low_bits): Likewise.
12883 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12884 handling rather than repeating the check.
12885 (emit_group_store): Likewise.
12886 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12887 * optabs.c (expand_absneg_bit): Likewise.
12888 (expand_copysign_absneg): Likewise.
12889 (expand_copysign_bit): Likewise.
12890 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12891 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12892 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12893 * var-tracking.c (prepare_call_arguments): Likewise.
12894 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12895 int_mode_for_mode instead of mode_for_size.
12896 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12899 Alan Hayward <alan.hayward@arm.com>
12900 David Sherwood <david.sherwood@arm.com>
12902 * machmode.h (int_mode_for_size): New function.
12903 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12904 instead of mode_for_size.
12905 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12907 * combine.c (expand_field_assignment): Use int_mode_for_size
12908 instead of mode_for_size.
12909 (make_extraction): Likewise.
12910 (simplify_shift_const_1): Likewise.
12911 (simplify_comparison): Likewise.
12912 * dojump.c (do_jump): Likewise.
12913 * dwarf2out.c (mem_loc_descriptor): Likewise.
12914 * emit-rtl.c (init_derived_machine_modes): Likewise.
12915 * expmed.c (flip_storage_order): Likewise.
12916 (convert_extracted_bit_field): Likewise.
12917 * expr.c (copy_blkmode_from_reg): Likewise.
12918 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12919 * internal-fn.c (expand_mul_overflow): Likewise.
12920 * lower-subreg.c (simple_move): Likewise.
12921 * optabs-libfuncs.c (init_optabs): Likewise.
12922 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12923 * tree.c (vector_type_mode): Likewise.
12924 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12925 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12926 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12927 * tree-vect-stmts.c (vectorizable_load): Likewise.
12928 (vectorizable_store): Likewise.
12930 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12931 Alan Hayward <alan.hayward@arm.com>
12932 David Sherwood <david.sherwood@arm.com>
12934 * coretypes.h (pod_mode): New type.
12935 (scalar_int_mode_pod): New typedef.
12936 * machmode.h (pod_mode): New class.
12937 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12938 * genmodes.c (emit_mode_int_n): Update accordingly.
12939 * lower-subreg.h (target_lower_subreg): Change type to
12940 scalar_int_mode_pod.
12941 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12942 scalar_int_mode_pod.
12944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12945 Alan Hayward <alan.hayward@arm.com>
12946 David Sherwood <david.sherwood@arm.com>
12948 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12949 machine_mode to scalar_int_mode.
12950 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12951 (rs6000_option_override_internal): Remove cast to int.
12952 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12953 machine_mode to scalar_int_mode.
12954 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12955 (rs6000_option_override_internal): Remove cast to int.
12956 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12957 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12959 * config/s390/s390.c (s390_expand_builtin): Likewise.
12960 * coretypes.h (scalar_int_mode): New type.
12961 (opt_scalar_int_mode): New typedef.
12962 * machmode.h (scalar_int_mode): New class.
12963 (scalar_int_mode::includes_p): New function.
12964 (byte_mode): Change type to scalar_int_mode.
12965 (word_mode): Likewise.
12966 (ptr_mode): Likewise.
12967 * emit-rtl.c (byte_mode): Likewise.
12968 (word_mode): Likewise.
12969 (ptr_mode): Likewise.
12970 (init_derived_machine_modes): Update accordingly.
12971 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12972 and MODE_PARTIAL_INT.
12973 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12974 opt_scalar_int_mode.
12976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12977 Alan Hayward <alan.hayward@arm.com>
12978 David Sherwood <david.sherwood@arm.com>
12980 * target.def (libgcc_floating_mode_supported_p): Take a
12982 * doc/tm.texi: Regenerate.
12983 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12985 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12986 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12990 Alan Hayward <alan.hayward@arm.com>
12991 David Sherwood <david.sherwood@arm.com>
12993 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12994 * doc/tm.texi: Regenerate.
12995 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12996 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12997 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12998 * targhooks.h (default_floatn_mode): Likewise.
12999 * targhooks.c (default_floatn_mode): Likewise.
13000 * tree.c (build_common_tree_nodes): Update accordingly.
13002 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13003 Alan Hayward <alan.hayward@arm.com>
13004 David Sherwood <david.sherwood@arm.com>
13006 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
13007 (mode_iterator::iterate_p): Likewise.
13008 (mode_iterator::get_wider): Likewise.
13009 * expr.c (init_expr_target): Use opt_scalar_float_mode.
13011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13012 Alan Hayward <alan.hayward@arm.com>
13013 David Sherwood <david.sherwood@arm.com>
13015 * coretypes.h (opt_scalar_float_mode): New typedef.
13016 * machmode.h (float_mode_for_size): New function.
13017 * emit-rtl.c (double_mode): Delete.
13018 (init_emit_once): Use float_mode_for_size.
13019 * stor-layout.c (layout_type): Likewise.
13020 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
13022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13023 Alan Hayward <alan.hayward@arm.com>
13024 David Sherwood <david.sherwood@arm.com>
13026 * output.h (assemble_real): Take a scalar_float_mode.
13027 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
13028 * config/arm/arm.md (consttable_4): Likewise.
13029 (consttable_8): Likewise.
13030 (consttable_16): Likewise.
13031 * config/mips/mips.md (consttable_float): Likewise.
13032 * config/s390/s390.c (s390_output_pool_entry): Likewise.
13033 * varasm.c (assemble_real): Take a scalar_float_mode.
13034 (output_constant_pool_2): Update accordingly.
13035 (output_constant): Likewise.
13037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13038 Alan Hayward <alan.hayward@arm.com>
13039 David Sherwood <david.sherwood@arm.com>
13041 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
13042 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
13043 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
13044 (native_encode_real): Likewise.
13045 (native_interpret_real): Likewise.
13046 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
13047 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13050 Alan Hayward <alan.hayward@arm.com>
13051 David Sherwood <david.sherwood@arm.com>
13053 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
13054 <scalar_float_mode>. Simplify.
13055 (gen_extend_conv_libfunc): Likewise.
13057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13058 Alan Hayward <alan.hayward@arm.com>
13059 David Sherwood <david.sherwood@arm.com>
13061 * coretypes.h (scalar_float_mode): New type.
13062 * machmode.h (mode_traits::from_int): Use machine_mode if
13063 USE_ENUM_MODES is defined.
13064 (is_a): New function.
13066 (dyn_cast): Likewise.
13067 (scalar_float_mode): New class.
13068 (scalar_float_mode::includes_p): New function.
13069 (is_float_mode): Likewise.
13070 * gdbhooks.py (MachineModePrinter): New class.
13071 (build_pretty_printer): Use it for scalar_float_mode.
13072 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
13073 (format_helper::format_helper): Turn into a template.
13074 * genmodes.c (get_mode_class): New function.
13075 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
13076 or machine_mode if none.
13077 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
13078 as_a <scalar_float_mode>.
13079 * dwarf2out.c (mem_loc_descriptor): Likewise.
13080 (insert_float): Likewise.
13081 (add_const_value_attribute): Likewise.
13082 * simplify-rtx.c (simplify_immed_subreg): Likewise.
13083 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
13084 (expand_unop): Update accordingly.
13085 (expand_abs_nojump): Likewise.
13086 (expand_copysign_absneg): Take a scalar_float_mode.
13087 (expand_copysign_bit): Likewise.
13088 (expand_copysign): Update accordingly.
13090 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13091 Alan Hayward <alan.hayward@arm.com>
13092 David Sherwood <david.sherwood@arm.com>
13094 * coretypes.h (opt_mode): New class.
13095 * machmode.h (opt_mode): Likewise.
13096 (opt_mode::else_void): New function.
13097 (opt_mode::require): Likewise.
13098 (opt_mode::exists): Likewise.
13099 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
13100 (GET_MODE_2XWIDER_MODE): Likewise.
13101 (mode_iterator::get_wider): Update accordingly.
13102 (mode_iterator::get_2xwider): Likewise.
13103 (mode_iterator::get_known_wider): Likewise, turning into a template.
13104 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
13105 forcing a wider mode to exist.
13106 * config/cr16/cr16.h (LONG_REG_P): Likewise.
13107 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13108 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
13109 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
13110 * lower-subreg.c (init_lower_subreg): Likewise.
13111 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
13112 on the final iteration.
13113 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
13114 a wider mode exists before asking for a move pattern.
13115 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
13116 forcing a wider mode to exist.
13117 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
13118 returning false if no such mode exists.
13119 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
13120 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
13121 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
13122 Avoid checking for a MODE_INT if we already know the mode is not a
13124 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
13125 forcing a wider mode to exist.
13126 (expmed_mult_highpart_optab): Likewise.
13127 (expmed_mult_highpart): Likewise.
13128 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
13130 * lto-streamer-in.c (lto_input_mode_table): Likewise.
13131 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
13132 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13133 * internal-fn.c (expand_mul_overflow): Update use of
13134 GET_MODE_2XWIDER_MODE.
13135 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13136 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
13137 GET_MODE_WIDER_MODE.
13138 (convert_plusminus_to_widen): Likewise.
13139 * tree-switch-conversion.c (array_value_type): Likewise.
13140 * var-tracking.c (emit_note_insn_var_location): Likewise.
13141 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
13142 Return false inside rather than outside the loop if no wider mode
13144 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
13145 and GET_MODE_2XWIDER_MODE
13146 (can_compare_p): Use else_void.
13147 * gdbhooks.py (OptMachineModePrinter): New class.
13148 (build_pretty_printer): Use it for opt_mode.
13150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13151 Alan Hayward <alan.hayward@arm.com>
13152 David Sherwood <david.sherwood@arm.com>
13154 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
13155 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
13157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13158 Alan Hayward <alan.hayward@arm.com>
13159 David Sherwood <david.sherwood@arm.com>
13161 * machmode.h (mode_traits): New structure.
13162 (get_narrowest_mode): New function.
13163 (mode_iterator::start): Likewise.
13164 (mode_iterator::iterate_p): Likewise.
13165 (mode_iterator::get_wider): Likewise.
13166 (mode_iterator::get_known_wider): Likewise.
13167 (mode_iterator::get_2xwider): Likewise.
13168 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
13169 (FOR_EACH_MODE): Likewise.
13170 (FOR_EACH_MODE_FROM): Likewise.
13171 (FOR_EACH_MODE_UNTIL): Likewise.
13172 (FOR_EACH_WIDER_MODE): Likewise.
13173 (FOR_EACH_2XWIDER_MODE): Likewise.
13174 * builtins.c (expand_builtin_strlen): Use new mode iterators.
13175 * combine.c (simplify_comparison): Likewise
13176 * config/i386/i386.c (type_natural_mode): Likewise.
13177 * cse.c (cse_insn): Likewise.
13178 * dse.c (find_shift_sequence): Likewise.
13179 * emit-rtl.c (init_derived_machine_modes): Likewise.
13180 (init_emit_once): Likewise.
13181 * explow.c (hard_function_value): Likewise.
13182 * expmed.c (extract_fixed_bit_field_1): Likewise.
13183 (extract_bit_field_1): Likewise.
13184 (expand_divmod): Likewise.
13185 (emit_store_flag_1): Likewise.
13186 * expr.c (init_expr_target): Likewise.
13187 (convert_move): Likewise.
13188 (alignment_for_piecewise_move): Likewise.
13189 (widest_int_mode_for_size): Likewise.
13190 (emit_block_move_via_movmem): Likewise.
13191 (copy_blkmode_to_reg): Likewise.
13192 (set_storage_via_setmem): Likewise.
13193 (compress_float_constant): Likewise.
13194 * omp-low.c (omp_clause_aligned_alignment): Likewise.
13195 * optabs-query.c (get_best_extraction_insn): Likewise.
13196 * optabs.c (expand_binop): Likewise.
13197 (expand_twoval_unop): Likewise.
13198 (expand_twoval_binop): Likewise.
13199 (widen_leading): Likewise.
13200 (widen_bswap): Likewise.
13201 (expand_parity): Likewise.
13202 (expand_unop): Likewise.
13203 (prepare_cmp_insn): Likewise.
13204 (prepare_float_lib_cmp): Likewise.
13205 (expand_float): Likewise.
13206 (expand_fix): Likewise.
13207 (expand_sfix_optab): Likewise.
13208 * postreload.c (move2add_use_add2_insn): Likewise.
13209 * reg-stack.c (reg_to_stack): Likewise.
13210 * reginfo.c (choose_hard_reg_mode): Likewise.
13211 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
13212 * stor-layout.c (mode_for_size): Likewise.
13213 (smallest_mode_for_size): Likewise.
13214 (mode_for_vector): Likewise.
13215 (finish_bitfield_representative): Likewise.
13216 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
13217 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
13218 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13219 * var-tracking.c (prepare_call_arguments): Likewise.
13221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13222 Alan Hayward <alan.hayward@arm.com>
13223 David Sherwood <david.sherwood@arm.com>
13225 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
13226 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
13227 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
13228 * machmode.h (mode_size): Move earlier in file.
13229 (mode_precision): Likewise.
13230 (mode_inner): Likewise.
13231 (mode_nunits): Likewise.
13232 (mode_unit_size): Likewise.
13233 (unit_unit_precision): Likewise.
13234 (mode_wider): Likewise.
13235 (mode_2xwider): Likewise.
13236 (machine_mode): New class.
13237 (mode_to_bytes): New function.
13238 (mode_to_bits): Likewise.
13239 (mode_to_precision): Likewise.
13240 (mode_to_inner): Likewise.
13241 (mode_to_unit_size): Likewise.
13242 (mode_to_unit_precision): Likewise.
13243 (mode_to_nunits): Likewise.
13244 (GET_MODE_SIZE): Use mode_to_bytes.
13245 (GET_MODE_BITSIZE): Use mode_to_bits.
13246 (GET_MODE_PRECISION): Use mode_to_precision.
13247 (GET_MODE_INNER): Use mode_to_inner.
13248 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
13249 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
13250 (GET_MODE_NUNITS): Use mode_to_nunits.
13251 * system.h (ALWAYS_INLINE): New macro.
13252 * config/powerpcspe/powerpcspe-c.c
13253 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
13254 int for arg1_mode and arg2_mode.
13256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13257 Alan Hayward <alan.hayward@arm.com>
13258 David Sherwood <david.sherwood@arm.com>
13260 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13261 Prefix mode names with E_ in case statements.
13262 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13263 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
13264 (aarch64_split_simd_move): Likewise.
13265 (aarch64_gen_storewb_pair): Likewise.
13266 (aarch64_gen_loadwb_pair): Likewise.
13267 (aarch64_gen_store_pair): Likewise.
13268 (aarch64_gen_load_pair): Likewise.
13269 (aarch64_get_condition_code_1): Likewise.
13270 (aarch64_constant_pool_reload_icode): Likewise.
13271 (get_rsqrte_type): Likewise.
13272 (get_rsqrts_type): Likewise.
13273 (get_recpe_type): Likewise.
13274 (get_recps_type): Likewise.
13275 (aarch64_gimplify_va_arg_expr): Likewise.
13276 (aarch64_simd_container_mode): Likewise.
13277 (aarch64_emit_load_exclusive): Likewise.
13278 (aarch64_emit_store_exclusive): Likewise.
13279 (aarch64_expand_compare_and_swap): Likewise.
13280 (aarch64_gen_atomic_cas): Likewise.
13281 (aarch64_emit_bic): Likewise.
13282 (aarch64_emit_atomic_swap): Likewise.
13283 (aarch64_emit_atomic_load_op): Likewise.
13284 (aarch64_evpc_trn): Likewise.
13285 (aarch64_evpc_uzp): Likewise.
13286 (aarch64_evpc_zip): Likewise.
13287 (aarch64_evpc_ext): Likewise.
13288 (aarch64_evpc_rev): Likewise.
13289 (aarch64_evpc_dup): Likewise.
13290 (aarch64_gen_ccmp_first): Likewise.
13291 (aarch64_gen_ccmp_next): Likewise.
13292 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
13293 (alpha_emit_xfloating_libcall): Likewise.
13294 (emit_insxl): Likewise.
13295 (alpha_arg_type): Likewise.
13296 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
13297 (arc_preferred_simd_mode): Likewise.
13298 (arc_secondary_reload): Likewise.
13299 (get_arc_condition_code): Likewise.
13300 (arc_print_operand): Likewise.
13301 (arc_legitimate_constant_p): Likewise.
13302 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13303 * config/arc/arc.md (casesi_load): Likewise.
13304 (casesi_compact_jump): Likewise.
13305 * config/arc/predicates.md (proper_comparison_operator): Likewise.
13306 (cc_use_register): Likewise.
13307 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13308 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13309 (arm_init_iwmmxt_builtins): Likewise.
13310 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
13311 (neon_expand_vector_init): Likewise.
13312 (arm_attr_length_move_neon): Likewise.
13313 (maybe_get_arm_condition_code): Likewise.
13314 (arm_emit_vector_const): Likewise.
13315 (arm_preferred_simd_mode): Likewise.
13316 (arm_output_iwmmxt_tinsr): Likewise.
13317 (thumb1_output_casesi): Likewise.
13318 (thumb2_output_casesi): Likewise.
13319 (arm_emit_load_exclusive): Likewise.
13320 (arm_emit_store_exclusive): Likewise.
13321 (arm_expand_compare_and_swap): Likewise.
13322 (arm_evpc_neon_vuzp): Likewise.
13323 (arm_evpc_neon_vzip): Likewise.
13324 (arm_evpc_neon_vrev): Likewise.
13325 (arm_evpc_neon_vtrn): Likewise.
13326 (arm_evpc_neon_vext): Likewise.
13327 (arm_validize_comparison): Likewise.
13328 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
13329 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
13330 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
13331 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
13332 (c6x_preferred_simd_mode): Likewise.
13333 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
13334 (epiphany_rtx_costs): Likewise.
13335 * config/epiphany/predicates.md (proper_comparison_operator):
13337 * config/frv/frv.c (condexec_memory_operand): Likewise.
13338 (frv_emit_move): Likewise.
13339 (output_move_single): Likewise.
13340 (output_condmove_single): Likewise.
13341 (frv_hard_regno_mode_ok): Likewise.
13342 (frv_matching_accg_mode): Likewise.
13343 * config/h8300/h8300.c (split_adds_subs): Likewise.
13344 (h8300_rtx_costs): Likewise.
13345 (h8300_print_operand): Likewise.
13346 (compute_mov_length): Likewise.
13347 (output_logical_op): Likewise.
13348 (compute_logical_op_length): Likewise.
13349 (compute_logical_op_cc): Likewise.
13350 (h8300_shift_needs_scratch_p): Likewise.
13351 (output_a_shift): Likewise.
13352 (compute_a_shift_length): Likewise.
13353 (compute_a_shift_cc): Likewise.
13354 (expand_a_rotate): Likewise.
13355 (output_a_rotate): Likewise.
13356 * config/i386/i386.c (classify_argument): Likewise.
13357 (function_arg_advance_32): Likewise.
13358 (function_arg_32): Likewise.
13359 (function_arg_64): Likewise.
13360 (function_value_64): Likewise.
13361 (ix86_gimplify_va_arg): Likewise.
13362 (ix86_legitimate_constant_p): Likewise.
13363 (put_condition_code): Likewise.
13364 (split_double_mode): Likewise.
13365 (ix86_avx256_split_vector_move_misalign): Likewise.
13366 (ix86_expand_vector_logical_operator): Likewise.
13367 (ix86_split_idivmod): Likewise.
13368 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
13369 (ix86_build_const_vector): Likewise.
13370 (ix86_build_signbit_mask): Likewise.
13371 (ix86_match_ccmode): Likewise.
13372 (ix86_cc_modes_compatible): Likewise.
13373 (ix86_expand_branch): Likewise.
13374 (ix86_expand_sse_cmp): Likewise.
13375 (ix86_expand_sse_movcc): Likewise.
13376 (ix86_expand_int_sse_cmp): Likewise.
13377 (ix86_expand_vec_perm_vpermi2): Likewise.
13378 (ix86_expand_vec_perm): Likewise.
13379 (ix86_expand_sse_unpack): Likewise.
13380 (ix86_expand_int_addcc): Likewise.
13381 (ix86_split_to_parts): Likewise.
13382 (ix86_vectorize_builtin_gather): Likewise.
13383 (ix86_vectorize_builtin_scatter): Likewise.
13384 (avx_vpermilp_parallel): Likewise.
13385 (inline_memory_move_cost): Likewise.
13386 (ix86_tieable_integer_mode_p): Likewise.
13387 (x86_maybe_negate_const_int): Likewise.
13388 (ix86_expand_vector_init_duplicate): Likewise.
13389 (ix86_expand_vector_init_one_nonzero): Likewise.
13390 (ix86_expand_vector_init_one_var): Likewise.
13391 (ix86_expand_vector_init_concat): Likewise.
13392 (ix86_expand_vector_init_interleave): Likewise.
13393 (ix86_expand_vector_init_general): Likewise.
13394 (ix86_expand_vector_set): Likewise.
13395 (ix86_expand_vector_extract): Likewise.
13396 (emit_reduc_half): Likewise.
13397 (ix86_emit_i387_round): Likewise.
13398 (ix86_mangle_type): Likewise.
13399 (ix86_expand_round_sse4): Likewise.
13400 (expand_vec_perm_blend): Likewise.
13401 (canonicalize_vector_int_perm): Likewise.
13402 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
13403 (expand_vec_perm_1): Likewise.
13404 (expand_vec_perm_interleave3): Likewise.
13405 (expand_vec_perm_even_odd_pack): Likewise.
13406 (expand_vec_perm_even_odd_1): Likewise.
13407 (expand_vec_perm_broadcast_1): Likewise.
13408 (ix86_vectorize_vec_perm_const_ok): Likewise.
13409 (ix86_expand_vecop_qihi): Likewise.
13410 (ix86_expand_mul_widen_hilo): Likewise.
13411 (ix86_expand_sse2_abs): Likewise.
13412 (ix86_expand_pextr): Likewise.
13413 (ix86_expand_pinsr): Likewise.
13414 (ix86_preferred_simd_mode): Likewise.
13415 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
13416 * config/i386/sse.md (*andnot<mode>3): Likewise.
13417 (<mask_codefor><code><mode>3<mask_name>): Likewise.
13418 (*<code><mode>3): Likewise.
13419 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13420 (ia64_expand_atomic_op): Likewise.
13421 (ia64_arg_type): Likewise.
13422 (ia64_mode_to_int): Likewise.
13423 (ia64_scalar_mode_supported_p): Likewise.
13424 (ia64_vector_mode_supported_p): Likewise.
13425 (expand_vec_perm_broadcast): Likewise.
13426 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13427 (iq2000_function_arg_advance): Likewise.
13428 (iq2000_function_arg): Likewise.
13429 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13430 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13431 (m68k_libcall_value): Likewise.
13432 (m68k_function_value): Likewise.
13433 (sched_attr_op_type): Likewise.
13434 * config/mcore/mcore.c (mcore_output_move): Likewise.
13435 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13437 (microblaze_function_arg): Likewise.
13438 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13439 (mips_print_operand): Likewise.
13440 (mips_mode_ok_for_mov_fmt_p): Likewise.
13441 (mips_vector_mode_supported_p): Likewise.
13442 (mips_preferred_simd_mode): Likewise.
13443 (mips_expand_vpc_loongson_even_odd): Likewise.
13444 (mips_expand_vec_unpack): Likewise.
13445 (mips_expand_vi_broadcast): Likewise.
13446 (mips_expand_vector_init): Likewise.
13447 (mips_expand_vec_reduc): Likewise.
13448 (mips_expand_msa_cmp): Likewise.
13449 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13450 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13451 (cc_flags_for_mode): Likewise.
13452 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13453 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13454 (nds32_output_casesi_pc_relative): Likewise.
13455 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13456 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13457 (nvptx_gen_unpack): Likewise.
13458 (nvptx_gen_pack): Likewise.
13459 (nvptx_gen_shuffle): Likewise.
13460 (nvptx_gen_wcast): Likewise.
13461 (nvptx_preferred_simd_mode): Likewise.
13462 * config/pa/pa.c (pa_secondary_reload): Likewise.
13463 * config/pa/predicates.md (base14_operand): Likewise.
13464 * config/powerpcspe/powerpcspe-c.c
13465 (altivec_resolve_overloaded_builtin): Likewise.
13466 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13468 (rs6000_preferred_simd_mode): Likewise.
13469 (output_vec_const_move): Likewise.
13470 (rs6000_expand_vector_extract): Likewise.
13471 (rs6000_split_vec_extract_var): Likewise.
13472 (reg_offset_addressing_ok_p): Likewise.
13473 (rs6000_legitimate_offset_address_p): Likewise.
13474 (rs6000_legitimize_address): Likewise.
13475 (rs6000_emit_set_const): Likewise.
13476 (rs6000_const_vec): Likewise.
13477 (rs6000_emit_move): Likewise.
13478 (spe_build_register_parallel): Likewise.
13479 (rs6000_darwin64_record_arg_recurse): Likewise.
13480 (swap_selector_for_mode): Likewise.
13481 (spe_init_builtins): Likewise.
13482 (paired_init_builtins): Likewise.
13483 (altivec_init_builtins): Likewise.
13484 (do_load_for_compare): Likewise.
13485 (rs6000_generate_compare): Likewise.
13486 (rs6000_expand_float128_convert): Likewise.
13487 (emit_load_locked): Likewise.
13488 (emit_store_conditional): Likewise.
13489 (rs6000_output_function_epilogue): Likewise.
13490 (rs6000_handle_altivec_attribute): Likewise.
13491 (rs6000_function_value): Likewise.
13492 (emit_fusion_gpr_load): Likewise.
13493 (emit_fusion_p9_load): Likewise.
13494 (emit_fusion_p9_store): Likewise.
13495 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13496 (fusion_gpr_mem_load): Likewise.
13497 (fusion_addis_mem_combo_load): Likewise.
13498 (fusion_addis_mem_combo_store): Likewise.
13499 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13500 (fusion_gpr_mem_load): Likewise.
13501 (fusion_addis_mem_combo_load): Likewise.
13502 (fusion_addis_mem_combo_store): Likewise.
13503 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13505 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13506 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13507 (rs6000_preferred_simd_mode): Likewise.
13508 (output_vec_const_move): Likewise.
13509 (rs6000_expand_vector_extract): Likewise.
13510 (rs6000_split_vec_extract_var): Likewise.
13511 (reg_offset_addressing_ok_p): Likewise.
13512 (rs6000_legitimate_offset_address_p): Likewise.
13513 (rs6000_legitimize_address): Likewise.
13514 (rs6000_emit_set_const): Likewise.
13515 (rs6000_const_vec): Likewise.
13516 (rs6000_emit_move): Likewise.
13517 (rs6000_darwin64_record_arg_recurse): Likewise.
13518 (swap_selector_for_mode): Likewise.
13519 (paired_init_builtins): Likewise.
13520 (altivec_init_builtins): Likewise.
13521 (rs6000_expand_float128_convert): Likewise.
13522 (emit_load_locked): Likewise.
13523 (emit_store_conditional): Likewise.
13524 (rs6000_output_function_epilogue): Likewise.
13525 (rs6000_handle_altivec_attribute): Likewise.
13526 (rs6000_function_value): Likewise.
13527 (emit_fusion_gpr_load): Likewise.
13528 (emit_fusion_p9_load): Likewise.
13529 (emit_fusion_p9_store): Likewise.
13530 * config/rx/rx.c (rx_gen_move_template): Likewise.
13531 (flags_from_mode): Likewise.
13532 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13533 (s390_slb_comparison): Likewise.
13534 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13535 (s390_vector_mode_supported_p): Likewise.
13536 (s390_cc_modes_compatible): Likewise.
13537 (s390_match_ccmode_set): Likewise.
13538 (s390_canonicalize_comparison): Likewise.
13539 (s390_emit_compare_and_swap): Likewise.
13540 (s390_branch_condition_mask): Likewise.
13541 (s390_rtx_costs): Likewise.
13542 (s390_secondary_reload): Likewise.
13543 (__SECONDARY_RELOAD_CASE): Likewise.
13544 (s390_expand_cs): Likewise.
13545 (s390_preferred_simd_mode): Likewise.
13546 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13547 * config/sh/sh.c (sh_print_operand): Likewise.
13548 (dump_table): Likewise.
13549 (sh_secondary_reload): Likewise.
13550 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13551 * config/sh/sh.md (casesi_worker_1): Likewise.
13552 (casesi_worker_2): Likewise.
13553 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13554 (fcc_comparison_operator): Likewise.
13555 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13556 (emit_soft_tfmode_cvt): Likewise.
13557 (sparc_preferred_simd_mode): Likewise.
13558 (output_cbranch): Likewise.
13559 (sparc_print_operand): Likewise.
13560 (sparc_expand_vec_perm_bmask): Likewise.
13561 (vector_init_bshuffle): Likewise.
13562 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13563 (spu_vector_mode_supported_p): Likewise.
13564 (spu_expand_insv): Likewise.
13565 (spu_emit_branch_or_set): Likewise.
13566 (spu_handle_vector_attribute): Likewise.
13567 (spu_builtin_splats): Likewise.
13568 (spu_builtin_extract): Likewise.
13569 (spu_builtin_promote): Likewise.
13570 (spu_expand_sign_extend): Likewise.
13571 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13572 (tilegx_simd_int): Likewise.
13573 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13574 (tilepro_simd_int): Likewise.
13575 * config/v850/v850.c (const_double_split): Likewise.
13576 (v850_print_operand): Likewise.
13577 (ep_memory_offset): Likewise.
13578 * config/vax/vax.c (vax_rtx_costs): Likewise.
13579 (vax_output_int_move): Likewise.
13580 (vax_output_int_add): Likewise.
13581 (vax_output_int_subtract): Likewise.
13582 * config/visium/predicates.md (visium_branch_operator): Likewise.
13583 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13584 (visium_print_operand_address): Likewise.
13585 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13586 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13587 (xtensa_expand_conditional_branch): Likewise.
13588 (xtensa_copy_incoming_a7): Likewise.
13589 (xtensa_output_literal): Likewise.
13590 * dfp.c (decimal_real_maxval): Likewise.
13591 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13594 Alan Hayward <alan.hayward@arm.com>
13595 David Sherwood <david.sherwood@arm.com>
13597 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13598 (mode_nunits_inline): Likewise.
13599 (mode_inner_inline): Likewise.
13600 (mode_unit_size_inline): Likewise.
13601 (mode_unit_precision_inline): Likewise.
13602 (emit_insn_modes_h): Likewise. Also emit a #define of the
13604 (emit_mode_wider): Add an E_ prefix to mode names.
13605 (emit_mode_complex): Likewise.
13606 (emit_mode_inner): Likewise.
13607 (emit_mode_adjustments): Likewise.
13608 (emit_mode_int_n): Likewise.
13609 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13610 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13611 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13612 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13613 (CRC32_BUILTIN, ENTRY): Likewise.
13614 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13615 (aarch64_pop_regs): Likewise.
13616 (aarch64_process_components): Likewise.
13617 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13618 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13619 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13620 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13621 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13622 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13624 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13625 (pcmpistr): Likewise.
13626 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13627 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13628 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13630 * config/rl78/rl78.c (mduc_regs): Likewise.
13631 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13632 (htm_expand_builtin): Likewise.
13633 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13634 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13635 * config/xtensa/xtensa.c (print_operand): Likewise.
13636 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13637 (NUM_MODE_VECTOR_INT): Likewise.
13638 * genoutput.c (null_operand): Likewise.
13639 (output_operand_data): Likewise.
13640 * genrecog.c (print_parameter_value): Likewise.
13641 * lra.c (debug_operand_data): Likewise.
13643 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13645 * dwarf2out.c (scompare_loc_descriptor_wide)
13646 (scompare_loc_descriptor_narrow): New functions, split out from...
13647 (scompare_loc_descriptor): ...here.
13648 * expmed.c (emit_store_flag_int): New function, split out from...
13649 (emit_store_flag): ...here.
13651 2017-08-30 Richard Biener <rguenther@suse.de>
13653 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13654 (dwarf2out_early_finish): Move setting of AT_pubnames from
13655 early debug output to early finish.
13657 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13659 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13660 and -mdata-region to the assembler.
13662 2017-08-30 Richard Biener <rguenther@suse.de>
13664 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13666 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13667 (dwarf2out_early_global_decl): Only generate a DIE for the
13668 abstract origin if it doesn't already exist or is a declaration DIE.
13669 (resolve_addr): Do not add the linkage name twice when
13670 generating a stub DIE for the DW_TAG_GNU_call_site target.
13672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13674 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13675 Use machine_mode rather than int for arg1_mode.
13677 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13680 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13681 that the second argument of the built-in functions to unpack
13682 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13683 switch statement instead a lot of if statements.
13684 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13685 Allow 64-bit values to be in Altivec registers as well as
13686 traditional floating point registers.
13687 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13689 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13691 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13692 MAX_REGS_PER_ADDRESS == 1.
13694 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13696 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13697 * config/i386/i386.c (ix86_option_override_internal): Simplify
13698 setting of opts->x_flag_entry.
13700 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13701 Jakub Jelinek <jakub@redhat.com>
13702 Richard Biener <rguenther@suse.de>
13704 PR tree-optimization/81503
13705 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13706 folded constant fits in the target type; reorder tests for clarity.
13708 2017-08-29 Martin Liska <mliska@suse.cz>
13710 * passes.def: Include pass_lower_switch.
13711 * stmt.c (dump_case_nodes): Remove and move to
13712 tree-switch-conversion.
13713 (case_values_threshold): Likewise.
13714 (expand_switch_as_decision_tree_p): Likewise.
13715 (emit_case_decision_tree): Likewise.
13716 (expand_case): Likewise.
13717 (balance_case_nodes): Likewise.
13718 (node_has_low_bound): Likewise.
13719 (node_has_high_bound): Likewise.
13720 (node_is_bounded): Likewise.
13721 (emit_case_nodes): Likewise.
13722 (struct simple_case_node): New struct.
13723 (add_case_node): Remove.
13724 (emit_case_dispatch_table): Use vector instead of case_list.
13725 (reset_out_edges_aux): Remove.
13726 (compute_cases_per_edge): Likewise.
13727 (expand_case): Build list of simple_case_node.
13728 (expand_sjlj_dispatch_table): Use it.
13729 * tree-switch-conversion.c (struct case_node): Moved from
13730 stmt.c and adjusted.
13731 (emit_case_nodes): Likewise.
13732 (node_has_low_bound): Likewise.
13733 (node_has_high_bound): Likewise.
13734 (node_is_bounded): Likewise.
13735 (case_values_threshold): Likewise.
13736 (reset_out_edges_aux): Likewise.
13737 (compute_cases_per_edge): Likewise.
13738 (add_case_node): Likewise.
13739 (dump_case_nodes): Likewise.
13740 (balance_case_nodes): Likewise.
13741 (expand_switch_as_decision_tree_p): Likewise.
13742 (emit_jump): Likewise.
13743 (emit_case_decision_tree): Likewise.
13744 (try_switch_expansion): Likewise.
13745 (do_jump_if_equal): Likewise.
13746 (emit_cmp_and_jump_insns): Likewise.
13747 (fix_phi_operands_for_edge): New function.
13748 (record_phi_operand_mapping): Likewise.
13749 (class pass_lower_switch): New pass.
13750 (pass_lower_switch::execute): New function.
13751 (make_pass_lower_switch): Likewise.
13752 (conditional_probability):
13753 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13754 * tree-pass.h: Add make_pass_lower_switch.
13756 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13759 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13762 2017-08-29 Richard Biener <rguenther@suse.de>
13764 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13765 we do not add a DW_AT_inline attribute twice.
13766 (gen_subprogram_die): Remove code setting DW_AT_inline on
13767 DECL_ABSTRACT_P nodes.
13769 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13771 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13772 calls to internal functions.
13773 (gimplify_modify_expr): Likewise.
13774 * tree-call-cdce.c (use_internal_fn): Likewise.
13775 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13776 (convert_to_divmod): Set the nothrow flag.
13777 * tree-if-conv.c (predicate_mem_writes): Likewise.
13778 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13779 (vectorizable_call): Likewise.
13780 (vectorizable_store): Likewise.
13781 (vectorizable_load): Likewise.
13782 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13783 (vect_recog_mask_conversion_pattern): Likewise.
13785 2017-08-29 Martin Liska <mliska@suse.cz>
13788 * gcc.c (driver_handle_option): Add new argument.
13789 * opts-common.c (handle_option): Pass
13790 target_option_override_hook.
13791 * opts-global.c (lang_handle_option): Add new option.
13792 (set_default_handlers): Add new argument.
13793 (decode_options): Likewise.
13794 * opts.c (target_handle_option): Likewise.
13795 (common_handle_option): Call target_option_override_hook.
13796 * opts.h (struct cl_option_handler_func): Add hook for
13797 target option override.
13798 (struct cl_option_handlers): Likewise.
13799 (set_default_handlers): Add new argument.
13800 (decode_options): Likewise.
13801 (common_handle_option): Likewise.
13802 (target_handle_option): Likewise.
13803 * toplev.c (toplev::main): Pass targetm.target_option.override
13806 2017-08-29 Richard Biener <rguenther@suse.de>
13807 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13809 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13810 life to the active subtree.
13812 2017-08-28 Jeff Law <law@redhat.com>
13814 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13815 derive_equivalences.
13816 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13817 Code moved into....
13818 (edge_info::derive_equivalences): New private member function
13820 * tree-ssa-dom.c (class edge_info): Changed from a struct
13821 to a class. Add ctor/dtor, methods and data members.
13822 (edge_info::edge_info): Renamed from allocate_edge_info.
13823 Initialize additional members.
13824 (edge_info::~edge_info): New.
13825 (free_dom_edge_info): Delete the edge info.
13826 (record_edge_info): Use new class & associated member functions.
13827 Tighten forms for testing for edge equivalences.
13828 (record_temporary_equivalences): Iterate over the simple
13829 equivalences rather than assuming there's only one per edge.
13830 (cprop_into_successor_phis): Iterate over the simple
13831 equivalences rather than assuming there's only one per edge.
13832 (optimize_stmt): Use operand_equal_p rather than pointer
13833 equality for mini-DSE code.
13835 2017-08-28 Nathan Sidwell <nathan@acm.org>
13837 * gcc.c (execute): Fold SIGPIPE handling into switch
13838 statement. Adjust internal error message.
13840 2017-08-28 Richard Biener <rguenther@suse.de>
13843 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13844 Do nothing for removed DIEs.
13846 2017-08-28 Richard Biener <rguenther@suse.de>
13848 PR tree-optimization/81977
13849 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13852 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13855 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13856 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13857 using targetm.gen_mem_thread_fence.
13859 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13862 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13863 predicate to register_operand. Reorder operands.
13864 (*btr<mode>): Ditto.
13865 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13866 (*btr<mode>_mask): Ditto.
13868 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13870 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13871 * config/rs6000/xmmintrin.h: New file.
13872 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13874 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13877 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13878 parameter and_insn and return it.
13879 (recombine_lvx_pattern): Insert a copy to ensure availability of
13880 the base register of the copied masking operation at the point of
13881 the instruction replacement.
13882 (recombine_stvx_pattern): Likewise.
13884 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13886 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13887 undocumented switches.
13888 (-mpower9-dform-vector): Likewise.
13889 (-mpower9-dform): Likewise.
13890 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13891 comments to delete references to -mpower9-dform* switches.
13892 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13893 Delete reference to -mpower9-dform* switches, test for
13894 -mpower9-vector instead.
13895 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13896 (OTHER_P9_VECTOR_MASKS): Likewise.
13897 (POWERPC_MASKS): Likewise.
13898 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13899 tests against -mpower9-dform* to -mpower9-vector. Delete code
13900 that checked for -mpower9-dform* consistancy with other options.
13901 Add test for -mpower9-misc to enable other power9 switches.
13902 (rs6000_init_hard_regno_mode_ok): Likewise.
13903 (rs6000_option_override_internal): Likewise.
13904 (rs6000_emit_prologue): Likewise.
13905 (rs6000_emit_epilogue): Likewise.
13906 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13907 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13908 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13910 (emit_fusion_p9_store): Likewise.
13911 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13912 resetting these macros if the assembler does not support ISA 3.0
13914 (TARGET_P9_DFORM_VECTOR): Likewise.
13915 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13916 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13918 2017-08-25 Alan Modra <amodra@gmail.com>
13920 PR rtl-optimization/81747
13921 * cse.c (cse_extended_basic_block): Don't attempt to record
13922 equivalences for degenerate conditional branches that branch
13923 to their fall-through.
13925 2017-08-24 Martin Sebor <msebor@redhat.com>
13927 PR middle-end/81908
13928 * gimple-fold.c (size_must_be_zero_p): New function.
13929 (gimple_fold_builtin_memory_op): Call it.
13931 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13933 * config/rs6000/mm_malloc.h: New file.
13935 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13937 PR tree-optimization/81913
13938 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13939 analysis when either IVs in condition can wrap.
13941 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13943 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13944 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13946 2017-08-24 Richard Biener <rguenther@suse.de>
13949 * targhooks.c (default_target_can_inline_p): Properly
13950 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13951 is present and always compare.
13952 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13953 infer -mfpmath=sse from TARGET_SSE_P.
13954 (ix86_can_inline_p): Properly use target_option_default_node when
13955 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13957 2017-08-24 Richard Biener <rguenther@suse.de>
13960 * dwarf2out.c (output_die): Handle flag_generate_offload like
13962 (output_comp_unit): Likewise.
13963 (gen_array_type_die): Likewise.
13964 (dwarf2out_early_finish): Likewise.
13965 (note_variable_value_in_expr): Likewise.
13966 (dwarf2out_finish): Likewise. Adjust assert.
13967 * cgraphunit.c (symbol_table::compile): Move setting of
13968 flag_generate_offload earlier ...
13969 (symbol_table::finalize_compilation_unit): ... here, before
13970 early debug finalization.
13972 2017-08-24 Richard Biener <rguenther@suse.de>
13974 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13975 and ipa-fnsummary.h.
13976 (ix86_can_inline_p): When ix86_fpmath flags do not match
13977 check whether the callee uses FP math at all.
13979 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13981 PR middle-end/81931
13982 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13983 instead of TYPE_PRECISION.
13985 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13986 Alan Hayward <alan.hayward@arm.com>
13987 David Sherwood <david.sherwood@arm.com>
13989 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13991 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13993 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13995 (ix86_frame::stack_realign_allocate): New field.
13996 (struct machine_frame_state): Modify comments.
13997 (machine_frame_state::sp_realigned_fp_end): New field.
13998 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13999 layout calculation.
14000 (sp_valid_at): Add assertion to assure no attempt to access invalid
14001 offset of a realigned stack.
14002 (fp_valid_at): Likewise.
14003 (choose_baseaddr): Modify comments.
14004 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
14005 ix86_expand_prologue.
14006 (ix86_expand_prologue): Modify stack realignment and allocation.
14007 (ix86_expand_epilogue): Modify comments.
14008 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
14009 avx2_runtime, avx512f, and avx512f_runtime.
14011 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
14013 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
14014 (mstackrealign): Do not init to -1.
14015 * config/i386/i386.c (ix86_option_override_internal):
14016 Check opts_set, not opts when setting default value of
14017 opts->x_ix86_force_align_arg_pointer.
14019 2017-08-23 Richard Biener <rguenther@suse.de>
14021 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
14022 lang_hooks.decl_printable_name.
14023 * print-rtl-function.c (print_rtx_function): Likewise.
14024 * tree-pretty-print.c (dump_function_header): Likewise.
14026 2017-08-23 Richard Biener <rguenther@suse.de>
14029 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
14030 -g0 at compile-time.
14032 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14034 PR middle-end/19706
14035 * doc/sourcebuild.texi (Other hardware attributes):
14038 2017-08-23 Tamar Christina <tamar.christina@arm.com>
14040 PR middle-end/19706
14041 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
14042 Fix single-use check.
14044 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
14046 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
14048 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14050 * doc/install.texi: Modify to add more details on running selected
14053 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
14055 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
14056 is combined with -mabi=ms.
14057 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
14060 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14062 PR tree-optimization/81488
14063 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
14064 and cached_basis fields.
14065 (MAX_SPREAD): New constant.
14066 (alloc_cand_and_find_basis): Initialize new fields.
14067 (clear_visited): New function.
14068 (create_phi_basis_1): Rename from create_phi_basis, set visited
14069 and cached_basis fields.
14070 (create_phi_basis): New wrapper function.
14071 (phi_add_costs_1): Rename from phi_add_costs, add spread
14072 parameter, set visited field, short-circuit when limits reached.
14073 (phi_add_costs): New wrapper function.
14074 (record_phi_increments_1): Rename from record_phi_increments, set
14076 (record_phi_increments): New wrapper function.
14077 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
14078 (phi_incr_cost): New wrapper function.
14079 (all_phi_incrs_profitable_1): Rename from
14080 all_phi_incrs_profitable, set visited field.
14081 (all_phi_incrs_profitable): New wrapper function.
14083 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14084 Alan Hayward <alan.hayward@arm.com>
14085 David Sherwood <david.sherwood@arm.com>
14087 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
14088 that takes the outer and inner modes.
14089 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
14090 comparison as the canonical test for a paradoxical subreg.
14091 * combine.c (simplify_set): Use paradoxical_subreg_p.
14092 (make_extraction): Likewise.
14093 (force_to_mode): Likewise.
14094 (rtx_equal_for_field_assignment_p): Likewise.
14095 (gen_lowpart_for_combine): Likewise.
14096 (simplify_comparison): Likewise.
14097 * cse.c (equiv_constant): Likewise.
14098 * expmed.c (store_bit_field_1): Likewise.
14099 * final.c (alter_subreg): Likewise.
14100 * fwprop.c (propagate_rtx): Likewise.
14101 (forward_propagate_subreg): Likewise.
14102 * ira-conflicts.c (ira_build_conflicts): Likewise.
14103 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
14104 * lra-constraints.c (curr_insn_transform): Likewise.
14105 (split_reg): Likewise.
14106 * lra-eliminations.c (move_plus_up): Likewise.
14107 (lra_eliminate_regs_1): Likewise.
14108 * recog.c (general_operand): Likewise.
14109 * ree.c (combine_reaching_defs): Likewise.
14110 * reload.c (push_reload): Likewise.
14111 (find_reloads): Likewise.
14112 * reload1.c (elimination_effects): Likewise.
14113 (compute_reload_subreg_offset): Likewise.
14114 (choose_reload_regs): Likewise.
14115 * rtlanal.c (subreg_lsb_1): Likewise.
14116 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
14117 (simplify_subreg): Likewise.
14118 * var-tracking.c (track_loc_p): Likewise.
14119 * emit-rtl.c (byte_lowpart_offset): Likewise.
14120 (paradoxical_subreg_p): Delete out-of-line definition.
14122 2017-08-22 Jeff Law <law@redhat.com>
14124 PR tree-optimization/81741
14125 PR tree-optimization/71947
14126 * tree-ssa-dom.c: Include tree-inline.h.
14127 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
14128 equivalences if one is more expensive to compute than the other.
14129 * tree-ssa-scopedtables.h (class const_or_copies): Make
14130 record_const_or_copy_raw method private.
14131 (class avail_exprs_stack): New method simplify_binary_operation.
14132 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
14133 avail_exprs_stack::simplify_binary_operation as needed.
14134 (avail_exprs_stack::simplify_binary_operation): New function.
14136 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
14138 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
14139 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
14140 (DOT_SYMBOLS): Likewise.
14141 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
14142 (RELOCATABLE_NEEDS_FIXUP): Likewise.
14143 (RS6000_ABI_NAME): Likewise.
14144 (TARGET_CMODEL): Likewise.
14145 (TOC_SECTION_ASM_OP): Likewise.
14146 (SET_CMODEL): New macro.
14147 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
14149 2017-08-22 Richard Biener <rguenther@suse.de>
14151 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
14152 to allow for free-lang-data replacements similar to verify_type_variant.
14154 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
14155 Alan Hayward <alan.hayward@arm.com>
14156 David Sherwood <david.sherwood@arm.com>
14158 * config/aarch64/aarch64.md (casesi): Use DImode rather than
14159 VOIDmode for the LABEL_REF.
14161 2017-08-22 Richard Biener <rguenther@suse.de>
14163 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
14165 2017-08-22 Richard Biener <rguenther@suse.de>
14167 * common.opt (feliminate-dwarf2-dups): Ignore.
14168 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
14169 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
14170 same_die_p_wrap, compute_section_prefix,
14171 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
14172 (comdat_symbol_id, comdat_symbol_number): Likewise.
14173 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
14175 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
14176 (output_die): Mark unreachable path unreachable.
14177 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
14178 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
14179 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
14180 (dwarf2out_early_finish): Likewise.
14182 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
14184 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
14186 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
14189 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
14190 not VAR_P. Filter attribute warnings with OPT_Wattributes.
14191 (avr_attribute_table) <io, io_low, address>: Initialize
14192 .decl_required with true.
14194 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
14196 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
14197 undocumented debugging options.
14198 (-mvsx-scalar-double): Likewise.
14199 (-mallow-df-permute): Likewise.
14200 (-mvectorize-builtins): Likewise.
14201 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
14202 (rs6000_builtin_vectorized_function): Likewise.
14203 (rs6000_builtin_md_vectorized_function): Likewise.
14204 (rs6000_opt_vars): Likewise.
14206 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
14209 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
14210 (*btrq_imm): Rename from *btrq.
14211 (*btcq_imm): Rename from *btcq.
14212 (btsc): New code attribute.
14213 (*<btsc><mode>): New insn pattern.
14214 (*btr<mode>): Ditto.
14215 (*<btsc><mode>_mask): New insn_and_split pattern.
14216 (*btr<mode>_mask): Ditto.
14218 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14219 Alan Hayward <alan.hayward@arm.com>
14220 David Sherwood <david.sherwood@arm.com>
14222 * function.c (pad_below): Simplify padding calculation.
14224 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14225 Alan Hayward <alan.hayward@arm.com>
14226 David Sherwood <david.sherwood@arm.com>
14228 * target.def (function_prologue): Remove frame size argument.
14229 (function_epilogue): Likewise.
14230 * doc/tm.texi: Regenerate.
14231 * final.c (final_start_function): Update call to function_prologue.
14232 (final_end_function): Update call to function_epilogue.
14233 (default_function_pro_epilogue): Remove frame size argument.
14234 * output.h (default_function_pro_epilogue): Likewise.
14235 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
14236 (arm_output_function_prologue): Likewise.
14237 * config/frv/frv.c (frv_function_prologue): Likewise.
14238 (frv_function_epilogue): Likewise.
14239 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
14240 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
14241 (ia64_output_function_epilogue): Likewise.
14242 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
14243 (m32r_output_function_epilogue): Likewise.
14244 * config/microblaze/microblaze.c (microblaze_function_prologue)
14245 (microblaze_function_epilogue): Likewise.
14246 * config/mips/mips.c (mips_output_function_prologue): Likewise.
14247 (mips_output_function_epilogue): Likewise.
14248 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
14249 (mmix_target_asm_function_epilogue): Likewise.
14250 * config/msp430/msp430.c (msp430_start_function): Likewise.
14251 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
14252 (nds32_asm_function_epilogue): Likewise.
14253 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
14254 * config/pa/pa.c (pa_output_function_prologue): Likewise.
14255 (pa_output_function_epilogue): Likewise.
14256 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
14257 (rs6000_output_function_epilogue): Likewise.
14258 * config/rl78/rl78.c (rl78_start_function): Likewise.
14259 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
14260 (rs6000_output_function_epilogue): Likewise.
14261 * config/rx/rx.c (rx_output_function_prologue): Likewise.
14262 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
14263 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
14264 (sparc_asm_function_epilogue): Likewise.
14266 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14268 * tree.h (type_has_mode_precision_p): New function.
14269 * convert.c (convert_to_integer_1): Use it.
14270 * expr.c (expand_expr_real_2): Likewise.
14271 (expand_expr_real_1): Likewise.
14272 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
14273 * match.pd: Likewise.
14274 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
14275 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
14276 * tree-tailcall.c (process_assignment): Likewise.
14277 * tree-vect-loop.c (vectorizable_reduction): Likewise.
14278 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
14279 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
14280 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
14281 (vectorizable_assignment): Likewise.
14282 (vectorizable_shift): Likewise.
14283 (vectorizable_operation): Likewise.
14284 * tree-vrp.c (register_edge_assert_for_2): Likewise.
14286 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
14288 * match.pd: Add pow (C, x) simplification.
14290 2017-08-21 Richard Biener <rguenther@suse.de>
14292 PR tree-optimization/81900
14293 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
14294 for blocks with abnormal predecessors.
14295 (compute_antic): Do not set visited flag prematurely.
14297 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
14300 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
14302 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14304 * stor-layout.h (vector_type_mode): Move to...
14305 * tree.h (vector_type_mode): ...here.
14306 * stor-layout.c (vector_type_mode): Move to...
14307 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
14309 2017-08-21 Richard Biener <rguenther@suse.de>
14311 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
14312 register_external_die hooks.
14313 (debug_false_tree_charstarstar_uhwistar): Declare.
14314 (debug_nothing_tree_charstar_uhwi): Likewise.
14315 * debug.c (do_nothing_debug_hooks): Adjust.
14316 (debug_false_tree_charstarstar_uhwistar): New do nothing.
14317 (debug_nothing_tree_charstar_uhwi): Likewise.
14318 * dbxout.c (dbx_debug_hooks): Adjust.
14319 (xcoff_debug_hooks): Likewise.
14320 * sdbout.c (sdb_debug_hooks): Likewise.
14321 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14322 * dwarf2out.c (macinfo_label_base): New global.
14323 (dwarf2out_register_external_die): New function for the
14324 register_external_die hook.
14325 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
14326 (dwarf2_debug_hooks): Use them.
14327 (dwarf2_lineno_debug_hooks): Adjust.
14328 (struct die_struct): Add with_offset flag.
14329 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
14330 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
14331 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
14332 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
14333 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
14334 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
14335 defining section names for the early LTO debug variants.
14336 (reset_indirect_string): New helper.
14337 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
14338 (print_dw_val): Add support for offsetted symbol references.
14339 (get_ultimate_context): Split out from is_cxx.
14340 (is_cxx): Use get_ultimate_context.
14341 (is_fortran): Add decl overload.
14342 (compute_comp_unit_symbol): Split out worker from
14343 compute_section_prefix.
14344 (compute_section_prefix): Call compute_comp_unit_symbol and
14345 set comdat_type_p here.
14346 (output_die): Skip DIE symbol output for the LTO added one.
14347 Handle DIE symbol references with offset.
14348 (output_comp_unit): Guard section name mangling properly.
14349 For LTO debug sections emit a symbol at the section beginning
14350 which we use to refer to its DIEs.
14351 (add_abstract_origin_attribute): For DIEs registered via
14352 dwarf2out_register_external_die directly refer to the early
14353 DIE rather than indirectly through the shadow one we created.
14354 Remove obsolete call to dwarf2out_abstract_function for
14355 non-function/block origins.
14356 (gen_array_type_die): When generating early LTO debug do
14357 not emit DW_AT_string_length.
14358 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
14359 late when in LTO. As suggested place a gcc_unreachable for
14360 the DECL_ABSTRACT_P case.
14361 (gen_subprogram_die): Avoid another specification DIE
14362 for early built declarations/definitions for the late LTO case.
14363 (gen_variable_die): Add type references for late duplicated VLA dies
14365 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
14366 we have the abstract instance already.
14367 (process_scope_var): Adjust decl DIE contexts in LTO which
14368 first puts them in limbo.
14369 (gen_decl_die): Do not generate type DIEs late apart from
14370 types for VLAs or for decls we do not yet have a DIE. Do not
14371 call dwarf2out_abstract_function late.
14372 (dwarf2out_early_global_decl): Make sure to create DIEs
14373 for abstract instances of a decl first.
14374 (dwarf2out_late_global_decl): Adjust comment.
14375 (output_macinfo_op): With multiple macro sections use
14376 macinfo_label_base to distinguish labels.
14377 (output_macinfo): Likewise. Update macinfo_label_base.
14378 Pass in the line info label.
14379 (note_variable_value_in_expr): When generating LTO resolve
14380 all variable values here by generating DIEs as needed.
14381 (init_sections_and_labels): Add early LTO debug flag parameter
14382 and generate different sections and names if set. Add generation
14383 counter for the labels so we can have multiple of them.
14384 (reset_dies): Helper to allow DIEs to be output multiple times.
14385 (dwarf2out_finish): When outputting DIEs to the fat part of an
14386 LTO object first reset DIEs.
14387 (dwarf2out_early_finish): Output early DIEs when generating LTO.
14388 (modified_type_die): Check for decl_ultimate_origin being self
14390 (gen_type_die_with_usage): Likewise.
14391 (gen_typedef_die): Allow decl_ultimate_origin being self.
14392 (set_decl_abstract_flags): Remove.
14393 (set_block_abstract_flags): Likewise.
14394 (dwarf2out_abstract_function): Treat the early generated DIEs
14395 as the abstract copy and only add DW_AT_inline and
14396 DW_AT_artificial here and call set_decl_origin_self.
14397 If the DIE has an abstract origin don't do anything.
14398 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
14399 if we have none yet (Go fails to build one, PR78628).
14400 (variably_modified_type_p): Prevent endless recursion for Ada
14401 cyclic pointer types.
14402 * lto-streamer-in.c: Include debug.h.
14403 (dref_queue): New global.
14404 (lto_read_tree_1): Stream in DIE references.
14405 (lto_input_tree): Register DIE references.
14406 (input_function): Stream DECL_DEBUG_ARGS.
14407 * lto-streamer-out.c: Include debug.h.
14408 (lto_write_tree_1): Output DIE references.
14409 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
14410 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
14411 (output_function): Stream DECL_DEBUG_ARGS.
14412 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
14413 Stream DECL_ABSTRACT_ORIGIN.
14414 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
14415 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
14416 DECL_CONTEXT for file-scope decls.
14417 * lto-streamer.h (struct dref_entry): Declare.
14418 (dref_queue): Likewise.
14419 * cfgexpand.c (pass_expand::execute): Do not call the
14420 outlining_inline_function hook here.
14421 * lto-wrapper.c (debug_obj): New global.
14422 (tool_cleanup): Unlink it if required.
14423 (debug_objcopy): New function.
14424 (run_gcc): Handle early debug sections in the IL files by
14425 extracting them to separate files, partially linkin them and
14426 feeding the result back as result to the linker.
14427 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14428 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14429 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14430 sections into a separate segment.
14431 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14433 (darwin_asm_dwarf_section): Likewise.
14434 (darwin_asm_output_dwarf_offset): Likewise.
14435 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14437 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14438 Alan Hayward <alan.hayward@arm.com>
14439 David Sherwood <david.sherwood@arm.com>
14441 * read-md.h (md_reader::record_potential_iterator_use): Replace
14442 pointer argument with an rtx and an index.
14443 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14444 (apply_mode_iterator): Likewise.
14445 (apply_code_iterator): Likewise.
14446 (apply_int_iterator): Likewise.
14447 (apply_subst_iterator): Likewise.
14448 (record_iterator_use): Likewise.
14449 (record_attribute_use): Likewise.
14450 (md_reader::record_potential_iterator_use): Likewise. Update calls
14451 to record_iterator_use and apply_iterator.
14452 (iterator_use): Replace ptr with x and index.
14453 (attribute_use): Likewise.
14454 (apply_attribute_uses): Update calls to apply_iterator.
14455 (apply_iterators): Likewise. Update initialization of iterator_use.
14456 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14457 and record_potential_iterator_use.
14458 (rtx_reader::read_rtx_operand): Likewise.
14460 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14461 Alan Hayward <alan.hayward@arm.com>
14462 David Sherwood <david.sherwood@arm.com>
14464 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14467 2017-08-21 Richard Biener <rguenther@suse.de>
14469 PR middle-end/81884
14470 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14471 at struct end conservatively when comparing common bases.
14473 2017-08-21 Richard Biener <rguenther@suse.de>
14475 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14476 (mem_ref_in_stmt): Remove.
14477 (determine_max_movement): Use ref index to get at the reference.
14478 (invariantness_dom_walker::before_dom_children): Deal with
14479 lim data already initialized.
14480 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14482 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14484 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14485 (TARGET_ISA_ROUND): Ditto.
14486 (TARGET_ROUND): Ditto.
14487 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14488 * config/i386/i386.md: Ditto.
14489 * config/i386/sse.md: Ditto.
14490 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14491 with OPTION_MASK_ISA_SSE4_1.
14493 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14496 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14497 __builtin_ia32_lzcnt_u16.
14499 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14502 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14503 (rs6000_set_current_function): Rewrite function to use it.
14505 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14508 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14509 and TYPE_WARN_IF_NOT_ALIGN.
14510 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14511 (handle_warn_if_not_align): New.
14512 (place_union_field): Call handle_warn_if_not_align.
14513 (place_field): Call handle_warn_if_not_align.
14514 Copy TYPE_WARN_IF_NOT_ALIGN.
14515 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14516 (layout_type): Likewise.
14517 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14519 (tree_decl_common): Add warn_if_not_align.
14520 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14521 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14522 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14523 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14524 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14525 * doc/extend.texi: Document warn_if_not_aligned attribute.
14526 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14528 2017-08-17 Martin Liska <mliska@suse.cz>
14531 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14532 (get_data_dependence): Use it as pointer type.
14533 (distribute_loop): Likewise.
14535 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14537 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14538 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14539 (p8_vmrgow_<mode>_direct): New define_insn.
14540 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14541 handle endianness for vmrgew and vmrgow permute patterns.
14543 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14545 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14546 * config/rs6000/rs6000-cpus.def: Remove comment.
14547 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14548 (POWERPC_MASKS): Likewise.
14549 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14550 use of TARGET_VSX_TIMODE.
14551 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14552 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14553 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14554 (rs6000_option_override_internal): Remove dead code.
14555 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14556 (rs6000_legitimize_reload_address): Likewise.
14557 (rs6000_legitimate_address_p): Likewise.
14558 (rs6000_opt_masks): Delete "vsx-timode".
14559 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14560 from function comment.
14561 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14562 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14563 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14565 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14566 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14567 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14568 (VSX_TI): Likewise.
14570 (define_peephole2): Likewise.
14572 2017-08-17 Martin Sebor <msebor@redhat.com>
14575 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14576 past the end of an array.
14577 (test_pp_format): Add test cases.
14579 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14581 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14582 missing ECF_NOTHROW flags.
14584 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14587 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14588 operands residing in integer registers.
14589 (*vsx_le_perm_load_<mode>): Likewise.
14590 (*vsx_le_perm_store_<mode>): Likewise.
14591 (define_peephole2): Add peepholes to optimize the above.
14593 2017-08-17 Marek Polacek <polacek@redhat.com>
14595 PR middle-end/81814
14596 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14597 to mimic what shorten_compare did. Change the return type to bool.
14598 (fold_cond_expr_with_comparison): Update call to
14599 operand_equal_for_comparison_p.
14600 (fold_ternary_loc): Likewise.
14602 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14604 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14606 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14608 2017-08-17 Richard Biener <rguenther@suse.de>
14610 * tree-ssa-structalias.c (solve_graph): When propagating
14611 to successors update the graphs succ edges and avoid duplicate work.
14613 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14616 * config/i386/i386.c (ix86_option_override_internal): Save target
14617 specific options after ix86_stack_protector_guard_reg was changed.
14619 2017-08-17 Richard Biener <rguenther@suse.de>
14621 PR tree-optimization/81827
14622 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14623 (new_var_info): Initialize it conservatively.
14624 (get_call_vi): Mark register vars.
14625 (new_scalar_tmp_constraint_exp): Likewise.
14626 (handle_rhs_call): Likewise.
14627 (handle_const_call): Likewise.
14628 (create_function_info_for): Likewise.
14629 (solve_constraints): Sort varinfos to separate register from
14630 non-register vars to pack points-to solution bitmaps during
14633 2017-08-17 Marek Polacek <polacek@redhat.com>
14635 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14637 2017-08-17 Richard Biener <rguenther@suse.de>
14639 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14640 to true when overflow is undefined and we saturated the result.
14642 2017-08-17 Alan Modra <amodra@gmail.com>
14645 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14646 Don't use store multiple if only one reg needs saving.
14647 (interesting_frame_related_regno): New function.
14648 (rs6000_frame_related): Don't emit frame info for regs that
14650 (rs6000_emit_epilogue): Likewise.
14652 2017-08-16 Nathan Sidwell <nathan@acm.org>
14654 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14655 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14656 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14657 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14658 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14659 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14660 (verify_type): Adjust for TYPE_BINFO move.
14661 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14662 process TYPE_BINFO directly.
14663 (hash_tree): Likewise.
14664 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14666 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14669 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14671 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14673 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14676 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14677 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14678 Add ix86_binary_operator_ok to insn constraint.
14679 (*iordi_1_bts): Ditto.
14680 (*xordi_1_btc): Ditto.
14681 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14682 Update corresponding peephole2 pattern.
14686 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14688 PR tree-optimization/81832
14689 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14690 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14692 2017-08-16 Marek Polacek <polacek@redhat.com>
14695 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14696 regarding size_zero_node.
14698 2017-08-16 Martin Liska <mliska@suse.cz>
14701 * config.gcc: Respect previously set extra_objs in case
14704 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14706 PR tree-optimization/81835
14707 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14708 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14709 not depend on the phi.
14711 2017-08-16 Alan Modra <amodra@gmail.com>
14713 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14716 2017-08-16 Alan Modra <amodra@gmail.com>
14718 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14719 (save_reg_p): ..into this. Update all callers.
14720 (first_reg_to_save): Simplify.
14722 2017-08-16 Alan Modra <amodra@gmail.com>
14724 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14727 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14731 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14732 constant count if that count is less than 32.
14734 2017-08-15 Nathan Sidwell <nathan@acm.org>
14736 * gcc.c (execute): Emit friendlier message if inferior is killed
14737 by an external cause.
14739 2017-08-15 Richard Biener <rguenther@suse.de>
14741 PR tree-optimization/81790
14742 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14743 CONSTRUCTORs from simplifying and VN.
14745 2017-08-14 Martin Sebor <msebor@redhat.com>
14747 * builtin-attrs.def: Add comments.
14749 2017-08-14 Martin Sebor <msebor@redhat.com>
14752 * doc/extend.texi (attribute nonstring): Document new attribute.
14754 2017-08-14 Martin Sebor <msebor@redhat.com>
14757 * tree-diagnostic.c (default_tree_printer): Handle %G.
14758 * gimple-pretty-print.h (percent_G_format): Declare new function.
14759 * gimple-pretty-print.c (percent_G_format): Define.
14760 * tree-pretty-print.c (percent_K_format): Add argument.
14762 2017-08-14 Martin Sebor <msebor@redhat.com>
14764 PR translation/79998
14765 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14766 Remove a stray space.
14768 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14771 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14772 (*iordi_1_bts): Ditto.
14773 (*xordi_1_btc): Ditto.
14775 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14778 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14779 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14781 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14782 quoted strings, and make more translator-friendly.
14783 (darwin_rs6000_override_options): Likewise.
14784 (rs6000_option_override_internal): Likewise.
14785 (rs6000_return_in_memory): Fix overlong line.
14786 (init_cmulative_args): Use quoted strings, and make more
14787 translator-friendly.
14788 (rs6000_pass_by_reference): Fix overlong line.
14789 (def_builtin): Use quoted strings.
14790 (altivec_expand_predicate_builtin): Use quoted strings, and make
14791 more translator-friendly.
14792 (htm_expand_builtin): Use quoted strings.
14793 (cpu_expand_builtin): Use quoted strings, and make more
14794 translator-friendly.
14795 (altivec_expand_builtin): Likewise.
14796 (paired_expand_predicate_builtin): Likewise.
14797 (rs6000_invalid_builtin): Likewise.
14798 (builtin_function_type): Use quoted strings.
14799 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14800 more translator-friendly.
14801 (rs6000_trampoline_init): Likewise.
14802 (rs6000_handle_altivec_attribute): Likewise.
14803 (rs6000_inner_target_options): Use quoted strings.
14804 (rs6000_disable_incompatible_switches): Likewise.
14805 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14806 strings, and make more translator-friendly.
14807 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14809 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14811 PR tree-optimization/81799
14812 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14813 cond_expr to simple gimple operand.
14815 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14817 PR middle-end/46932
14818 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14820 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14824 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14825 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14826 TARGET_GASISR_PROLOGUES.
14827 * config/avr/avr.c (avr_option_override): Same.
14828 (avr_pass_pre_proep::execute): Same.
14830 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14833 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14834 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14836 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14838 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14839 define_insn_and_split. Split to a memory load from 0 in
14840 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14841 using PTR mode iterator.
14842 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14843 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14844 (*add_tp_<mode>): Redefine as define_insn_and_split.
14845 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14846 address space. Merge with *add_tp_x32 using PTR mode iterator.
14847 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14848 Split to an add with a memory load from 0 in
14849 DEFAULT_TLS_SEG_REG address space.
14851 2017-08-12 Andrew Pinski <apinski@cavium.com>
14853 * config/aarch64/aarch64-option-extensions.def (rdma):
14854 Fix feature string to what Linux prints out in /proc/cpuinfo.
14856 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14859 * dwarf2out.c (modified_type_die): For C typedef types that have
14860 an ultimate origin, process the ultimate origin instead of the
14862 (gen_typedef_die): Assert that input DECLs have no ultimate
14864 (gen_type_die_with_usage): For typedef variants that have an
14865 ultimate origin, just call gen_decl_die on the original DECL.
14866 (process_scope_var): Avoid creating DIEs for local typedefs and
14867 concrete static variables.
14869 2017-08-12 Alan Modra <amodra@gmail.com>
14873 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14874 match gnu-user.h startfile.
14875 (ENDFILE_LINUX_SPEC): Similarly.
14877 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14880 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14882 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14884 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14885 * config/aarch64/aarch64.md (mov<mode>): Change.
14886 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14887 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14888 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14890 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14892 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14893 for storage order barriers.
14895 2017-08-11 Martin Liska <mliska@suse.cz>
14898 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14899 variables of void type.
14901 2017-08-11 Martin Liska <mliska@suse.cz>
14903 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14904 TARGET_SUPPORTS_ALIASES.
14905 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14906 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14907 (optimize_weakref): Likewise.
14908 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14909 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14910 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14912 2017-08-11 Martin Liska <mliska@suse.cz>
14915 * config/i386/i386.c (make_resolver_func): Do complete
14916 refactoring of the function.
14918 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14921 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14922 * config/i386/i386.c (ix86_stack_protect_guard): Use
14923 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14924 * doc/invoke.texi (x86 Options): Document
14925 -mstack-protector-guard-symbol= option.
14927 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14929 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14930 * config/i386/i386.c (ix86_split_stack_guard): New function.
14931 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14932 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14933 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14934 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14935 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14936 (split_stack_space_check): Call ix86_split_stack_guard.
14938 2017-08-10 Martin Sebor <msebor@redhat.com>
14940 * print-tree.c (print_node): Print location using the established
14942 Replace spaces with colons.
14943 (debug_raw, debug): Ditto.
14945 2017-08-10 Martin Sebor <msebor@redhat.com>
14948 * pretty-print.c (pp_format): Correct the handling of %s precision.
14950 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14953 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14955 (ix86_finalize_stack_frame_flags): This. Also clear
14956 frame_pointer_needed if -fno-omit-frame-pointer is used without
14958 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14959 with ix86_finalize_stack_frame_flags.
14960 (ix86_expand_epilogue): Likewise.
14961 (ix86_expand_split_stack_prologue): Likewise.
14962 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14964 2017-08-10 Martin Liska <mliska@suse.cz>
14967 * c-attribs.c (handle_target_attribute):
14968 Report warning for an empty string argument of target attribute.
14970 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14973 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14975 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14976 or DECL_NONLOCAL labels.
14977 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14978 or DECL_NONLOCAL labels here.
14980 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14982 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14983 to indicate when early gimple folding has been disabled.
14984 (rs6000_gimple_fold_builtin): Add debug content.
14985 (rs6000_invalid_builtin): Fix whitespace.
14986 (rs6000_expand_builtin): Fix whitespace.
14987 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14989 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14992 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14993 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14995 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14997 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14998 (qdf24xx): Likewise.
14999 * config/aarch64/aarch64.md: Include falkor.md.
15000 * config/aarch64/falkor.md: New.
15002 2017-08-09 Marek Polacek <polacek@redhat.com>
15005 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
15006 * diagnostic.c (emit_diagnostic): Add a comment.
15007 (emit_diagnostic_valist): New function.
15009 2017-08-09 Marek Polacek <polacek@redhat.com>
15012 * input.c (make_location): New overload.
15013 * input.h (make_location): Declare.
15015 2017-08-08 Alan Modra <amodra@gmail.com>
15016 H.J. Lu <hongjiu.lu@intel.com>
15019 * gcc.c (NO_PIE_SPEC): Delete.
15020 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
15022 (LINK_PIE_SPEC): ..to here.
15023 (LINK_COMMAND_SPEC): Support -no-pie.
15024 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
15025 chain of crtbegin*.o selection, update for PIE_SPEC changes and
15027 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
15028 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
15029 (ENDFILE_CRTEND_SPEC): Similarly.
15031 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
15034 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
15035 (mstack-protector-guard-offset=): Ditto.
15036 * config/i386/i386.c (ix86_option_override): Handle
15037 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
15039 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
15040 ix86_stack_protect_guard_offset variables.
15041 (TARGET_STACK_PROTECT_GUARD): Always define.
15042 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
15043 and -mstack-protector-guard-offset= options.
15045 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15047 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
15048 boundary case for the last candidate.
15050 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15052 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
15053 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
15055 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15057 PR middle-end/19706
15058 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
15059 * config/aarch64/aarch64-builtins.c
15060 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
15061 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
15062 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
15064 2017-08-08 Tamar Christina <tamar.christina@arm.com>
15065 Andrew Pinski <pinskia@gmail.com>
15067 PR middle-end/19706
15068 * internal-fn.def (XORSIGN): New.
15069 * optabs.def (xorsign_optab): New.
15070 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
15071 (convert_expand_mult_copysign): New.
15072 (pass_optimize_widening_mul::execute): Call
15073 convert_expand_mult_copysign.
15075 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15077 PR tree-optimization/81354
15078 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
15079 Insert on edges rather than explicitly creating landing pads.
15080 (analyze_candidates_and_replace): Commit edge inserts.
15082 2017-08-08 Richard Biener <rguenther@suse.de>
15084 PR middle-end/81719
15085 * tree-ssa-loop-niter.c: Include tree-dfa.h.
15086 (expand_simple_operations): Also look through ADDR_EXPRs with
15087 MEM_REF bases treating them as POINTER_PLUS_EXPR.
15089 2017-08-08 Richard Biener <rguenther@suse.de>
15091 PR tree-optimization/81723
15092 * tree-vect-slp.c (struct bst_traits): New hash traits.
15093 (bst_fail): New global.
15094 (vect_build_slp_tree_2): New worker, split out from ...
15095 (vect_build_slp_tree): ... this now wrapping it with using
15096 bst_fail set to cache SLP tree build fails. Properly handle
15098 (vect_analyze_slp_instance): Allocate and free bst_fail.
15100 2017-08-08 Martin Liska <mliska@suse.cz>
15103 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
15104 LABEL_DECLs that can be from a different function.
15106 2017-08-08 Bin Cheng <bin.cheng@arm.com>
15108 PR tree-optimization/81744
15109 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
15110 loop's number of iterations.
15112 2017-08-08 Martin Liska <mliska@suse.cz>
15114 * asan.c: Include header files.
15115 * attribs.c (build_decl_attribute_variant): New function moved
15117 (build_type_attribute_qual_variant): Likewise.
15118 (cmp_attrib_identifiers): Likewise.
15119 (simple_cst_list_equal): Likewise.
15120 (omp_declare_simd_clauses_equal): Likewise.
15121 (attribute_value_equal): Likewise.
15122 (comp_type_attributes): Likewise.
15123 (build_type_attribute_variant): Likewise.
15124 (lookup_ident_attribute): Likewise.
15125 (remove_attribute): Likewise.
15126 (merge_attributes): Likewise.
15127 (merge_type_attributes): Likewise.
15128 (merge_decl_attributes): Likewise.
15129 (merge_dllimport_decl_attributes): Likewise.
15130 (handle_dll_attribute): Likewise.
15131 (attribute_list_equal): Likewise.
15132 (attribute_list_contained): Likewise.
15133 * attribs.h (lookup_attribute): New function moved from tree.[ch].
15134 (lookup_attribute_by_prefix): Likewise.
15135 * bb-reorder.c: Include header files.
15136 * builtins.c: Likewise.
15137 * calls.c: Likewise.
15138 * cfgexpand.c: Likewise.
15139 * cgraph.c: Likewise.
15140 * cgraphunit.c: Likewise.
15141 * convert.c: Likewise.
15142 * dwarf2out.c: Likewise.
15143 * final.c: Likewise.
15144 * fold-const.c: Likewise.
15145 * function.c: Likewise.
15146 * gimple-expr.c: Likewise.
15147 * gimple-fold.c: Likewise.
15148 * gimple-pretty-print.c: Likewise.
15149 * gimple.c: Likewise.
15150 * gimplify.c: Likewise.
15151 * hsa-common.c: Likewise.
15152 * hsa-gen.c: Likewise.
15153 * internal-fn.c: Likewise.
15154 * ipa-chkp.c: Likewise.
15155 * ipa-cp.c: Likewise.
15156 * ipa-devirt.c: Likewise.
15157 * ipa-fnsummary.c: Likewise.
15158 * ipa-inline.c: Likewise.
15159 * ipa-visibility.c: Likewise.
15161 * lto-cgraph.c: Likewise.
15162 * omp-expand.c: Likewise.
15163 * omp-general.c: Likewise.
15164 * omp-low.c: Likewise.
15165 * omp-offload.c: Likewise.
15166 * omp-simd-clone.c: Likewise.
15167 * opts-global.c: Likewise.
15168 * passes.c: Likewise.
15169 * predict.c: Likewise.
15170 * sancov.c: Likewise.
15171 * sanopt.c: Likewise.
15172 * symtab.c: Likewise.
15173 * toplev.c: Likewise.
15174 * trans-mem.c: Likewise.
15175 * tree-chkp.c: Likewise.
15176 * tree-eh.c: Likewise.
15177 * tree-into-ssa.c: Likewise.
15178 * tree-object-size.c: Likewise.
15179 * tree-parloops.c: Likewise.
15180 * tree-profile.c: Likewise.
15181 * tree-ssa-ccp.c: Likewise.
15182 * tree-ssa-live.c: Likewise.
15183 * tree-ssa-loop.c: Likewise.
15184 * tree-ssa-sccvn.c: Likewise.
15185 * tree-ssa-structalias.c: Likewise.
15186 * tree-ssa.c: Likewise.
15187 * tree-streamer-in.c: Likewise.
15188 * tree-vectorizer.c: Likewise.
15189 * tree-vrp.c: Likewise.
15190 * tsan.c: Likewise.
15191 * ubsan.c: Likewise.
15192 * varasm.c: Likewise.
15193 * varpool.c: Likewise.
15194 * tree.c: Remove functions moved to attribs.[ch].
15195 * tree.h: Likewise.
15196 * config/aarch64/aarch64.c: Add attrs.h header file.
15197 * config/alpha/alpha.c: Likewise.
15198 * config/arc/arc.c: Likewise.
15199 * config/arm/arm.c: Likewise.
15200 * config/avr/avr.c: Likewise.
15201 * config/bfin/bfin.c: Likewise.
15202 * config/c6x/c6x.c: Likewise.
15203 * config/cr16/cr16.c: Likewise.
15204 * config/cris/cris.c: Likewise.
15205 * config/darwin.c: Likewise.
15206 * config/epiphany/epiphany.c: Likewise.
15207 * config/fr30/fr30.c: Likewise.
15208 * config/frv/frv.c: Likewise.
15209 * config/ft32/ft32.c: Likewise.
15210 * config/h8300/h8300.c: Likewise.
15211 * config/i386/winnt.c: Likewise.
15212 * config/ia64/ia64.c: Likewise.
15213 * config/iq2000/iq2000.c: Likewise.
15214 * config/lm32/lm32.c: Likewise.
15215 * config/m32c/m32c.c: Likewise.
15216 * config/m32r/m32r.c: Likewise.
15217 * config/m68k/m68k.c: Likewise.
15218 * config/mcore/mcore.c: Likewise.
15219 * config/microblaze/microblaze.c: Likewise.
15220 * config/mips/mips.c: Likewise.
15221 * config/mmix/mmix.c: Likewise.
15222 * config/mn10300/mn10300.c: Likewise.
15223 * config/moxie/moxie.c: Likewise.
15224 * config/msp430/msp430.c: Likewise.
15225 * config/nds32/nds32-isr.c: Likewise.
15226 * config/nds32/nds32.c: Likewise.
15227 * config/nios2/nios2.c: Likewise.
15228 * config/nvptx/nvptx.c: Likewise.
15229 * config/pa/pa.c: Likewise.
15230 * config/pdp11/pdp11.c: Likewise.
15231 * config/powerpcspe/powerpcspe.c: Likewise.
15232 * config/riscv/riscv.c: Likewise.
15233 * config/rl78/rl78.c: Likewise.
15234 * config/rx/rx.c: Likewise.
15235 * config/s390/s390.c: Likewise.
15236 * config/sh/sh.c: Likewise.
15237 * config/sol2.c: Likewise.
15238 * config/sparc/sparc.c: Likewise.
15239 * config/spu/spu.c: Likewise.
15240 * config/stormy16/stormy16.c: Likewise.
15241 * config/tilegx/tilegx.c: Likewise.
15242 * config/tilepro/tilepro.c: Likewise.
15243 * config/v850/v850.c: Likewise.
15244 * config/vax/vax.c: Likewise.
15245 * config/visium/visium.c: Likewise.
15246 * config/xtensa/xtensa.c: Likewise.
15248 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15251 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
15252 constraints since the -mupper-regs-* switches have been
15254 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
15255 into a vector from a double word element that was extracted from
15256 another vector, and eliminate extra XXPERMDI instructions.
15257 (vsx_concat_<mode>_2): Likewise.
15258 (vsx_concat_<mode>_3): Likewise.
15259 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
15260 concat to allow optimizing inserts from previous extracts.
15262 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
15264 * config/i386/i386.c (ix86_stack_protect_guard): Generate
15265 memory reference to a SSP offset in TLS address space.
15266 (ix86_print_operand) <case '@'>: Remove.
15267 (ix86_print_operand_punct_valid_p): Remove '@' code.
15268 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
15269 UNSPEC_SP_TLS_TEST.
15270 (stack_tls_protect_set_<mode>): Remove.
15271 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
15272 (stack_tls_protect_test_<mode>): Remove.
15273 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
15275 2017-08-07 Olivier Hainque <hainque@adacore.com>
15278 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
15280 2017-08-07 Douglas Rupp <rupp@adacore.com>
15282 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
15283 variable was referenced as multidir in command.
15285 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15288 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
15291 2017-08-07 Martin Liska <mliska@suse.cz>
15293 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
15294 * config/rl78/rl78.c: Add include of attribs.h.
15295 * config/sh/sh.c: Likewise.
15296 * config/v850/v850.c: Likewise.
15298 2017-08-07 Tom de Vries <tom@codesourcery.com>
15300 PR middle-end/78266
15301 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
15303 2017-08-07 Martin Liska <mliska@suse.cz>
15305 * config/mips/mips.c: Include attribs.h.
15307 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15310 * doc/invoke.texi: Document change in behvaior for -Ofast for
15313 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
15315 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
15317 (aarch64_pop_regs): Likewise.
15318 (aarch64_gen_load_pair): Likewise.
15319 (aarch64_save_callee_saves): Likewise.
15320 (aarch64_restore_callee_saves): Likewise.
15322 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15324 * config/i386/i386.c: Revert the last change.
15326 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15329 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
15331 (ix86_finalize_stack_frame_flags): This. Also clear
15332 frame_pointer_needed if -fno-omit-frame-pointer is used without
15334 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
15335 with ix86_finalize_stack_frame_flags.
15336 (ix86_expand_epilogue): Likewise.
15337 (ix86_expand_split_stack_prologue): Likewise.
15339 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
15342 * config/i386/i386.c (get_builtin_code_for_version): Set priority
15343 to P_AES for Westmere.
15345 2017-08-07 Jonathan Yong <10walls@gmail.com>
15347 * config/i386/mingw.opt (fset-stack-executable): Removed.
15348 * config/i386/cygming.opt (fset-stack-executable): Moved
15350 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
15352 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
15354 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
15356 2017-08-07 Marek Polacek <polacek@redhat.com>
15358 PR middle-end/81737
15359 * fold-const.c (fold_indirect_ref_1): Check type_domain.
15361 2017-08-07 Martin Liska <mliska@suse.cz>
15363 * attribs.h (canonicalize_attr_name): New function.
15364 (cmp_attribs): Move from c-format.c and adjusted.
15365 (is_attribute_p): Moved from tree.h.
15366 * tree-inline.c: Add new includes.
15367 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
15368 (private_is_attribute_p): Remove.
15369 (private_lookup_attribute): Likewise.
15370 (private_lookup_attribute_by_prefix): Simplify.
15371 (remove_attribute): Use is_attribute_p.
15372 * tree.h: Remove removed declarations.
15374 2017-08-07 Jakub Jelinek <jakub@redhat.com>
15376 PR middle-end/81698
15377 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
15378 instead of computing it in the function. Formatting fix.
15379 (expand_case): Don't rely on default_edge being the first edge,
15380 clear it if removing it, pass default_edge to
15381 emit_case_dispatch_table.
15382 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
15385 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
15387 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
15388 insn in the function, emit NOP after the insn.
15390 2017-08-06 Tom de Vries <tom@codesourcery.com>
15392 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
15395 2017-08-06 Tom de Vries <tom@codesourcery.com>
15397 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
15400 2017-08-04 Yury Gribov <tetra2005@gmail.com>
15402 PR tree-optimization/57371
15403 * match.pd: New pattern.
15405 2017-08-04 Marek Polacek <polacek@redhat.com>
15407 PR middle-end/81695
15408 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
15409 perform the computation in offset_int.
15411 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15413 PR tree-optimization/81136
15414 * tree-vectorizer.h: Include tree-hash-traits.h.
15415 (vec_base_alignments): New typedef.
15416 (vec_info): Add a base_alignments field.
15417 (vect_record_base_alignments): Declare.
15418 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15420 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15421 (create_data_ref): Add an is_conditional_in_stmt argument.
15422 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15423 the is_conditional_in_stmt field.
15424 (data_ref_loc): Add an is_conditional_in_stmt field.
15425 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15426 (find_data_references_in_stmt): Update call to create_data_ref.
15427 (graphite_find_data_references_in_stmt): Likewise.
15428 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15429 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15430 (vect_record_base_alignment): New function.
15431 (vect_record_base_alignments): Likewise.
15432 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15433 for nested statements even if we fail to compute a misalignment.
15434 Use pooled base alignments for unconditional references.
15435 (vect_find_same_alignment_drs): Compare base addresses instead
15437 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15438 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15440 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15442 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15443 Add an explicit name for the enum. Use auto_vec for slp_instances
15444 and grouped_stores.
15445 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15447 (_bb_vec_info): Add a constructor and destructor.
15448 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15449 (destroy_loop_vec_info): Delete.
15450 (vect_destroy_datarefs): Likewise.
15451 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15452 (vec_info::vec_info): New function.
15453 (vec_info::~vec_info): Likewise.
15454 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15455 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15456 destroy_loop_vec_info.
15457 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15458 (_loop_vec_info::_loop_vec_info): ...this.
15459 (destroy_loop_vec_info): Replace with...
15460 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15461 the stmt_vec_infos. Leave handling of vec_info information to its
15462 destructor. Remove explicit vector releases.
15463 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15464 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15465 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15466 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15467 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15468 (destroy_bb_vec_info): Replace with...
15469 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15470 information to its destructor.
15471 (vect_slp_analyze_bb_1): Use new and delete instead of
15472 new_bb_vec_info and destroy_bb_vec_info.
15473 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15476 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15478 * tree-data-ref.h (subscript): Add access_fn field.
15479 (data_dependence_relation): Add could_be_independent_p.
15480 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15481 (same_access_functions): Move to tree-data-ref.c.
15482 * tree-data-ref.c (ref_contains_union_access_p): New function.
15483 (access_fn_component_p): Likewise.
15484 (access_fn_components_comparable_p): Likewise.
15485 (dr_analyze_indices): Add a reference to access_fn_component_p.
15486 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15488 (constant_access_functions): Likewise.
15489 (add_other_self_distances): Likewise.
15490 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15491 (initialize_data_dependence_relation): Use XCNEW and remove
15492 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15493 of access functions that have the same type. Allow the
15494 subsequence to end with different bases in some circumstances.
15495 Record the chosen access functions in SUB_ACCESS_FN.
15496 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15497 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15498 (subscript_dependence_tester_1): Likewise dra and drb.
15499 (build_classic_dist_vector): Update calls accordingly.
15500 (subscript_dependence_tester): Likewise.
15501 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15502 DDR_COULD_BE_INDEPENDENT_P.
15503 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15504 comp_alias_ddrs instead of may_alias_ddrs.
15505 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15507 (vect_analyze_data_ref_dependence): Use it if
15508 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15509 distance vectors if that fails.
15510 (dependence_distance_ge_vf): New function.
15511 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15512 LOOP_VINFO_MAY_ALIAS_DDRS.
15514 2017-08-04 Richard Biener <rguenther@suse.de>
15516 PR middle-end/81705
15517 * fold-const.c (fold_binary_loc): Properly restrict
15518 minus_var0 && minus_var1 case when associating undefined overflow
15521 2017-08-04 Tom de Vries <tom@codesourcery.com>
15523 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15525 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15527 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15528 Don't start diagnostic messages with a capital letter.
15529 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15531 (rs6000_invalid_builtin): Likewise.
15532 (rs6000_trampoline_init): Likewise.
15534 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15537 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15538 after setting changeable df flags.
15540 2017-08-03 Richard Biener <rguenther@suse.de>
15542 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15543 up if the use is in USE - X.
15545 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15547 * toplev.c (dumpfile.h): New include.
15548 (internal_error_reentered): New static function. Use it...
15549 (internal_error_function): ...here to handle reentered internal_error.
15551 2017-08-03 Richard Biener <rguenther@suse.de>
15553 PR middle-end/81148
15554 * fold-const.c (split_tree): Add minus_var and minus_con
15555 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15556 here but always use minus_*.
15557 (associate_trees): Assert we never associate with MINUS_EXPR
15558 and NULL first operand. Do not recurse for PLUS_EXPR operands
15559 when associating as MINUS_EXPR either.
15560 (fold_binary_loc): Track minus_var and minus_con.
15562 2017-08-03 Tom de Vries <tom@codesourcery.com>
15565 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15566 ACCEL_COMPILER, apply finish_options on
15567 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15569 2017-08-03 Tom de Vries <tom@codesourcery.com>
15572 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15573 function_entry_patch_area_size > 0.
15575 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15578 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15579 instead of 10??LU, perform unit multiplication in wide_int,
15580 don't change alloc_object_size_limit if the limit is larger
15583 PR tree-optimization/81655
15584 PR tree-optimization/81588
15585 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15586 the case when ranges[i].low and high are 1 for unsigned type with
15589 PR middle-end/81052
15590 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15591 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15593 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15595 * tree-vrp.h: Add include guard.
15597 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15600 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15601 (ud2): New insn pattern.
15602 * config/i386/i386.c (ix86_expand_epilogue):
15603 For naked functions, generate ud2 instead of trap insn.
15605 2017-08-02 Marek Polacek <polacek@redhat.com>
15608 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15610 2017-08-02 Tom de Vries <tom@codesourcery.com>
15611 Cesar Philippidis <cesar@codesourcery.com>
15613 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15614 Add missing edge probabilities.
15616 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15618 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15619 Correct endianness.
15621 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15623 PR middle-end/79499
15624 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15625 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15626 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15627 prologue_seq sequences - if any.
15629 2017-08-02 Richard Biener <rguenther@suse.de>
15631 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15632 via vectors if supported, integer extracts via punning if supported
15633 or otherwise vector extracts.
15635 2017-08-02 Richard Biener <rguenther@suse.de>
15637 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15639 (bitmap_insert_into_set): ... this.
15641 2017-08-02 Richard Biener <rguenther@suse.de>
15643 PR tree-optimization/81633
15645 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15647 PR tree-optimization/71752
15648 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15650 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15652 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15653 (machine_function::call_ms2sysv_pad_out): Remove field.
15654 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15655 (ix86_compute_frame_layout): Likewise.
15657 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15660 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15661 attribute with interrupt attribute.
15663 2017-08-01 Andrew Pinski <apinski@cavium.com>
15665 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15666 BIT_INSERT_EXPR's operand 1
15667 to see if the types precision matches.
15669 2017-08-01 Martin Liska <mliska@suse.cz>
15671 PR middle-end/70140
15672 * builtins.c (expand_builtin_memcpy_args): Remove.
15673 (expand_builtin_memcpy): Call newly added function
15674 expand_builtin_memory_copy_args.
15675 (expand_builtin_memcpy_with_bounds): Likewise.
15676 (expand_builtin_mempcpy): Remove last argument.
15677 (expand_builtin_mempcpy_with_bounds): Likewise.
15678 (expand_builtin_memory_copy_args): New function created from
15679 expand_builtin_mempcpy_args with small modifications.
15680 (expand_builtin_mempcpy_args): Remove.
15681 (expand_builtin_stpcpy): Remove unused argument.
15682 (expand_builtin): Likewise.
15683 (expand_builtin_with_bounds): Likewise.
15685 2017-08-01 Martin Liska <mliska@suse.cz>
15688 Make mempcpy more optimal (PR middle-end/70140).
15690 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15693 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15694 __builtin_vec_cmpne verify both arguments are compatible vectors
15695 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15696 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15697 move computation of aligned to after checking the argument types.
15701 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15702 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15705 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15706 Doug Rupp <rupp@adacore.com>
15707 Olivier Hainque <hainque@adacore.com>
15709 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15710 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15712 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15713 for TARGET_OS_CPP_BUILTIN.
15714 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15715 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15717 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15718 passing required abi options to the assembler for EABI configurations.
15719 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15720 of .text.hot and .text.unlikely sections for kernel modules when
15721 using ARM style exceptions.
15722 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15723 options. Add EXTRA_CC1_SPEC.
15724 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15726 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15728 (ARM_TARGET2_DWARF_FORMAT): Define.
15729 * config/arm/t-vxworks: Adjust multilib control to removal of the
15730 Diab command line options.
15732 2017-08-01 Martin Liska <mliska@suse.cz>
15734 PR gcov-profile/81561
15735 * gcov.c (unblock): Make unblocking safe as we need to preserve
15736 index correspondence of blocks and block_lists.
15738 2017-08-01 Richard Biener <rguenther@suse.de>
15740 PR tree-optimization/81181
15741 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15742 (compute_antic): ... end of iteration here.
15744 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15746 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15747 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15748 (ftree-slp-vectorize): Likewise.
15749 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15750 can no longer be set independent of flag_tree_loop_vectorize.
15751 * omp-general.c (emp_max_vf): Likewise.
15752 * opts.c (enable_fdo_optimizations): Remove references to
15753 flag_tree_vectorize, these are now implicit.
15754 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15755 and leave it for the options machinery.
15757 2017-08-01 Martin Liska <mliska@suse.cz>
15759 PR middle-end/70140
15760 * builtins.c (expand_builtin_memcpy_args): Remove.
15761 (expand_builtin_memcpy): Call newly added function
15762 expand_builtin_memory_copy_args.
15763 (expand_builtin_memcpy_with_bounds): Likewise.
15764 (expand_builtin_mempcpy): Remove last argument.
15765 (expand_builtin_mempcpy_with_bounds): Likewise.
15766 (expand_builtin_memory_copy_args): New function created from
15767 expand_builtin_mempcpy_args with small modifications.
15768 (expand_builtin_mempcpy_args): Remove.
15769 (expand_builtin_stpcpy): Remove unused argument.
15770 (expand_builtin): Likewise.
15771 (expand_builtin_with_bounds): Likewise.
15773 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15776 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15777 print "ds:" only for immediates in generic address space.
15779 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15782 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15783 (ix86_function_ok_for_sibcall): Return false for naked functions.
15785 2017-08-01 Richard Biener <rguenther@suse.de>
15787 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15788 (compute_antic): Seed worklist with exit block predecessors.
15789 * cfganal.c (dfs_find_deadend): For a cycle return the source
15790 of the edge closing it.
15792 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15794 * config/aarch64/aarch64.c
15795 (aarch64_can_const_movi_rtx_p): Move 0 check.
15797 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15799 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15800 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15802 * match.pd: Ditto in address comparison pattern.
15804 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15806 PR tree-optimization/81627
15807 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15808 closed ssa form for store-store chain.
15810 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15812 PR tree-optimization/81620
15813 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15814 for store-store chain.
15816 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15818 PR tree-optimization/81588
15819 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15820 ranges[i].in_p, invert comparison code ccode. For >/>=,
15821 swap rhs1 and rhs2 and comparison code unconditionally,
15822 for </<= don't do that. Don't swap rhs1/rhs2 again if
15823 ranges[i].in_p, instead invert comparison code ccode if
15824 opcode or oe->rank is BIT_IOR_EXPR.
15827 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15828 a direct optab to conversion optab.
15829 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15830 with GET_MODE_INNER as last argument instead of optab_handler.
15831 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15832 vector extraction if possible and optab is available.
15833 * expr.c (store_constructor): Use convert_optab_handler instead
15834 of optab_handler. Use vector initialization from smaller
15835 vectors if possible and optab is available.
15836 * tree-vect-stmts.c (vectorizable_load): Likewise.
15837 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15838 now have two modes.
15839 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15840 of vec_init from half-sized vectors with the same element mode.
15841 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15842 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15843 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15844 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15845 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15846 after mode in gen_vec_extract* calls.
15847 (vec_extract<mode>): Renamed to ...
15848 (vec_extract<mode><ssescalarmodelower>): ... this.
15849 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15850 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15851 element mode after mode in gen_vec_init* calls.
15852 (VEC_INIT_HALF_MODE): New mode iterator.
15853 (vec_init<mode>): Renamed to ...
15854 (vec_init<mode><ssescalarmodelower>): ... this.
15855 (vec_init<mode><ssehalfvecmodelower>): New expander.
15856 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15857 (vec_extractv2sfsf): ... this.
15858 (vec_initv2sf): Renamed to ...
15859 (vec_initv2sfsf): ... this.
15860 (vec_extractv2si): Renamed to ...
15861 (vec_extractv2sisi): ... this.
15862 (vec_initv2si): Renamed to ...
15863 (vec_initv2sisi): ... this.
15864 (vec_extractv4hi): Renamed to ...
15865 (vec_extractv4hihi): ... this.
15866 (vec_initv4hi): Renamed to ...
15867 (vec_initv4hihi): ... this.
15868 (vec_extractv8qi): Renamed to ...
15869 (vec_extractv8qiqi): ... this.
15870 (vec_initv8qi): Renamed to ...
15871 (vec_initv8qiqi): ... this.
15872 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15873 (vec_init<mode>): Renamed to ...
15874 (vec_init<mode><VEC_base_l>): ... this.
15875 (vec_extract<mode>): Renamed to ...
15876 (vec_extract<mode><VEC_base_l>): ... this.
15877 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15878 (vec_initv2sfsf): ... this.
15879 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15880 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15881 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15882 element mode after mode in gen_vec_init* calls.
15883 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15884 (vec_init<mode><Vel>): ... this.
15885 (vec_extract<mode>): Renamed to ...
15886 (vec_extract<mode><Vel>): ... this.
15887 * config/aarch64/iterators.md (Vel): New mode attribute.
15888 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15889 Add element mode after mode in gen_vec_extract* calls.
15890 * config/s390/vector.md (non_vec_l): New mode attribute.
15891 (vec_extract<mode>): Renamed to ...
15892 (vec_extract<mode><non_vec_l>): ... this.
15893 (vec_init<mode>): Renamed to ...
15894 (vec_init<mode><non_vec_l>): ... this.
15895 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15896 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15898 * config/arm/iterators.md (V_elem_l): New mode attribute.
15899 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15900 (vec_extract<mode><V_elem_l>): ... this.
15901 (vec_extractv2di): Renamed to ...
15902 (vec_extractv2didi): ... this.
15903 (vec_init<mode>): Renamed to ...
15904 (vec_init<mode><V_elem_l>): ... this.
15905 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15906 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15907 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15908 Add element mode after gen_vec_extract* calls.
15909 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15910 (vec_init<mode><unitmode>): ... this.
15911 (vec_extract<mode>): Renamed to ...
15912 (vec_extract<mode><unitmode>): ... this.
15913 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15914 (vec_init<mode><unitmode>): ... this.
15915 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15916 (vec_initv2sfsf): ... this.
15917 (vec_extractv2sf): Renamed to ...
15918 (vec_extractv2sfsf): ... this.
15919 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15920 Add element mode after gen_vec_extract* calls.
15921 * config/mips/mips.md (unitmode): New mode iterator.
15922 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15923 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15924 * config/spu/spu.md (inner_l): New mode attribute.
15925 (vec_init<mode>): Renamed to ...
15926 (vec_init<mode><inner_l>): ... this.
15927 (vec_extract<mode>): Renamed to ...
15928 (vec_extract<mode><inner_l>): ... this.
15929 * config/sparc/sparc.md (veltmode): New mode iterator.
15930 (vec_init<VMALL:mode>): Renamed to ...
15931 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15932 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15933 (vec_initv2sisi): ... this.
15934 (vec_initv2sf): Renamed to ...
15935 (vec_initv2sfsf): ... this.
15936 (vec_extractv2sf): Renamed to ...
15937 (vec_extractv2sfsf): ... this.
15938 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15939 (vec_init<mode>): Renamed to ...
15940 (vec_init<mode><VEC_base_l>): ... this.
15941 (vec_extract<mode>): Renamed to ...
15942 (vec_extract<mode><VEC_base_l>): ... this.
15943 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15944 (vec_initv2sfsf): ... this.
15945 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15946 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15947 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15948 gen_vec_init* calls.
15950 2017-08-01 Richard Biener <rguenther@suse.de>
15952 PR tree-optimization/81297
15953 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15954 TREE_OVERFLOW from INTEGER_CSTs.
15956 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15958 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15960 2017-07-31 Carl Love <cel@us.ibm.com>
15962 * config/rs6000/rs6000-c: Add support for built-in functions
15963 vector signed char vec_xl_be (signed long long, signed char *);
15964 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15965 vector signed int vec_xl_be (signed long long, signed int *);
15966 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15967 vector signed long long vec_xl_be (signed long long, signed long long *);
15968 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15969 vector signed short vec_xl_be (signed long long, signed short *);
15970 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15971 vector double vec_xl_be (signed long long, double *);
15972 vector float vec_xl_be (signed long long, float *);
15973 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15974 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15975 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15977 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15978 (altivec_expand_builtin): Add switch statement to call
15979 altivec_expand_xl_be for each builtin.
15980 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15981 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15982 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15983 __builtin_vsx_le_be_v16qi.
15984 * doc/extend.texi: Update the built-in documentation file for the
15985 new built-in functions.
15987 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15990 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15992 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15994 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15996 * config.gcc: Add z14.
15997 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15998 CPU model numbers for z13s and z14.
15999 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
16001 * config/s390/s390-opts.h (enum processor_type): Rename
16002 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16003 * config/s390/s390.c (processor_table): Add field for CPU name to
16004 be passed to Binutils.
16005 (s390_asm_output_machine_for_arch): Use the new field in
16006 processor_table for Binutils.
16007 (s390_expand_builtin): Replace arch12 with z14.
16008 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
16009 (s390_get_sched_attrmask): Likewise.
16010 (s390_get_unit_mask): Likewise.
16011 * config/s390/s390.opt: Add z14 to processor_type enum.
16013 2017-07-31 Martin Jambor <mjambor@suse.cz>
16016 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
16017 regardless of optimization level.
16019 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
16020 Martin Liska <mliska@suse.cz>
16022 * predict.def: Remove old comment and adjust probability.
16023 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
16024 PREDICT statements.
16026 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
16029 * config/i386/i386.c (ix86_function_naked): New function.
16030 (ix86_can_use_return_insn_p): Return false for naked functions.
16031 (ix86_expand_prologue): Skip prologue for naked functions.
16032 (ix86_expand_epilogue): Skip epilogue for naked functions
16033 and emit trap instruction.
16034 (ix86_warn_func_return): New function.
16035 (ix86_attribute_table): Add "naked" attribute specification.
16036 (TARGET_WARN_FUNC_RETURN): Define.
16037 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
16039 2017-07-31 Martin Liska <mliska@suse.cz>
16041 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
16042 (dump_gimple_bb_header): Always dump BB info.
16043 (pp_cfg_jump): Do not append info about BB when dumping a jump.
16045 2017-07-31 Martin Liska <mliska@suse.cz>
16048 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
16049 also with current_function_decl non-null equality.
16051 2017-07-31 Jakub Jelinek <jakub@redhat.com>
16054 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
16055 change type to the element type, instead add eltype variable and
16056 use it where we are interested in the element type.
16058 PR tree-optimization/81603
16059 * ipa-polymorphic-call.c
16060 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
16061 offset arithmetic in offset_int, bail out if the resulting bit offset
16062 doesn't fit into shwi.
16064 2017-07-31 Martin Liska <mliska@suse.cz>
16066 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
16067 (gimplify_save_expr): Fix comment.
16069 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
16072 * config/i386/i386.c (ix86_function_arg): Update arguments for
16074 (ix86_compute_frame_layout): Set the initial stack offset to
16075 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
16076 INCOMING_FRAME_SP_OFFSET.
16077 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
16078 stack before exception handler returns.
16079 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
16080 the 'ERROR_CODE' for exception handler.
16082 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
16084 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
16085 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
16086 (ASM_OUTPUT_REG_POP): Ditto.
16087 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
16088 instead of asm_fprintf to output pure string.
16090 2017-07-29 Jakub Jelinek <jakub@redhat.com>
16092 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
16093 to imported_module_or_decl hook.
16094 (debug_nothing_tree_tree_tree_bool): Remove.
16095 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
16096 * debug.c (do_nothing_debug_hooks): Use
16097 debug_nothing_tree_tree_tree_bool_bool instead of
16098 debug_nothing_tree_tree_tree_bool.
16099 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
16100 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
16101 * sdbout.c (sdb_debug_hooks): Likewise.
16102 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
16103 (gen_namespace_die): Add DW_AT_export_symbols attribute if
16105 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
16106 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
16107 attribute, don't add anything.
16109 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16111 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
16112 (fold_build2_stat_loc): Likewise.
16113 (fold_build3_stat_loc): Likewise.
16114 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
16115 (fold_build1_loc): Remove macro.
16116 (fold_build2_loc): Likewise.
16117 (fold_build3_loc): Likewise.
16119 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16121 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
16122 (gimple_build_debug_bind_source_stat): Likewise.
16123 * gimple.h (gimple_build_debug_bind): Remove macro.
16124 (gimple_build_debug_bind_source): Likewise.
16126 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16128 * bitmap.c (bitmap_alloc): Adjust.
16129 (bitmap_gc_alloc): Likewise.
16130 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
16132 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16134 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
16135 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
16136 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
16137 (bitmap_gc_alloc_stat): Likewise.
16138 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
16140 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16142 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
16143 * rtl.h (shallow_copy_rtx): Remove macro.
16145 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16147 * emit-rtl.c (gen_raw_REG): Adjust.
16148 * gengenrtl.c (gendef): Likewise.
16149 * rtl.c (rtx_alloc_stat): Remove _stat from name.
16150 * rtl.h (rtx_alloc): Remove macro.
16152 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16154 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
16155 (build_tree_list_stat): Likewise.
16156 * tree.h (build_tree_list): Remove macro.
16157 (build_tree_list_vec): Likewise.
16159 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16161 * tree.c (make_vector_stat): Remove _stat from name.
16162 (build_vector_stat): Likewise.
16163 * tree.h (make_vector_stat): Remove macro.
16164 (build_vector_stat): Likewise.
16166 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16168 * tree.h (build_var_debug_value): Remove prototype.
16170 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16172 * tree.c (tree_cons_stat): Remove _stat from name.
16173 * tree.h (tree_cons): Remove macro.
16175 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16177 * tree.c (build_vl_exp_stat): Remove _stat from name.
16178 * tree.h (build_vl_exp): Remove macro.
16180 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16182 * tree.c (build_decl_stat): Remove _stat from name.
16183 * tree.h (build_decl): Remove macro.
16185 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16187 * gimple.c (gimple_build_with_ops_stat): Adjust.
16188 (gimple_alloc_stat): Remove _stat from name.
16189 * gimple.h (gimple_alloc): Remove macro.
16191 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16193 * tree.c (make_tree_vec_stat): Remove _stat from name.
16194 (grow_tree_vec_stat): Likewise.
16195 * tree.h (make_tree_vec_stat): Adjust prototype.
16196 (grow_tree_vec_stat): Likewise.
16197 (make_tree_vec): Remove macro.
16198 (grow_tree_vec): Likewise.
16200 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16202 * fold-const.c (fold_build1_stat_loc): Adjust.
16203 (fold_build2_stat_loc): Likewise.
16204 (fold_build3_stat_loc): Likewise.
16205 * tree.c (build0_stat): Remove _stat from name.
16206 (build1_stat): Likewise.
16207 (build2_stat): Likewise.
16208 (build3_stat): Likewise.
16209 (build4_stat): Likewise.
16210 (build5_stat): Likewise.
16211 * tree.h (build1_loc): Remove macro, and rename _stat function
16213 (build2_loc): Likewise.
16214 (build3_loc): Likewise.
16215 (build4_loc): Likewise.
16216 (build5_loc): Likewise.
16218 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16220 * tree.c (make_int_cst_stat): Remove _stat from name.
16221 * tree.h (make_int_cst_stat): Adjust prototype.
16222 (make_int_cst): Remove macro.
16224 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16226 * tree.c (make_tre_binfo_stat): Remove _stat from name.
16227 * tree.h (make_tree_binfo_stat): Adjust prototype.
16228 (make_tree_binfo): Remove.
16230 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16232 * tree.c (copy_node_stat): Rename to copy_node.
16233 (build_distinct_type_copy): Adjust.
16234 * tree.h (copy_node_stat): Adjust prototype.
16235 (copy_node): Remove macro.
16237 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
16239 * tree.c (make_node_stat): rename to make_node.
16240 (build_tree_list_stat): Adjust.
16241 (build0_stat): Likewise.
16242 (build2_stat): Likewise.
16243 (build3_stat): Likewise.
16244 (build4_stat): Likewise.
16245 (build5_stat): Likewise.
16246 (build_decl_stat): Likewise.
16247 * tree.h (make_node_stat): Adjust prototype.
16248 (make_node): remove macro.
16250 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
16252 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
16253 (PPC_FEATURE2_SCV): Likewise.
16254 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
16256 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16258 * config/aarch64/aarch64.c
16259 (aarch64_internal_mov_immediate): Add new special pattern.
16260 * config/aarch64/aarch64.md (*movdi_aarch64):
16261 Add reg/32bit const mov case.
16263 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16264 Richard Sandiford <richard.sandiford@linaro.org>
16266 * config/aarch64/aarch64.md (mov<mode>): Generalize.
16267 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
16268 Add integer and movi cases.
16269 (movi-split-hf-df-sf split, fp16): New.
16270 (enabled): Added TARGET_FP_F16INST.
16271 * config/aarch64/iterators.md (GPF_HF): New.
16272 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
16274 2017-07-28 Tamar Christina <tamar.christina@arm.com>
16276 * config/aarch64/aarch64.c
16277 (aarch64_simd_container_mode): Add prototype.
16278 (aarch64_expand_mov_immediate): Add HI support.
16279 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
16280 (aarch64_can_const_movi_rtx_p): New.
16281 (aarch64_preferred_reload_class):
16282 Remove restrictions of using FP registers for certain SIMD operations.
16283 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
16284 (aarch64_valid_floating_const): Add integer move validation.
16285 (aarch64_simd_imm_scalar_p): Remove.
16286 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
16287 (aarch64_legitimate_constant_p): Expand list of supported cases.
16288 * config/aarch64/aarch64-protos.h
16289 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
16290 (aarch64_reinterpret_float_as_int): New.
16291 (aarch64_simd_imm_scalar_p): Remove.
16292 * config/aarch64/constraints.md (Uvi): New.
16293 (Dd): Split into Ds and new Dd.
16294 * config/aarch64/aarch64.md (*movsi_aarch64):
16296 (*movdi_aarch64): Add SIMD mov case.
16298 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16300 * tree-predcom.c: (struct chain): Handle store-store chain in which
16301 stores for elimination only store loop invariant values.
16302 (execute_pred_commoning_chain): Ditto.
16303 (prepare_initializers_chain_store_elim): Ditto.
16304 (prepare_finalizers): Ditto.
16305 (is_inv_store_elimination_chain): New function.
16306 (initialize_root_vars_store_elim_1): New function.
16308 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16310 * tree-predcom.c: Revise general description of the pass.
16311 (enum chain_type): New enum type for store elimination.
16312 (struct chain): New field supporting store elimination.
16313 (struct component): Ditto.
16314 (dump_chain): Dump store-stores chain.
16315 (release_chain): Release resources.
16316 (split_data_refs_to_components): Compute and create component
16317 contains only stores for elimination.
16318 (get_chain_last_ref_at): New function.
16319 (make_invariant_chain): Initialization.
16320 (make_rooted_chain): Specify chain type in parameter and record it.
16321 (add_looparound_copies): Skip for store-stores chain.
16322 (determine_roots_comp): Compute type of chain and pass it to
16324 (initialize_root_vars_store_elim_2): New function.
16325 (finalize_eliminated_stores): New function.
16326 (remove_stmt): Handle store for elimination.
16327 (execute_pred_commoning_chain): Execute predictive commoning on
16328 store-store chains.
16329 (determine_unroll_factor): Skip unroll for store-stores chain.
16330 (prepare_initializers_chain_store_elim): New function.
16331 (prepare_initializers_chain): Hanlde store-store chain.
16332 (prepare_finalizers_chain, prepare_finalizers): New function.
16333 (tree_predictive_commoning_loop): Return integer value indicating
16334 if loop is unrolled or lcssa form is corrupted.
16335 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
16337 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16339 * tree-predcom.c (initialize_root): Delete.
16340 (execute_pred_commoning_chain): Initialize root vars and replace
16341 reference of non-combined chain directly, rather than call above
16344 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16346 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
16347 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
16349 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16351 * tree-predcom.c (struct chain): New field init_seq.
16352 (release_chain): Release init_seq.
16353 (prepare_initializers_chain): Record intialization stmts in above
16355 (insert_init_seqs): New function.
16356 (tree_predictive_commoning_loop): Call insert_init_seqs.
16358 2017-07-28 Bin Cheng <bin.cheng@arm.com>
16360 * tree-predcom.c (determine_roots_comp): Skip trivial components.
16362 2017-07-28 Richard Biener <rguenther@suse.de>
16364 * match.pd: Remove superfluous :c.
16365 * genmatch.c (simplify::id): Add member.
16366 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
16368 (current_id): New global.
16369 (dt_node::parent): Move from ...
16370 (dt_operand::parent): ... here. Add for_id member.
16371 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
16372 (decision_tree::find_node): Relax order requirement when
16373 merging DT_TRUE nodes to ones inbetween the current simplify
16374 and the one we try to merge with. Add diagnostic whenever
16375 we need to enforce pattern order by not merging.
16376 (decision_tree::insert): Set current_id.
16377 (decision_tree::print_node): Dump parent node and for_id.
16378 (parser::last_id): Add member.
16379 (parser::push_simplify): Assign unique id.
16380 (parser::parser): Initialize last_id.
16382 2017-07-28 Martin Liska <mliska@suse.cz>
16385 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
16386 gimple_build_debug_bind.
16388 2017-07-28 Richard Biener <rguenther@suse.de>
16390 PR tree-optimization/81502
16391 * match.pd: Add pattern combining BIT_INSERT_EXPR with
16393 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
16395 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
16396 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
16397 for BIT_FIELD_REF args.
16398 * fold-const.c (make_bit_field_ref): Likewise.
16399 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
16401 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16404 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
16405 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
16406 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
16407 Or it into SANITIZER_UNDEFINED.
16408 * ubsan.c: Include gimple-fold.h and varasm.h.
16409 (ubsan_expand_ptr_ifn): New function.
16410 (instrument_pointer_overflow): New function.
16411 (maybe_instrument_pointer_overflow): New function.
16412 (instrument_object_size): Formatting fix.
16413 (pass_ubsan::execute): Call instrument_pointer_overflow
16414 and maybe_instrument_pointer_overflow.
16415 * internal-fn.c (expand_UBSAN_PTR): New function.
16416 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
16417 * sanitizer.def (__ubsan_handle_pointer_overflow,
16418 __ubsan_handle_pointer_overflow_abort): New builtins.
16419 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16420 * internal-fn.def (UBSAN_PTR): New internal function.
16421 * opts.c (sanitizer_opts): Add pointer-overflow.
16422 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16423 * fold-const.c (build_range_check): Compute pointer range check in
16424 integral type if pointer arithmetics would be needed. Formatting
16427 2017-07-28 Martin Liska <mliska@suse.cz>
16430 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16431 parameters that are of a variable-length.
16433 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16435 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16437 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16438 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16439 (CRT_CALL_STATIC_FUNCTION): Likewise.
16440 (ASM_DEFAULT_SPEC): New define.
16441 (ASM_SPEC32): Likewise.
16442 (ASM_SPEC64): Likewise.
16443 (ASM_SPEC_COMMON): Likewise.
16444 (ASM_SPEC): Likewise.
16445 (INVALID_64BIT): Likewise.
16446 (LINK_OS_DEFAULT_SPEC): Likewise.
16447 (LINK_OS_SPEC32): Likewise.
16448 (LINK_OS_SPEC64): Likewise.
16449 (POWERPC_LINUX): Likewise.
16450 (PTRDIFF_TYPE): Likewise.
16451 (RESTORE_FP_PREFIX): Likewise.
16452 (RESTORE_FP_SUFFIX): Likewise.
16453 (SAVE_FP_PREFIX): Likewise.
16454 (SAVE_FP_SUFFIX): Likewise.
16455 (SIZE_TYPE): Likewise.
16456 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16457 (TARGET_64BIT): Likewise.
16458 (TARGET_64BIT): Likewise.
16459 (TARGET_AIX): Likewise.
16460 (WCHAR_TYPE_SIZE): Likewise.
16461 (WCHAR_TYPE): Undefine.
16462 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16463 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16464 (CPP_OS_RTEMS_SPEC): Delete.
16465 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16466 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16468 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16470 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16472 PR tree-optimization/81578
16473 * tree-parloops.c (build_new_reduction): Bail out if
16474 reduction_code isn't one of the standard OpenMP reductions.
16475 Move the details printing after that decision.
16477 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16479 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16480 related to reload_in_progress.
16481 (splat_input_operand): Likewise.
16482 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16484 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16486 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16487 (TARGET_INSTANTIATE_DECLS): Likewise.
16488 (legitimate_indexed_address_p): Delete reload_in_progress code.
16489 (rs6000_debug_legitimate_address_p): Likewise.
16490 (rs6000_eliminate_indexed_memrefs): Likewise.
16491 (rs6000_emit_le_vsx_store): Likewise.
16492 (rs6000_emit_move_si_sf_subreg): Likewise.
16493 (rs6000_emit_move): Likewise.
16494 (register_to_reg_type): Likewise.
16495 (rs6000_pre_atomic_barrier): Likewise.
16496 (rs6000_machopic_legitimize_pic_address): Likewise.
16497 (rs6000_allocate_stack_temp): Likewise.
16498 (rs6000_address_for_fpconvert): Likewise.
16499 (rs6000_address_for_altivec): Likewise.
16500 (rs6000_secondary_memory_needed_rtx): Delete function.
16501 (rs6000_check_sdmode): Likewise.
16502 (rs6000_alloc_sdmode_stack_slot): Likewise.
16503 (rs6000_instantiate_decls): Likewise.
16504 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16505 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16506 Delete reload_in_progress.
16507 (*vec_reload_and_plus_<mptrsize>): Likewise.
16508 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16509 (vsx_div_v2di): Likewise.
16510 (vsx_udiv_v2di): Likewise.
16512 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16514 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16515 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16516 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16517 (rs6000_debug_reg_global): Delete print of LRA status.
16518 (rs6000_option_override_internal): Delete dead LRA related code.
16519 (rs6000_lra_p): Delete function.
16520 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16522 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16524 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16525 * config/riscv/rtems.h: New file.
16527 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16528 Sudakshina Das <sudi.das@arm.com>
16530 * config/aarch64/aarch64.md
16531 (define_split for and<mode>3nr_compare): Move
16532 non aarch64_logical_operand to a register.
16533 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16534 register immediate operand to a register.
16535 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16537 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16539 PR middle-end/81564
16540 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16542 2017-07-27 Richard Biener <rguenther@suse.de>
16544 PR tree-optimization/81573
16545 PR tree-optimization/81494
16546 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16547 multi defuse cycle case.
16549 2017-07-27 Richard Biener <rguenther@suse.de>
16551 PR tree-optimization/81571
16552 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16555 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16557 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16558 earlier and only if MASK_FPU is set. Adjust formatting.
16560 2017-07-27 Martin Liska <mliska@suse.cz>
16562 * opt-functions.awk: Add validation of value of Init.
16563 * optc-gen.awk: Pass new argument.
16565 2017-07-27 Martin Liska <mliska@suse.cz>
16567 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16568 Fix wrong condition.
16570 2017-07-27 Martin Liska <mliska@suse.cz>
16572 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16573 BBs and edges seen by autoFDO.
16575 2017-07-27 Richard Biener <rguenther@suse.de>
16577 PR tree-optimization/81502
16578 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16579 with incompatible but same sized type.
16580 (execute_update_addresses_taken): Likewise.
16582 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16584 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16585 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16587 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16590 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16591 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16592 Change s_operand to memory_operand.
16594 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16596 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16597 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16598 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16599 Emit instructions rather than returning an expression. Handle TFmode
16600 and KFmode by casting to TImode.
16601 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16602 (rs6000_emit_le_vsx_store): Likewise.
16603 * config/rs6000/vsx.md (VSX_TI): New iterator.
16604 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16605 (*vsx_le_undo_permute_<mode>): Likewise.
16606 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16607 emit the split sequence.
16608 (*vsx_le_perm_store_<mode>): Likewise.
16610 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16612 PR tree-optimization/81555
16613 PR tree-optimization/81556
16614 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16615 if true, force CHANGED for the recursive invocation.
16616 (reassociate_bb): Remember original length of ops array, pass
16617 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16619 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16620 attributes for noipa attribute. For naked attribute use
16621 lookup_attribute first before lookup_attribute_spec.
16622 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16624 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16625 for functions with noipa attribute.
16626 (cgraph_externally_visible_p): Return true for functions with noipa
16628 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16629 for functions with noipa attribute.
16630 * doc/extend.texi: Document noipa function attribute.
16631 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16632 also for functions with noipa attribute.
16633 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16635 2017-07-26 Andrew Pinski <apinski@cavium.com>
16637 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16638 vec_unalign_load_cost and vec_unalign_store_cost.
16640 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16642 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16643 -mvsx-small-integer option.
16644 (ISA_3_0_MASKS_IEEE): Likewise.
16645 (OTHER_VSX_VECTOR_MASKS): Likewise.
16646 (POWERPC_MASKS): Likewise.
16647 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16648 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16649 code, only testing for DImode being allowed in non-VSX floating
16651 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16652 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16654 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16655 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16656 TARGET_P8_VECTOR test.
16657 (rs6000_secondary_reload_simple_move): Likewise.
16658 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16659 since TARGET_P9_VECTOR was already tested.
16660 (rs6000_opt_masks): Remove -mvsx-small-integer.
16661 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16662 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16664 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16665 test for TARGET_VEXTRACTUB was used, and that uses
16667 (p9 extract splitter): Likewise.
16668 (vsx_extract_<mode>_di_p9): Likewise.
16669 (vsx_extract_<mode>_store_p9): Likewise.
16670 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16671 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16672 the elimination of TARGET_VSX_SMALL_INTEGER.
16673 (vsx_extract_<mode>_p8): Likewise.
16674 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16675 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16676 (vsx_set_<mode>_p9): Likewise.
16677 (vsx_set_v4sf_p9): Likewise.
16678 (vsx_set_v4sf_p9_zero): Likewise.
16679 (vsx_insert_extract_v4sf_p9): Likewise.
16680 (vsx_insert_extract_v4sf_p9_2): Likewise.
16681 * config/rs6000/rs6000.md (sign extend splitter): Change
16682 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16683 (floatsi<mode>2_lfiwax_mem): Likewise.
16684 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16685 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16686 since a test for TARGET_P9_VECTOR was used.
16687 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16688 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16689 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16690 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16691 TARGET_P8_VECTOR test.
16692 (fix_trunc<mode>si2_stfiwx): Likewise.
16693 (fix_trunc<mode>si2_internal): Likewise.
16694 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16695 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16697 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16698 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16699 TARGET_P8_VECTOR test.
16700 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16701 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16702 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16704 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16705 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16706 since a test for TARGET_P9_VECTOR was used.
16707 (splitter for loading small constants): Likewise.
16709 2017-07-26 Andrew Pinski <apinski@cavium.com>
16711 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16714 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16717 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16718 (fp_valid_at): Likewise.
16720 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16722 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16723 (qdf24xx_addrcost_table): Likewise.
16724 (cortexa57_tunings): Update to use generic_branch_cost.
16725 (cortexa72_tunings): Likewise.
16726 (cortexa73_tunings): Likewise.
16727 (qdf24xx_tunings): Likewise.
16729 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16731 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16732 (thunderx2t99_branch_cost): Likewise.
16733 (cortexa35_tunings): Update to use generic_branch_cost.
16734 (cortexa53_tunings): Likewise.
16735 (cortexa57_tunings): Likewise.
16736 (cortexa72_tunings): Likewise.
16737 (cortexa73_tunings): Likewise.
16738 (thunderx2t99_tunings): Likewise.
16740 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16742 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16743 (sparc_option_override): Honour MASK_FSMULD.
16744 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16745 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16746 * config/sparc/sparc.opt (mfsmuld): New option.
16747 * doc/invoke.texi (mfsmuld): Document option.
16749 2017-07-26 Marek Polacek <polacek@redhat.com>
16751 PR middle-end/70992
16752 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16754 2017-07-26 Richard Biener <rguenther@suse.de>
16756 * gimple-match-head.c (do_valueize): Return OP if valueize
16758 (get_def): New helper to get at the def stmt of a SSA name
16759 if valueize allows.
16760 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16761 do_valueize to get at the def stmt.
16762 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16764 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16766 PR middle-end/46932
16767 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16769 2017-07-26 Martin Liska <mliska@suse.cz>
16772 * function.c (expand_function_start): Make expansion of
16773 nonlocal_goto_save_area after parm_birth_insn.
16775 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16777 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16778 from all CPU target flags enable members.
16780 2017-07-26 Richard Biener <rguenther@suse.de>
16782 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16783 (decision_tree::gen): Make 'type' const.
16784 (write_predicate): Likewise.
16786 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16788 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16789 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16790 (rs6000_option_override_internal): Likewise.
16791 (rs6000_expand_vector_set): Likewise.
16792 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16793 (TARGET_UPPER_REGS_SF): Likewise.
16794 (TARGET_UPPER_REGS_DI): Likewise.
16795 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16796 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16797 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16798 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16799 (Splitters for DI constants in Altivec registers): Likewise.
16800 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16801 (vsx_set_v4sf_p9): Likewise.
16802 (vsx_set_v4sf_p9_zero): Likewise.
16803 (vsx_insert_extract_v4sf_p9): Likewise.
16804 (vsx_insert_extract_v4sf_p9_2): Likewise.
16806 2017-07-25 Carl Love <cel@us.ibm.com>
16808 * doc/extend.texi: Update the built-in documentation file for the
16809 existing built-in functions
16810 vector signed char vec_cnttz (vector signed char);
16811 vector unsigned char vec_cnttz (vector unsigned char);
16812 vector signed short vec_cnttz (vector signed short);
16813 vector unsigned short vec_cnttz (vector unsigned short);
16814 vector signed int vec_cnttz (vector signed int);
16815 vector unsigned int vec_cnttz (vector unsigned int);
16816 vector signed long long vec_cnttz (vector signed long long);
16817 vector unsigned long long vec_cnttz (vector unsigned long long);
16819 2017-07-25 Andrew Pinski <apinski@cavium.com>
16821 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16822 accesses where the use is for the first operand of a BIT_INSERT.
16824 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16827 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16828 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16830 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16832 * config/i386/gstabs.h: Delete.
16833 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16835 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16837 * config/i386/i386.c (ix86_decompose_address): Do not check for
16838 register RTX when looking at index_reg or base_reg.
16839 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16841 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16843 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16844 to update EH info here.
16846 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16848 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16850 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16852 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16854 2017-07-25 Torsten Duwe <duwe@suse.de>
16856 * common.opt: Introduce -fpatchable-function-entry
16857 command line option, and its variables function_entry_patch_area_size
16858 and function_entry_patch_area_start.
16859 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16860 including a two-value parser.
16861 * target.def (print_patchable_function_entry): New target hook.
16862 * targhooks.h (default_print_patchable_function_entry): New function.
16863 * targhooks.c (default_print_patchable_function_entry): Likewise.
16864 * toplev.c (process_options): Switch off IPA-RA if
16865 patchable function entries are being generated.
16866 * varasm.c (assemble_start_function): Look at the
16867 patchable-function-entry command line switch and current
16868 function attributes and maybe generate NOP instructions by
16869 calling the print_patchable_function_entry hook.
16870 * doc/extend.texi: Document patchable_function_entry attribute.
16871 * doc/invoke.texi: Document -fpatchable_function_entry
16872 command line option.
16873 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16875 * doc/tm.texi: Re-generate.
16877 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16880 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16881 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16883 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16885 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16887 2017-07-25 Richard Biener <rguenther@suse.de>
16889 PR tree-optimization/81455
16890 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16891 not walk in cycles when looking for guards.
16893 2017-07-25 Richard Biener <rguenther@suse.de>
16895 PR tree-optimization/81529
16896 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16897 when optimizing backedge uses.
16899 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16901 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16903 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16904 to dl_section_ref. On AIX, append an expression to subtract
16905 the size of the section length to dl_section_ref.
16907 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16909 * configure.ac: If any of the config.* scripts fail, exit 1.
16910 * configure: Regenerate.
16912 2017-07-25 Richard Biener <rguenther@suse.de>
16914 PR middle-end/81546
16915 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16916 of immediate uses, be more verbose on errors.
16918 2017-07-25 Richard Biener <rguenther@suse.de>
16920 PR tree-optimization/81510
16921 * tree-vect-loop.c (vect_is_simple_reduction): When the
16922 reduction stmt is not inside the loop bail out.
16924 2017-07-25 Richard Biener <rguenther@suse.de>
16926 PR tree-optimization/81303
16927 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16928 profitability check against LOOP_VINFO_NITERSM1.
16930 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16932 * domwalk.c (cmp_bb_postorder): Simplify.
16933 (sort_bbs_postorder): New function. Use it...
16934 (dom_walker::walk): ...here to optimize common cases.
16936 2017-07-25 Martin Liska <mliska@suse.cz>
16939 * ipa-visibility.c (function_and_variable_visibility): Make the
16940 redirection just on target that supports aliasing.
16941 Fix GNU coding style.
16943 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16946 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16948 * config/arm/rtems.h: Likewise.
16949 * config/bfin/rtems.h: Likewise.
16950 * config/i386/rtemself.h: Likewise.
16951 * config/lm32/rtems.h: Likewise.
16952 * config/m32c/rtems.h: Likewise.
16953 * config/m68k/rtemself.h: Likewise.
16954 * config/microblaze/rtems.h: Likewise.
16955 * config/mips/rtems.h: Likewise.
16956 * config/moxie/rtems.h: Likewise.
16957 * config/nios2/rtems.h: Likewise.
16958 * config/powerpcspe/rtems.h: Likewise.
16959 * config/rs6000/rtems.h: Likewise.
16960 * config/rtems.h: Likewise.
16961 * config/sh/rtems.h: Likewise.
16962 * config/sh/rtemself.h: Likewise.
16963 * config/sparc/rtemself.h: Likewise.
16965 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16968 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16969 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16970 * tree-ssa-structalias.c (alias_get_name): Same.
16972 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16975 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16976 instructions if no du chain is found.
16978 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16980 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16982 2017-07-25 Richard Biener <rguenther@suse.de>
16984 PR middle-end/81505
16985 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16988 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16990 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16991 upper-regs options.
16992 (ISA_2_7_MASKS_SERVER): Likewise.
16993 (ISA_3_0_MASKS_IEEE): Likewise.
16994 (OTHER_P8_VECTOR_MASKS): Likewise.
16995 (OTHER_VSX_VECTOR_MASKS): Likewise.
16996 (POWERPC_MASKS): Likewise.
16997 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16998 duplicate list of options.
16999 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
17000 explicit -mupper-regs options.
17001 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
17002 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
17003 alias for -mupper-regs-df.
17004 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17005 (rs6000_init_hard_regno_mode_ok): Likewise.
17006 (rs6000_option_override_internal): Likewise.
17007 (rs6000_opt_masks): Likewise.
17008 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
17009 options in terms of whether -mvsx or -mpower8-vector was used.
17010 (TARGET_UPPER_REGS_DI): Likewise.
17011 (TARGET_UPPER_REGS_SF): Likewise.
17012 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
17013 -mupper-regs-* options.
17015 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17017 * passes.c (emergency_dump_function): Print some empty lines and a
17018 header before the RTL dump.
17020 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
17022 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
17024 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
17027 * config/aarch64/aarch64.c (aarch64_classify_symbol):
17028 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
17030 2017-07-24 Carl Love <cel@us.ibm.com>
17032 * config/rs6000/rs6000-c.c: Add support for built-in functions
17033 vector float vec_extract_fp32_from_shorth (vector unsigned short);
17034 vector float vec_extract_fp32_from_shortl (vector unsigned short);
17035 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
17036 vec_extract_fp_from_shortl): Add defines for the two builtins.
17037 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
17038 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
17040 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
17041 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
17042 * doc/extend.texi: Update the built-in documentation file for the
17043 new built-in function.
17045 2017-07-24 Jakub Jelinek <jakub@redhat.com>
17048 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
17050 * doc/generic.texi: Likewise.
17051 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
17052 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
17054 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
17056 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
17057 (aarch64_mls_elt_merge<mode>): Likewise.
17059 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
17061 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17062 having __cxa_atexit.
17064 2017-07-23 Michael Collison <michael.collison@arm.com>
17066 * config/arm/arm.c (arm_option_override): Deprecate
17067 use of -mstructure-size-boundary.
17068 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
17069 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
17071 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17074 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
17075 Reduce cost estimate for direct moves.
17077 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
17080 * config/i386/i386.c (ix86_option_override_internal): Disable
17081 BMI, BMI2 and TBM instructions for -m16.
17083 2017-07-21 Carl Love <cel@us.ibm.com>
17085 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17086 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17087 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17088 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17089 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17090 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17091 VMULOSW): New enum "unspec" values.
17092 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17093 altivec_vmulosw): New patterns.
17094 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17095 VMULOSW): Add definitions.
17097 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
17099 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
17100 (qdf24xx): Likewise.
17101 * config/aarch64/aarch64-options-extensions.def (rdma); New.
17102 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
17103 (AARCH64_FL_V8_1): Renumber.
17104 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
17105 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
17106 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
17107 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
17108 rdma to feature modifiers list.
17110 2017-07-21 Yury Gribov <tetra2005@gmail.com>
17112 PR middle-end/56727
17113 * ipa-visibility (function_and_variable_visibility): Convert
17114 recursive PLT call to direct call if appropriate.
17116 2017-07-21 Andrew Pinski <apinski@cavium.com>
17118 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
17119 operand 1 to see if the types precision matches.
17120 * fold-const.c (operand_equal_p): Likewise.
17122 2017-07-21 Richard Biener <rguenther@suse.de>
17124 PR tree-optimization/81303
17125 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
17126 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
17127 (vect_peeling_hash_get_lowest_cost): Adjust.
17128 (vect_enhance_data_refs_alignment): Likewise. Use
17129 vect_get_peeling_costs_all_drs to compute the penalty for no
17130 peeling to match up costs.
17132 2017-07-21 Richard Biener <rguenther@suse.de>
17134 PR tree-optimization/81500
17135 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
17136 we didn't identify a reduction path.
17138 2017-07-21 Tom de Vries <tom@codesourcery.com>
17139 Cesar Philippidis <cesar@codesourcery.com>
17141 PR gcov-profile/81442
17142 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
17145 2017-07-21 Tom de Vries <tom@codesourcery.com>
17148 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
17150 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
17151 nvptx_override_options_after_change.
17153 2017-07-21 Ulrich Drepper <drepper@redhat.com>
17155 * dwarf2out.c (output_file_names): Avoid double testing for
17156 dwarf_version >= 5.
17158 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
17160 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
17162 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17164 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
17166 (try_crossjump_to_edge): Do not punt on partitioned functions.
17168 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
17170 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
17171 Put all BBs reachable only via paths crossing cold region to cold
17173 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
17175 2016-07-21 Richard Biener <rguenther@suse.de>
17177 PR tree-optimization/81303
17178 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
17179 into account prologue and epilogue iterations when raising
17180 min_profitable_iters to sth at least covering one vector iteration.
17182 2017-07-21 Tamar Christina <tamar.christina@arm.com>
17184 * config/arm/arm.c (arm_test_cpu_arch_dat):
17187 2017-07-20 Nathan Sidwell <nathan@acm.org>
17189 Remove TYPE_METHODS.
17190 * tree.h (TYPE_METHODS): Delete.
17191 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
17192 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
17193 (dbxout_type_methods): Scan TYPE_FIELDS.
17194 (dbxout_type): Don't check TYPE_METHODS here.
17195 * function.c (use_register_for_decl): Always ignore register for
17196 class types when not optimizing.
17197 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
17198 * tree.c (free_lang_data_in_type): Stitch out member functions and
17199 templates from TYPE_FIELDS.
17200 (build_distinct_type_copy, verify_type_variant,
17201 verify_type): Member fns are on TYPE_FIELDS.
17202 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
17203 * tree-pretty-print.c (dump_generic_node): Likewise.
17205 2017-07-20 Jakub Jelinek <jakub@redhat.com>
17208 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
17209 V2TImode and V4TImode.
17210 (ix86_expand_vector_extract): Likewise.
17211 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
17212 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
17213 (ssescalarmode): Handle V4TImode and V2TImode.
17214 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
17215 (*vec_extractv2ti, *vec_extractv4ti): New insns.
17216 (VEXTRACTI128_MODE): New mode iterator.
17217 (splitter for *vec_extractv?ti first element): New.
17218 (VEC_INIT_MODE): New mode iterator.
17219 (vec_init<mode>): Consolidate 3 expanders into one using
17220 VEC_INIT_MODE mode iterator.
17222 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17224 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
17225 non_spilled_static_chain_regno_p.
17227 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
17229 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
17231 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
17233 * bb-reorder.c (connect_traces): Allow copying of blocks within
17236 2017-07-20 Richard Biener <rguenther@suse.de>
17238 * gimple.h (gimple_phi_result): Add gphi * overload.
17239 (gimple_phi_result_ptr): Likewise.
17240 (gimple_phi_arg): Likewise. Adjust index assert to only
17241 allow actual argument accesses rather than all slots available
17243 (gimple_phi_arg_def): Add gphi * overload.
17244 * tree-phinodes.c (make_phi_node): Initialize only actual
17246 (resize_phi_node): Clear memory not covered by old node,
17247 do not initialize excess argument slots.
17248 (reserve_phi_args_for_new_edge): Initialize new argument slot
17251 2017-07-20 Bin Cheng <bin.cheng@arm.com>
17253 PR tree-optimization/81388
17255 2016-07-21 Bin Cheng <bin.cheng@arm.com>
17257 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
17258 by removing computation of may_be_zero.
17260 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17261 Tom de Vries <tom@codesourcery.com>
17263 PR middle-end/81030
17264 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
17265 when gimple level profile disagrees with what RTL expander did.
17267 2017-07-20 Richard Biener <rguenther@suse.de>
17269 PR tree-optimization/61171
17270 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
17271 (vect_analyze_stmt): Add slp instance parameter.
17272 (vectorizable_reduction): Likewise.
17273 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
17274 (vect_is_simple_reduction): Deal with chains not detected
17275 as SLP reduction chain, specifically not properly associated
17276 chains containing a mix of plus/minus.
17277 (get_reduction_op): Remove.
17278 (get_initial_defs_for_reduction): Simplify, pass in whether
17279 this is a reduction chain, pass in the SLP node for the PHIs.
17280 (vect_create_epilog_for_reduction): Get the SLP instance as
17282 (vectorizable_reduction): Get the SLP instance as arg.
17283 During analysis remember the SLP node with the PHIs in the
17284 instance. Simplify getting at the vectorized reduction PHIs.
17285 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
17286 through SLP instance.
17287 (vect_slp_analyze_operations): Likewise.
17288 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
17289 (vect_transform_stmt): Likewise.
17291 2017-07-20 Tom de Vries <tom@codesourcery.com>
17293 PR tree-optimization/81489
17294 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
17295 read of phi arg location to before loop that modifies phi.
17297 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17299 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
17302 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17304 PR middle-end/81331
17305 * except.c (execute): Fix ordering issue.
17307 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17309 PR rtl-optimization/81423
17310 * combine.c (make_compound_operation_int): Don't try to optimize
17311 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
17313 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
17315 PR rtl-optimization/81423
17316 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
17317 with a constant that is -1 in the truncated to mode.
17319 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17321 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
17322 (determine_unlikely_bbs): ... here.
17323 * predict.h (propagate_unlikely_bbs_forward): Declare.
17324 * cfgexpand.c (pass_expand::execute): Use it.
17325 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
17327 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
17328 propagate_unlikely_bbs_forward.
17330 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
17332 PR middle-end/81331
17333 * except.c (maybe_add_nop_after_section_switch): New function.
17336 2017-07-19 Tom de Vries <tom@codesourcery.com>
17338 * gimple.h (gimple_phi_set_arg): Make assert more strict.
17340 2017-07-19 Tom de Vries <tom@codesourcery.com>
17342 * gimple.h (gimple_phi_arg): Make assert more strict.
17344 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
17346 * config.gcc (powerpc*-*-*): Add mmintrin.h.
17347 * config/rs6000/mmintrin.h: New file.
17348 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
17350 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17352 PR tree-optimization/81346
17353 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
17355 2017-07-19 Tom de Vries <tom@codesourcery.com>
17357 * config/nvptx/nvptx.md (VECIM): Add V2DI.
17359 2017-07-19 Tom de Vries <tom@codesourcery.com>
17361 * config/nvptx/nvptx-modes.def: Add V2DImode.
17362 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
17363 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
17364 (nvptx_output_mov_insn): Handle lack of mov.b128.
17365 (nvptx_print_operand): Handle 'H' and 'L' codes.
17366 (nvptx_vector_mode_supported): Allow V2DImode.
17367 (nvptx_preferred_simd_mode): New function.
17368 (nvptx_data_alignment): New function.
17369 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
17370 nvptx_preferred_simd_mode.
17371 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
17373 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
17375 2017-07-19 Tom de Vries <tom@codesourcery.com>
17377 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
17378 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
17379 (nvptx_vector_mode_supported): New function. Allow V2SImode.
17380 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
17381 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
17382 (mov<VECIM>_insn): New define_insn.
17383 (define_expand "mov<VECIM>): New define_expand.
17385 2017-07-19 Tom de Vries <tom@codesourcery.com>
17387 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
17389 2017-07-19 Jakub Jelinek <jakub@redhat.com>
17391 PR tree-optimization/81346
17392 * fold-const.h (fold_div_compare, range_check_type): Declare.
17393 * fold-const.c (range_check_type): New function.
17394 (build_range_check): Use range_check_type.
17395 (fold_div_compare): No longer static, rewritten into
17396 a match.pd helper function.
17397 (fold_comparison): Don't call fold_div_compare here.
17398 * match.pd (X / C1 op C2): New optimization using fold_div_compare
17399 as helper function.
17401 2017-07-19 Nathan Sidwell <nathan@acm.org>
17403 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
17404 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
17405 * tree.c (find_decls_types_r, verify_type): Use
17406 TYPE_{MIN,MAX}_VALUE_RAW.
17407 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
17408 (hash_tree): Likewise.
17409 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
17411 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
17414 2017-07-18 Tom de Vries <tom@codesourcery.com>
17416 PR middle-end/81464
17417 * omp-expand.c (expand_omp_for_static_chunk): Handle
17418 equal-argument loop exit phi.
17420 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17423 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17424 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17425 operand 2 predicate.
17426 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17427 operand 2 predicate.
17428 (ror,rol -> rorx splitters): Use const_int_operand as
17429 operand 2 predicate.
17431 2017-06-18 Richard Biener <rguenther@suse.de>
17433 PR tree-optimization/81410
17434 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17435 the gap in the ! slp_perm SLP case after each group.
17437 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17439 PR middle-end/81463
17440 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17443 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17445 PR middle-end/81462
17446 * predict.c (set_even_probabilities): Cleanup; do not affect
17447 probabilities that are already known.
17448 (combine_predictions_for_bb): Call even when count is set.
17450 2017-07-18 Nathan Sidwell <nathan@acm.org>
17452 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17455 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17458 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17459 optimization for loop niter analysis.
17461 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17464 * config/avr/avr.c (avr_optimize_casesi): Don't use
17465 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17467 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17469 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17470 body_cost_vec from _vect_peel_extended_info.
17471 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17472 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17475 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17477 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17479 2017-07-18 Richard Biener <rguenther@suse.de>
17481 PR tree-optimization/80620
17482 PR tree-optimization/81403
17483 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17484 info when re-using a VN table entry.
17486 2017-07-18 Richard Biener <rguenther@suse.de>
17488 PR tree-optimization/81418
17489 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17490 vectype_in. Verify that with lane-reducing reduction operations
17491 we have a single def-use cycle.
17493 2017-07-17 Carl Love <cel@us.ibm.com>
17495 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17497 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17498 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17499 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17500 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17501 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17502 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17503 VMULOSW): New enum "unspec" values.
17504 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17505 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17506 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17507 altivec_vmulosw): New patterns.
17508 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17509 VMULOSW): Add definitions.
17510 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17512 * config/alpha/alpha.c: Include predict.h.
17514 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17516 * tree-vrp.c (compare_assert_loc): Fix comparison function
17517 to return predictable results.
17519 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17521 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17523 (subdi3): Likewise.
17524 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17525 * doc/invoke.texi (mexpand-adddi): Update text.
17527 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17529 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17530 that also clobbers the CC register. The old expand code is moved
17532 (*arc_clzsi2): ... here.
17533 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17534 the CC register. The old expand code is moved to ...
17535 (arc_ctzsi2): ... here.
17537 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17539 * config/arc/arc.opt (mindexed-loads): Use initial value
17540 TARGET_INDEXED_LOADS_DEFAULT.
17541 (mauto-modify-reg): Use initial value
17542 TARGET_AUTO_MODIFY_REG_DEFAULT.
17543 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17544 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17545 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17546 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17548 2017-07-17 Martin Liska <mliska@suse.cz>
17551 * opts.c (finish_options): Do not allow -fgnu-tm
17552 w/ -fsanitize={kernel-,}address. Say sorry.
17554 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17557 * tree-loop-distribution.c (classify_partition): Only assert on
17558 numer of iterations.
17559 (merge_dep_scc_partitions): Delete prameter. Update function call.
17560 (distribute_loop): Remove code handling loop with unknown niters.
17561 (pass_loop_distribution::execute): Skip loop with unknown niters.
17563 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17566 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17567 function sort_partitions_by_post_order.
17569 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17571 PR tree-optimization/81374
17572 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17573 the max index of basic blocks, rather than number of basic blocks.
17575 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17577 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17579 (arc_legitimate_pic_operand_p): Likewise.
17580 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17582 (arc_needs_pcl_p): Likewise.
17583 (arc_legitimate_pc_offset_p): Likewise.
17584 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17585 function is also used in constrains.md.
17586 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17587 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17588 PLUS. Only return true/false in known cases, otherwise assert.
17589 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17590 is already called in arc_legitimate_constant_p.
17591 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17593 (LEGITIMATE_PIC_OPERAND_P): Use
17594 arc_raw_symbolic_reference_mentioned_p function.
17595 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17600 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17601 Andrew Burgess <andrew.burgess@embecosm.com>
17603 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17604 (arc_return_address_register): New function.
17605 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17606 (arc_handle_fndecl_attribute): Add naked attribute.
17607 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17608 (TARGET_WARN_FUNC_RETURN): Likewise.
17609 (arc_allocate_stack_slots_for_args): New function.
17610 (arc_warn_func_return): Likewise.
17611 (machine_function): Change type fn_type.
17612 (arc_compute_function_type): Consider new naked function type,
17613 change function return type.
17614 (arc_must_save_register): Adapt to handle new
17615 arc_compute_function_type's return type.
17616 (arc_expand_prologue): Likewise.
17617 (arc_expand_epilogue): Likewise.
17618 (arc_return_address_regs): Delete.
17619 (arc_return_address_register): New function.
17620 (arc_epilogue_uses): Use above function.
17621 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17622 (arc_function_type): Change encoding, add naked type.
17623 (ARC_INTERRUPT_P): Change to handle the new encoding.
17624 (ARC_FAST_INTERRUPT_P): Likewise.
17625 (ARC_NORMAL_P): Define.
17626 (ARC_NAKED_P): Likewise.
17627 (arc_compute_function_type): Delete prototype.
17628 * config/arc/arc.md (in_ret_delay_slot): Use
17629 arc_return_address_register function.
17630 (simple_return): Likewise.
17631 (p_return_i): Likewise.
17633 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17635 PR tree-optimization/81428
17636 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17637 can't be built for those types.
17639 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17641 Remove stuff dead since r239246.
17643 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17644 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17645 (avr_inform_devices): Remove dead stuff.
17647 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17649 * config/arm/arm_neon.h: Fix softp typo.
17651 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17653 PR tree-optimization/81365
17654 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17655 aggregate moves onto bb predecessor edges, make sure there are no
17656 loads that could alias the lhs in between the start of bb and the
17659 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17662 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17663 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17664 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17666 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17668 PR tree-optimization/81396
17669 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17670 (init_symbolic_number): Initialize it to 1.
17671 (perform_symbolic_merge): Add n_ops from both operands into the new
17673 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17674 without base_addr as useless if they need more than one operation.
17675 (bswap_replace): Handle !bswap case for NULL base_addr.
17677 2017-07-17 Tom de Vries <tom@codesourcery.com>
17680 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17683 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17685 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17686 conditional builtin define __FIX_LEON3FT_B2BST.
17688 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17690 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17691 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17694 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17696 PR rtl-optimization/81424
17697 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17698 to remove potential trapping from operands if -fnon-call-exceptions.
17700 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17702 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17703 profile_proability for scalling.
17704 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17706 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17708 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17710 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17712 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17713 fixpoint arithmetics.
17715 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17717 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17718 fixpoint arithmetics.
17720 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17722 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17723 fixpoint arithmetics.
17725 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17727 * profile-count.h (profile_probability::from_reg_br_prob_note,
17728 profile_probability::to_reg_br_prob_note): New functions.
17729 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17730 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17731 * predict.c (probability_reliable_p): Update.
17732 (edge_probability_reliable_p): Update.
17733 (br_prob_note_reliable_p): Update.
17734 (invert_br_probabilities): Update.
17735 (add_reg_br_prob_note): New function.
17736 (combine_predictions_for_insn): Update.
17737 * asan.c (asan_clear_shadow): Update.
17738 * cfgbuild.c (compute_outgoing_frequencies): Update.
17739 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17740 (update_br_prob_note): Update.
17741 (rtl_verify_edges): Update.
17742 (purge_dead_edges): Update.
17743 (fixup_reorder_chain): Update.
17744 * emit-rtl.c (try_split): Update.
17745 * ifcvt.c (cond_exec_process_insns): Update.
17746 (cond_exec_process_if_block): Update.
17747 (dead_or_predicable): Update.
17748 * internal-fn.c (expand_addsub_overflow): Update.
17749 (expand_neg_overflow): Update.
17750 (expand_mul_overflow): Update.
17751 * loop-doloop.c (doloop_modify): Update.
17752 * loop-unroll.c (compare_and_jump_seq): Update.
17753 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17754 * predict.h: Update.
17755 * reorg.c (mostly_true_jump): Update.
17757 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17758 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17759 * config/arc/arc.c: (emit_unlikely_jump): Update.
17760 * config/arm/arm.c: (emit_unlikely_jump): Update.
17761 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17762 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17763 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17764 (ix86_print_operand): Update.
17765 (ix86_split_fp_branch): Update.
17766 (predict_jump): Update.
17767 * config/ia64/ia64.c (ia64_print_operand): Update.
17768 * config/mmix/mmix.c (mmix_print_operand): Update.
17769 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17770 (rs6000_expand_split_stack_prologue): Update.
17771 * config/rs6000/rs6000.c: Update.
17772 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17773 (s390_expand_vec_movstr): Update.
17774 (s390_expand_cs_tdsi): Update.
17775 (s390_expand_split_stack_prologue): Update.
17776 * config/sh/sh.c (sh_print_operand): Update.
17777 (expand_cbranchsi4): Update.
17778 (expand_cbranchdi4): Update.
17779 * config/sparc/sparc.c (output_v9branch): Update.
17780 * config/spu/spu.c (get_branch_target): Update.
17781 (ea_load_store_inline): Update.
17782 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17783 * config/tilepro/tilepro.c: Update.
17785 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17787 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17788 (gimplify_save_expr): Likewise.
17790 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17792 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17794 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17796 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17798 * timevar.def (TV_IPA_FNSUMMARY): Define.
17800 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17802 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17803 to back store errata sensitive sequence from being generated.
17804 (sqrtdf2_fix): Likewise.
17806 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17808 * tree-ssa-threadupdate.c (compute_path_counts,
17809 update_joiner_offpath_counts): Use profile_probability.
17811 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17814 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17816 * config/arm/arm-c.c (arm_cpu_builtins): Define
17817 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17819 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17821 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17822 array entries to represent __ieee128 versions of the
17823 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17824 scalar_extract_sig, and scalar_insert_exp built-in functions.
17825 (altivec_resolve_overloaded_builtin): Add special case handling
17826 for the __builtin_scalar_insert_exp function, as represented by
17827 the P9V_BUILTIN_VEC_VSIEDP constant.
17828 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17829 exponent support for __ieee128 argument.
17830 (VSESQP): Add scalar extract signature support for __ieee128
17832 (VSTDCNQP): Add scalar test negative support for __ieee128
17834 (VSIEQP): Add scalar insert exponent support for __int128 argument
17835 with __ieee128 result.
17836 (VSIEQPF): Add scalar insert exponent support for __ieee128
17837 argument with __ieee128 result.
17838 (VSTDCQP): Add scalar test data class support for __ieee128
17840 (VSTDCNQP): Add overload support for scalar test negative with
17841 __ieee128 argument.
17842 (VSTDCQP): Add overload support for scalar test data class
17843 __ieee128 argument.
17844 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17845 UNSPEC_VSX_SXSIGDP.
17846 (UNSPEC_VSX_SIEXPQP): New constant.
17847 (xsxexpqp): New insn for VSX scalar extract exponent quad
17849 (xsxsigqp): New insn for VSX scalar extract significand quad
17851 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17852 precision with floating point argument.
17853 (xststdcqp): New expand for VSX scalar test data class quad
17855 (xststdcnegqp): New expand for VSX scalar test negative quad
17857 (xststdcqp): New insn to match expansions for VSX scalar test data
17858 class quad precision and VSX scalar test negative quad precision.
17859 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17860 special case operand checking to enforce that second operand of
17861 VSX scalar test data class with quad precision argument is a 7-bit
17863 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17864 prototypes and descriptions of __ieee128 versions of
17865 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17866 scalar_test_data_class, and scalar_test_neg built-in functions.
17868 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17870 PR tree-optimization/81162
17871 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17872 replace a negate with an add.
17874 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17876 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17878 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17880 * config/arm/arm-c.c (arm_cpu_builtins): Define
17881 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17883 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17885 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17886 (armv8-r): Set ARM Cortex-R52 as default CPU.
17887 * config/arm/arm-tables.opt: Regenerate.
17888 * config/arm/arm-tune.md: Regenerate.
17889 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17891 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17892 extension for -mcpu=cortex-r52.
17894 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17896 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17897 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17898 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17899 (fp-armv8): Define it as FP_ARMv8 only.
17900 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17901 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17903 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17904 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17905 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17906 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17907 than TARGET_FPU_ARMV8.
17908 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17909 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17910 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17912 * config/arm/neon.md (neon_vrint): Likewise.
17913 (neon_vcvt): Likewise.
17914 (neon_<fmaxmin_op><mode>): Likewise.
17915 (<fmaxmin><mode>3): Likewise.
17916 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17917 * config/arm/predicates.md (arm_cond_move_operator): Check against
17918 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17920 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17922 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17923 to top of function.
17925 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17927 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17928 loop in comment with memset.
17930 2017-07-14 Martin Liska <mliska@suse.cz>
17932 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17933 * dwarf2out.c (is_java): Remove the function.
17934 (output_pubname): Remove usage of the function.
17935 (lower_bound_default): Remove usage of DW_LANG_Java.
17936 (gen_compile_unit_die): Likewise.
17937 * gcc.c: Remove compiler defaults for .java and .zip files.
17938 * gimple-expr.c (remove_suffix): Change as there's no longer
17939 extension than 4-letter one.
17940 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17941 (gimplify_save_expr): Likewise.
17942 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17943 as it's possible even for other languages than Java.
17944 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17945 * lto-opts.c (lto_write_options): Remove reference to Java.
17946 * opts.c (strip_off_ending): Update file extension handling.
17947 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17948 * tree-eh.c (lower_resx): Likewise.
17949 * tree.c (free_lang_data_in_type): Remove dead code.
17950 (find_decls_types_r): Likewise.
17951 (build_common_builtin_nodes): Remove Java from a comment.
17952 (verify_type): Remove dead code.
17953 * varasm.c (assemble_external): Remove Java from a comment.
17955 2017-07-14 Martin Liska <mliska@suse.cz>
17957 * opts.c (finish_options): Add quotes.
17958 (common_handle_option): Likewise.
17960 2017-07-14 Martin Liska <mliska@suse.cz>
17962 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17963 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17964 Remove N_SO_PASCAL.
17965 * dwarf2out.c (lower_bound_default): Do not handle
17967 (gen_compile_unit_die): Likewise.
17968 * gcc.c: Remove default extension binding for GNU Pascal.
17969 * stmt.c: Remove Pascal language from a comment.
17970 * xcoffout.c: Likewise.
17972 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17975 * diagnostic-show-locus.c (fixit_cmp): New function.
17976 (layout::layout): Sort m_fixit_hints.
17977 (column_range::column_range): Assert that the values are valid.
17978 (struct char_span): New struct.
17979 (correction::overwrite): New method.
17980 (struct source_line): New struct.
17981 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17982 calls in terms of classes source_line and char_span, and
17983 correction::overwrite.
17984 (selftest::test_overlapped_fixit_printing_2): New function.
17985 (selftest::diagnostic_show_locus_c_tests): Call it.
17987 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17989 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17990 early if there is no lhs.
17992 2017-07-13 Martin Liska <mliska@suse.cz>
17994 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17995 (gen_reference_type_die): Likewise.
17996 * stor-layout.c: Remove Pascal-related comment.
17998 2017-07-13 Martin Liska <mliska@suse.cz>
18000 * opts.c (finish_options): Add quotes to error messages.
18001 (parse_sanitizer_options): Likewise.
18003 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
18005 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
18007 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
18009 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
18011 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
18013 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
18015 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
18017 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
18020 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
18021 provides the hardware capability bits, define the macro
18022 __BUILTIN_CPU_SUPPORTS__.
18023 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
18024 if GLIBC does not provide the hardware capability bits. Add a
18025 gcc_unreachable call if the built-in cpu function is neither
18026 __builtin_cpu_is nor __builtin_cpu_supports.
18027 (rs6000_get_function_versions_dispatcher): Change the warning
18028 that an old GLIBC is used which does not export the capability
18029 bits to be an error.
18030 * doc/extend.texi (target_clones attribute): Document the
18031 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
18032 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
18033 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
18034 the macros defined by GCC if the newer GLIBC is available.
18036 2017-07-12 Jeff Law <law@redhat.com>
18038 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
18039 remaining includes slightly.
18040 * config/riscv/riscv-builtins.c: Include profile-count.h.
18042 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18045 * config/avr/avr.c (avr_set_current_function): In diagnostic
18046 messages: Quote keywords and (parts of) identifiers.
18047 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
18050 2017-07-12 Carl Love <cel@us.ibm.com>
18052 * config/rs6000/rs6000-c.c: Add support for built-in functions
18053 vector bool char vec_revb (vector bool char);
18054 vector bool short vec_revb (vector short char);
18055 vector bool int vec_revb (vector bool int);
18056 vector bool long long vec_revb (vector bool long long);
18057 * doc/extend.texi: Update the built-in documentation file for the
18058 new built-in functions.
18060 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18062 * config/s390/s390.md: Remove movcc splitter.
18064 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18066 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
18067 load/store on condition.
18069 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
18072 * config/avr/avr.c (avr_encode_section_info)
18073 [progmem && !TREE_READONLY]: Error if progmem object needs
18076 2017-07-11 Michael Collison <michael.collison@arm.com>
18078 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
18081 2017-07-11 Carl Love <cel@us.ibm.com>
18083 * config/rs6000/rs6000-c.c: Add support for builtins
18084 vector unsigned int vec_parity_lsbb (vector signed int);
18085 vector unsigned int vec_parity_lsbb (vector unsigned int);
18086 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
18087 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
18088 vector unsigned long long vec_parity_lsbb (vector signed long long);
18089 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
18090 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
18091 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
18092 * doc/extend.texi: Update the built-in documentation file for the
18093 new built-in functions.
18095 2017-07-11 David Malcolm <dmalcolm@redhat.com>
18097 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
18098 (layout::m_primary_loc): New field.
18099 (layout::layout): Initialize new field. Move location filtering
18100 logic from here to...
18101 (layout::maybe_add_location_range): ...this new method. Add
18102 support for filtering to just the lines already specified by other
18104 (layout::will_show_line_p): New method.
18105 (gcc_rich_location::add_location_if_nearby): New method.
18106 (selftest::test_add_location_if_nearby): New test function.
18107 (selftest::diagnostic_show_locus_c_tests): Call it.
18108 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
18111 2017-07-11 Tom de Vries <tom@codesourcery.com>
18113 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
18114 (bb_first_real_insn): New function.
18115 (nvptx_single): Add extra initialization of broadcasted condition
18118 2017-07-11 Nathan Sidwell <nathan@acm.org>
18120 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
18122 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
18124 * doc/extend.texi (AVR Function Attributes): Remove weblink to
18125 Binutils doc as TEXI will mess them up.
18126 * doc/invoke.texi (AVR Options): Same here.
18128 2017-07-11 Daniel Cederman <cederman@gaisler.com>
18130 * config/sparc/sparc.opt (mfix-ut700): New option.
18131 (mfix-gr712rc): Likewise.
18132 (sparc_fix_b2bst): New variable.
18133 * doc/invoke.texi (SPARC options): Document them.
18134 (ARM options): Fix warnings.
18135 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
18136 instructions to prevent sequences that can trigger the store-store
18137 errata for certain LEON3FT processors.
18138 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
18139 (sparc_option_override): Set sparc_fix_b2bst appropriately.
18140 * config/sparc/sparc.md (fix_b2bst): New attribute.
18141 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
18143 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
18146 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
18148 (*rsqrtsf2_sse): Ditto.
18150 (div<mode>3): Macroize insn from divdf3 and divsf3
18151 using MODEF mode iterator.
18153 2017-07-10 Martin Sebor <msebor@redhat.com>
18155 PR tree-optimization/80397
18156 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
18157 instead of testing for equality to INTEGER_TYPE.
18159 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
18161 * config.gcc: Remove uclibc from arc target spec.
18163 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
18165 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
18167 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
18170 * lto-wrapper.c (remove_option): New function.
18171 (merge_and_complain): Merge PIC/PIE options more realistically.
18173 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18175 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
18179 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
18180 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
18181 * config.in: Regenerate.
18182 * configure: Regenerate.
18183 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
18184 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
18185 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
18186 (TARGET_GASISR_PROLOGUES): ...target mask.
18187 * common/config/avr/avr-common.c
18188 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
18189 Set -mgas-isr-prologues.
18190 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
18191 INSERT_PASS_BEFORE for it.
18192 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
18193 * config/avr/avr.c (avr_option_override)
18194 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
18195 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
18196 (avr_attribute_table) <no_gccisr>: Add new function attribute.
18197 (avr_set_current_function) <is_no_gccisr>: Init machine field.
18198 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
18200 (make_avr_pass_pre_proep): New function.
18201 (emit_push_sfr) <treg>: Add argument to function and use it
18202 instead of TMP_REG.
18203 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
18204 and set machine->gasisr.yes.
18205 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
18206 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
18207 __gcc_isr.n_pushed to .L__stack_usage.
18208 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
18209 (avr_asm_final_postscan_insn): ...this new static function.
18210 * config/avr/avr.h (machine_function)
18211 <is_no_gccisr, use_L__stack_usage>: New fields.
18212 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
18213 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
18214 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
18215 (gasisr, *gasisr): New expander and insn.
18216 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
18217 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
18218 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
18220 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
18222 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
18225 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
18227 Move jump-tables out of .text again.
18230 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
18231 (ASM_OUTPUT_ADDR_VEC): New function.
18232 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
18233 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
18234 INSN_ADDRESSes as asm comment.
18235 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
18236 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
18237 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
18238 * config/avr/avr.md (*tablejump): Adjust comment.
18239 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
18240 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
18242 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
18243 (avr_output_addr_vec): New proto.
18244 (avr_log_t) <insn_addresses>: New field.
18246 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
18249 * config/i386/i386.c (ix86_function_arg_advance): Set
18250 outgoing_args_on_stack to true if there are outgoing arguments
18252 (ix86_function_arg): Likewise.
18253 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
18254 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
18255 * config/i386/i386.h (machine_function): Add
18256 outgoing_args_on_stack.
18258 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
18260 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
18261 supporting pthreds.
18262 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
18264 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
18266 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
18267 (REAL_H): Remove $(MACHMODE_H).
18268 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
18270 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
18271 $(MACHMODE_H) and double-int.h.
18272 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
18274 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
18277 2017-07-07 Andrew Pinski <apinski@cavium.com>
18279 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
18280 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
18282 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
18284 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
18285 Add warning if GCC was not configured to link against a GLIBC that
18286 exports the hardware capability bits.
18287 (make_resolver_func): Make resolver function private and not a
18288 COMDAT function. Create the name with clone_function_name instead
18289 of make_unique_name.
18292 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
18293 correct operand in doing the split.
18295 2017-07-07 Carl Love <cel@us.ibm.com>
18297 * config/rs6000/rs6000-c: Add support for built-in function
18298 vector unsigned short vec_pack_to_short_fp32 (vector float,
18300 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
18301 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
18302 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
18303 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
18304 (convert_4f32_8i16): Add define_expand.
18305 * doc/extend.texi: Update the built-in documentation file for the
18306 new built-in function.
18308 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18310 * config/sparc/m8.md: New file.
18311 * config/sparc/sparc.md: Include m8.md.
18313 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18315 * config/sparc/sparc.opt: New option -mvis4b.
18316 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
18317 (sparc_option_override): Handle VIS4B.
18318 (enum sparc_builtins): Define
18319 SPARC_BUILTIN_DICTUNPACK{8,16,32},
18320 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
18321 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
18322 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
18323 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
18324 (check_constant_argument): New function.
18325 (sparc_vis_init_builtins): Define builtins
18326 __builtin_vis_dictunpack{8,16,32},
18327 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
18328 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
18329 __builtin_vis_fpcmpde{8,16,32}shl and
18330 __builtin_vis_fpcmpur{8,16,32}shl.
18331 (sparc_expand_builtin): Check that the constant operands to
18332 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
18333 constant and in range.
18334 * config/sparc/sparc-c.c (sparc_target_macros): Handle
18336 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
18337 (SPARC_IMM5_P): Likewise.
18338 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
18339 (enabled): Handle vis4b.
18340 (UNSPEC_DICTUNPACK): New unspec.
18341 (UNSPEC_FPCMPSHL): Likewise.
18342 (UNSPEC_FPUCMPSHL): Likewise.
18343 (UNSPEC_FPCMPDESHL): Likewise.
18344 (UNSPEC_FPCMPURSHL): Likewise.
18345 (cpu_feature): New CPU feature `vis4b'.
18346 (dictunpack{8,16,32}): New insns.
18347 (FPCSMODE): New mode iterator.
18348 (fpcscond): New code iterator.
18349 (fpcsucond): Likewise.
18350 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
18351 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
18352 (fpcmpde{8,16,32}{si,di}shl): Likewise.
18353 (fpcmpur{8,16,32}{si,di}shl): Likewise.
18354 * config/sparc/constraints.md: Define constraints `q' for unsigned
18355 2-bit integer constants and `t' for unsigned 5-bit integer
18357 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
18359 (imm5_operand_dictunpack16): Likewise.
18360 (imm5_operand_dictunpack32): Likewise.
18361 (imm2_operand): Likewise.
18362 * doc/invoke.texi (SPARC Options): Document -mvis4b.
18363 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
18364 ditunpack* and fpcmp*shl builtins.
18366 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18368 * config.gcc: Handle m8 in --with-{cpu,tune} options.
18369 * config.in: Add HAVE_AS_SPARC6 define.
18370 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
18372 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
18374 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
18375 (CPP_CPU_SPEC): Handle m8.
18376 (ASM_CPU_SPEC): Likewise.
18377 * config/sparc/sparc-opts.h (enum processor_type): Add
18379 * config/sparc/sparc.c (m8_costs): New struct.
18380 (sparc_option_override): Handle TARGET_CPU_m8.
18381 (sparc32_initialize_trampoline): Likewise.
18382 (sparc64_initialize_trampoline): Likewise.
18383 (sparc_issue_rate): Likewise.
18384 (sparc_register_move_cost): Likewise.
18385 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
18386 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
18387 (ASM_CPU64_DEFAULT_SPEC): Likewise.
18388 (CPP_CPU_SPEC): Handle M8.
18389 (ASM_CPU_SPEC): Likewise.
18390 (AS_M8_FLAG): Define.
18391 * config/sparc/sparc.md: Add m8 to the cpu attribute.
18392 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
18393 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
18395 * configure: Regenerate.
18396 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
18399 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18401 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
18403 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
18404 ("*movdi_insn_sp32"): Do not set v3pipe.
18405 ("*movsi_insn"): Likewise.
18406 ("*movdi_insn_sp64"): Likewise.
18407 ("*movsf_insn"): Likewise.
18408 ("*movdf_insn_sp32"): Likewise.
18409 ("*movdf_insn_sp64"): Likewise.
18410 ("*zero_extendsidi2_insn_sp64"): Likewise.
18411 ("*sign_extendsidi2_insn"): Likewise.
18412 ("*mov<VM32:mode>_insn"): Likewise.
18413 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18414 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18415 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18416 ("<vlop:code><VL:mode>3"): Likewise.
18417 ("*not_<vlop:code><VL:mode>3"): Likewise.
18418 ("*nand<VL:mode>_vis"): Likewise.
18419 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18420 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18421 ("one_cmpl<VL:mode>2"): Likewise.
18422 ("faligndata<VM64:mode>_vis"): Likewise.
18423 ("alignaddrsi_vis"): Likewise.
18424 ("alignaddrdi_vis"): Likweise.
18425 ("alignaddrlsi_vis"): Likewise.
18426 ("alignaddrldi_vis"): Likewise.
18427 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18428 ("bmaskdi_vis"): Likewise.
18429 ("bmasksi_vis"): Likewise.
18430 ("bshuffle<VM64:mode>_vis"): Likewise.
18431 ("cmask8<P:mode>_vis"): Likewise.
18432 ("cmask16<P:mode>_vis"): Likewise.
18433 ("cmask32<P:mode>_vis"): Likewise.
18434 ("pdistn<P:mode>_vis"): Likewise.
18435 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18437 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18439 * config/sparc/sparc.md ("subtype"): New insn attribute.
18440 ("*wrgsr_sp64"): Set insn subtype.
18441 ("*rdgsr_sp64"): Likewise.
18442 ("alignaddrsi_vis"): Likewise.
18443 ("alignaddrdi_vis"): Likewise.
18444 ("alignaddrlsi_vis"): Likewise.
18445 ("alignaddrldi_vis"): Likewise.
18446 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18447 ("fexpand_vis"): Likewise.
18448 ("fpmerge_vis"): Likewise.
18449 ("faligndata<VM64:mode>_vis"): Likewise.
18450 ("bshuffle<VM64:mode>_vis"): Likewise.
18451 ("cmask8<P:mode>_vis"): Likewise.
18452 ("cmask16<P:mode>_vis"): Likewise.
18453 ("cmask32<P:mode>_vis"): Likewise.
18454 ("fchksm16_vis"): Likewise.
18455 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18456 ("fmean16_vis"): Likewise.
18457 ("fp<plusminus_insn>64_vis"): Likewise.
18458 ("<plusminus_insn>v8qi3"): Likewise.
18459 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18460 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18461 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18462 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18463 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18464 ("*movqi_insn"): Likewise.
18465 ("*movhi_insn"): Likewise.
18466 ("*movsi_insn"): Likewise.
18467 ("movsi_pic_gotdata_op"): Likewise.
18468 ("*movdi_insn_sp32"): Likewise.
18469 ("*movdi_insn_sp64"): Likewise.
18470 ("movdi_pic_gotdata_op"): Likewise.
18471 ("*movsf_insn"): Likewise.
18472 ("*movdf_insn_sp32"): Likewise.
18473 ("*movdf_insn_sp64"): Likewise.
18474 ("*zero_extendhisi2_insn"): Likewise.
18475 ("*zero_extendqihi2_insn"): Likewise.
18476 ("*zero_extendqisi2_insn"): Likewise.
18477 ("*zero_extendqidi2_insn"): Likewise.
18478 ("*zero_extendhidi2_insn"): Likewise.
18479 ("*zero_extendsidi2_insn_sp64"): Likewise.
18480 ("ldfsr"): Likewise.
18481 ("prefetch_64"): Likewise.
18482 ("prefetch_32"): Likewise.
18483 ("tie_ld32"): Likewise.
18484 ("tie_ld64"): Likewise.
18485 ("*tldo_ldub_sp32"): Likewise.
18486 ("*tldo_ldub1_sp32"): Likewise.
18487 ("*tldo_ldub2_sp32"): Likewise.
18488 ("*tldo_ldub_sp64"): Likewise.
18489 ("*tldo_ldub1_sp64"): Likewise.
18490 ("*tldo_ldub2_sp64"): Likewise.
18491 ("*tldo_ldub3_sp64"): Likewise.
18492 ("*tldo_lduh_sp32"): Likewise.
18493 ("*tldo_lduh1_sp32"): Likewise.
18494 ("*tldo_lduh_sp64"): Likewise.
18495 ("*tldo_lduh1_sp64"): Likewise.
18496 ("*tldo_lduh2_sp64"): Likewise.
18497 ("*tldo_lduw_sp32"): Likewise.
18498 ("*tldo_lduw_sp64"): Likewise.
18499 ("*tldo_lduw1_sp64"): Likewise.
18500 ("*tldo_ldx_sp64"): Likewise.
18501 ("*mov<VM32:mode>_insn"): Likewise.
18502 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18503 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18505 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18507 * config/sparc/sparc.md ("type"): New insn type viscmp.
18508 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18510 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18511 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18512 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18513 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18515 ("n7_vis_logical_11cycle"): Likewise.
18516 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18517 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18518 * config/sparc/niagara.md ("niag_vis"): Likewise.
18519 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18520 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18522 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18524 * config/sparc/sparc.md: New instruction type `bmask'.
18525 (bmaskdi_vis): Use the `bmask' type.
18526 (bmasksi_vis): Likewise.
18527 * config/sparc/ultra3.md (us3_array): Likewise.
18528 * config/sparc/niagara7.md (n7_array): Likewise.
18529 * config/sparc/niagara4.md (n4_array): Likewise.
18530 * config/sparc/niagara2.md (niag2_vis): Likewise.
18531 (niag3_vis): Likewise.
18532 * config/sparc/niagara.md (niag_vis): Likewise.
18534 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18536 * ipa-comdats.c: Remove optimize check from gate.
18537 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18538 for functions not optimized.
18539 (ipa_fn_summary_read): Skip optimize check.
18540 (ipa_fn_summary_write): Likewise.
18541 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18543 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18545 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18546 (check_callers): Give up on caller which is not optimized.
18547 (inline_small_functions): Likewise.
18548 (ipa_inline): Do not give up when not optimizing.
18549 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18550 away unoptimizes cdtors.
18551 (whole_program_function_and_variable_visibility): Do
18552 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18553 * ipa.c (process_references): Do not check optimize.
18554 (symbol_table::remove_unreachable_nodes): Update optimize check.
18555 (set_writeonly_bit): Update optimize check.
18556 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18557 (pass_ipa_single_use::gate): Remove.
18559 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18561 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18562 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18563 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18564 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18565 permute_load, permute_store, adjust_extract, adjust_splat,
18566 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18567 replace_swap_with_copy, dump_swap_insn_table,
18568 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18569 recombine_lvx_pattern, recombine_stvx_pattern,
18570 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18571 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18572 to file rs6000-p8swap.c.
18573 * config/rs6000/rs6000-p8swap.c: New file.
18574 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18575 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18576 and rs6000*-*-* targets.
18578 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18580 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18582 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18584 * lto-wrapper.c (merge_and_complain): Do not merge
18585 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18586 fsigned_zeros, ftrapping_math, fwrapv.
18587 (append_compiler_options): Do not track these options.
18588 (append_linker_options): Likewie
18590 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18592 * cgraphunit.c (cgraph_node::finalize_function): When
18593 !flag_toplevel_reorde set no_reorder flag.
18594 (varpool_node::finalize_decl): Likewise.
18595 (symbol_table::compile): Drop no toplevel reorder path.
18597 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18599 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18600 edges; zero probability is not better than uninitialized.
18602 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18604 * asan.h (asan_sanitize_allocas_p): Declare.
18605 * asan.c (asan_sanitize_allocas_p): New function.
18606 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18607 (handle_builtin_alloca): Likewise.
18608 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18609 if !asan_sanitize_allocas_p.
18610 * params.def (asan-instrument-allocas): Add new option.
18611 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18612 * opts.c (common_handle_option): Disable allocas sanitization for
18615 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18617 * asan.c: Include gimple-fold.h.
18618 (get_last_alloca_addr): New function.
18619 (handle_builtin_stackrestore): Likewise.
18620 (handle_builtin_alloca): Likewise.
18621 (asan_emit_allocas_unpoison): Likewise.
18622 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18623 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18624 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18625 (instrument_builtin_call): Pass gimple iterator to
18626 get_mem_refs_of_builtin_call.
18627 (last_alloca_addr): New global.
18628 * asan.h (asan_emit_allocas_unpoison): Declare.
18629 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18630 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18631 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18632 if function calls alloca.
18633 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18634 * gimple-fold.h (replace_call_with_value): Declare.
18635 * internal-fn.c: Include asan.h.
18636 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18637 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18639 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18641 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18642 (C_SELFTEST_FLAGS): New.
18643 (CPP_SELFTEST_FLAGS): New.
18644 (SELFTEST_DEPS): New, from deps of s-selftest.
18645 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18646 (CPP_SELFTEST_DEPS): New.
18647 (selftest): Add dependency on s-selftest-c++.
18648 (s-selftest): Rename to...
18649 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18650 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18651 than SELFTEST_FLAGS.
18652 (selftest-gdb): Rename to...
18653 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18655 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18656 (selftest-valgrind): Rename to...
18657 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18659 (selftest-valgrind): Reintroduce as an alias for
18660 selftest-c-valgrind.
18661 (s-selftest-c++): New.
18662 (selftest-c++-gdb): New.
18663 (selftest-c++-valgrind): New.
18665 2017-07-06 Olivier Hainque <hainque@adacore.com>
18667 * gcc.c (process_command): When deciding if undefined variables
18668 should be ignored when processing specs, accept "gcc -v" as well.
18670 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18672 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18673 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18675 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18677 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18678 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18679 * config/arm/arm-tables.opt: Regenerate.
18680 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18682 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18684 2017-07-06 Carl Love <cel@us.ibm.com>
18686 * ChangeLog: Clean up from mid air collision
18688 2017-07-06 Carl Love <cel@us.ibm.com>
18690 * config/rs6000/rs6000-c.c: Add support for built-in functions
18691 vector signed int vec_subc (vector signed int, vector signed int);
18692 vector signed __int128 vec_subc (vector signed __int128,
18693 vector signed __int128);
18694 vector unsigned __int128 vec_subc (vector unsigned __int128,
18695 vector unsigned __int128);
18696 vector signed int vec_sube (vector signed int, vector signed int,
18697 vector signed int);
18698 vector unsigned int vec_sube (vector unsigned int,
18699 vector unsigned int,
18700 vector unsigned int);
18701 vector signed __int128 vec_sube (vector signed __int128,
18702 vector signed __int128,
18703 vector signed__int128);
18704 vector unsigned __int128 vec_sube (vector unsigned __int128,
18705 vector unsigned __int128,
18706 vector unsigned __int128);
18707 vector signed int vec_subec (vector signed int, vector signed int,
18708 vector signed int);
18709 vector unsigned int vec_subec (vector unsigned int,
18710 vector unsigned int,
18711 vector unsigned int);
18712 vector signed __int128 vec_subec (vector signed __int128,
18713 vector signed __int128,
18714 vector signed__int128);
18715 vector unsigned __int128 vec_subec (vector unsigned __int128,
18716 vector unsigned __int128,
18717 vector unsigned __int128);
18718 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18719 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18720 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18721 BU_ALTIVEC_OVERLOAD_X definitions.
18722 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18723 * doc/extend.texi: Update the built-in documentation file for the new
18724 built-in functions.
18726 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18729 * diagnostic-show-locus.c (layout::layout): Use start and finish
18730 spelling location for the start and finish of each range.
18731 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18732 Add unused aspect param.
18733 * input.c (expand_location_1): Add "aspect" param, and use it
18734 to access the correct part of the location.
18735 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18737 (expand_location_to_spelling_point): Likewise.
18738 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18739 param, and pass it to expand_location_1.
18741 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18743 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18744 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18745 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18746 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18747 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18748 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18749 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18750 _mm_maskz_getmant_ss): New intrinsics.
18751 (__builtin_ia32_getexpss128_mask): Changed to ...
18752 __builtin_ia32_getexpss128_round ... this.
18753 (__builtin_ia32_getexpsd128_mask): Changed to ...
18754 __builtin_ia32_getexpsd128_round ... this.
18755 * config/i386/i386-builtin-types.def
18756 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18757 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18758 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18759 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18760 __builtin_ia32_getmantss_mask_round): New builtins.
18761 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18762 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18763 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18764 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18765 * config/i386/sse.md
18766 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18767 avx512f_sgetexp<mode><mask_scalar_name>
18768 <round_saeonly_scalar_name> ... this.
18769 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18770 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18771 vgetexp<ssescalarmodesuffix>
18772 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18773 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18774 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18775 avx512f_vgetmant<mode><mask_scalar_name>
18776 <round_saeonly_scalar_name> ... this.
18777 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18778 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18779 vgetmant<ssescalarmodesuffix>
18780 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18781 %0<mask_scalar_operand4>, %1, %2
18782 <round_saeonly_scalar_mask_op4>, %3} ... this.
18783 * config/i386/subst.md (mask_scalar_operand4,
18784 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18785 round_saeonly_scalar_nimm_predicate): New subst attributes.
18787 2017-07-06 Julia Koval <julia.koval@intel.com>
18789 * config/i386/i386.c (ix86_erase_embedded_rounding):
18790 Remove code for old rounding pattern.
18792 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18794 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18796 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18798 * doc/sourcebuild.texi (Test Directives, Variants of
18799 dg-require-support): Add documentation for dg-require-stack-check.
18801 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18803 * config/i386/subst.md (mask_scalar, round_scalar,
18804 round_saeonly_scalar): New meta-templates.
18805 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18806 round_scalar_mask_operand3, round_scalar_mask_op3,
18807 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18808 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18809 round_saeonly_scalar_constraint,
18810 round_saeonly_scalar_prefix): New subst attribute.
18811 * config/i386/sse.md
18812 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18813 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18814 <round_scalar_name> ... this.
18815 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18816 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18817 <round_scalar_name> ... this.
18818 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18819 <sse>_vm<code><mode>3<mask_scalar_name>
18820 <round_saeonly_scalar_name> ... this.
18821 (v<plusminus_mnemonic><ssescalarmodesuffix>
18822 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18823 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18824 v<plusminus_mnemonic><ssescalarmodesuffix>
18825 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18826 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18827 (v<multdiv_mnemonic><ssescalarmodesuffix>
18828 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18829 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18830 v<multdiv_mnemonic><ssescalarmodesuffix>
18831 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18832 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18833 (v<maxmin_float><ssescalarmodesuffix>
18834 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18835 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18836 v<maxmin_float><ssescalarmodesuffix>
18837 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18838 %0<mask_scalar_operand3>, %1, %<iptr>2
18839 <round_saeonly_scalar_mask_op3>} ... this.
18841 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18843 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18844 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18846 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18847 Alan Hayward <alan.hayward@arm.com>
18848 David Sherwood <david.sherwood@arm.com>
18850 * combine.c (simplify_if_then_else): Remove "enum" before
18852 * compare-elim.c (can_eliminate_compare): Likewise.
18853 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18855 (aarch64_lookup_simd_builtin_type): Likewise.
18856 (aarch64_simd_builtin_type): Likewise.
18857 (aarch64_init_simd_builtin_types): Likewise.
18858 (aarch64_simd_expand_args): Likewise.
18859 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18861 (aarch64_reverse_mask): Likewise.
18862 (aarch64_simd_emit_reg_reg_move): Likewise.
18863 (aarch64_gen_adjusted_ldpstp): Likewise.
18864 (aarch64_ccmp_mode_to_code): Likewise.
18865 (aarch64_operands_ok_for_ldpstp): Likewise.
18866 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18867 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18869 (aarch64_min_divisions_for_recip_mul): Likewise.
18870 (aarch64_reassociation_width): Likewise.
18871 (aarch64_get_condition_code_1): Likewise.
18872 (aarch64_simd_emit_reg_reg_move): Likewise.
18873 (aarch64_simd_attr_length_rglist): Likewise.
18874 (aarch64_reverse_mask): Likewise.
18875 (aarch64_operands_ok_for_ldpstp): Likewise.
18876 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18877 (aarch64_gen_adjusted_ldpstp): Likewise.
18878 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18880 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18881 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18882 (arm_lookup_simd_builtin_type): Likewise.
18883 (arm_simd_builtin_type): Likewise.
18884 (arm_init_simd_builtin_types): Likewise.
18885 (arm_expand_builtin_args): Likewise.
18886 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18887 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18888 (ft32_setup_incoming_varargs): Likewise.
18889 (ft32_function_arg): Likewise.
18890 (ft32_function_arg_advance): Likewise.
18891 (ft32_pass_by_reference): Likewise.
18892 (ft32_arg_partial_bytes): Likewise.
18893 (ft32_valid_pointer_mode): Likewise.
18894 (ft32_addr_space_pointer_mode): Likewise.
18895 (ft32_addr_space_legitimate_address_p): Likewise.
18896 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18898 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18899 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18900 (iamcu_alignment): Likewise.
18901 (canonicalize_vector_int_perm): Likewise.
18902 (ix86_noce_conversion_profitable_p): Likewise.
18903 (ix86_mpx_bound_mode): Likewise.
18904 (ix86_operands_ok_for_move_multiple): Likewise.
18905 * config/microblaze/microblaze-protos.h
18906 (microblaze_expand_conditional_branch_reg): Likewise.
18907 * config/microblaze/microblaze.c
18908 (microblaze_expand_conditional_branch_reg): Likewise.
18909 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18911 (rs6000_reassociation_width): Likewise.
18912 (rs6000_invalid_binary_op): Likewise.
18913 (fusion_p9_p): Likewise.
18914 (emit_fusion_p9_load): Likewise.
18915 (emit_fusion_p9_store): Likewise.
18916 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18918 (riscv_hard_regno_mode_ok_p): Likewise.
18919 (riscv_address_insns): Likewise.
18920 (riscv_split_symbol): Likewise.
18921 (riscv_legitimize_move): Likewise.
18922 (riscv_function_value): Likewise.
18923 (riscv_hard_regno_nregs): Likewise.
18924 (riscv_expand_builtin): Likewise.
18925 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18926 (riscv_build_integer): Likewise.
18927 (riscv_split_integer): Likewise.
18928 (riscv_legitimate_constant_p): Likewise.
18929 (riscv_cannot_force_const_mem): Likewise.
18930 (riscv_regno_mode_ok_for_base_p): Likewise.
18931 (riscv_valid_base_register_p): Likewise.
18932 (riscv_valid_offset_p): Likewise.
18933 (riscv_valid_lo_sum_p): Likewise.
18934 (riscv_classify_address): Likewise.
18935 (riscv_legitimate_address_p): Likewise.
18936 (riscv_address_insns): Likewise.
18937 (riscv_load_store_insns): Likewise.
18938 (riscv_force_binary): Likewise.
18939 (riscv_split_symbol): Likewise.
18940 (riscv_force_address): Likewise.
18941 (riscv_legitimize_address): Likewise.
18942 (riscv_move_integer): Likewise.
18943 (riscv_legitimize_const_move): Likewise.
18944 (riscv_legitimize_move): Likewise.
18945 (riscv_address_cost): Likewise.
18946 (riscv_subword): Likewise.
18947 (riscv_output_move): Likewise.
18948 (riscv_canonicalize_int_order_test): Likewise.
18949 (riscv_emit_int_order_test): Likewise.
18950 (riscv_function_arg_boundary): Likewise.
18951 (riscv_pass_mode_in_fpr_p): Likewise.
18952 (riscv_pass_fpr_single): Likewise.
18953 (riscv_pass_fpr_pair): Likewise.
18954 (riscv_get_arg_info): Likewise.
18955 (riscv_function_arg): Likewise.
18956 (riscv_function_arg_advance): Likewise.
18957 (riscv_arg_partial_bytes): Likewise.
18958 (riscv_function_value): Likewise.
18959 (riscv_pass_by_reference): Likewise.
18960 (riscv_setup_incoming_varargs): Likewise.
18961 (riscv_print_operand): Likewise.
18962 (riscv_elf_select_rtx_section): Likewise.
18963 (riscv_save_restore_reg): Likewise.
18964 (riscv_for_each_saved_reg): Likewise.
18965 (riscv_register_move_cost): Likewise.
18966 (riscv_hard_regno_mode_ok_p): Likewise.
18967 (riscv_hard_regno_nregs): Likewise.
18968 (riscv_class_max_nregs): Likewise.
18969 (riscv_memory_move_cost): Likewise.
18970 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18971 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18972 (rl78_addr_space_address_mode): Likewise.
18973 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18975 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18976 (rs6000_reassociation_width): Likewise.
18977 (rs6000_invalid_binary_op): Likewise.
18978 (fusion_p9_p): Likewise.
18979 (emit_fusion_p9_load): Likewise.
18980 (emit_fusion_p9_store): Likewise.
18981 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18982 (ok_for_simple_move_operands): Likewise.
18983 (ok_for_simple_move_strict_operands): Likewise.
18984 (ok_for_simple_arith_logic_operands): Likewise.
18985 (visium_legitimize_reload_address): Likewise.
18986 (visium_select_cc_mode): Likewise.
18987 (output_cbranch): Likewise.
18988 (visium_split_double_move): Likewise.
18989 (visium_expand_copysign): Likewise.
18990 (visium_expand_int_cstore): Likewise.
18991 (visium_expand_fp_cstore): Likewise.
18992 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18993 (visium_function_arg): Likewise.
18994 (visium_function_arg_advance): Likewise.
18995 (visium_libcall_value): Likewise.
18996 (visium_setup_incoming_varargs): Likewise.
18997 (visium_legitimate_constant_p): Likewise.
18998 (visium_legitimate_address_p): Likewise.
18999 (visium_legitimize_address): Likewise.
19000 (visium_secondary_reload): Likewise.
19001 (visium_register_move_cost): Likewise.
19002 (visium_memory_move_cost): Likewise.
19003 (prepare_move_operands): Likewise.
19004 (ok_for_simple_move_operands): Likewise.
19005 (ok_for_simple_move_strict_operands): Likewise.
19006 (ok_for_simple_arith_logic_operands): Likewise.
19007 (visium_function_value_1): Likewise.
19008 (rtx_ok_for_offset_p): Likewise.
19009 (visium_legitimize_reload_address): Likewise.
19010 (visium_split_double_move): Likewise.
19011 (visium_expand_copysign): Likewise.
19012 (visium_expand_int_cstore): Likewise.
19013 (visium_expand_fp_cstore): Likewise.
19014 (visium_split_cstore): Likewise.
19015 (visium_select_cc_mode): Likewise.
19016 (visium_split_cbranch): Likewise.
19017 (output_cbranch): Likewise.
19018 (visium_print_operand_address): Likewise.
19019 * expmed.c (flip_storage_order): Likewise.
19020 * expmed.h (emit_cstore): Likewise.
19021 (flip_storage_order): Likewise.
19022 * genrecog.c (validate_pattern): Likewise.
19023 * hsa-gen.c (gen_hsa_addr): Likewise.
19024 * internal-fn.c (expand_arith_overflow): Likewise.
19025 * ira-color.c (allocno_copy_cost_saving): Likewise.
19026 * lra-assigns.c (find_hard_regno_for_1): Likewise.
19027 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
19028 (process_invariant_for_inheritance): Likewise.
19029 * lra-eliminations.c (move_plus_up): Likewise.
19030 * omp-low.c (lower_oacc_reductions): Likewise.
19031 * simplify-rtx.c (simplify_subreg): Likewise.
19032 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
19033 (TARGET_CHKP_BOUND_MODE): Likewise..
19034 * targhooks.c (default_chkp_bound_mode): Likewise.
19035 (default_setup_incoming_vararg_bounds): Likewise.
19036 * targhooks.h (default_chkp_bound_mode): Likewise.
19037 (default_setup_incoming_vararg_bounds): Likewise.
19038 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
19039 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
19040 (have_whole_vector_shift): Likewise.
19041 * tree-vect-stmts.c (vectorizable_load): Likewise.
19042 * doc/tm.texi: Regenerate.
19044 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19046 Graceful degrade if Binutils PR21472 is not available.
19049 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
19050 .rodata in flash test fails.
19051 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
19052 * confgure: Regenerate.
19053 * config.in: Regenerate.
19054 * config/avr/avr.c (avr_asm_named_section)
19055 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
19056 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
19057 (avr_asm_init_sections): Same.
19059 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19061 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
19062 (fma<VH:mode>4_intrinsic): Likewise.
19063 (*fmsub<VCVTF:mode>4): Likewise.
19064 (*fmsub<VH:mode>4_intrinsic): Likewise.
19066 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
19069 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
19070 Don't depend on "optimize > 0".
19071 (out_movhi_r_mr, out_movqi_mr_r): Same.
19072 (out_movhi_mr_r, out_movqi_r_mr): Same.
19073 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
19074 io_address_operand on "optimize > 0".
19076 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19078 * tree-loop-distribution.c: Add general explanantion on the pass.
19079 (generate_loops_for_partition): Mark distributed loop.
19080 (pg_add_dependence_edges): New parameter. Handle alias data
19081 dependence specially and record it in the parameter if asked.
19082 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
19083 (init_partition_graph_vertices, add_partition_graph_edge): New.
19084 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
19085 (free_partition_graph_vdata, build_partition_graph): New.
19086 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
19087 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
19088 (data_ref_segment_size, latch_dominated_by_data_ref): New.
19089 (compute_alias_check_pairs, version_loop_by_alias_check): New.
19090 (version_for_distribution_p, finalize_partitions): New.
19091 (distribute_loop): Handle alias data dependence specially. Factor
19092 out loop fusion code as functions and call these functions.
19094 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19096 * tree-loop-distribution.c (classify_partition): New parameter and
19097 better handle reduction statement.
19098 (rdg_build_partitions): Revise comment.
19099 (distribute_loop): Compute statements in all partitions and pass it
19100 to classify_partition.
19102 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19104 * tree-loop-distribution.c (enum partition_type): New.
19105 (struct partition): New field type.
19106 (partition_merge_into): Add parameter. Update partition type.
19107 (data_dep_in_cycle_p, update_type_for_merge): New functions.
19108 (build_rdg_partition_for_vertex): Compute partition type.
19109 (rdg_build_partitions): Dump partition type.
19110 (distribute_loop): Update calls to partition_merge_into.
19112 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19114 * tree-loop-distribution.c (struct ddr_hasher): New.
19115 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
19117 (classify_partition): Call get_data_dependence.
19118 (pg_add_dependence_edges): Ditto.
19119 (distribute_loop): Release data dependence hash table.
19121 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19123 * tree-loop-distribution.c (ref_base_address): Delete.
19124 (similar_memory_accesses): Rename ...
19125 (share_memory_accesses): ... to this. Check if partitions access
19126 the same memory reference.
19127 (distribute_loop): Call share_memory_accesses.
19129 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19131 * tree-loop-distribution.c (struct partition): New field recording
19132 its data reference.
19133 (partition_alloc, partition_free): Init and release data refs.
19134 (partition_merge_into): Merge data refs.
19135 (build_rdg_partition_for_vertex): Collect data refs for partition.
19136 (pg_add_dependence_edges): Change parameters from vector to bitmap.
19138 (distribute_loop): Remve data refs from vertice data of partition
19141 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19143 * tree-loop-distribution.c (params.h): Include header file.
19144 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
19145 (datarefs_vec): New global var.
19146 (create_rdg_vertices): Use datarefs_vec directly.
19147 (free_rdg): Don't free data references.
19148 (build_rdg): Update use. Don't free data references.
19149 (distribute_loop): Compute global variable for data references.
19150 Bail out if there are too many data references.
19152 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19154 * tree-loop-distribution.c (loop_nest): New global var.
19155 (build_rdg): Use loop directly, rather than loop nest.
19156 (pg_add_dependence_edges): Remove loop nest parameter. Use global
19158 (distribute_loop): Compute global variable loop nest. Update use.
19160 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19162 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
19163 (partition_merge_into): New parameter. Dump reason for fusion.
19164 (distribute_loop): Update use of partition_merge_into.
19166 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19168 * tree-loop-distribution.c (bb_top_order_index): New.
19169 (bb_top_order_index_size, bb_top_order_cmp): New.
19170 (stmts_from_loop): Use topological order.
19171 (pass_loop_distribution::execute): Compute and release topological
19172 order for basic blocks.
19174 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19176 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
19179 2017-07-05 Bin Cheng <bin.cheng@arm.com>
19181 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
19182 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
19183 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
19184 * internal-fn.def (LOOP_DIST_ALIAS): New.
19185 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
19186 (fold_loop_internal_call): ... this.
19187 (vect_loop_dist_alias_call): New function.
19188 (set_uid_loop_bbs): Call fold_loop_internal_call.
19189 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
19192 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19195 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
19196 Require dead FLAGS_REG at the beginning of a peephole.
19198 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
19201 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
19202 arguments in the call to __builtin_ia32_sbb_u32.
19203 (_subborrow_u64): Swap _X and _Y arguments in the call to
19204 __builtin_ia32_sbb_u64.
19206 2017-07-04 Jakub Jelinek <jakub@redhat.com>
19209 * tree-vrp.c (compare_assert_loc): Turn into a function template
19210 with stable template parameter. Only test if a->e is NULL,
19211 !a->e == !b->e has been verified already. Use e == NULL or
19212 e != NULL instead of e or ! e tests. If stable is true, don't use
19213 iterative_hash_expr, on the other side allow a or b or both NULL
19214 and sort the NULLs last.
19215 (process_assert_insertions): Sort using compare_assert_loc<false>
19216 instead of compare_assert_loc, later sort using
19217 compare_assert_loc<true> before calling process_assert_insertions_for
19218 in a loop. Use break instead of continue once seen NULL pointer.
19220 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19222 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
19223 Cortex-R7 and Cortex-R8 processors.
19225 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19227 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
19228 uninitialized while src is not.
19230 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
19232 * common/config/arm/arm-common.c: Adjust include path for
19234 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
19235 (arm-cpu.h): Create in build directory. Adjust dependency rules.
19236 (arm-cpu-data.h): Likewise.
19237 (arm-cpu-cdata.h): Likewise.
19238 * config/arm/arm-cpu.h: Delete.
19239 * config/arm/arm-cpu-cdata.h: Delete.
19240 * config/arm/arm-cpu-data.h: Delete.
19242 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
19244 * config/arm/arm-cpus.in (cortex-a55): New.
19245 (cortex-a75): Likewise.
19246 (cortex-a75.cortex-a55): Likewise.
19247 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
19249 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
19250 * config/arm/arm-cpu-cdata.h: Regenerate.
19251 * config/arm/arm-cpu-data.h: Regenerate.
19252 * config/arm/arm-cpu.h: Regenerate.
19253 * config/arm/arm-tables.opt: Regenerate.
19254 * config/arm/arm-tune.md: Regenerate.
19256 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19258 * haifa-sched.c (sched_create_recovery_edges): Update profile.
19260 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
19262 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
19265 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
19267 PR tree-optimization/81292
19268 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
19269 full_string_p, also call adjust_related_strinfos if the adjustment
19270 is simple, otherwise invalidate related strinfos.
19272 2017-07-04 Martin Liska <mliska@suse.cz>
19275 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
19276 newly created variable as DECL_IGNORED_P.
19278 2017-07-04 Martin Liska <mliska@suse.cz>
19281 * ipa-inline.c (inline_small_functions):
19282 Use xstrdup_for_dump.
19284 2017-07-04 Tom de Vries <tom@codesourcery.com>
19286 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
19288 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19291 * config/darwin.c (darwin_function_switched_text_sections):
19294 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
19296 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
19298 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19300 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
19302 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19304 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
19305 min_profitable_iters, and th as inclusive lower bounds.
19306 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
19307 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
19308 for min_profitable_iters and min_profitable_estimate.
19309 (vect_transform_loop): Treat th as an inclusive lower bound.
19310 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
19312 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
19315 * config/darwin.c (darwin_function_switched_text_sections):
19316 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
19317 in two pieces, and suppress the use of buf.
19319 2017-07-03 Nathan Sidwell <nathan@acm.org>
19321 * hash-table.h (hash_table_mod1): Fix indentation.
19323 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19325 PR middle-end/81290
19326 * predict.c (force_edge_cold): Be more careful about propagation
19328 * profile-count.h (profile_probability::guessed,
19329 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
19331 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
19333 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
19335 * doc/invoke.texi (rcpc architecture extension): Document it.
19337 2017-07-03 Richard Biener <rguenther@suse.de>
19339 PR tree-optimization/60510
19340 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
19341 the scalar reduction PHI and use it.
19342 (vectorizable_reduction): Properly guard the single_defuse_cycle
19343 path for non-SLP reduction chains where we cannot use it.
19344 Rework reduc_def/index and vector type deduction. Rework
19345 vector operand gathering during reduction op code-gen.
19346 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
19347 chains dissolve the chain and leave it to non-SLP reduction
19350 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19352 * tree-data-ref.h (dr_alignment): Declare.
19353 * tree-data-ref.c (dr_alignment): New function.
19354 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
19355 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
19357 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
19359 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19361 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
19362 and base_misalignment fields.
19363 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
19364 * tree-data-ref.c: Include builtins.h.
19365 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
19366 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
19367 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
19368 * tree-vect-data-refs.c: Include tree-cfg.h.
19369 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
19370 fields instead of calculating an alignment here.
19371 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
19372 innermost_loop_behavior fields.
19374 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19376 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
19378 (DR_STEP_ALIGNMENT): New macro.
19379 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
19380 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
19381 (create_data_ref): Print it.
19382 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
19383 to tell whether the step preserves vector (mis)alignment.
19384 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19385 Move the check for an integer step and generalise to all INTEGER_CST.
19386 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
19387 Print the outer step alignment.
19389 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19391 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
19392 with offset_alignment.
19393 (DR_ALIGNED_TO): Delete.
19394 (DR_OFFSET_ALIGNMENT): New macro.
19395 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
19396 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
19397 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
19398 (create_data_ref): Likewise.
19399 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
19400 (vect_analyze_data_refs): Likewise.
19401 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
19402 creating dummy innermost behavior.
19404 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19406 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
19407 with a "innermost_loop_behavior *" and refeence tree.
19408 * tree-data-ref.c (dr_analyze_innermost): Likewise.
19409 (create_data_ref): Update call accordingly.
19410 * tree-predcom.c (find_looparound_phi): Likewise.
19412 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19414 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
19415 fields with dr_wrt_vec_loop.
19416 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
19417 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
19418 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
19419 (vect_dr_behavior): New function.
19420 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19421 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19422 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19423 track whether the step preserves the misalignment.
19424 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19425 Use vect_dr_behavior.
19426 (vect_setup_realignment): Update call accordingly.
19427 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19428 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19429 call to vect_create_addr_base_for_vector_ref.
19430 (vect_create_cond_for_align_checks): Likewise.
19431 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19432 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19433 (vect_recog_mask_conversion_pattern): Likewise.
19434 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19435 (new_stmt_vec_info): Remove redundant zeroing.
19437 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19439 * common/config/arm/arm-common.c (arm_be8_option): New function.
19440 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19441 (ISA_ARMv6): Add isa_bit_be8.
19442 * config/arm/arm.h (arm_be8_option): Add prototype.
19443 (BE8_SPEC_FUNCTION): New define.
19444 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19445 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19446 (mlittle-endian): Similarly.
19447 (mbe8, mbe32): New options.
19448 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19449 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19451 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19453 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19455 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19457 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19458 (cleanup_tree_cfg_bb): Use it.
19459 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19461 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19463 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19466 * loop-doloop.c (add_test): Update profile.
19468 2017-07-03 Martin Liska <mliska@suse.cz>
19471 * sanopt.c (rewrite_usage_of_param): New function.
19472 (sanitize_rewrite_addressable_params): Likewise.
19473 (pass_sanopt::execute): Call rewrite_usage_of_param.
19475 2017-07-03 Richard Biener <rguenther@suse.de>
19477 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19478 back to using VIEW_CONVERT_EXPR.
19480 2017-07-03 Martin Liska <mliska@suse.cz>
19483 * doc/extend.texi: Document when a resolver function is
19484 generated for target_clones.
19486 2017-07-03 Martin Liska <mliska@suse.cz>
19488 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19489 and shadow memory of auto variables which are subject of
19490 use-after-scope sanitization.
19491 (asan_expand_mark_ifn): Add do set only when is_poison.
19493 2016-07-03 Richard Biener <rguenther@suse.de>
19495 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19497 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19498 (vectorizable_reduction): Perform reduction PHI creation when
19499 visiting a reduction PHI and adjust and simplify code generation
19500 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19501 (vect_transform_loop): Visit reduction PHIs.
19502 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19503 defs into the SLP tree.
19504 (vect_build_slp_tree): Reduction defs terminate the recursion.
19505 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19507 (vect_get_vec_defs_for_stmt_copy): Export.
19508 (vect_get_vec_defs): Likewise.
19509 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19511 (vect_get_vec_defs_for_stmt_copy): Declare.
19512 (vect_get_vec_defs): Likewise.
19514 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19516 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19517 parameter with a "loop" parameter and use it instead of the
19518 loop containing DR_STMT. Don't check simple_iv when doing
19519 BB analysis. Describe the two analysis modes in the comment.
19521 2017-07-03 Tom de Vries <tom@codesourcery.com>
19523 PR tree-optimization/69468
19524 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19525 (find_same_succ_bb): Handle ignore_edge_flags.
19527 2017-07-03 Tom de Vries <tom@codesourcery.com>
19529 PR tree-optimization/81192
19530 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19532 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19534 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19536 2017-07-03 Tom de Vries <tom@codesourcery.com>
19538 PR tree-optimization/81192
19539 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19540 BB_SAME_SUCC (bb) == NULL.
19542 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19544 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19547 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19549 * dumpfile.c: Include profile-count.h
19550 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19552 (insert_cond_bb): Update profile.
19553 * tree-cfg.h (insert_cond_bb): Update prototype.
19554 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19555 * tree-dump.c: Do not include tree-cfg.
19557 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19559 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19561 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19563 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19566 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19568 * tree-complex.c (expand_complex_div_wide): update profile.
19570 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19571 Alan Hayward <alan.hayward@arm.com>
19572 David Sherwood <david.sherwood@arm.com>
19574 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19575 (CORETYPES_H): New define.
19576 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19577 (insn-modes-inline.h, s-modes-inline-h): New rules.
19578 (generated_files): Add insn-modes-inline.h.
19579 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19580 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19581 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19582 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19583 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19584 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19585 (build/gencodes.o, build/genconditions.o): Likewise.
19586 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19587 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19588 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19589 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19590 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19591 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19592 * coretypes.h: Include everything up to real.h for generators.
19593 Include insn-modes.h first. Include wide-int-print.h after
19594 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19595 * machmode.h: Don't include insn-modes.h here.
19596 * function-tests.c: Remove includes of signop.h, machmode.h,
19597 double-int.h and wide-int.h.
19599 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19601 * optc-save-gen.awk: Likewise.
19602 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19603 * godump.c: Remove include of wide-int-print.h.
19604 * pretty-print.h: Likewise.
19605 * wide-int-print.cc: Likewise.
19606 * wide-int.cc: Likewise.
19607 * hash-map-tests.c: Remove include of signop.h.
19608 * hash-set-tests.c: Likewise.
19609 * rtl-tests.c: Likewise.
19610 * mkconfig.sh: Remove include of machmode.h.
19611 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19613 (emit_insn_modes_inline_h): ...this new function. Emit the code
19614 into an insn-modes-inline.h header file, adding appropriate
19615 include guards and end comments.
19616 (emit_insn_modes_c_header): Remove include of machmode.h.
19617 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19619 (main): Handle -i flag and call emit_insn_modes_inline_h when
19622 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19624 * tree-ssa-strlen.c (strinfo): Rename the length field to
19625 nonzero_chars. Add a full_string_p field.
19626 (compare_nonzero_chars, zero_length_string_p): New functions.
19627 (get_addr_stridx): Add an offset_out parameter.
19628 Use compare_nonzero_chars.
19629 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19630 (new_strinfo): Update after above changes to strinfo.
19631 (set_endptr_and_length): Set full_string_p.
19632 (get_string_length): Update after above changes to strinfo.
19633 (unshare_strinfo): Update call to new_strinfo.
19634 (maybe_invalidate): Likewise.
19635 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19636 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19637 as a uhwi instead of an shwi. Update after above changes to
19638 strinfo and new_strinfo.
19639 (zero_length_string): Assert that chainsi contains full strings.
19640 Use zero_length_string_p. Update call to new_strinfo.
19641 (adjust_related_strinfos): Update after above changes to strinfo.
19642 Copy full_string_p from origsi.
19643 (adjust_last_stmt): Use zero_length_string_p.
19644 (handle_builtin_strlen): Update after above changes to strinfo and
19645 new_strinfo. Install the lhs as the string length if the previous
19646 entry didn't describe a full string.
19647 (handle_builtin_strchr): Update after above changes to strinfo
19649 (handle_builtin_strcpy): Likewise.
19650 (handle_builtin_strcat): Likewise.
19651 (handle_builtin_malloc): Likewise.
19652 (handle_pointer_plus): Likewise.
19653 (handle_builtin_memcpy): Likewise. Track nonzero characters
19654 that aren't necessarily followed by a nul terminator.
19655 (handle_char_store): Likewise.
19657 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19659 PR tree-optimization/80769
19660 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19661 for malloc and calloc. Document the new invariant that all related
19662 strinfos have delayed lengths or none do.
19663 (verify_related_strinfos): Move earlier in file.
19664 (set_endptr_and_length): New function, split out from...
19665 (get_string_length): ...here. Also set the lengths of related
19667 (zero_length_string): Assert that chainsi has known (rather than
19669 (adjust_related_strinfos): Likewise.
19671 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19673 PR tree-optimization/81136
19674 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19675 assert that two references with the same misalignment have the same
19676 compile-time misalignment if those compile-time misalignments
19679 2017-07-01 Andi Kleen <ak@linux.intel.com>
19681 * print-tree.c (print_node): Print all attributes.
19683 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19685 * cfg.c (scale_bbs_frequencies): New function.
19686 * cfg.h (scale_bbs_frequencies): Declare it.
19687 * cfgloopanal.c (single_likely_exit): Cleanup.
19688 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19690 (scale_loop_profile): Likewise.
19691 (loop_version): Likewise.
19692 (create_empty_loop_on_edge): Update.
19693 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19694 scale_loop_frequencies, scale_loop_profile, loopify,
19695 loop_version): Update prototypes.
19696 * modulo-sched.c (sms_schedule): Update.
19697 * predict.c (unlikely_executed_edge_p): Also check probability.
19698 (probably_never_executed_edge_p): Fix typo.
19699 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19700 * tree-parloops.c (gen_parallel_loop): Update.
19701 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19702 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19703 * tree-ssa-loop-split.c (split_loop): Update.
19704 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19705 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19706 (vect_loop_versioning): Update.
19707 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19709 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19711 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19713 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19715 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19716 to keep profile consistent.
19718 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19720 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19721 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19722 * profile-count.h (max_safe_multiplier): Make unsigned.
19723 (profile_count::guessed_zero): New.
19725 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19727 * bb-reorder.c (fix_up_crossing_landing_pad,
19728 fix_crossing_conditional_branches): Use make_single_succ_edge
19729 to keep profile consistent.
19731 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19733 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19736 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19739 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19740 the right scopes, make sure cond_jump isn't preserved between multiple
19741 iterations. Search for fallthru edge whenever there are 3+ edges and
19742 use find_fallthru_edge for it.
19744 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19746 Patch by Alexander Monakov <amonakov@ispras.ru>
19747 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19748 probabilities consistently.
19750 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19752 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19753 emit_cmp_and_jump_insns.
19755 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19758 * tree-inline.c (expand_call_inline): Combine profile statuses.
19760 2017-06-30 Andrew Pinski <apinski@cavium.com>
19762 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19763 fold_stmt returned true.
19765 2017-06-30 Nathan Sidwell <nathan@acm.org>
19767 * ggc.h (empty_string): Delete.
19768 * cfgexpand.c (expand_asm_stmt): Use plain "".
19769 * optabs.c (expand_asm_memory_barrier): Likewise.
19770 * stringpool.c (empty_string): Delete.
19771 (digit_vector, digit_string): Delete.
19772 (ggc_alloc_string): Use plain "", don't optimize single digit
19773 strings. Use ggc_alloc_atomic.
19775 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19777 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19778 comparison set and one other set, use the cost of the non-comparison
19781 2017-06-30 Nathan Sidwell <nathan@acm.org>
19783 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19786 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19788 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19789 loops. Remove now unneeded calls to gimple_switch_set_label() that
19790 just set removed labels to NULL_TREE.
19792 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19794 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19795 (set_range_info): ...here. Only call set_range_info_raw if domain
19797 (set_nonzero_bits): Call set_range_info_raw.
19798 * tree-ssanames.h (set_range_info_raw): New.
19800 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19803 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19804 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19805 of nonimmediate_operand and <store_mask_constraint> instead of m
19806 for the input operand. For V8FI iterator, always split if input
19807 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19808 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19809 <store_mask_predicate> instead of register_operand and
19810 <store_mask_constraint> instead of v for the input operand. Make
19811 sure both operands aren't MEMs for if not <mask_applied>.
19813 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19815 * lto-wrapper.c (copy_file) Close both file descriptors before
19818 2017-06-30 Martin Liska <mliska@suse.cz>
19821 * multiple_target.c (create_dispatcher_calls): Make ifunc
19822 also for function that don't have calls or are not referenced.
19824 2017-06-30 Richard Biener <rguenther@suse.de>
19826 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19827 analyze the first scalar stmt. Move vector type computation
19828 for the BB case here from ...
19829 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19830 live operation processing in the SLP case properly.
19832 2017-06-30 Richard Biener <rguenther@suse.de>
19834 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19836 2017-06-30 Martin Liska <mliska@suse.cz>
19839 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19840 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19842 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19844 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19847 2017-06-30 Richard Biener <rguenther@suse.de>
19849 PR tree-optimization/81249
19850 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19851 condition reduction result to original scalar type.
19853 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19855 * profile-count.h (enum profile_quality): Fix typos and whitespace
19858 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19860 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19861 type for branch probabilities.
19863 2017-06-29 Julian Brown <julian@codesourcery.com>
19864 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19866 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19867 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19868 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19869 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19871 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19873 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19874 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19875 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19876 CC usage from generic code to here.
19877 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19878 CC usage into the target macros.
19880 2017-06-29 Maya Rashish <coypu@sdf.org>
19882 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19885 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19887 * arm/arm-builtins.c: Include profile-count.h
19888 * except.c (sjlj_emit_function_enter): Use
19889 profile_probability::unlikely.
19891 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19893 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19894 and tocrel_offset be pointer args rather than implicitly using
19896 (legitimate_constant_pool_address_p, rs6000_emit_move,
19897 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19898 tocrel_offset and use in toc_relative_expr_p call.
19899 (print_operand, print_operand_address): Use static tocrel_base_oac
19900 and tocrel_offset_oac.
19901 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19904 2017-06-29 Maya Rashish <coypu@sdf.org>
19906 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19908 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19910 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19911 objects, take into account only the alignment of 'op0' and 'mode1' if
19914 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19916 * ccmp.c (ccmp_tree_comparison_p): New function.
19917 (ccmp_candidate_p): Update to use above function.
19918 (get_compare_parts): New function.
19919 (expand_ccmp_next): Update to use new functions.
19920 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19922 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19923 take mode as argument.
19924 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19925 * expr.c (expand_expr_real_1): Pass mode as argument.
19927 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19929 * combine.c (combine_instructions): Print insns to dump_file, together
19932 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19934 * asan.c (asan_emit_stack_protection): Update.
19935 (create_cond_insert_point): Update.
19936 * auto-profile.c (afdo_propagate_circuit): Update.
19937 * basic-block.h (struct edge_def): Turn probability to
19938 profile_probability.
19939 (EDGE_FREQUENCY): Update.
19940 * bb-reorder.c (find_traces_1_round): Update.
19941 (better_edge_p): Update.
19942 (sanitize_hot_paths): Update.
19943 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19944 (make_single_succ_edge): Update.
19945 (check_bb_profile): Update.
19946 (dump_edge_info): Update.
19947 (update_bb_profile_for_threading): Update.
19948 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19949 probabilitycount to 0.
19950 * cfgbuild.c (compute_outgoing_frequencies): Update.
19951 * cfgcleanup.c (try_forward_edges): Update.
19952 (outgoing_edges_match): Update.
19953 (try_crossjump_to_edge): Update.
19954 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19955 (expand_gimple_tailcall): Update.
19956 (construct_init_block): Use make_single_succ_edge.
19957 (construct_exit_block): Use make_single_succ_edge.
19958 * cfghooks.c (verify_flow_info): Update.
19959 (redirect_edge_succ_nodup): Update.
19960 (split_edge): Update.
19961 (account_profile_record): Update.
19962 * cfgloopanal.c (single_likely_exit): Update.
19963 * cfgloopmanip.c (scale_loop_profile): Update.
19964 (set_zero_probability): Remove.
19965 (duplicate_loop_to_header_edge): Update.
19966 * cfgloopmanip.h (loop_version): Update prototype.
19967 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19968 (force_nonfallthru_and_redirect): Update.
19969 (update_br_prob_note): Update.
19970 (rtl_verify_edges): Update.
19971 (purge_dead_edges): Update.
19972 (rtl_lv_add_condition_to_bb): Update.
19973 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19974 * cgraphunit.c (init_lowered_empty_function): Update.
19975 (cgraph_node::expand_thunk): Update.
19976 * cilk-common.c: Include profile-count.h
19977 * dojump.c (inv): Remove.
19978 (jumpifnot): Update.
19979 (jumpifnot_1): Update.
19980 (do_jump_1): Update.
19982 (do_jump_by_parts_greater_rtx): Update.
19983 (do_compare_rtx_and_jump): Update.
19984 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19985 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19986 * dwarf2cfi.c: Include profile-count.h
19987 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19988 (sjlj_emit_dispatch_table): Likewise.
19989 * explow.c: Include profile-count.h
19990 * expmed.c (emit_store_flag_force): Update.
19991 (do_cmp_and_jump): Update.
19992 * expr.c (compare_by_pieces_d::generate): Update.
19993 (compare_by_pieces_d::finish_mode): Update.
19994 (emit_block_move_via_loop): Update.
19995 (store_expr_with_bounds): Update.
19996 (store_constructor): Update.
19997 (expand_expr_real_2): Update.
19998 (expand_expr_real_1): Update.
19999 * expr.h (try_casesi, try_tablejump): Update prototypes.
20000 * gimple-pretty-print.c (dump_probability): Update.
20001 (dump_profile): New.
20002 (dump_gimple_label): Update.
20003 (dump_gimple_bb_header): Update.
20004 * graph.c (draw_cfg_node_succ_edges): Update.
20005 * hsa-gen.c (convert_switch_statements): Update.
20006 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
20007 (find_if_case_1): Update.
20008 (find_if_case_2): Update.
20009 * internal-fn.c (expand_arith_overflow_result_store): Update.
20010 (expand_addsub_overflow): Update.
20011 (expand_neg_overflow): Update.
20012 (expand_mul_overflow): Update.
20013 (expand_vector_ubsan_overflow): Update.
20014 * ipa-cp.c (good_cloning_opportunity_p): Update.
20015 * ipa-split.c (split_function): Use make_single_succ_edge.
20016 * ipa-utils.c (ipa_merge_profiles): Update.
20017 * loop-doloop.c (add_test): Update.
20018 (doloop_modify): Update.
20019 * loop-unroll.c (compare_and_jump_seq): Update.
20020 (unroll_loop_runtime_iterations): Update.
20021 * lra-constraints.c (lra_inheritance): Update.
20022 * lto-streamer-in.c (input_cfg): Update.
20023 * lto-streamer-out.c (output_cfg): Update.
20024 * mcf.c (adjust_cfg_counts): Update.
20025 * modulo-sched.c (sms_schedule): Update.
20026 * omp-expand.c (expand_omp_for_init_counts): Update.
20027 (extract_omp_for_update_vars): Update.
20028 (expand_omp_ordered_sink): Update.
20029 (expand_omp_for_ordered_loops): Update.
20030 (expand_omp_for_generic): Update.
20031 (expand_omp_for_static_nochunk): Update.
20032 (expand_omp_for_static_chunk): Update.
20033 (expand_cilk_for): Update.
20034 (expand_omp_simd): Update.
20035 (expand_omp_taskloop_for_outer): Update.
20036 (expand_omp_taskloop_for_inner): Update.
20037 * omp-simd-clone.c (simd_clone_adjust): Update.
20038 * optabs.c (expand_doubleword_shift): Update.
20039 (expand_abs): Update.
20040 (emit_cmp_and_jump_insn_1): Update.
20041 (expand_compare_and_swap_loop): Update.
20042 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
20043 * predict.c (predictable_edge_p): Update.
20044 (edge_probability_reliable_p): Update.
20045 (set_even_probabilities): Update.
20046 (combine_predictions_for_insn): Update.
20047 (combine_predictions_for_bb): Update.
20048 (propagate_freq): Update.
20049 (estimate_bb_frequencies): Update.
20050 (force_edge_cold): Update.
20051 * profile-count.c (profile_count::dump): Add missing space into dump.
20052 (profile_count::debug): Add newline.
20053 (profile_count::differs_from_p): Explicitly convert to unsigned.
20054 (profile_count::stream_in): Update.
20055 (profile_probability::dump): New member function.
20056 (profile_probability::debug): New member function.
20057 (profile_probability::differs_from_p): New member function.
20058 (profile_probability::differs_lot_from_p): New member function.
20059 (profile_probability::stream_in): New member function.
20060 (profile_probability::stream_out): New member function.
20061 * profile-count.h (profile_count_quality): Rename to ...
20062 (profile_quality): ... this one.
20063 (profile_probability): New.
20064 (profile_count): Update.
20065 * profile.c (compute_branch_probabilities): Update.
20066 * recog.c (peep2_attempt): Update.
20067 * sched-ebb.c (schedule_ebbs): Update.
20068 * sched-rgn.c (find_single_block_region): Update.
20069 (compute_dom_prob_ps): Update.
20070 (schedule_region): Update.
20071 * sel-sched-ir.c (compute_succs_info): Update.
20072 * stmt.c (struct case_node): Update.
20073 (do_jump_if_equal): Update.
20074 (get_outgoing_edge_probs): Update.
20075 (conditional_probability): Update.
20076 (emit_case_dispatch_table): Update.
20077 (expand_case): Update.
20078 (expand_sjlj_dispatch_table): Update.
20079 (emit_case_nodes): Update.
20080 * targhooks.c: Update.
20081 * tracer.c (better_p): Update.
20082 (find_best_successor): Update.
20083 * trans-mem.c (expand_transaction): Update.
20084 * tree-call-cdce.c: Update.
20085 * tree-cfg.c (gimple_split_edge): Upate.
20086 (move_sese_region_to_fn): Upate.
20087 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
20088 * tree-eh.c (lower_resx): Upate.
20089 (cleanup_empty_eh_move_lp): Upate.
20090 * tree-if-conv.c (version_loop_for_if_conversion): Update.
20091 * tree-inline.c (copy_edges_for_bb): Update.
20092 (copy_cfg_body): Update.
20093 * tree-parloops.c (gen_parallel_loop): Update.
20094 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
20095 (gimple_gen_time_profiler): Update.
20096 * tree-ssa-dce.c (remove_dead_stmt): Update.
20097 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20098 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
20099 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
20100 (unloop_loops): Update.
20101 (try_peel_loop): Update.
20102 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20103 * tree-ssa-loop-split.c (connect_loops): Update.
20104 (split_loop): Update.
20105 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
20106 (hoist_guard): Update.
20107 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
20108 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
20109 (value_replacement): Update.
20110 * tree-ssa-reassoc.c (branch_fixup): Update.
20111 * tree-ssa-tail-merge.c (replace_block_by): Update.
20112 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
20113 (create_edge_and_update_destination_phis): Update.
20114 (compute_path_counts): Update.
20115 (recompute_probabilities): Update.
20116 (update_joiner_offpath_counts): Update.
20117 (freqs_to_counts_path): Update.
20118 (duplicate_thread_path): Update.
20119 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
20120 (struct switch_conv_info): Update.
20121 (gen_inbound_check): Update.
20122 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20123 (vect_do_peeling): Update.
20124 (vect_loop_versioning): Update.
20125 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20126 (optimize_mask_stores): Update.
20127 * ubsan.c (ubsan_expand_null_ifn): Update.
20128 * value-prof.c (gimple_divmod_fixed_value): Update.
20129 (gimple_divmod_fixed_value_transform): Update.
20130 (gimple_mod_pow2): Update.
20131 (gimple_mod_pow2_value_transform): Update.
20132 (gimple_mod_subtract): Update.
20133 (gimple_mod_subtract_transform): Update.
20134 (gimple_ic): Update.
20135 (gimple_stringop_fixed_value): Update.
20136 (gimple_stringops_transform): Update.
20137 * value-prof.h: Update.
20139 2017-06-29 Carl Love <cel@us.ibm.com>
20141 * config/rs6000/rs6000-c.c: Add support for built-in functions
20142 vector signed int vec_signed (vector float);
20143 vector signed long long vec_signed (vector double);
20144 vector signed int vec_signed2 (vector double, vector double);
20145 vector signed int vec_signede (vector double);
20146 vector signed int vec_signedo (vector double);
20147 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
20148 instruction generator.
20149 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20150 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
20151 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
20153 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
20154 vunsignede_v2df): Add define_expands.
20155 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
20156 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
20157 VEC_UNSIGNEDO): Add definitions.
20158 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
20159 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
20160 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
20161 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
20162 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
20163 * config/rs6000/altivec.h (vec_signed, vec_signed2,
20164 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
20165 vec_unsignede, vec_unsignedo): Add builtin defines.
20166 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
20168 * doc/extend.texi: Update the built-in documentation file for the
20169 new built-in functions.
20171 2017-06-29 Richard Biener <rguenther@suse.de>
20173 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
20174 reduction chains to LOOP_VINFO_REDUCTIONS.
20175 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
20176 SLP reductions after processing reduction chains.
20178 2017-06-29 Nathan Sidwell <nathan@acm.org>
20180 * builtins.c (fold_builtin_FUNCTION): Use
20181 lang_hooks.decl_printable_name.
20183 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
20185 PR middle-end/81194
20186 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
20187 with only one label.
20188 * stmt.c (expand_case): Assert NCASES is greater than one.
20190 2017-06-29 Richard Biener <rguenther@suse.de>
20192 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
20194 (group_case_labels): Likewise.
20195 (find_taken_edge): Push sanity checking on val to workers...
20196 (find_taken_edge_cond_expr): ... here
20197 (find_taken_edge_switch_expr): ... and here, handle cases
20198 with just a default label.
20199 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
20200 (group_case_labels): Likewise.
20201 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
20202 group_case_labels does anything cleanup the CFG again.
20204 2017-06-29 Bin Cheng <bin.cheng@arm.com>
20206 PR tree-optimization/81196
20207 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
20208 exit condition comparing two IVs.
20210 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
20212 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
20213 profile to the dummy entry at the end of the list of architectures.
20214 * config/arm/arm-cpu-cdata.h: Regenerated.
20216 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20217 Michael Collison <michael.collison@arm.com>
20220 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
20222 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
20223 (*aarch64_reg_<mode>3_minus_mask): New pattern.
20224 (*aarch64_<optab>_reg_di3_mask2): New pattern.
20225 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
20226 of shift when the shift amount is masked with constant equal to
20227 the size of the mode.
20228 * config/aarch64/predicates.md (subreg_lowpart_operator): New
20231 2017-06-29 Martin Liska <mliska@suse.cz>
20233 * config/i386/i386.opt: Change range from [1,5] to [0,5].
20235 2017-06-29 Yury Gribov <tetra2005@gmail.com>
20238 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
20241 (edge_growth_cache_entry::edge_growth_cache_entry): New
20243 (reset_edge_growth_cache): Update to use constructor.
20245 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20247 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
20248 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20249 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
20251 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
20253 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
20254 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
20256 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
20258 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
20259 (*-linux-uclibc*): Add t-uclibc tmake_file.
20260 * config/t-musl: New.
20261 * config/t-uclibc: New.
20263 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
20265 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
20267 (gen_comm_data): Emit architectural setting of arch_prof.
20268 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
20270 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
20271 (armv8-m.base, armv8-m.main): Likewise.
20272 * arm-protos.h (arm_build_target): Add profile field.
20273 (arch_option): Likewise.
20274 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
20276 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
20277 arm_active_target.profile.
20279 2017-06-28 Richard Biener <rguenther@suse.de>
20281 PR middle-end/81227
20282 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
20283 TYPE_OVERFLOW_WRAPS.
20284 * match.pd (negate_expr_p): Likewise.
20285 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
20286 fold_build2, not fold_binary.
20288 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20290 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
20291 Convert memory address to Pmode.
20292 (aarch64_print_operand): Assert MEM operands are always Pmode.
20294 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
20297 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
20298 Remove redundant if.
20299 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
20300 * config/arm/aarch-common-protos.h
20301 (aarch_forward_to_shift_is_not_shifted_re): Remove.
20302 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
20304 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
20307 * multiple_target.c (create_dispatcher_calls): Set the default
20308 clone to be static, not public.
20310 2017-06-28 Richard Biener <rguenther@suse.de>
20312 * tree-vect-loop.c (vectorizable_reduction): Move special
20313 cond reduction IV var creation ...
20314 (vect_create_epilog_for_reduction): ... here. Remove induction_index
20315 parameter. Use STMT_VINFO_VECTYPE.
20316 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
20319 2017-06-28 Martin Liska <mliska@suse.cz>
20322 * ipa-visibility.c (non_local_p): Handle visibility.
20324 2017-06-28 Martin Liska <mliska@suse.cz>
20327 * common.opt: Add IntegerRange to various options.
20328 * opt-functions.awk (integer_range_info): New function.
20329 * optc-gen.awk: Add integer_range_info to cl_options struct.
20330 * opts-common.c (decode_cmdline_option): Handle
20331 CL_ERR_INT_RANGE_ARG.
20332 (cmdline_handle_error): Likewise.
20333 * opts.c (print_filtered_help): Show valid interval in
20334 when --help is provided.
20335 * opts.h (struct cl_option): Add range_min and range_max fields.
20336 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
20338 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
20340 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
20341 (x * C EQ/NE y * C): New transformation.
20343 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
20345 * genmultilib (combination_space): Accept '+' in option names.
20347 2017-06-28 Martin Liska <mliska@suse.cz>
20350 * asan.c (instrument_derefs): Bail out inner references
20351 that are hard register variables.
20353 2017-06-28 Jakub Jelinek <jakub@redhat.com>
20356 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
20357 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
20359 2017-06-28 Richard Biener <rguenther@suse.de>
20361 * tree-vectorizer.h (vect_get_vec_defs): Remove.
20362 (vect_get_slp_defs): Adjust.
20363 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
20365 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
20367 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
20368 get_initial_defs_for_reduction instead of vect_get_vec_defs.
20369 (vectorizable_reduction): Adjust.
20370 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
20372 (vect_get_slp_defs): Likewise.
20373 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
20374 (vectorizable_bswap): Adjust.
20375 (vectorizable_call): Likewise.
20376 (vectorizable_conversion): Likewise.
20377 (vectorizable_assignment): Likewise.
20378 (vectorizable_shift): Likewise.
20379 (vectorizable_operation): Likewise.
20380 (vectorizable_store): Likewise.
20381 (vectorizable_condition): Likewise.
20382 (vectorizable_comparison): Likewise.
20384 2017-06-28 Michael Collison <michael.collison@arm.com>
20387 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
20389 (arm_gen_movmemqi): Removed unused variable 'i'.
20390 Convert 'for' loop into 'while' loop.
20391 (arm_expand_prologue): Remove last unnecessary set of insn.
20392 (thumb_pop): Remove unused variable 'pushed_words'.
20393 (thumb_exit): Remove last unnecessary set of regs_to_pop.
20395 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20397 * config/s390/predicates.md: Use s390_rel_address_ok_p.
20398 * config/s390/s390-protos.h: Add prototype of
20399 s390_rel_address_ok_p.
20400 * config/s390/s390.c (s390_got_symbol): New function.
20401 (s390_rel_address_ok_p): New function.
20402 (legitimize_pic_address): Use s390_rel_address_ok_p.
20403 (s390_load_got): Use s390_got_symbol.
20404 (s390_option_override): Issue error if
20405 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
20406 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
20408 * config/s390/s390.opt: New option mpic-data-is-text-relative.
20410 2017-06-27 Andrew Pinski <apinski@cavium.com>
20412 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
20413 (X * copysign (1.0, X)): New pattern.
20414 (X * copysign (1.0, -X)): New pattern.
20415 (copysign (-1.0, CST)): New pattern.
20417 2017-06-27 Joseph Myers <joseph@codesourcery.com>
20419 * genmultilib (combination_space): Remove variable.
20420 Validate reuse rules against regular expression for any sequence
20421 of multilib options in any order.
20423 2017-06-27 Michael Collison <michael.collison@arm.com>
20425 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20426 call aarch64_split_simd_combine.
20427 * (aarch64_combine_internal<mode>): Delete pattern.
20428 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20429 Allow register and subreg operands.
20431 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20433 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20434 specific need, just fallback on defaults.
20435 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20437 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20438 Olivier Hainque <hainque@adacore.com>
20440 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20442 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20443 targets. Pick a default if no particular attempt applied.
20444 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20447 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20449 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20450 (x86_64-wrs-vxworks7): Likewise.
20452 2017-06-27 Marek Polacek <polacek@redhat.com>
20455 * ubsan.c (instrument_null): Check get_base_address's result for null.
20457 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20459 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20461 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20463 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20464 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20465 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20466 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20467 New function types.
20468 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20469 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20470 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20471 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20472 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20473 BUILT_IN_FEUPDATEENV): New builtins.
20474 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20475 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20476 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20477 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20479 (builtin_structptr_types): Adjust size.
20480 * tree.c (builtin_structptr_types): Add four entries.
20482 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20483 Olivier Hainque <hainque@adacore.com>
20485 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20486 (TLS_SYM): New local macro, forcing reference to __tls__ on
20487 link command lines for VxWorks 7 RTPs, triggering initialization
20489 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20490 OS features TLS support, true for RTPs on VxWorks 7.
20491 * config/vxworks.c (vxworks_override_options): Setup emutls
20494 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20496 * predict.c (test_prediction_value_range): Use -1U instead of -1
20497 to avoid narrowing conversion warning.
20498 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20499 to avoid narrowing conversion warning.
20500 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20502 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20504 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20506 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20507 64bit configurations.
20508 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20509 (SIZE_TYPE): Likewise.
20510 * config/vxworks.c (vxworks_emutls_var_fields): Use
20511 long_unsigned_type_node instead of unsigned_type_node as the offset
20512 field type, which is "pointer" mode in emutls.c.
20514 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20517 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20519 PR middle-end/81207
20520 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20521 gimple_vuse copying separately from gimple_vdef copying.
20523 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20525 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20528 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20529 Olivier Hainque <hainque@adacore.com>
20531 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20532 for all vxworks7 targets.
20533 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20534 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20535 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20536 variations for VX6/VX7 and 32/64bits later on in ...
20537 (VXWORKS_LIB_SPEC): Leverage new macros.
20538 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20539 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20541 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20543 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20544 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20546 2017-06-26 Carl Love <cel@us.ibm.com>
20548 * config/rs6000/rs6000-c.c: Add support for built-in functions
20549 vector bool char vec_reve (vector bool char);
20550 vector signed char vec_reve (vector signed char);
20551 vector unsigned char vec_reve (vector unsigned char);
20552 vector bool int vec_reve (vector bool int);
20553 vector signed int vec_reve (vector signed int);
20554 vector unsigned int vec_reve (vector unsigned int);
20555 vector bool long long vec_reve (vector bool long long);
20556 vector signed long long vec_reve (vector signed long long);
20557 vector unsigned long long vec_reve (vector unsigned long long);
20558 vector bool short vec_reve (vector bool short);
20559 vector signed short vec_reve (vector signed short);
20560 vector double vec_reve (vector double);
20561 vector float vec_reve (vector float);
20562 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20563 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20564 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20565 (altivec_vreve): New pattern.
20566 * config/rs6000/altivec.h (vec_reve): New define.
20567 * doc/extend.texi (vec_rev): Update the built-in documentation file
20568 for the new built-in functions.
20570 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20572 PR tree-optimization/71815
20573 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20575 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20577 (slsr_process_phi): Likewise.
20578 (replace_uncond_cands_and_profitable_phis): Don't replace a
20579 multiply candidate with a stride of 1 (copy or cast).
20580 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20582 (lowest_cost_path): Likewise.
20583 (total_savings): Likewise.
20585 2017-06-26 Richard Biener <rguenther@suse.de>
20588 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20589 Use def_builtin_pure for all gather builtins.
20591 2017-06-26 Richard Biener <rguenther@suse.de>
20593 PR tree-optimization/81203
20594 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20595 non-dominating BBs.
20597 2017-06-26 Marek Polacek <polacek@redhat.com>
20600 * doc/invoke.texi: Document -Wmultistatement-macros.
20602 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20604 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20605 arm_neon_ok_no_float_abi effective target.
20607 2017-06-26 Richard Biener <rguenther@suse.de>
20609 PR tree-optimization/80928
20610 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20611 (copy_bbs): Set BB_DUPLICATED flag early.
20612 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20614 (execute_on_shrinking_pred): Likewise.
20615 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20616 BB_DUPLICATED blocks.
20617 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20618 iterate over all PHIs considering removal of *gsi.
20620 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20622 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20625 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20627 * config/rs6000/rs6000-string.c: (expand_block_clear,
20628 do_load_for_compare, select_block_compare_mode,
20629 compute_current_alignment, expand_block_compare,
20630 expand_strncmp_align_check, expand_strn_compare,
20631 expand_block_move, rs6000_output_load_multiple)
20632 Move functions related to string/block move/compare
20633 to a separate file.
20634 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20635 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20636 for this function which is now used in two files.
20637 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20638 * config.gcc: Add rs6000-string.o to extra_objs for
20639 targets powerpc*-*-* and rs6000*-*-*.
20641 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20644 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20645 32-bit, since indexed is not valid for DImode.
20646 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20647 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20648 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20649 (define_peephole2 for Altivec d-form store): Likewise.
20652 * multiple_target.c (create_dispatcher_calls): Only create the
20653 dispatcher call if the function is the default clone of a
20654 versioned function.
20656 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20658 PR middle-end/80902
20659 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20660 a call, force the call to not be a tail call.
20662 2017-06-23 Jeff Law <law@redhat.com>
20664 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20667 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20669 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20670 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20671 handling for early expansion of vector shifts (sl,sr,sra,rl).
20672 (builtin_function_type): Add vector shift right instructions
20673 to the unsigned argument list.
20675 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20677 rtl-optimizatoin/79286
20678 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20679 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20680 trap. PIC register plus a const unspec without offset can never trap.
20682 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20684 * tree.h (builtin_structptr_type): New type.
20685 (builtin_structptr_types): Declare new array.
20686 * tree.c (builtin_structptr_types): New array.
20687 (free_lang_data, build_common_tree_nodes): Use it.
20689 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20692 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20695 2017-06-22 Matt Turner <mattst88@gmail.com>
20697 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20698 Lake models to skylake case. Assume skylake for unknown
20699 models with clflushopt.
20701 2017-06-22 Jeff Law <law@redhat.com>
20703 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20704 frame sizes that do not satisfy aarch64_uimm12_shift.
20706 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20708 * profile-count.h (apply_probability,
20709 apply_scale, probability_in): Fix checks for zero.
20711 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20713 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20714 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20716 2016-06-22 Richard Biener <rguenther@suse.de>
20718 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20719 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20720 REDUC_MAX_EXPR support.
20721 (vectorizable_reduction): Likewise.
20722 (vect_create_epilog_for_reduction): Likewise.
20724 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20726 * match.pd (A / (1 << B) -> A >> B): New.
20727 * generic-match-head.c: Include optabs-tree.h.
20728 * gimple-match-head.c: Likewise.
20729 * optabs-tree.h (target_supports_op_p): New.
20730 * optabs-tree.c (target_supports_op_p): New.
20732 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20734 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20735 $gcc_cv_ld --help output.
20736 (gcc_cv_ld_demangle): Likewise.
20737 (gcc_cv_ld_eh_frame_hdr): Likewise.
20738 (gcc_cv_ld_pie): Likewise.
20739 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20740 (gcc_cv_ld_buildid): Likewise.
20741 (gcc_cv_ld_sysroot): Likewise.
20742 (ld_bndplt_support): Likewise.
20743 (ld_pushpopstate_support): Likewise.
20744 * configure: Regenerate.
20745 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20747 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20750 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20751 operands indexes to avoid gap between operands and match_dups.
20753 2017-06-21 Andrew Pinski <apinski@cavium.com>
20755 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20756 Increment Arith_shift and Arith_shift_reg by 1.
20757 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20759 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20760 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20761 (aarch64_strip_extend): Add new argument and test for it.
20762 (aarch64_cheap_mult_shift_p): New function.
20763 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20764 add a cost if it is true.
20765 Update calls to aarch64_strip_extend.
20766 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20768 2017-06-21 Andrew Pinski <apinski@cavium.com>
20770 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20772 (thunderxt88): Likewise.
20773 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20774 (thunderx_prefetch_tune): New variable.
20775 (thunderx2t99_prefetch_tune): Update for the correct values.
20776 (thunderxt88_tunings): New variable.
20777 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20778 generic_prefetch_tune.
20779 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20781 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20783 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20784 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20785 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20786 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20787 (aarch64_atomic_cas<mode>, GPI): Likewise.
20789 2017-06-21 Martin Liska <mliska@suse.cz>
20791 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20792 statements on cold and hot labels.
20793 * predict.c (tree_estimate_probability_bb): Remove the
20794 prediction from this place.
20796 2017-06-21 Martin Liska <mliska@suse.cz>
20798 PR tree-optimization/79489
20799 * gimplify.c (maybe_add_early_return_predict_stmt): New
20801 (gimplify_return_expr): Call the function.
20802 * predict.c (tree_estimate_probability_bb): Remove handling
20804 * predict.def: Update comment about early return predictor.
20805 * gimple-predict.h (is_gimple_predict): New function.
20806 * predict.def: Change default value of early return to 66.
20807 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20809 * passes.def: Put pass_strip_predict_hints to the beginning of
20812 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20814 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20815 FUNCTION_DECL declarations.
20816 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20818 (dwaf2out_decl): Likewise.
20819 * godump.c (go_early_global_decl): Skip call to the real debug hook
20820 for FUNCTION_DECL declarations.
20821 * passes.c (rest_of_decl_compilation): Skip call to the
20822 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20823 -fdump-go-spec is passed.
20825 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20827 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20828 Reorder for compactness.
20829 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20830 (def_builtin_pure, def_builtin_pure2): New functions.
20831 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20833 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20835 * match.pd (nop_convert): New predicate.
20836 ((A +- CST1) +- CST2): Allow some NOP conversions.
20838 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20841 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20842 with ctors/dtors if GOVD_SHARED is set.
20844 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20846 * config/aarch64/aarch64.md (movti_aarch64):
20847 Emit mov rather than orr.
20848 (movtf_aarch64): Likewise.
20849 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20850 Emit mov rather than orr.
20852 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20854 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20855 Swap alternatives, make integer dup more expensive.
20857 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20859 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20860 Return true for non-tls symbols.
20862 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20864 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20865 (cortex-a75): Likewise.
20866 (cortex-a75.cortex-a55): Likewise.
20867 * config/aarch64/aarch64-tune.md: Regenerate.
20868 * doc/invoke.texi (-mtune): Document new values for -mtune.
20870 2017-06-21 Tom de Vries <tom@codesourcery.com>
20872 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20873 stack_size feature.
20874 (Effective-Target Keywords, Other attributes): Suggest using
20875 dg-add-options stack_size feature to get stack limit in stack_size
20876 effective target documentation.
20878 2017-06-21 Julian Brown <julian@codesourcery.com>
20879 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20881 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20882 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20883 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20885 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20886 attribute type list for neon_multiply.
20887 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20888 attribute type list for neon_multiply.
20889 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20890 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20891 attribute type list for neon_multiply.
20892 * config/arm/types.md (crypto_pmull): Add.
20893 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20894 attribute type list.
20896 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20898 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20901 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20903 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20904 to make sure not to dereference a NULL cost_classes_ptr pointer.
20906 2017-06-20 Carl Love <cel@us.ibm.com>
20908 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20909 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20910 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20911 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20912 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20913 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20914 VMULOSW): New enum "unspec" values.
20915 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20916 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20917 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20918 altivec_vmulosw): New patterns.
20919 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20920 VMULOSW): Add definitions.
20922 2017-06-20 Julia Koval <julia.koval@intel.com>
20924 * config/i386/i386.c: Fix rounding expand for new pattern.
20925 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20927 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20929 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20930 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20932 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20934 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20937 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20939 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20940 architecture, then by implementer ID.
20941 * config/aarch64/aarch64-tune.md: Regenerate.
20943 2017-06-20 Richard Biener <rguenther@suse.de>
20945 PR middle-end/81097
20946 * fold-const.c (split_tree): Fold to type before negating.
20948 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20950 * diagnostic-show-locus.c
20951 (selftest::test_fixit_deletion_affecting_newline): New function.
20952 (selftest::diagnostic_show_locus_c_tests): Call it.
20954 2017-06-20 Andreas Schwab <schwab@suse.de>
20957 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20960 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20962 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20963 __ARM_FEATURE_COPROC according to support.
20965 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20967 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20968 Rewritten to avoid overflow for > 32-bit pointers.
20971 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20972 by removing enum keyword.
20973 (ubsan_type_descriptor): Likewise. Formatting fix.
20976 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20977 splitter): Require TARGET_SSE2 in the condition.
20979 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20982 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20983 for doing vector set of SFmode on ISA 3.0.
20984 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20985 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20987 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20988 SFmode value into a V4SF variable that was extracted from another
20989 V4SF variable without converting the element to double precision
20990 and back to single precision vector format.
20991 (vsx_insert_extract_v4sf_p9_2): Likewise.
20993 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20995 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20996 in UWHI to avoid undefined overflow.
20999 * ubsan.h (enum ubsan_encode_value_phase): New.
21000 (ubsan_encode_value): Change second argument to
21001 enum ubsan_encode_value_phase with default value of
21002 UBSAN_ENCODE_VALUE_GENERIC.
21003 * ubsan.c (ubsan_encode_value): Change second argument to
21004 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
21005 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
21006 create_tmp_var_raw instead of create_tmp_var and use a
21008 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
21009 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
21010 ubsan_encode_value callers.
21013 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
21014 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
21015 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
21017 2017-06-19 Richard Biener <rguenther@suse.de>
21019 PR middle-end/81118
21020 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
21021 estimates if we changed anything.
21023 2017-06-19 Richard Biener <rguenther@suse.de>
21025 PR tree-optimization/80887
21026 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
21027 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
21028 simplified lookups, then reset mprts_hook.
21029 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
21031 (try_to_simplify): Likewise.
21033 2017-06-19 Martin Liska <mliska@suse.cz>
21036 * gimplify.c (gimplify_switch_expr):
21037 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
21039 2017-06-19 Martin Liska <mliska@suse.cz>
21041 * doc/install.texi: Document that PGO runs in 4 stages.
21043 2017-06-19 Martin Liska <mliska@suse.cz>
21046 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
21047 to dispatcher function name.
21048 * multiple_target.c (replace_function_decl): New function.
21049 (create_dispatcher_calls): Redirect both edges and references.
21051 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
21053 * profile-count.c (profile_count::dump): Dump quality.
21054 (profile_count::differs_from_p): Update for unsigned val.
21055 * profile-count.h (profile_count_quality): New enum.
21056 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
21058 2017-06-19 Richard Biener <rguenther@suse.de>
21060 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
21061 struct function as arg.
21062 (estimate_numbers_of_iterations): Export overload with loop arg.
21063 (free_numbers_of_iterations_estimates_loop): Use an overload of
21064 free_numbers_of_iterations_estimates instead.
21065 * tree-cfg.c (remove_bb): Adjust.
21066 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
21067 * tree-parloops.c (gen_parallel_loop): Likewise.
21068 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
21070 (tree_unroll_loops_completely): Likewise.
21071 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
21072 Use an overload instead and export.
21073 (estimated_loop_iterations): Adjust.
21074 (max_loop_iterations): Likewise.
21075 (likely_max_loop_iterations): Likewise.
21076 (estimate_numbers_of_iterations): Take struct function as arg
21078 (loop_exits_before_overflow): Adjust.
21079 (free_numbers_of_iterations_estimates_loop): Use an overload.
21080 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
21081 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
21083 2017-06-19 Richard Biener <rguenther@suse.de>
21086 * ipa-prop.c (find_constructor_constant_at_offset): Handle
21087 RANGE_EXPR conservatively.
21089 2017-06-16 Carl Love <cel@us.ibm.com>
21091 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21092 definitions for vec_float, vec_float2, vec_floato,
21093 vec_floate built-ins.
21094 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
21095 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
21097 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
21098 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
21099 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
21100 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
21101 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
21102 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
21103 vec_floato): Add builtin defines.
21104 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
21105 Update the built-in documentation file for the new built-in
21108 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21110 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
21111 (mthumb): Mark as the negative of -marm.
21113 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21115 * doc/invoke.texi (ARM Options, -mcpu): Document supported
21117 (ARM Options, -mtune): Document that this accepts the same
21118 extension options as -mcpu.
21119 (ARM Options, -mfpu): Document addition of -mfpu=auto.
21121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21123 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
21124 permitted extensions.
21126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21128 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
21129 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
21130 (armv8-m.main): Add option +nodsp.
21131 * config/arm/arm-cpu-cdata.h: Regenerated.
21133 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21135 * config/arm/t-fuchsia: New file.
21136 * config.gcc (arm*-*-fuchsia*): Use it.
21138 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21140 * config/arm/t-symbian: Rewrite for new option infrastructure.
21142 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21144 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
21145 (MULTILIB_REQUIRED): Likewise.
21147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21149 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
21150 (MULTILIB_RESUE): Likewise.
21151 (MULTILIB_MATCHES): Likewise.
21152 (MULTLIB_REQUIRED): Likewise.
21154 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21156 * config/arm/t-rtems: Rewrite for new option framework.
21158 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21160 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
21161 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
21162 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
21163 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
21164 * config/arm/t-multilib: ... here.
21165 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
21166 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
21167 armv7-a and armv8*-a when A-profile libraries have not been built.
21168 * config/arm/t-rmprofile: Rewrite.
21170 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21172 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
21173 with a backslash. Remove the backslash after substituting unescaped
21175 * doc/fragments.texi (MULTILIB_REUSE): Document it.
21177 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21179 * config.gcc: (arm*-*-*): When building a-profile libraries, force
21180 the driver to pass through the default setting of -mfloat-abi.
21181 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
21183 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
21184 (all_feat_combs): New rule.
21185 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
21187 * config/arm/t-aprofile: Rewrite.
21189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21191 * config/arm/arm.h (FPUTYPE_AUTO): Define.
21192 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
21193 fpu is not specified by the user/command-line.
21194 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
21195 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
21196 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
21197 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
21198 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
21199 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
21201 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21203 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
21204 * config/arm/t-arm-elf: Rewritten.
21206 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21208 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
21209 have some floating-point instructions.
21210 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
21211 (TARGET_MAYBE_HARD_FLOAT): New macro.
21212 * config/arm/arm-builtins.c (arm_init_builtins): Use
21213 TARGET_MAYBE_HARD_FLOAT.
21214 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
21216 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21218 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
21219 (configargs.h): Include it.
21220 (arm_print_hint_for_fpu_option): New function.
21221 (arm_parse_fpu_option): New function.
21222 (candidate_extension): New class.
21223 (arm_canon_for_multilib): New function.
21224 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
21225 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
21226 (ARCH_CANONICAL_SPECS): New macro.
21227 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
21229 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21231 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
21232 are set after handling multilib fragments. Set target_cpu_default2
21235 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21237 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
21239 (arm*-*-*): Set target_cpu_default2 to a quoted string.
21240 * config/arm/parsecpu.awk (check_cpu): Validate any extension
21242 (check_arch): Likewise.
21243 * config/arm/arm.c (arm_configure_build_target): Handle
21244 TARGET_CPU_DEFAULT being a string constant. Scan any feature
21245 options in the default.
21247 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21249 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
21250 when an option is an alias of another.
21251 * config/arm/parsecpu.awk (optalias): New parser token.
21252 (gen_comm_data): Mark non-alias options as such. Emit entries
21253 for extension aliases.
21254 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
21255 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
21256 (armv6kz, armv6zk, armv6t2): Likewise.
21257 (armv7): Make vfpv3-d16 an alias.
21258 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
21260 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
21261 Sort in canonical order.
21262 (armv8-a): Sort in canonical order.
21263 (armv8.1-a, armv8.2-a): Likewise.
21264 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
21266 (cortex-a9): Sort in canonical order.
21267 * config/arm/arm.c (selftests.h): Include it.
21268 (arm_test_cpu_arch_data): New function.
21269 (arm_run_self_tests): New function.
21270 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
21271 (targetm): Move declaration to the end of the file.
21272 * arm-cpu-cdata.h: Regenerated.
21274 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21276 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
21277 call to target_mode_check describing the type of option passed.
21278 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
21279 (arm_target_thumb_only): Use arm_parse_arch_option_name or
21280 arm_parse_cpu_option_name to match parameters against list of
21282 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
21283 arm_arch_core_flags data structure.
21284 * config/arm/arm-cpu_cdata.h: Regenerated.
21286 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21288 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
21290 (arm_print_hint_for_cpu_option): Likewise.
21291 (arm_print_hint_for_arch_option): Likewise.
21292 (arm_parse_cpu_option_name): Likewise.
21293 (arm_parse_arch_option_name): Likewise.
21294 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
21295 of entries in the all_fpus list.
21296 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
21297 (arm_parse_cpu_option_name): Declare.
21298 (arm_parse_arch_option_name): Declare.
21299 (arm_parse_option_features): Declare.
21300 (arm_intialize_isa): Declare.
21301 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
21303 (gen_comm_data): ... here. Make definitions non-static.
21304 * config/arm/arm-cpu-data.h: Regenerated.
21305 * config/arm/arm-cpu-cdata.h: Regenerated.
21307 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21309 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
21310 (cpu_arch_extension): New structure.
21311 (cpu_arch_option, arch_option, cpu_option): New structures.
21312 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
21313 architecture types.
21314 (gen_data): Generate new format data tables.
21315 * config/arm/arm.c (cpu_tune): New structure.
21316 (cpu_option, processors): Delete.
21317 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
21318 (arm_print_hint_for_cpu_option): ... this and ...
21319 (arm_print_hint_for_arch_option): ... this.
21320 (arm_parse_arch_cpu_name): Delete. Replace with ...
21321 (arm_parse_cpu_option_name): ... this and ...
21322 (arm_parse_arch_option_name): ... this.
21323 (arm_unrecognized_feature): Change type of target parameter to
21325 (arm_parse_arch_cpu_features): Delete. Replace with ...
21326 (arm_parse_option_features): ... this.
21327 (arm_configure_build_target): Rework to use new configuration data
21329 (arm_print_tune_info): Rework for new configuration data tables.
21330 * config/arm/arm-cpu-data.h: Regenerated.
21331 * config/arm/arm-cpu.h: Regenerated.
21333 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21335 * Makefile.in (OBJS): Move sbitmap.o from here ...
21336 (OBJS-libcommon): ... to here.
21338 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21340 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
21341 (ISA_ALL_CRYPTO): New macro.
21342 (ISA_ALL_SIMD): New macro
21343 (ISA_ALL_FP): New macro.
21344 * config/arm/arm.c (fpu_bitlist): Update initializer.
21345 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
21347 (arm9e): Add fpu. Add option for nofp
21348 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
21349 (arm926ej-s, arm1026ej-s): Likewise.
21350 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
21351 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
21352 neon-fp16, neon-vfpv4, nofp and nosimd.
21353 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
21354 (cortex-a8): Add fpu. Add option for nofp.
21355 (cortex-a9): Add fpu. Add options for nosimd and nofp.
21356 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
21357 (cortex-r4f): Add fpu.
21358 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
21359 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
21361 (cortex-r8): Likewise.
21362 (cortex-m4): Add fpu. Add option for nofp.
21363 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
21364 (cortex-a17.cortex-a7): Likewise.
21365 (cortex-a32): Add fpu. Add options for crypto and nofp.
21366 (cortex-a35, cortex-a53): Likewise.
21367 (cortex-a57): Add fpu. Add option for crypto.
21368 (cortex-a72, cortex-a73): Likewise.
21369 (exynos-m1): Likewise.
21370 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
21371 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
21372 (cortex-m33): Add fpu. Add option for nofp.
21373 * config/arm/arm-cpu-cdata.h: Regenerated
21374 * config/arm/arm-cpu-data.h: Regenerated.
21376 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21378 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
21379 (armv5te, armv5tej): Likewise.
21380 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
21381 (armv7): Add options fp and vfpv3-d16.
21382 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
21383 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
21385 (armv7ve): Likewise.
21386 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
21387 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
21388 (armv8-a): Add nocrypto option.
21389 (armv8.1-a, armv8.2-a): Likewise.
21390 (armv8-m.main): add options fp, fp.dp and nofp.
21392 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21394 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
21396 (armv8-a+crc): Delete.
21397 (armv8.1-a): Add options simd, crypto and nofp.
21398 (armv8.2-a): Add options fp16, simd, crypto and nofp.
21399 (armv8.2-a+fp16): Delete.
21400 (armv8-m.main): Add option dsp.
21401 (armv8-m.main+dsp): Delete.
21402 (cortex-a8): Add fpu. Add nofp option.
21403 (cortex-a9): Add fpu. Add nofp and nosimd options.
21404 * config/arm/parsecpu.awk (gen_data): Generate option tables and
21405 link to main cpu and architecture data structures.
21406 (gen_comm_data): Only put isa attributes from the main architecture
21408 (option): New statement for architecture and CPU entries.
21409 * arm.c (struct cpu_option): New structure.
21410 (struct processors): Add entry for options.
21411 (arm_unrecognized_feature): New function.
21412 (arm_parse_arch_cpu_name): Ignore any characters after the first
21414 (arm_parse_arch_cpu_feature): New function.
21415 (arm_configure_build_target): Separate out any CPU and architecture
21416 features and parse separately. Don't error out if -mfpu=auto is
21417 used with only an architecture string.
21418 (arm_print_asm_arch_directives): New function.
21419 (arm_file_start): Call it.
21420 * config/arm/arm-cpu-cdata.h: Regenerated.
21421 * config/arm/arm-cpu-data.h: Likewise.
21422 * config/arm/arm-tables.opt: Likewise.
21424 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21426 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21427 assembler when it is not -mfpu=auto.
21429 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21431 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21432 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21433 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21434 (ASM_CPU_SPEC): Rewrite.
21435 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21436 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21437 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21438 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21439 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21440 copied string is NUL-terminated. Also strip any characters prefixed
21442 (arm_rewrite_selected_arch): New function.
21443 (arm_rewrite_march): New function.
21445 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21447 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21448 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21449 (march, mcpu, mtune): Convert to string-based options.
21450 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21451 (arm_parse_arch_cpu_name): New function.
21452 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21453 identify selected architecture or CPU.
21454 (arm_option_save): New function.
21455 (TARGET_OPTION_SAVE): Redefine.
21456 (arm_option_restore): Restore string options.
21457 (arm_option_print): Print string options.
21459 2017-06-16 Martin Sebor <msebor@redhat.com>
21461 PR tree-optimization/80933
21462 PR tree-optimization/80934
21463 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21464 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21465 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21466 (gimple_fold_builtin): Call them.
21468 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21470 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21471 as unlikely; update profile.
21473 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21475 * predict.c (force_edge_cold): Handle declaring edges impossible
21478 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21480 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21482 (try_unroll_loop_completely): Fix reporting.
21484 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21486 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21488 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21491 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21492 if given a non-constant argument for an intrinsic which requires a
21495 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21497 * profile.c (compare_freqs): New function.
21498 (branch_prob): Sort edge list.
21499 (find_spanning_tree): Assume that the list is priority sorted.
21501 2017-06-16 Richard Biener <rguenther@suse.de>
21503 PR tree-optimization/81090
21504 * passes.def (pass_record_bounds): Remove.
21505 * tree-pass.h (make_pass_record_bounds): Likewise.
21506 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21507 make_pass_record_bounds): Likewise.
21508 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21509 not free niter estimates at the beginning but at the end.
21510 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21512 2017-06-16 Richard Biener <rguenther@suse.de>
21514 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21515 initializer to workaround ICE in host GCC 4.8.
21517 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21519 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21521 (clone_inlined_nodes): Update.
21523 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21525 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21526 prefetch settings, and enable prefetching by default at -O3.
21528 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21530 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21531 Set flag_prefetch_loop_arrays according to tuning data.
21533 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21535 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21536 New tune structure.
21537 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21538 [Unrelated to main purpose of the patch] Place the pointer field last
21539 to enable type checking errors when tune structure are wrongly merged.
21540 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21541 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21542 (thunderx2t99_prefetch_tune): New tune constants.
21543 (tune_params *_tunings): Update all tunings (no functional change).
21544 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21545 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21546 from tunings structures.
21548 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21551 * ubsan.c (instrument_null): Add T argument, use it instead
21552 of computing it based on IS_LHS.
21553 (instrument_object_size): Likewise.
21554 (pass_ubsan::execute): Adjust instrument_null and
21555 instrument_object_size callers to pass gimple_get_lhs or
21556 gimple_assign_rhs1 result to it. Use instrument_null instead of
21557 calling get_base_address and instrument_mem_ref. Handle
21558 aggregate call arguments for object-size sanitization.
21560 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21562 PR tree-optimization/81089
21563 * tree-vrp.c (is_masked_range_test): Validate operands of
21566 2017-06-15 Martin Sebor <msebor@redhat.com>
21569 * dumpfile.c (dump_register): Avoid calling memset to initialize
21570 a class with a default ctor.
21571 * gcc.c (struct compiler): Remove const qualification.
21572 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21573 * hash-table.h: Ditto.
21574 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21576 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21577 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21579 * params.h (struct param_info): Make struct members non-const.
21580 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21581 with default initialization.
21582 * vec.h (vec_copy_construct, vec_default_construct): New helper
21584 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21585 with vec_copy_construct.
21586 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21587 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21588 * doc/invoke.texi (-Wclass-memaccess): Document.
21590 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21592 * emit-rtl.h (is_leaf): Update comment about local
21593 register allocator.
21595 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21598 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21599 for a variable to have a section before checking if the section has a
21601 Set section to.persistent if persistent attribute is set.
21602 Warn if .persistent attribute is used on an automatic variable.
21604 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21606 PR rtl-optimization/80474
21607 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21609 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21611 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21614 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21616 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21617 (casesi): Emit bounds checking as RTL.
21618 (casesi_internal_mips16_<mode>): Remove bounds checking.
21620 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21622 * config/xtensa/xtensa.c (xtensa_option_override): Append
21623 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21624 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21625 xtensa_doloop_hooks): Define unconditionally.
21626 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21628 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21629 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21630 for it in xtensa_option_override.
21631 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21632 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21634 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21636 * doc/cppopts.texi: Document '-' special value to -MF.
21638 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21640 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21641 (cortex_a53_fconst): Likewise.
21642 (cortex_a53_fpmul): Likewise.
21643 (cortex_a53_f_load_64): Likewise.
21644 (cortex_a53_f_load_many): Likewise.
21645 (cortex_a53_advsimd_alu): Likewise.
21646 (cortex_a53_advsimd_alu_q): Likewise.
21647 (cortex_a53_advsimd_mul): Likewise.
21648 (cortex_a53_advsimd_mul_q): Likewise.
21649 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21650 Add missing fmul, r2f_cvt and fconst cases.
21652 2017-06-14 Richard Biener <rguenther@suse.de>
21654 PR middle-end/81088
21655 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21657 (fold_binary_loc): When associating do not treat pre-existing
21658 TREE_OVERFLOW on literal constants as a reason to allow
21659 TREE_OVERFLOW on associated literal constants.
21661 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21663 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21664 (MASK_FEATURES): New macro.
21665 * config/sparc/sparc.c (sparc_option_override): Remove the special
21666 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21668 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21670 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21671 a division of 0 if non-call exceptions are enabled.
21673 2017-06-14 Andrew Pinski <apinski@cavium.com>
21674 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21677 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21678 Improve vector initialization code gen for only variable case.
21680 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21682 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21684 2017-06-14 Richard Biener <rguenther@suse.de>
21686 PR tree-optimization/81083
21687 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21690 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21692 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21693 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21694 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21695 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21696 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21697 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21699 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21701 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21702 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21704 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21706 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21708 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21710 * config/rs6000/t-rtems: Don't handle SPE.
21712 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21714 * config/rs6000/t-linux: Don't handle SPE.
21716 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21718 * config/rs6000/eabispe.h: Delete file.
21720 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21722 * config/rs6000/t-spe: Delete file.
21724 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21726 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21727 (rs6000_legitimate_offset_address_p): Return false for anything in
21728 V2SImode or V2SFmode.
21730 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21732 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21733 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21734 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21736 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21737 (rs6000_legitimate_offset_address_p): Ditto.
21738 (rs6000_emit_move): Ditto.
21739 (rs6000_init_builtins): Remove V4HI_type_node.
21741 2017-06-13 Martin Liska <mliska@suse.cz>
21744 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21745 (gate_asan): Likewise.
21746 * asan.h (asan_no_sanitize_address_p): Remove the function.
21747 (sanitize_flags_p): New function.
21748 * builtins.def: Fix coding style.
21749 * common.opt: Use renamed enum value.
21750 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21751 * doc/extend.texi: Document no_sanitize attribute.
21752 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21753 to SANITIZE_UNDEFINED_NONDEFAULT.
21754 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21755 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21756 Use sanitize_flags_p.
21757 * gimplify.c (gimplify_function_tree): Likewise.
21758 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21759 * opts.c (parse_no_sanitize_attribute): New function.
21760 (common_handle_option): Use renamed enum value.
21761 * opts.h (parse_no_sanitize_attribute): Declare.
21762 * tree.c (sanitize_flags_p): New function.
21763 * tree.h: Declared here.
21764 * tsan.c: Use sanitize_flags_p.
21765 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21766 (instrument_mem_ref): Likewise.
21767 (instrument_bool_enum_load): Likewise.
21768 (do_ubsan_in_current_function): Remove the function.
21769 (pass_ubsan::execute): Use sanitize_flags_p.
21770 * ubsan.h: Remove do_ubsan_in_current_function
21771 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21772 (dump_function_to_file): Use it here.
21774 2017-06-13 Martin Jambor <mjambor@suse.cz>
21776 PR tree-optimization/80803
21777 PR tree-optimization/81063
21778 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21779 (propagate_subaccesses_across_link): Enqueue subtree whenever
21780 necessary instead of relying on the caller.
21782 2017-06-13 Martin Jambor <mjambor@suse.cz>
21784 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21785 that have a first_link.
21786 (sort_and_splice_var_accesses): Do not check first_link before
21788 (subtree_mark_written_and_enqueue): Likewise.
21789 (propagate_all_subaccesses): Likewise and do not stop at first
21790 parent with a first_link.
21792 2017-06-13 Martin Jambor <mjambor@suse.cz>
21794 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21797 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21799 * match.pd: New pattern.
21801 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21803 * tree-vrp.c (is_masked_range_test): New function.
21804 (register_edge_assert_for): Determine ranges for
21807 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21809 PR tree-optimization/67328
21810 * fold-const.c (maskable_range_p): New function.
21811 (build_range_check): Generate bittests if possible.
21813 2017-06-13 Martin Liska <mliska@suse.cz>
21815 * gimple-pretty-print.c (dump_probability): Add new argument.
21816 (dump_edge_probability): Dump both probability and count.
21817 (dump_gimple_label): Likewise.
21818 (dump_gimple_bb_header): Likewise.
21820 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21823 * config/avr/avr-devices.c: Fix indentation.
21824 * config/avr/gen-avr-mmcu-specs.c: Dito.
21826 2017-06-13 Richard Biener <rguenther@suse.de>
21828 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21829 instead get vector type from stmt_info.
21830 (vectorizable_reduction): Adjust. Remove dead code.
21832 2017-06-13 Richard Biener <rguenther@suse.de>
21834 PR middle-end/81065
21835 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21836 case of C * (x * C2 + C3).
21837 (fold_addr_of_array_ref_difference): Properly fold index difference.
21839 2017-06-12 David S. Miller <davem@davemloft.net>
21842 * config/sparc/sparc.md (return expander): Emit frame blockage if
21843 function uses alloca.
21845 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21847 * combine.c (make_field_assignment): Check len rather than the mode
21848 precision when calling force_to_mode.
21850 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21852 Support multilibs and devices that see flash in RAM address range.
21855 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21856 (avr_mcu_t) <flash_pm_offset>: New field.
21857 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21858 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21859 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21860 (AVR_TINY_PM_OFFSET): Remove macro.
21861 * config/avr/avr.opt (-mshort-calls): New option.
21862 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21863 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21864 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21865 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21866 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21867 instead of avr_arch->have_jmp_call.
21868 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21869 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21870 avr_arch->flash_pm_offset to define.
21871 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21872 new field flash_pm_offset. Add entry for avrxmega3.
21873 (avr_texinfo): Add entry for avrxmega3.
21874 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21875 attiny212, attiny214,
21876 attiny412, attiny414, attiny416, attiny417,
21877 attiny814, attiny816, attiny817,
21878 attiny1614, attiny1616, attiny1617,
21879 attiny3214, attiny3216, attiny3217.
21880 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21881 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21882 (avr_print_operand_address) [AVR_TINY]: Same.
21883 (avr_asm_init_sections) <readonly_data_section>: Only patch
21884 callback if avr_arch->flash_pm_offset = 0.
21885 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21886 for rodata if avr_arch->flash_pm_offset != 0.
21887 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21888 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21889 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21890 (m_options): Append opt_rcall.
21891 (m_dirnames): Append dir_rcall.
21892 * config/avr/t-multilib: Regenerate.
21894 * configure.ac [target=avr]: Check whether avrxmega3 default
21895 linker description file works as needed.
21896 * configure: Regenerate.
21897 * doc/avr-mmcu.texi: Regenerate.
21898 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21899 <__AVR_ARCH__>: Document avrxmega3 and 103.
21900 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21901 <__AVR_SHORT_CALLS__>: Document it.
21902 <__AVR_PM_BASE_ADDRESS__>: Document it.
21903 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21904 (AVR Variable Attributes) <progmem>: Document this is
21905 not needed for avrxmega3.
21906 (AVR Named Address Spaces) <__flash>: Dito.
21908 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21910 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21912 2017-06-12 Doug Rupp <rupp@adacore.com>
21914 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21915 Append vxworks-stdint.h to the tm_file list.
21916 * config/vxworks-stdint.h: New file.
21918 2017-06-12 Martin Liska <mliska@suse.cz>
21920 PR tree-optimization/81041
21921 * tree-profile.c (gimple_gen_ic_func_profiler):
21922 Create an extra BB in profile-generate
21923 (gimple_gen_time_profiler): Likewise.
21925 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21927 PR tree-optimization/81003
21928 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21929 (update_range_test): Use it instead of force_gimple_operand_gsi.
21931 2017-06-12 Richard Biener <rguenther@suse.de>
21933 PR tree-optimization/81053
21934 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21935 with backedge value not defined in loop. Simplify def stmt
21938 2017-06-11 Tom de Vries <tom@codesourcery.com>
21941 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21943 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21944 nvptx_cannot_force_const_mem.
21946 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21948 * opts.c (finish_options): Move test for flag_split_stack after
21949 it has been initialized.
21951 2017-06-11 Jason Merrill <jason@redhat.com>
21953 * tree.h (id_equal): New.
21954 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21955 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21956 instead of strcmp of IDENTIFIER_POINTER.
21958 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21960 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21961 (mark_all_inlined_calls_cdtor): Fix formating.
21962 (inline_transform): Rescale profile before inlining.
21964 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21966 * cgraph.h (cgraph_edge::clone): Update prototype.
21967 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21968 (cgraph_node::create_clone): Update.
21969 (cgraph_node::create_version_clone): Update.
21970 * tree-inline.c (copy_bb): Update.
21971 (expand_call_inline): Update.
21973 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21975 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21976 factored out from ...
21977 (rs6000_emit_prologue): ... here.
21979 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21981 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21982 factored out from ...
21983 (rs6000_emit_prologue): ... here.
21985 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21987 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21989 (handle_missing_profiles): Fix computation of tp_first_run.
21990 (counts_to_freqs): Do not touch freqs when count is 0.
21992 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21994 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21997 2017-06-10 Tom de Vries <tom@codesourcery.com>
21999 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
22000 attributes): Document signal effective target.
22002 2017-06-10 Tom de Vries <tom@codesourcery.com>
22004 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22005 Document effective target stack_size.
22007 2017-06-09 David Malcolm <dmalcolm@redhat.com>
22009 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
22010 to the edit_context if they can be auto-applied.
22012 2017-06-9 Ian Lance Taylor <iant@golang.org>
22014 * opts.c (finish_options): If -fsplit-stack, disable implicit
22015 -forder-blocks-and-partition.
22016 * doc/invoke.texi (Optimize Options): Document that when using
22017 -fsplit-stack -forder-blocks-and-partition is not implicitly
22020 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22022 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
22023 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
22024 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
22025 * builtins.def (abort, trap, unreachable): Declare cold.
22026 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
22027 * tree-core.h (ECF_COLD): New.
22028 * tree.c (set_call_expr_flags): Handle ECF_COLD.
22029 (build_common_builtin_nodes): Mark unreachable and abort as cold.
22031 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22033 * predict.c (unlikely_executed_stmt_p): Cleanup.
22035 2017-06-09 Richard Biener <rguenther@suse.de>
22037 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
22038 model if the ref is always written to.
22040 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22042 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
22044 2017-06-09 Tamar Christina <tamar.christina@arm.com>
22046 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
22049 2017-06-09 Tom de Vries <tom@codesourcery.com>
22052 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
22053 "target cannot support label values" when encountering LABEL_REF.
22055 2017-06-09 Martin Liska <mliska@suse.cz>
22057 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
22058 (gimple_gen_ic_func_profiler): Emit direct comparison
22059 of __gcov_indirect_call_callee with NULL.
22060 (gimple_gen_time_profiler): Change probability from
22061 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
22063 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22065 * profile.c (edge_gcov_counts): Turn to pointer.
22066 (compute_branch_probabilities, compute_branch_probabilities): Update.
22067 (branch_prob): Do not clear edge_gcov_count.
22068 * profile.h (edge_gcov_counts): Turn to pointer.
22069 (edge_gcov_count): Update.
22071 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
22073 * gimple.h (gimple_check_failed): Mark cold.
22075 2017-06-09 Richard Biener <rguenther@suse.de>
22077 PR tree-optimization/66623
22078 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
22079 refactor check_reduction into two parts, properly computing
22080 whether we have to check reduction validity for outer loop
22083 2017-06-09 Richard Biener <rguenther@suse.de>
22085 PR tree-optimization/79483
22086 * graphite-scop-detection.c (order): New global.
22087 (get_order): Compute bb to order mapping that satisfies code
22088 generation constraints.
22089 (cmp_pbbs): New helper.
22090 (build_scops): Start domwalk at entry block, sort generated
22093 2017-06-09 Richard Biener <rguenther@suse.de>
22095 PR middle-end/81007
22096 * ipa-polymorphic-call.c
22097 (ipa_polymorphic_call_context::restrict_to_inner_class):
22098 Skip FIELD_DECLs with error_mark_node type.
22099 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
22102 2017-06-09 Martin Liska <mliska@suse.cz>
22104 * predict.c (struct branch_predictor): New struct.
22105 (test_prediction_value_range): New test.
22106 (predict_c_tests): New function.
22107 * selftest-run-tests.c (selftest::run_tests): Run the function.
22108 * selftest.h: Declare new tests.
22110 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
22113 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
22114 gen_add3_insn did not fail.
22115 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
22116 r0, construct that number in a temporary reg and add that reg to r0.
22117 If asked to put the result in r0 as well, fail.
22119 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
22121 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22122 for early expansion of vec_eqv.
22124 2017-06-08 Jakub Jelinek <jakub@redhat.com>
22126 PR middle-end/81005
22127 * ubsan.c (instrument_null): Avoid pointless code temporary.
22128 (pass_ubsan::execute): Instrument aggregate arguments of calls.
22130 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
22134 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
22137 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
22140 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22142 * predict.c (unlikely_executed_edge_p): Move ahead.
22143 (probably_never_executed_edge_p): Use it.
22145 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22147 PR middle-end/79988
22148 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
22149 gimple_call_builtin_p call.
22151 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22153 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
22154 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
22155 rtl_check_failed_type2, rtl_check_failed_code1,
22156 rtl_check_failed_code2, rtl_check_failed_code_mode,
22157 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
22158 rtvec_check_failed_bounds, rtl_check_failed_flag,
22159 _fatal_insn_not_found, _fatal_insn): Likewise.
22160 * tree.h (tree_contains_struct_check_failed,
22161 tree_check_failed, tree_not_check_failed,
22162 tree_class_check_failed, tree_range_check_failed,
22163 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
22164 tree_vec_elt_check_failed, phi_node_elt_check_failed,
22165 tree_operand_check_failed, omp_clause_check_failed,
22166 omp_clause_operand_check_failed, omp_clause_range_check_failed):
22169 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22171 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
22172 flag_branch_probabilities.
22173 * ipa-inline.c (edge_badness): Likewise.
22174 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
22175 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
22176 * predict.c (maybe_hot_frequency_p): Likewise.
22177 (probably_never_executed): Likewise.
22178 * sched-ebb.c (schedule_ebbs): Likewise.
22179 * sched-rgn.c (find_single_block_region): Likewise.
22180 * tracer.c (tail_duplicate): Likewise.
22182 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22184 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
22185 longer requires x_flag_profile_use.
22187 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22189 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
22190 instead of flag_reorder_blocks_and_partition.
22191 * dbxout.c (dbxout_function_end): Likewise.
22192 * dwarf2out.c (gen_subprogram_die): Likewise.
22193 * haifa-sched.c (sched_create_recovery_edges): Likewise.
22194 * hw-doloop.c (reorg_loops): Likewise.
22195 * varasm.c (assemble_start_function,
22196 assemble_end_function): Likewise.
22197 (decide_function_section): Do not check for
22198 flag_reorder_blocks_and_partition.
22200 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
22202 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
22204 (chkp_get_hard_register_fake_addr_expr): Ditto.
22205 (chkp_build_addr_expr): Add check for hard reg case.
22206 (chkp_parse_array_and_component_ref): Ditto.
22207 (chkp_find_bounds_1): Ditto.
22208 (chkp_process_stmt): Don't generate bounds store for
22211 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22213 * predict.c (maybe_hot_bb_p): Do not check profile status.
22214 (maybe_hot_edge_p): Likewise.
22215 (probably_never_executed): Check for zero counts even if profile
22217 (unlikely_executed_edge_p): New function.
22218 (unlikely_executed_stmt_p): New function.
22219 (unlikely_executed_bb_p): New function.
22220 (set_even_probabilities): Use unlikely predicates.
22221 (combine_predictions_for_bb): Likewise.
22222 (predict_paths_for_bb): Likewise.
22223 (predict_paths_leading_to_edge): Likewise.
22224 (determine_unlikely_bbs): New function.
22225 (estimate_bb_frequencies): Use it.
22226 (compute_function_frequency): Use zero counts even if profile is
22228 * profile-count.h: Fix typo.
22230 2017-08-08 Julia Koval <julia.koval@intel.com>
22232 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
22233 _mm512_mask_cvtsepi16_storeu_epi8,
22234 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
22235 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
22236 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
22237 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
22238 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
22239 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
22240 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
22241 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
22242 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
22243 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
22244 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
22245 __builtin_ia32_pmovuswb256mem_mask,
22246 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
22247 __builtin_ia32_pmovwb512mem_mask): New builtins.
22249 2017-08-08 Julia Koval <julia.koval@intel.com>
22251 PR target/73350,80862
22252 * config/i386/subst.md (round): Fix round pattern.
22253 * config/i386/i386.c (ix86_erase_embedded_rounding):
22254 Fix erasing rounding for the fixed pattern.
22256 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
22258 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
22260 2017-06-08 Martin Liska <mliska@suse.cz>
22262 PR gcov-profile/80911
22263 * gcov.c (block_info::block_info): New constructor.
22265 2017-06-07 Carl Love <cel@us.ibm.com>
22267 * config/rs6000/rs6000-c: The return type of the following
22268 built-in functions was implemented as int not long long. Fix sign
22269 of return value for the unsigned version of vec_mulo and vec_mule.
22270 vector unsigned long long vec_bperm (vector unsigned long long,
22271 vector unsigned char)
22272 vector signed long long vec_mule (vector signed int,
22274 vector unsigned long long vec_mule (vector unsigned int,
22275 vector unsigned int)
22276 vector signed long long vec_mulo (vector signed int,
22278 vector unsigned long long vec_mulo (vector unsigned int,
22279 vector unsigned int)
22280 * doc/extend.texi: Fix the documentation for the built-in
22283 2017-06-07 Carl Love <cel@us.ibm.com>
22286 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
22289 2017-06-07 Carl Love <cel@us.ibm.com>
22291 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
22292 support, Generate doublehv for signed int/float for BE case only.
22294 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
22296 * doc/invoke.texi (mcx16): Rewrite.
22298 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22300 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
22301 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
22302 *mov<mode>_softfloat, and an anonymous splitter): Use
22303 nonimmediate_operand instead of rs6000_nonimmediate_operand.
22305 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22307 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
22309 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22310 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
22311 (rs6000_debug_reg_global): Adjust.
22312 (rs6000_init_hard_regno_mode_ok): Adjust.
22313 (rs6000_dbx_register_number): Adjust.
22314 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
22315 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22316 Remove SPE_ACC and SPEFSCR.
22317 (REG_ALLOC_ORDER): Ditto.
22318 (FRAME_POINTER_REGNUM): Change to 111.
22319 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
22320 (REG_CLASS_NAMES): Ditto.
22321 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
22322 (REGISTER_NAMES): Ditto.
22323 (ADDITIONAL_REG_NAMES): Ditto.
22324 (rs6000_reg_names): Ditto.
22325 * config/rs6000/rs6000.md: Renumber some register number
22328 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22330 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
22332 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
22333 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
22335 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
22336 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
22337 Delete the SPE high registers.
22338 (REG_ALLOC_ORDER): Ditto.
22339 (enum reg_class): Remove SPE_HIGH_REGS.
22340 (REG_CLASS_NAMES): Ditto.
22341 (REG_CLASS_CONTENTS): Delete the SPE high registers.
22342 (REGISTER_NAMES): Ditto.
22343 (rs6000_reg_names): Ditto.
22344 * doc/tm.texi.in: Remove SPE as example.
22345 * doc/tm.texi: Regenerate.
22347 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22349 * config/rs6000/8540.md (ppc8540_brinc): Delete.
22350 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
22351 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
22352 * config/rs6000/rs6000.md (type): Remove "brinc".
22354 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22356 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
22357 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
22358 * config/rs6000/linuxspe.h: Delete file.
22359 * config/rs6000/rs6000.md: Don't include spe.md.
22360 * config/rs6000/spe.h: Delete file.
22361 * config/rs6000/spe.md: Delete file.
22362 * config/rs6000/t-rs6000: Remove spe.md.
22364 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22366 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
22367 (reg_or_none500mem_operand): Delete.
22368 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
22369 instead of reg_or_none500mem_operand.
22371 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22373 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22374 handling of SPE flags.
22375 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
22377 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22379 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22381 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
22382 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
22383 paired_divv2sf3): Similar.
22384 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
22385 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
22386 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
22388 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
22389 Rename the paired_* instruction patterns.
22390 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
22392 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
22393 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
22394 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
22395 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
22396 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
22397 PAIRED_VECTOR_MODE.
22398 (struct machine_function): Delete field spe_insn_chain_scanned_p.
22399 (spe_func_has_64bit_regs_p): Delete.
22400 (spe_expand_predicate_builtin): Delete.
22401 (spe_expand_evsel_builtin): Delete.
22402 (TARGET_DWARF_REGISTER_SPAN): Do not define.
22403 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
22404 (invalid_e500_subreg): Delete.
22405 (rs6000_legitimize_address): Always force_reg op2 as well, for
22406 paired single memory accesses.
22407 (rs6000_member_type_forces_blk): Delete.
22408 (rs6000_spe_function_arg): Delete.
22409 (rs6000_expand_unop_builtin): Delete SPE handling.
22410 (rs6000_expand_binop_builtin): Ditto.
22411 (spe_expand_stv_builtin): Delete.
22412 (bdesc_2arg_spe): Delete.
22413 (spe_expand_builtin): Delete.
22414 (spe_expand_predicate_builtin): Delete.
22415 (spe_expand_evsel_builtin): Delete.
22416 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
22417 (spe_init_builtins): Delete.
22418 (spe_func_has_64bit_regs_p): Delete.
22419 (savres_routine_name): Delete "info" parameter. Adjust callers.
22420 (rs6000_emit_stack_reset): Ditto.
22421 (rs6000_dwarf_register_span): Delete.
22422 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22423 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22424 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22426 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22428 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22429 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22430 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22431 mulv2sf3, divv2sf3): Delete expanders.
22433 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22435 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22437 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22439 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22440 * config/rs6000/rs6000.c: Ditto.
22442 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22444 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22445 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22446 comparison_operator.
22448 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22450 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22451 * config/rs6000/rs6000.opt: Ditto.
22452 * config/rs6000/t-rtems: Ditto.
22454 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22456 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22457 TARGET_E500_SINGLE by 0, simplify.
22458 * config/rs6000/rs6000.c: Ditto.
22459 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22460 (spe_build_register_parallel): Delete.
22461 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22462 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22463 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22464 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22465 (E500_CONVERT): Delete.
22466 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22468 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22470 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22471 * config/rs6000/dfp.md: Ditto.
22472 (negdd2, *negdd2_fpr): Merge.
22473 (absdd2, *absdd2_fpr): Merge.
22474 (negtd2, *negtd2_fpr): Merge.
22475 (abstd2, *abstd2_fpr): Merge.
22476 * config/rs6000/e500.h: Delete file.
22477 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22478 TARGET_FPRS by 1 and simplify.
22479 * config/rs6000/rs6000-c.c: Ditto.
22480 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22481 TARGET_DF_SPE by 0.
22482 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22484 * config/rs6000/rs6000.md: Ditto.
22485 (floatdidf2, *floatdidf2_fpr): Merge.
22486 (move_from_CR_gt_bit): Delete.
22487 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22488 (E500_CR_IOR_COMPARE): Delete.
22489 (All patterns that require !TARGET_FPRS): Delete.
22490 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22492 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22494 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22496 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22498 * graphds.c (add_edge): Intitialize edge's attached data.
22499 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22500 pointer parameter. Call pointed function on each edge during
22501 graph traversing. Skip traversing the edge when the function
22503 (graphds_dfs, graphds_scc): Ditto.
22504 (for_each_edge): New parameter. Pass the new parameter to callback
22506 * graphds.h (skip_edge_callback): New function pointer type.
22507 (graphds_dfs, graphds_scc): New function pointer parameter.
22508 (graphds_edge_callback, for_each_edge): New parameter.
22510 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22512 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22513 out code checking if runtime alias check is possible to below ...
22514 Call the new function.
22515 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22516 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22518 2017-06-07 Marek Polacek <polacek@redhat.com>
22521 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22522 TYPE_OVERFLOW_WRAPS check.
22524 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22526 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22527 if versioning is required.
22528 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22529 peeling with the check for versioning.
22531 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22533 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22534 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22535 Set true to new parameter if new ssa variable is defined.
22536 (vect_gen_vector_loop_niters): Refactor. Set range information
22537 for the new vector loop bound variable.
22538 (vect_do_peeling): Ditto.
22540 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22542 * tree-affine.c (ssa.h): Include header file.
22543 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22544 has wrapping overflow behavior.
22546 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22548 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22550 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22552 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22553 (tree_to_aff_combination): ... here.
22555 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22557 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22558 reg_pressure model function.
22559 (ivopts_global_cost_for_size): Delete.
22560 (determine_set_costs, iv_ca_recount_cost): Call new model function
22561 ivopts_estimate_reg_pressure.
22563 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22565 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22566 expensive than udiv. Remove floating point cases from mod.
22568 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22570 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22571 Increase idiv cost.
22573 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22575 * config/aarch64/aarch64.md
22576 (copysignsf3): Fix mask generation.
22578 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22580 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22582 (class dump_manager): Add register_dumps method.
22583 * dumpfile.c: Include langhooks.h.
22584 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22585 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22586 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22587 (dump_manager::dump_register): Start with 512 entries instead of 32.
22588 (dump_manager::register_dumps): New method.
22589 * toplev.c (general_init): Instead of invoking register_dumps
22590 langhook, invoke register_dumps method on the dump manager.
22591 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22594 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22596 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22597 Say that # requires an associated define_split to exist, and that
22598 the define_split must be suitable for use after register allocation.
22600 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22602 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22603 (compute_outgoing_frequencies): Also initialize zero counts.
22604 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22605 around loops; preserve more of profile when nothing changes.
22607 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22609 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22611 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22612 * config/arm/arm-cpu-cdata.h: Regenerate.
22613 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22614 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22615 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22616 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22617 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22619 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22621 * config/arm/t-rmprofile: Likewise.
22622 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22624 2017-06-06 David S. Miller <davem@davemloft.net>
22627 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22628 blockage if function uses alloca.
22630 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22632 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22633 New "uid" fields to hold pretty-print IDs of group and ref.
22634 Memory references are now identified as <group_id>:<ref_id>
22635 instead of using [random] addresses.
22636 (dump_mem_details): Simplify, no functional change.
22637 (dump_mem_ref): Simplify and make output more concise.
22638 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22639 (find_or_create_group): Initialize group uid.
22640 (record_ref): Initialize ref uid. Improve debug output.
22641 (prune_group_by_reuse, should_issue_prefetch_p,)
22642 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22643 (mark_nontemporal_store, determine_loop_nest_reuse):
22644 Improve debug output.
22646 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22648 * dbgcnt.def (prefetch): New debug counter.
22649 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22650 (schedule_prefetches): Stop issueing prefetches if debug counter
22653 2017-06-06 Tom de Vries <tom@codesourcery.com>
22655 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22656 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22658 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22660 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22661 Use aarch64_reg_or_zero predicate for operand 4.
22662 (aarch64_compare_and_swap<mode> define_insn_and_split):
22663 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22664 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22666 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22668 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22669 (arm_compute_save_core_reg_mask): This.
22670 (thumb1_compute_save_reg_mask): Rename into ...
22671 (thumb1_compute_save_core_reg_mask): This.
22672 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22673 (arm_compute_frame_layout): Likewise.
22675 2017-06-06 Richard Biener <rguenther@suse.de>
22677 PR tree-optimization/80974
22678 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22679 keep or clear leaders SSA info.
22681 2017-06-06 Tom de Vries <tom@codesourcery.com>
22683 * config/nvptx/nvptx.c (split_mode_p): New function.
22684 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22686 2017-06-06 Tom de Vries <tom@codesourcery.com>
22688 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22690 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22693 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22696 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22698 * shrink-wrap.c (handle_simple_exit): Update profile.
22699 (try_shrink_wrapping): Upate profile.
22701 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22703 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22704 (tree_guess_outgoing_edge_probabilities): New.
22705 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22706 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22708 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22710 * ipa-split.c (split_function): Initialize return bb profile.
22712 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22714 * profile.c (compute_branch_probabilities): Also initialize
22715 EXIT_BLOCK profile.
22717 2017-06-06 Richard Biener <rguenther@suse.de>
22719 PR tree-optimization/80928
22720 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22721 (vect_analyze_loop_operations): Properly guard analysis for
22723 (vect_transform_loop): Likewise.
22724 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22725 (vect_model_induction_cost): Do not cost for pure SLP.
22726 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22727 of induction in inner loop vectorization.
22728 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22729 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22730 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22732 (vect_analyze_slp_cost_1): Cost induction.
22733 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22734 (vect_get_slp_vect_defs): Likewise.
22735 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22736 (vect_transform_stmt): Handle SLP reductions.
22737 * tree-vectorizer.h (vectorizable_induction): Adjust.
22739 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22741 * config/rs6000/rs6000.c (make_resolver_func): Update
22742 init_lowered_empty_function call.
22744 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22746 * doc/invoke.texi: Document the -fprofile-abs-path option.
22747 * common.opt (fprofile-abs-path): New option.
22748 * gcov-io.h (gcov_write_filename): Declare.
22749 * gcov-io.c (gcov_write_filename): New function.
22750 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22751 * profile.c (output_location): Likewise.
22753 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22755 * shring-wrap.c: Revert accidental commit.
22757 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22759 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22761 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22763 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22765 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22766 profile in callgraph edge.
22767 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22768 (apply_scale): Likewise.
22769 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22770 Also scale profile when inlining function with zero profile.
22771 (initialize_cfun): Update exit block profile even when it is zero.
22772 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22773 when profile is read.
22775 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22777 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22778 (CLONE_*): New constants to define the processors we can generate
22779 code for with the target_clone attribute.
22780 (rs6000_clone_map): New array to identify which clone processors
22781 the current program is running on.
22782 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22783 target_clone attribute.
22784 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22785 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22786 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22787 (cpu_expand_builtin): Add support for target_clone attribute.
22788 (rs6000_valid_attribute_p): Allow "default" attribute.
22789 (get_decl_name): New debug function to simplify printing the
22790 current function name in debugging statements.
22791 (rs6000_clone_priority): New functions to support the target_clone
22792 attribute, and be able to generate code to switch between ISA 2.05
22793 through ISA 3.0 (power6 through power9).
22794 (rs6000_compare_version_priority): Likewise.
22795 (rs6000_get_function_versions_dispatcher): Likewise.
22796 (make_resolver_func): Likewise.
22797 (add_condition_to_bb): Likewise.
22798 (dispatch_function_versions): Likewise.
22799 (rs6000_generate_version_dispatcher_body): Likewise.
22800 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22801 (fusion_gpr_load_p): Fix a spacing issue.
22802 * doc/extend.texi (Common Function Attributes): Document that the
22803 PowerPC supports the target_clone attribute.
22805 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22807 * config/arm/arm.h: explain F symbol found in description of ARM
22808 register allocation in its legend.
22810 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22812 * config/mips/frame-header-opt.c: Include profile-count.h.
22813 * config/riscv/riscv.c: Include profile-count.h
22815 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22817 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22819 (sm_set_flag_if_changed): Add bbs field.
22820 (execute_sm_if_changed_flag_set): Pass BBS.
22821 (execute_sm): Update.
22823 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22825 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22828 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22830 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22831 (peephole2): New peephole2 to emit the above.
22832 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22834 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22836 * config/aarch64/aarch64.c (define_peephole2 above
22837 *sub_<shift>_<mode>): New peephole.
22839 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22841 * config/i386/i386.c (make_resolver_func): Update.
22842 * Makefile.in: Add profile-count.h and profile-count.o
22843 * auto-profile.c (afdo_indirect_call): Update to new API.
22844 (afdo_set_bb_count): Update.
22845 (afdo_propagate_edge): Update.
22846 (afdo_propagate_circuit): Update.
22847 (afdo_calculate_branch_prob): Update.
22848 (afdo_annotate_cfg): Update.
22849 * basic-block.h: Include profile-count.h
22850 (struct edge_def): Turn count to profile_count.
22851 (struct basic_block_def): Likewie.
22852 (REG_BR_PROB_BASE): Move to profile-count.h
22853 (RDIV): Move to profile-count.h
22854 * bb-reorder.c (max_entry_count): Turn to profile_count.
22855 (find_traces): Update.
22856 (rotate_loop):Update.
22857 (connect_traces):Update.
22858 (sanitize_hot_paths):Update.
22859 * bt-load.c (migrate_btr_defs): Update.
22860 * cfg.c (RDIV): Remove.
22861 (init_flow): Use alloc_block.
22862 (alloc_block): Uninitialize count.
22863 (unchecked_make_edge): Uninitialize count.
22864 (check_bb_profile): Update.
22865 (dump_edge_info): Update.
22866 (dump_bb_info): Update.
22867 (update_bb_profile_for_threading): Update.
22868 (scale_bbs_frequencies_int): Update.
22869 (scale_bbs_frequencies_gcov_type): Update.
22870 (scale_bbs_frequencies_profile_count): New.
22871 * cfg.h (update_bb_profile_for_threading): Update.
22872 (scale_bbs_frequencies_profile_count): Declare.
22873 * cfgbuild.c (compute_outgoing_frequencies): Update.
22874 (find_many_sub_basic_blocks): Update.
22875 * cfgcleanup.c (try_forward_edges): Update.
22876 (try_crossjump_to_edge): Update.
22877 * cfgexpand.c (expand_gimple_tailcall): Update.
22878 (construct_exit_block): Update.
22879 * cfghooks.c (verify_flow_info): Update.
22880 (dump_bb_for_graph): Update.
22881 (split_edge): Update.
22882 (make_forwarder_block): Update.
22883 (duplicate_block): Update.
22884 (account_profile_record): Update.
22885 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22886 (get_estimated_loop_iterations): Update.
22887 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22888 (single_likely_exit): Update.
22889 * cfgloopmanip.c (scale_loop_profile): Update.
22891 (set_zero_probability): Update.
22892 (lv_adjust_loop_entry_edge): Update.
22893 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22894 (purge_dead_edges): Update.
22895 (rtl_account_profile_record): Update.
22896 * cgraph.c (cgraph_node::create): Uninitialize count.
22897 (symbol_table::create_edge): Uninitialize count.
22898 (cgraph_update_edges_for_call_stmt_node): Update.
22899 (cgraph_edge::dump_edge_flags): Update.
22900 (cgraph_node::dump): Update.
22901 (cgraph_edge::maybe_hot_p): Update.
22902 * cgraph.h: Include profile-count.h
22903 (create_clone), create_edge, create_indirect_edge): Update.
22904 (cgraph_node): Turn count to profile_count.
22905 (cgraph_edge0: Likewise.
22906 (make_speculative, clone): Update.
22907 (create_edge): Update.
22908 (init_lowered_empty_function): Update.
22909 * cgraphclones.c (cgraph_edge::clone): Update.
22910 (duplicate_thunk_for_node): Update.
22911 (cgraph_node::create_clone): Update.
22912 * cgraphunit.c (cgraph_node::analyze): Update.
22913 (cgraph_node::expand_thunk): Update.
22914 * final.c (dump_basic_block_info): Update.
22915 * gimple-streamer-in.c (input_bb): Update.
22916 * gimple-streamer-out.c (output_bb): Update.
22917 * graphite.c (print_global_statistics): Update.
22918 (print_graphite_scop_statistics): Update.
22919 * hsa-brig.c: Include basic-block.h.
22920 * hsa-dump.c: Include basic-block.h.
22921 * hsa-gen.c (T sum_slice): Update.
22922 (convert_switch_statements):Update.
22923 * hsa-regalloc.c: Include basic-block.h.
22924 * ipa-chkp.c (chkp_produce_thunks): Update.
22925 * ipa-cp.c (struct caller_statistics): Update.
22926 (init_caller_stats): Update.
22927 (gather_caller_stats): Update.
22928 (ipcp_cloning_candidate_p): Update.
22929 (good_cloning_opportunity_p): Update.
22930 (get_info_about_necessary_edges): Update.
22931 (dump_profile_updates): Update.
22932 (update_profiling_info): Update.
22933 (update_specialized_profile): Update.
22934 (perhaps_add_new_callers): Update.
22935 (decide_about_value): Update.
22936 (ipa_cp_c_finalize): Update.
22937 * ipa-devirt.c (struct odr_type_warn_count): Update.
22938 (struct decl_warn_count): Update.
22939 (struct final_warning_record): Update.
22940 (possible_polymorphic_call_targets): Update.
22941 (ipa_devirt): Update.
22942 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22943 * ipa-icf.c (sem_function::merge): Update.
22944 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22945 * ipa-inline.c (compute_uninlined_call_time): Update.
22946 (compute_inlined_call_time): Update.
22947 (want_inline_small_function_p): Update.
22948 (want_inline_self_recursive_call_p): Update.
22949 (edge_badness): Update.
22950 (lookup_recursive_calls): Update.
22951 (recursive_inlining): Update.
22952 (inline_small_functions): Update.
22953 (dump_overall_stats): Update.
22954 (dump_inline_stats): Update.
22955 * ipa-profile.c (ipa_profile_generate_summary): Update.
22956 (ipa_propagate_frequency): Update.
22957 (ipa_profile): Update.
22958 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22959 * ipa-utils.c (ipa_merge_profiles): Update.
22960 * loop-doloop.c (doloop_modify): Update.
22961 * loop-unroll.c (report_unroll): Update.
22962 (unroll_loop_runtime_iterations): Update.
22963 * lto-cgraph.c (lto_output_edge): Update.
22964 (lto_output_node): Update.
22965 (input_node): Update.
22966 (input_edge): Update.
22967 (merge_profile_summaries): Update.
22968 * lto-streamer-in.c (input_cfg): Update.
22969 * lto-streamer-out.c (output_cfg): Update.
22970 * mcf.c (create_fixup_graph): Update.
22971 (adjust_cfg_counts): Update.
22972 (sum_edge_counts): Update.
22973 * modulo-sched.c (sms_schedule): Update.
22974 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22975 * predict.c (maybe_hot_count_p): Update.
22976 (probably_never_executed): Update.
22977 (dump_prediction): Update.
22978 (combine_predictions_for_bb): Update.
22979 (propagate_freq): Update.
22980 (handle_missing_profiles): Update.
22981 (counts_to_freqs): Update.
22982 (rebuild_frequencies): Update.
22983 (force_edge_cold): Update.
22984 * predict.h: Include profile-count.h
22985 (maybe_hot_count_p, counts_to_freqs): UPdate.
22986 * print-rtl-function.c: Do not include cfg.h
22987 * print-rtl.c: Include basic-block.h
22988 * profile-count.c: New file.
22989 * profile-count.h: New file.
22990 * profile.c (is_edge_inconsistent): Update.
22991 (correct_negative_edge_counts): Update.
22992 (is_inconsistent): Update.
22993 (set_bb_counts): Update.
22994 (read_profile_edge_counts): Update.
22995 (compute_frequency_overlap): Update.
22996 (compute_branch_probabilities): Update; Initialize and deinitialize
22998 (branch_prob): Update.
22999 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
23000 (edge_gcov_count): New.
23001 (bb_gcov_count): New.
23002 * shrink-wrap.c (try_shrink_wrapping): Update.
23003 * tracer.c (better_p): Update.
23004 * trans-mem.c (expand_transaction): Update.
23005 (ipa_tm_insert_irr_call): Update.
23006 (ipa_tm_insert_gettmclone_call): Update.
23007 * tree-call-cdce.c: Update.
23008 * tree-cfg.c (gimple_duplicate_sese_region): Update.
23009 (gimple_duplicate_sese_tail): Update.
23010 (gimple_account_profile_record): Update.
23011 (execute_fixup_cfg): Update.
23012 * tree-inline.c (copy_bb): Update.
23013 (copy_edges_for_bb): Update.
23014 (initialize_cfun): Update.
23015 (freqs_to_counts): Update.
23016 (copy_cfg_body): Update.
23017 (expand_call_inline): Update.
23018 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
23019 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
23020 (try_unroll_loop_completely): Update.
23021 (try_peel_loop): Update.
23022 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23023 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
23024 * tree-ssa-loop-split.c (connect_loops): Update.
23025 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
23026 * tree-ssa-reassoc.c (branch_fixup): Update.
23027 * tree-ssa-tail-merge.c (replace_block_by): Update.
23028 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
23029 (compute_path_counts): Update.
23030 (update_profile): Update.
23031 (recompute_probabilities): Update.
23032 (update_joiner_offpath_counts): Update.
23033 (estimated_freqs_path): Update.
23034 (freqs_to_counts_path): Update.
23035 (clear_counts_path): Update.
23036 (ssa_fix_duplicate_block_edges): Update.
23037 (duplicate_thread_path): Update.
23038 * tree-switch-conversion.c (case_bit_test_cmp): Update.
23039 (struct switch_conv_info): Update.
23040 * tree-tailcall.c (decrease_profile): Update.
23041 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
23042 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23043 * value-prof.c (check_counter): Update.
23044 (gimple_divmod_fixed_value): Update.
23045 (gimple_mod_pow2): Update.
23046 (gimple_mod_subtract): Update.
23047 (gimple_ic_transform): Update.
23048 (gimple_stringop_fixed_value): Update.
23049 * value-prof.h (gimple_ic): Update.
23051 2017-06-02 Carl Love <cel@us.ibm.com>
23053 * config/rs6000/rs6000-c: Add support for built-in functions
23054 vector double vec_doublee (vector signed int);
23055 vector double vec_doublee (vector unsigned int);
23056 vector double vec_doublee (vector float);
23057 vector double vec_doubleh (vector signed int);
23058 vector double vec_doubleh (vector unsigned int);
23059 vector double vec_doubleh (vector float);
23060 vector double vec_doublel (vector signed int);
23061 vector double vec_doublel (vector unsigned int);
23062 vector double vec_doublel (vector float);
23063 vector double vec_doubleo (vector signed int);
23064 vector double vec_doubleo (vector unsigned int);
23065 vector double vec_doubleo (vector float);.
23066 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
23067 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
23069 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
23070 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
23071 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
23073 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
23074 vec_doublel, vec_doubleh.
23075 * doc/extend.texi: Update the built-in documentation file for the
23076 new built-in functions.
23078 2017-06-02 David Malcolm <dmalcolm@redhat.com>
23081 * ipa-inline-analysis.c (free_growth_caches): Set
23082 edge_removal_hook_holder to NULL after removing it.
23084 2017-06-02 Sudakshina Das <sudi.das@arm.com>
23086 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
23087 comparision with zero.
23089 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23090 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23091 for early expansion of vec_min and vec_max builtins.
23092 (builtin_function_type): Add min/max unsigned variants to those
23093 identified as having unsigned arguments.
23095 2017-06-02 Olivier Hainque <hainque@adacore.com>
23097 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
23099 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23101 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
23102 Use VALL_F16 iterator rather than VALL.
23104 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23106 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
23107 Emit CBNZ inside loop when doing a strong exchange and comparing
23108 against zero. Generate the CC flags after the loop.
23110 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
23112 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
23113 (dl_section_ref): New.
23114 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
23115 On AIX, append an expression to subtract the size of the
23116 section length to dl_section_ref.
23118 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
23120 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
23121 for early expansion of vector absolute builtins.
23123 2017-06-02 Richard Biener <rguenther@suse.de>
23125 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
23126 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
23128 2017-06-02 Richard Biener <rguenther@suse.de>
23130 PR tree-optimization/80948
23131 * tree-tailcall.c (find_tail_calls): Track stmts to move in
23132 stmt order as well.
23134 2017-06-02 Richard Biener <rguenther@suse.de>
23136 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
23138 * tree-vect-stmts.c (process_use): Do not mark backedge defs
23139 for inductions as relevant.
23141 2017-06-02 Richard Biener <rguenther@suse.de>
23143 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
23144 (vectorizable_induction): ... this. Remove dead code.
23146 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
23148 * builtins. (expand_builtin_alloca): Remove second parameter and
23149 infer its value from the first parameter instead.
23150 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
23152 2017-06-02 Jakub Jelinek <jakub@redhat.com>
23154 PR rtl-optimization/80903
23155 * loop-doloop.c (add_test): Unshare sequence.
23157 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
23159 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
23161 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
23163 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
23165 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
23166 xlogue_layout::get_instance, logue_layout::xlogue_layout,
23167 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
23168 (xlogue_layout::get_stub_rtx): Make static.
23169 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
23170 (xlogue_layout::compute_stub_managed_regs): Rename to...
23171 (xlogue_layout::count_stub_managed_regs): ...this.
23172 (xlogue_layout::is_stub_managed_reg): New function.
23173 (xlogue_layout::m_stub_names): Rename to...
23174 (xlogue_layout::s_stub_names): ...this, make static.
23175 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
23176 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
23177 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
23178 xlogue_layout::s_stub_names): Instantiate statics.
23179 (stub_managed_regs): Remove.
23180 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
23181 (disable_call_ms2sysv_xlogues): Rename to...
23182 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
23183 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
23185 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
23186 change after reload_completed.
23187 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
23189 (ix86_expand_prologue): Likewise.
23190 (ix86_expand_epilogue): Likewise.
23191 (ix86_expand_split_stack_prologue): Likewise.
23192 (ix86_compute_frame_layout): Remove frame parameter ...
23193 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
23194 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
23196 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
23197 (ix86_frame): Move from here ...
23198 * config/i386/i386.h (ix86_frame): ... to here.
23199 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
23200 complete ix86_frame data structure instead. Remove some_ld_name.
23202 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23204 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
23205 symbols that hold a DECL_VALUE_EXPR.
23207 2017-06-01 Martin Jambor <mjambor@suse.cz>
23209 PR tree-optimization/80898
23210 * tree-sra.c (process_subtree_disqualification): Removed.
23211 (disqualify_candidate): Do not acll
23212 process_subtree_disqualification.
23213 (subtree_mark_written_and_enqueue): New function.
23214 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
23215 RHS has been disqualified and re-queue LHS if necessary. Apart
23216 from that, ignore disqualified RHS.
23218 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23220 * config/s390/s390.c (s390_emit_epilogue): Disable early return
23221 address fetch for z10 or later.
23223 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23225 * config/arc/arc.md (tst_movb): Add guard when splitting.
23227 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23229 * config/arc/arc.c (arc_can_eliminate): Test against
23230 arc_frame_pointer_needed.
23232 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23234 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
23235 to prevent store reordering.
23236 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
23237 (type): Add block type.
23238 (stack_tie): Define special instruction to be used in
23241 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23243 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
23244 constraint. It is not valid for the pattern.
23245 (noncommutative_binary_comparison): Likewise.
23247 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23249 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
23252 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23254 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
23255 be used by the reg-alloc.
23257 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23259 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
23260 reg-alloc when having mul64 or mul32x16 instructions.
23261 (mulsidi3): Likewise.
23262 (umulsidi3): Likewise.
23263 (mulsi32x16): New pattern.
23264 (mulsi64): Likewise.
23265 (mulsidi64): Likewise.
23266 (umulsidi64): Likewise.
23267 (MUL32x16_REG): Define.
23268 (mul64_600): Use MUL32x16_REG.
23269 (mac64_600): Likewise.
23270 (umul64_600): Likewise.
23271 (umac64_600): Likewise.
23273 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
23275 * config/arc/arc.md (mulsi3_700): Make it commutative.
23277 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
23279 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
23281 (*sign_extendsidi2_insn): Likewise for movstosw.
23283 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
23285 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
23286 the type of the input discriminant value. Convert the
23287 discriminant value of signedness vary.
23289 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
23291 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
23292 Add to -Wall section.
23294 2017-06-01 Richard Biener <rguenther@suse.de>
23296 PR middle-end/66313
23297 * fold-const.c (fold_plusminus_mult_expr): If the factored
23298 factor may be zero use a wrapping type for the inner operation.
23299 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
23300 and handle moved defs.
23301 (process_assignment): Properly guard the unary op case. Return a
23302 tri-state indicating that moving the stmt before the call may allow
23303 to continue. Pass through to_move.
23304 (find_tail_calls): Handle moving unrelated defs before
23307 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
23310 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
23311 splitter result in the canonical way.
23313 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
23315 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
23316 also for 32bit target. Update insn attributes.
23317 (zero-extendsidi2 splitter): Allow all registers for operand 1.
23319 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
23321 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
23322 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
23323 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
23324 (_mm_maskz_min_ss): New intrinsics.
23326 2017-05-31 Martin Liska <mliska@suse.cz>
23328 * tree-vect-loop.c (vect_create_epilog_for_reduction):
23329 Change comment style to one we normally use.
23330 (vectorizable_reduction): Likewise.
23331 (vectorizable_induction): Likewise.
23332 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
23333 (vectorizable_call): Likewise.
23334 (vectorizable_simd_clone_call): Likewise.
23335 (vectorizable_conversion): Likewise.
23336 (vectorizable_assignment): Likewise.
23337 (vectorizable_shift): Likewise.
23338 (vectorizable_operation): Likewise.
23339 (vectorizable_store): Likewise.
23340 (vectorizable_load): Likewise.
23341 * tree-vectorizer.h: Likewise.
23343 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
23345 * passes.c (emergency_dump_function): New.
23346 * tree-pass.h (emergency_dump_function): Declare.
23347 * plugin.c (plugins_internal_error_function): Remove.
23348 * plugin.h (plugins_internal_error_function): Remove declaration.
23349 * toplev.c (internal_error_function): New static function. Use it...
23350 (general_init): ...here.
23352 2017-05-31 Graham Markall <graham.markall@embecosm.com>
23354 * config/arc/arc.c (arc_print_operand): Handle constant operands.
23355 (arc_rtx_costs): Add costs for new patterns.
23356 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
23357 * config/arc/predicates.md: Add _1_2_3_operand predicate.
23359 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23361 * tree-ssa-strlen.c (get_next_strinfo): New function.
23362 (get_stridx_plus_constant): Use it.
23363 (zero_length_string): Likewise.
23364 (adjust_related_strinfos): Likewise.
23365 (adjust_last_stmt): Likewise.
23367 2017-05-31 Richard Biener <rguenther@suse.de>
23370 * config/i386/i386.c (ix86_expand_builtin): Remove assert
23371 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
23373 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
23375 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
23376 loop_vinfo argument and use of dependence distance vectors.
23377 Check instead whether the two references differ only in their
23378 initial value and assume that they have the same alignment if the
23379 difference is a multiple of the vector alignment.
23380 (vect_analyze_data_refs_alignment): Update call accordingly.
23382 2017-05-31 Martin Liska <mliska@suse.cz>
23385 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
23387 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23389 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
23390 (create_intersect_range_checks): Move from ...
23391 * tree-data-ref.c (create_intersect_range_checks_index)
23392 (create_intersect_range_checks): ... to here.
23393 (create_runtime_alias_checks): New function factored from ...
23394 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
23395 here. Call above function.
23396 * tree-data-ref.h (create_runtime_alias_checks): New function.
23398 2017-05-31 Bin Cheng <bin.cheng@arm.com>
23400 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
23401 segment length for dr_b and compute it in wide_int.
23403 2017-05-31 Richard Biener <rguenther@suse.de>
23405 PR tree-optimization/80906
23406 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
23407 and pass through iv_map.
23408 (copy_bb_and_scalar_dependences): Adjust.
23409 (translate_pending_phi_nodes): Likewise.
23410 (copy_loop_close_phi_args): Handle code-generating IVs instead
23413 2017-05-30 David Malcolm <dmalcolm@redhat.com>
23415 * diagnostic-color.c (color_dict): Add "type-diff".
23416 (parse_gcc_colors): Update comment.
23417 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
23418 -fdiagnostics-show-template-tree and -fno-elide-type.
23419 (GCC_COLORS): Add type-diff to example.
23421 (-fdiagnostics-show-template-tree): New.
23422 (-fno-elide-type): New.
23423 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23424 the pp_format_decoder callback. Call any m_format_postprocessor's
23426 (pretty_printer::pretty_printer): Initialize
23427 m_format_postprocessor.
23428 (pretty_printer::~pretty_printer): Delete any
23429 m_format_postprocessor.
23430 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23431 (class format_postprocessor): New class.
23432 (struct pretty_printer::format_decoder): Document the new parameters.
23433 (struct pretty_printer::m_format_postprocessor): New field.
23434 * tree-diagnostic.c (default_tree_printer): Update for new
23435 bool and const char ** params.
23436 * tree-diagnostic.h (default_tree_printer): Likewise.
23438 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23440 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23441 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23442 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23443 rs6000_gen_cell_microcode code.
23444 (rs6000_final_prescan_insn): Delete.
23445 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23446 "warn-cell-microcode" entries.
23447 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23448 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23449 throughout. Change cc_reg_not_micro_cr0_operand to
23450 cc_reg_not_cr0_operand throughout.
23451 (*extendhi<mode>2_noload): Delete.
23452 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23453 (mwarn-cell-microcode): Delete.
23454 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23455 -mgen-cell-microcode and -mwarn-cell-microcode.
23457 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23460 * config/i386/constraints.md (Yd): New constraint.
23462 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23463 and (?Yd, r) alternatives. Update insn attributes.
23464 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23465 and (?*Yd, r) alternatives. Update insn attributes.
23466 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23468 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23470 * gimplify.c (gimplify_modify_expr): Don't create a
23471 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23474 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23476 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23478 2017-05-30 Richard Biener <rguenther@suse.de>
23480 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23481 and reduc_def fields.
23482 (STMT_VINFO_REDUC_TYPE): New define.
23483 (STMT_VINFO_REDUC_DEF): Likewise.
23484 (vect_force_simple_reduction): Adjust prototype.
23485 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23486 (vect_is_simple_reduction): Remove check_reduction argument.
23487 (vect_force_simple_reduction): Adjust and set
23488 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23489 (vectorizable_reduction): Do not re-do reduction analysis
23490 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23491 * tree-parloops.c (gather_scalar_reductions): Adjust.
23493 2017-05-30 Richard Biener <rguenther@suse.de>
23495 PR middle-end/80901
23496 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23499 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23501 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23502 Introduce unknown_misalignment parameter and remove vf.
23503 (vect_peeling_hash_get_lowest_cost):
23504 Pass unknown_misalignment parameter.
23505 (vect_enhance_data_refs_alignment):
23506 Fix unsupportable data ref treatment.
23508 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23510 * tree-vect-data-refs.c (vect_get_data_access_cost):
23511 Workaround for SLP handling.
23512 (vect_enhance_data_refs_alignment):
23513 Compute costs for doing no peeling at all, compare to the best
23514 peeling costs so far and avoid peeling if cheaper.
23516 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23518 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23519 Return peeling info and set costs to zero for unlimited cost
23521 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23522 with unknown misalignment. Compute and costs for unknown
23523 misalignment, compare them to the costs for known misalignment
23524 and choose the cheapest for peeling.
23526 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23528 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23529 (vect_get_peeling_costs_all_drs): Create function.
23530 (vect_peeling_hash_get_lowest_cost):
23531 Use vect_get_peeling_costs_all_drs.
23532 (vect_peeling_supportable): Create function.
23533 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23535 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23537 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23538 DR_HAS_NEGATIVE_STEP.
23539 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23540 (vect_enhance_data_refs_alignment): Use.
23541 (vect_duplicate_ssa_name_ptr_info): Use.
23542 * tree-vectorizer.h (dr_misalignment): Use.
23543 (known_alignment_for_access_p): Use.
23545 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23548 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23550 (has_section_name): New function.
23552 2017-05-30 Martin Liska <mliska@suse.cz>
23555 * auto-profile.c (get_function_decl_from_block): Fix
23558 2017-05-30 Richard Biener <rguenther@suse.de>
23560 PR middle-end/80876
23561 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23563 2017-05-30 Martin Liska <mliska@suse.cz>
23565 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23566 * dumpfile.h (struct dump_file_info): Remove ctors.
23568 2017-05-30 Martin Liska <mliska@suse.cz>
23570 * predict.def: Fix GNU coding style.
23572 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23574 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23575 Mark 'to' argument with ATTRIBUTE_UNUSED.
23577 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23579 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23580 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23581 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23584 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23586 * doc/install.texi (Options specification): Restore entry of
23587 --enable-sjlj-exceptions.
23589 2017-05-27 Michael Eager <eager@eagercon.com>
23592 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23594 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23596 * config/microblaze/microblaze.h
23597 (FIXED_REGISTERS): Update in macro.
23598 (CALL_USED_REGISTERS): Update in macro.
23600 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23602 * doc/install.texi: Add links to macOS binary distributions.
23604 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23608 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23610 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23612 2017-05-26 Martin Liska <mliska@suse.cz>
23614 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23616 2017-05-26 Martin Liska <mliska@suse.cz>
23618 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23619 always leading ';; '.
23620 (dump_bb_info): Likewise.
23621 (brief_dump_cfg): Likewise.
23622 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23623 * dumpfile.c: Remove usage of TDF_VERBOSE.
23624 * dumpfile.h (enum dump_kind): Likewise.
23625 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23626 * print-tree.c (debug_verbose): Remove.
23627 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23628 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23629 * tree-diagnostic.c (default_tree_printer): Replace
23630 TDF_DIAGNOSTIC with TDF_SLIM.
23632 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23634 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23635 in parameter loop, rather than loop_vinfo.
23636 (create_intersect_range_checks): Ditto.
23637 (vect_create_cond_for_alias_checks): Update call to above functions.
23639 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23641 PR tree-optimization/80815
23642 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23643 for merging runtime alias checks. Handle negative DR_STEPs.
23645 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23647 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23649 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23650 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23651 out code pruning runtime alias checks.
23652 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23653 factored out from above.
23654 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23656 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23658 (prune_runtime_alias_test_list): New decalaration.
23660 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23662 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23663 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23664 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23665 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23666 (operator==, comp_dr_with_seg_len_pair): Ditto.
23667 (vect_prune_runtime_alias_test_list): Ditto.
23669 2017-05-26 Martin Liska <mliska@suse.cz>
23672 * params.def: Bound partial-inlining-entry-probability param.
23674 2017-05-26 Marek Polacek <polacek@redhat.com>
23677 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23680 2017-05-26 Richard Biener <rguenther@suse.de>
23682 PR tree-optimization/80842
23683 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23686 2017-05-26 Richard Biener <rguenther@suse.de>
23688 PR tree-optimization/80844
23689 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23691 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23693 * doc/md.texi (Machine Constraints): Update x86 family
23694 machine constraints section to match 'config/i386/constraints.md'.
23696 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23698 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23700 2017-05-25 Nathan Sidwell <nathan@acm.org>
23702 * doc/invoke.texi (--enable-languages): Update documentation.
23704 2017-05-25 Martin Liska <mliska@suse.cz>
23706 * dumpfile.c: Add TDF_FOLDING.
23707 * dumpfile.h (enum dump_kind): Likewise.
23708 * genmatch.c (dt_simplify::gen_1): Use it.
23710 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23712 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23714 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23716 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23717 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23719 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23721 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23722 * match.pd (X == C): Rewrite it here.
23723 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23724 with_certain_nonzero_bits2): New predicates.
23725 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23727 2017-05-24 Nathan Sidwell <nathan@acm.org>
23729 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23732 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23735 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23737 * config/powerpcspe: New port. Files are copied from the rs6000
23738 port, with "rs6000" in filenames replaced by "powerpcspe".
23740 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23742 PR rtl-optimization/80754
23743 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23745 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23747 * config/sparc/sparc.md (length): Return the correct value for -mflat
23748 sibcalls to match output_sibcall.
23750 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23754 * config/rs6000/rs6000.c (struct machine_function): Add new field
23756 (rs6000_get_separate_components): Init that field, use it.
23757 (rs6000_components_for_bb): Use the field.
23759 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23761 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23763 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23765 PR middle-end/80823
23766 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23768 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23771 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23772 address against address_operand predicate.
23773 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23775 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23777 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23778 parameters passed indirectly.
23780 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23782 * config/i386/i386.md (*movdi_internal): Remove SSE4
23783 alternative 18 (?r, *v). Update insn attributes.
23784 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23785 Update insn attributes.
23786 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23787 Update insn attributes.
23788 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23789 alternative 1 (r, v). Remove isa attribute.
23790 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23791 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23792 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23794 2017-05-23 Tom de Vries <tom@codesourcery.com>
23796 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23799 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23801 * cgraphunit.c (symbol_table::process_new_functions): Update.
23802 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23803 (inline_generate_summary): Rename to ...
23804 (ipa_fn_summary_generate): ... this one.
23805 (inline_read_summary): Rename to ...
23806 (ipa_fn_summary_read): ... this one.
23807 (inline_write_summary): Rename to ...
23808 (ipa_fn_summary_write): ... this one.
23809 (inline_free_summary): Rename to ...
23810 (ipa_free_fn_summary): ... this one.
23811 (pass_data_local_fn_summary, pass_local_fn_summary,
23812 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23813 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23814 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23815 make_pass_ipa_fn_summary): New.
23816 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23817 inline_write_summary, inline_free_summary): Remove.
23818 (ipa_free_fn_summary) : New.
23819 * ipa-inline.c (ipa_inline): Update.
23820 (pass_ipa_inline): Do not generate summaries.
23821 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23823 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23824 and add pass_ipa_fn_summary.
23825 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23827 (make_pass_inline_parameters): Remove.
23829 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23831 * omp-low.c (struct omp_context): Remove "default_kind" member.
23834 * omp-offload.c (execute_oacc_device_lower): Remove the
23835 parallelism dimensions function attributes for unparallelized
23836 OpenACC kernels constructs.
23838 2017-05-23 Martin Liska <mliska@suse.cz>
23840 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23842 (cgraph_edge::make_speculative): Likewise.
23843 (cgraph_edge::resolve_speculation): Likewise.
23844 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23845 (cgraph_node::dump): Likewise.
23846 * cgraph.h: Likewise.
23847 * cgraphunit.c (analyze_functions): Likewise.
23848 (symbol_table::compile): Likewise.
23849 * ipa-cp.c (print_all_lattices): Likewise.
23850 (determine_versionability): Likewise.
23851 (initialize_node_lattices): Likewise.
23852 (ipcp_verify_propagated_values): Likewise.
23853 (estimate_local_effects): Likewise.
23854 (update_profiling_info): Likewise.
23855 (create_specialized_node): Likewise.
23856 (perhaps_add_new_callers): Likewise.
23857 (decide_about_value): Likewise.
23858 (decide_whether_version_node): Likewise.
23859 (identify_dead_nodes): Likewise.
23860 (ipcp_store_bits_results): Likewise.
23861 * ipa-devirt.c (dump_targets): Likewise.
23862 (ipa_devirt): Likewise.
23863 * ipa-icf.c (sem_item::dump): Likewise.
23864 (sem_function::equals): Likewise.
23865 (sem_variable::equals): Likewise.
23866 (sem_item_optimizer::read_section): Likewise.
23867 (sem_item_optimizer::execute): Likewise.
23868 (congruence_class::dump): Likewise.
23869 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23870 (dump_inline_summary): Likewise.
23871 (estimate_node_size_and_time): Likewise.
23872 (inline_analyze_function): Likewise.
23873 * ipa-inline-transform.c (inline_call): Likewise.
23874 * ipa-inline.c (report_inline_failed_reason): Likewise.
23875 (want_early_inline_function_p): Likewise.
23876 (edge_badness): Likewise.
23877 (update_edge_key): Likewise.
23878 (inline_small_functions): Likewise.
23879 * ipa-profile.c (ipa_profile): Likewise.
23880 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23881 (ipa_make_edge_direct_to_target): Likewise.
23882 (remove_described_reference): Likewise.
23883 (ipa_impossible_devirt_target): Likewise.
23884 (propagate_controlled_uses): Likewise.
23885 (ipa_print_node_params): Likewise.
23886 (ipcp_transform_function): Likewise.
23887 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23888 (propagate_pure_const): Likewise.
23889 * ipa-reference.c (generate_summary): Likewise.
23890 (read_write_all_from_decl): Likewise.
23891 (propagate): Likewise.
23892 (ipa_reference_read_optimization_summary): Likewise.
23893 * ipa-utils.c (ipa_merge_profiles): Likewise.
23894 * ipa.c (walk_polymorphic_call_targets): Likewise.
23895 (symbol_table::remove_unreachable_nodes): Likewise.
23896 (ipa_single_use): Likewise.
23897 * passes.c (execute_todo): Likewise.
23898 * predict.c (drop_profile): Likewise.
23899 * symtab.c (symtab_node::get_dump_name): New function.
23900 (symtab_node::dump_name): Likewise.
23901 (symtab_node::dump_asm_name): Likewise.
23902 (symtab_node::dump_references): Likewise.
23903 (symtab_node::dump_referring): Likewise.
23904 (symtab_node::dump_base): Likewise.
23905 (symtab_node::debug_symtab): Likewise.
23906 * tree-sra.c (convert_callers_for_node): Likewise.
23907 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23908 * value-prof.c (init_node_map): Likewise.
23910 2017-05-23 Martin Liska <mliska@suse.cz>
23912 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23913 and symtab_node::debug_symtab to symbol_table::debug.
23914 * cgraphunit.c (analyze_functions): Use the renamed function.
23915 (symbol_table::compile): Likewise.
23916 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23917 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23918 * passes.c (execute_todo): Likewise.
23919 * symtab.c (symbol_table::dump): New function.
23920 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23922 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23924 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23925 that nonconst implies exec.
23927 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23929 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23930 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23931 (inline_edge_summary_vec): Turn into ...
23932 (ipa_call_summaries): ... this one.
23933 (redirect_to_unreachable, edge_set_predicate,
23934 evaluate_properties_for_edge, inline_summary_alloc,
23935 reset_ipa_call_summary, reset_inline_summary,
23936 inline_summary_t::duplicate): Update.
23937 (inline_edge_duplication_hook): Turn to ...
23938 (ipa_call_summary_t::duplicate): ... this one.
23939 (inline_edge_removal_hook): Turn to ...
23940 (ipa_call_summary_t::remove): ... this one.
23941 (dump_inline_edge_summary): Turn to ...
23942 (dump_ipa_call_summary): ... this one.
23943 (estimate_function_body_sizes): Update.
23944 (inline_update_callee_summaries): Update.
23945 (remap_edge_change_prob): Update.
23946 (remap_edge_summaries): Update.
23947 (inline_merge_summary): Update.
23948 (do_estimate_edge_time): Update.
23949 (inline_generate_summary): Update.
23950 (inline_read_section): Update.
23951 (inline_read_summary): Update.
23952 (inline_free_summary): Update.
23953 * ipa-inline.c (can_inline_edge_p): Update.
23954 (compute_inlined_call_time): Update.
23955 (want_inline_small_function_p): Update.
23956 (edge_badness): Update.
23957 (early_inliner): Update.
23958 * ipa-inline.h (inline_edge_summary): Turn to ...
23959 (ipa_call_summary): ... this one.
23960 (ipa_call_summary_t): New class.
23961 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23962 (ipa_call_summaries): New.
23963 (inline_edge_summary): Remove.
23964 (estimate_edge_growth): Update.
23965 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23966 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23967 * ipa-split.c (execute_split_functions): Update.
23968 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23970 2017-05-23 Tom de Vries <tom@codesourcery.com>
23972 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23973 attributes): Document rdrand effective target.
23975 2017-05-23 Tom de Vries <tom@codesourcery.com>
23977 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23978 attributes): Sort alphabetically.
23980 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23982 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23984 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23987 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23988 V2DF/V2DI splat into two separate patterns, one that handles
23989 registers, and the other that only handles memory. Drop support
23990 for splatting from a GPR on ISA 2.07 and then splitting the
23991 splat into direct move and splat.
23992 (vsx_splat_<mode>_reg): Likewise.
23993 (vsx_splat_<mode>_mem): Likewise.
23995 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23997 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23999 2017-05-22 Jakub Jelinek <jakub@redhat.com>
24001 PR middle-end/80809
24002 * omp-low.c (finish_taskreg_remap): New function.
24003 (finish_taskreg_scan): If unit size of ctx->record_type
24004 is non-constant, unshare the size expression and replace
24005 decls in it with possible outer var refs.
24007 PR middle-end/80809
24008 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
24009 GOVD_SHARED rather than GOVD_PRIVATE with it.
24010 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
24011 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
24013 PR middle-end/80853
24014 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
24015 as last argument to build_outer_var_ref for pointer bases of array
24016 section reductions.
24018 2017-05-19 Martin Sebor <msebor@redhat.com>
24020 * print-tree.c (print_node): Print DECL_READ_P flag.
24022 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24024 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
24025 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
24026 * cgraph.c: Likewise.
24027 * cgraphunit.c: Likewise.
24028 * gengtype.c: Likewise.
24029 * ipa-cp.c: Likewise.
24030 * ipa-devirt.c: Likewise.
24031 * ipa-icf.c: Likewise.
24032 * ipa-predicate.c: Likewise.
24033 * ipa-profile.c: Likewise.
24034 * ipa-prop.c: Likewise.
24035 * ipa-split.c: Likewise.
24037 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
24038 edge_predicate_pool, dump_inline_hints,
24039 inline_summary::account_size_time, redirect_to_unreachable,
24040 edge_set_predicate, set_hint_predicate,
24041 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
24042 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
24043 inline_summary_t::remove, remap_hint_predicate_after_duplication,
24044 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
24045 ipa_call_summary_t::remove, initialize_growth_caches,
24046 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
24047 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
24048 mark_modified, unmodified_parm_1, unmodified_parm,
24049 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
24050 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
24051 compute_bb_predicates, will_be_nonconstant_expr_predicate,
24052 will_be_nonconstant_predicate, record_modified_bb_info,
24053 get_minimal_bb, record_modified, param_change_prob,
24054 phi_result_unknown_predicate, predicate_for_phi_result,
24055 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
24056 estimate_function_body_sizes, compute_inline_parameters,
24057 compute_inline_parameters_for_curren, pass_data_inline_parameters,
24058 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
24059 inline_update_callee_summaries, remap_edge_change_prob,
24060 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
24061 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
24062 inline_analyze_function, inline_summary_t::insert,
24063 inline_generate_summary, read_ipa_call_summary, inline_read_section,
24064 inline_read_summary, write_ipa_call_summary, inline_write_summary,
24065 inline_free_summary): Move to ipa-fnsummary.h
24066 (predicate_t): Remove.
24067 * ipa-fnsummary.c: New file.
24068 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
24069 (enum inline_hints_vals, inline_hints, agg_position_info,
24070 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
24071 inline_summaries, ipa_call_summary, ipa_call_summary_t,
24072 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
24073 dump_inline_summary, dump_inline_hints, inline_generate_summary,
24074 inline_read_summary, inline_write_summary, inline_free_summary,
24075 inline_analyze_function, initialize_inline_failed,
24076 inline_merge_summary, inline_update_overall_summary,
24077 compute_inline_parameters): Move to ipa-fnsummary.h
24078 * ipa-fnsummary.h: New file.
24079 * ipa-inline-transform.h: Include ipa-inline.h.
24080 * ipa-inline.c: LIkewise.
24082 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24084 * ipa-inline.c (edge_badness): Use inlined_time instead of
24085 inline_summaries->get.
24087 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24089 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
24091 2017-05-22 Nathan Sidwell <nathan@acm.org>
24093 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
24094 (fdump-lang): Document 'raw' option.
24095 * dumpfile.h (TDI_tu): Delete.
24096 * dumpfile.c (dump_files): Remove translation-unit.
24097 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
24099 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
24101 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
24102 command option from $(AWK) call.
24103 * config/avr/genmultilib.awk: Simplify and rewrite so that it
24104 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
24105 [FORMAT]: Remove handling of variable.
24106 * config/avr/t-multilib: Regenerate.
24108 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24110 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
24112 (dump_inline_summary): Do not print self_time.
24113 (estimate_function_body_sizes): Do not set self_time.
24114 (compute_inline_parameters): Likewise.
24115 (inline_read_section, inline_write_summary): Do not stream self_time.
24116 * ipa-inline.h (inline_summary): Drop self_time.
24118 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
24120 * ipa-inline-analysis.c (account_size_time): Rename to ...
24121 (inline_summary::account_size_time): ... this one.
24122 (reset_ipa_call_summary): Turn to ...
24123 (ipa_call_summary::reset): ... this one.
24124 (reset_inline_summary): Turn to ...
24125 (inline_summary::reset): ... this one.
24126 (inline_summary_t::remove): Update.
24127 (inline_summary_t::duplicate): Update.
24128 (ipa_call_summary_t::remove): Update.
24129 (dump_inline_summary): Update.
24130 (estimate_function_body_sizes): Update.
24131 (compute_inline_parameters): Update.
24132 (estimate_node_size_and_time): Update.
24133 (inline_merge_summary): Update.
24134 (inline_update_overall_summary): Update.
24135 (inline_read_section): Update.
24136 (inline_write_summary): Update.
24137 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
24138 add account_size_time and reset member functions.
24139 (ipa_call_summary): Add reset function.
24140 * ipa-predicate.h (predicate::operator &): Constify.
24142 2017-05-22 Richard Biener <rguenther@suse.de>
24144 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
24146 2017-05-19 Jason Merrill <jason@redhat.com>
24148 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
24150 2017-05-19 Marek Polacek <polacek@redhat.com>
24153 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
24154 TYPE_OVERFLOW_WRAPS checks.
24156 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
24158 * tree-core.h (enum omp_clause_default_kind): Add
24159 "OMP_CLAUSE_DEFAULT_PRESENT".
24160 * tree-pretty-print.c (dump_omp_clause): Handle it.
24161 * gimplify.c (enum gimplify_omp_var_data): Add
24162 "GOVD_MAP_FORCE_PRESENT".
24163 (gimplify_adjust_omp_clauses_1): Map it to
24164 "GOMP_MAP_FORCE_PRESENT".
24165 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
24167 * gimplify.c (oacc_default_clause): Clarify.
24169 2017-05-19 Nathan Sidwell <nathan@acm.org>
24171 LANG_HOOK_REGISTER_DUMPS
24172 * toplev.c (general_init): Call register dump lang hook.
24173 * doc/invoke.texi: Document -fdump-lang option family.
24174 * dumpfile.c (dump_files): Remove class dump here.
24175 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
24176 * dumpfile.h (tree_dump_index): Remove TDI_class.
24177 * langhooks-def.h (lhd_register_dumps): Declare.
24178 (LANG_HOOKS_REGISTER_DUMPS): Define.
24179 (LANG_HOOKS_INITIALIZER): Add it.
24180 * langhooks.c (lhd_register_dumps): Define.
24181 * langhooks.h (struct lang_hooks): Add register_dumps.
24183 2017-05-19 Nathan Sidwell <nathan@acm.org>
24185 * context.h (context::set_passes): New.
24186 * context.c (context::context): Do not create pass manager.
24187 * toplev.c (general_init): Create pass manager here.
24189 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
24191 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
24192 use this splitter if two add or or instructions would also work for
24193 the constant we want to generate.
24195 2017-05-19 Richard Biener <rguenther@suse.de>
24198 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
24199 predicate evaluation.
24201 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24203 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
24205 * ipa-inline.c (want_inline_small_function_p): Do not cast to
24208 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
24210 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
24211 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
24212 (inline_edge_summary_vec): Turn into ...
24213 (ipa_call_summaries): ... this one.
24214 (redirect_to_unreachable, edge_set_predicate,
24215 evaluate_properties_for_edge, inline_summary_alloc,
24216 reset_ipa_call_summary, reset_inline_summary,
24217 inline_summary_t::duplicate): Update.
24218 (inline_edge_duplication_hook): Turn to ...
24219 (ipa_call_summary_t::duplicate): ... this one.
24220 (inline_edge_removal_hook): Turn to ...
24221 (ipa_call_summary_t::remove): ... this one.
24222 (dump_inline_edge_summary): Turn to ...
24223 (dump_ipa_call_summary): ... this one.
24224 (estimate_function_body_sizes): Update.
24225 (inline_update_callee_summaries): Update.
24226 (remap_edge_change_prob): Update.
24227 (remap_edge_summaries): Update.
24228 (inline_merge_summary): Update.
24229 (do_estimate_edge_time): Update.
24230 (inline_generate_summary): Update.
24231 (inline_read_section): Update.
24232 (inline_read_summary): Update.
24233 (inline_free_summary): Update.
24234 * ipa-inline.c (can_inline_edge_p): Update.
24235 (compute_inlined_call_time): Update.
24236 (want_inline_small_function_p): Update.
24237 (edge_badness): Update.
24238 (early_inliner): Update.
24239 * ipa-inline.h (inline_edge_summary): Turn to ...
24240 (ipa_call_summary): ... this one.
24241 (ipa_call_summary_t): New class.
24242 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
24243 (ipa_call_summaries): New.
24244 (inline_edge_summary): Remove.
24245 (estimate_edge_growth): Update.
24246 * ipa-profile.c (ipa_propagate_frequency_1): Update.
24247 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
24248 * ipa-split.c (execute_split_functions): Update.
24249 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
24251 2017-05-19 Richard Biener <rguenther@suse.de>
24253 PR middle-end/80764
24254 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
24256 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
24258 * config/rs6000/rs6000.c (struct machine_function): Add field
24259 fpr_is_wrapped_separately.
24260 (rs6000_get_separate_components): Use 64 components. Handle the
24261 new FPR components.
24262 (rs6000_components_for_bb): Handle the FPR components.
24263 (rs6000_emit_prologue_components): Handle the FPR components.
24264 (rs6000_emit_epilogue_components): Handle the FPR components.
24265 (rs6000_set_handled_components): Handle the FPR components.
24266 (rs6000_emit_prologue): Don't output prologue code for those FPRs
24267 that are already separately shrink-wrapped.
24268 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
24269 that are already separately shrink-wrapped.
24271 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24274 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
24277 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
24278 (define_peephole2 for Altivec d-form load): Add peepholes to catch
24279 cases where the register allocator uses a move and an offsettable
24280 memory operation to/from a FPR register on ISA 2.06/2.07.
24281 (define_peephole2 for Altivec d-form store): Likewise.
24283 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
24286 * config/i386/mmx.md (*mov<mode>_internal): Enable
24287 alternatives 11, 12, 13 and 14 also for 32bit targets.
24288 Remove alternatives 15, 16, 17 and 18.
24289 * config/i386/sse.md (vec_concatv2di): Change
24290 alternative (!x, *y) to (x, ?!*Yn).
24292 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
24294 * dumpfile.h (enum dump_kind): Remove stray comma.
24296 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24298 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
24299 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
24300 predicate::num_conditions
24301 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
24302 (CHANGED): turn into predicate::changed.
24303 (agg_position_info): Move to ipa-predicate.h
24304 (add_condition, predicate::add_clause, predicate::operator &=,
24305 predicate::or_with, predicate::evaluate, predicate::probability,
24306 dump_condition, dump_clause, predicate::dump,
24307 predicate::remap_after_duplication, predicate::remap_after_inlining,
24308 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
24309 (evaluate_conditions_for_known_args): Update.
24310 (set_cond_stmt_execution_predicate): Update.
24311 * ipa-inline.h: Include ipa-predicate.h
24312 (condition, inline_param_summary, conditions, agg_position_info,
24313 predicate): Move to ipa-predicate.h
24314 * ipa-predicate.c: New file.
24315 * ipa-predicate.h: New file.
24317 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
24319 * final.c (leaf_function_p): Check we are not in a sequence.
24321 2017-05-18 Martin Liska <mliska@suse.cz>
24323 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
24324 * dumpfile.c (dump_register): Use new enum dump_kind.
24325 (get_dump_file_name): Likewise.
24326 (dump_enable_all): Likewise.
24327 (dump_switch_p_1): Likewise.
24328 (enable_rtl_dump_file): Remove usage of TDF_RTL.
24329 * dumpfile.h (enum dump_kind): New enum type.
24330 (struct dump_file_info): Create constructor and
24331 format fields and comments.
24332 * passes.c (pass_manager::register_one_dump_file):
24334 * statistics.c (statistics_early_init): Likewise.
24335 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
24336 TDF_TREE with TDF_SLIM.
24337 (gather_memory_references_ref): Likewise.
24339 2017-05-18 Martin Liska <mliska@suse.cz>
24341 * vec.h (struct vnull): Use it.
24343 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
24345 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
24346 (true_predicate, false_predicate, true_predicate_p,
24347 false_predicate_p): Remove.
24348 (single_cond_predicate, not_inlined_predicate): Turn to member function
24350 (add_condition): Update.
24351 (add_clause): Turn to...
24352 (predicate::add_clause): ... this one; update; allow passing NULL
24354 (and_predicates): Turn to ...
24355 (predicate::operator &=): ... this one.
24356 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
24357 (or_predicates): Turn to ...
24358 (predicate::or_with): ... this one.
24359 (evaluate_predicate): Turn to ...
24360 (predicate::evaluate): ... this one.
24361 (predicate_probability): Turn to ...
24362 (predicate::probability): ... this one.
24363 (dump_condition): Update.
24364 (dump_predicate): Turn to ...
24365 (predicate::dump): ... this one.
24366 (account_size_time): Update.
24367 (edge_set_predicate): Update.
24368 (set_hint_predicate): UPdate.
24369 (evaluate_conditions_for_known_args): Update.
24370 (evaluate_properties_for_edge): Update.
24371 (remap_predicate_after_duplication): Turn to...
24372 (predicate::remap_after_duplication): ... this one.
24373 (remap_hint_predicate_after_duplication): Update.
24374 (inline_summary_t::duplicate): UPdate.
24375 (dump_inline_edge_summary): Update.
24376 (dump_inline_summary): Update.
24377 (set_cond_stmt_execution_predicate): Update.
24378 (set_switch_stmt_execution_predicate): Update.
24379 (compute_bb_predicates): Update.
24380 (will_be_nonconstant_expr_predicate): Update.
24381 (will_be_nonconstant_predicate): Update.
24382 (phi_result_unknown_predicate): Update.
24383 (predicate_for_phi_result): Update.
24384 (array_index_predicate): Update.
24385 (estimate_function_body_sizes): Update.
24386 (estimate_node_size_and_time): Update.
24387 (estimate_ipcp_clone_size_and_time): Update.
24388 (remap_predicate): Rename to ...
24389 (predicate::remap_after_inlining): ... this one.
24390 (remap_hint_predicate): Update.
24391 (inline_merge_summary): Update.
24392 (inline_update_overall_summary): Update.
24393 (estimate_size_after_inlining): Update.
24394 (read_predicate): Rename to ...
24395 (predicate::stream_in): ... this one.
24396 (read_inline_edge_summary): Update.
24397 (write_predicate): Rename to ...
24398 (predicate::stream_out): ... this one.
24399 (write_inline_edge_summary): Update.
24400 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
24401 (clause_t): Turn to uint32_t
24402 (predicate): Turn to class; implement constructor and operators
24404 (size_time_entry): Update.
24405 (inline_summary): Update.
24406 (inline_edge_summary): Update.
24408 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
24410 * fold-const.c (fold_binary_loc): Move transformation...
24411 * match.pd (C - X CMP X): ... here.
24413 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
24415 * config/sparc/sparc.c (sparc_option_override): Set function
24416 alignment for -mcpu=niagara7 to 64 to match the I$ line.
24417 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
24419 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24421 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24423 2017-05-18 Marek Polacek <polacek@redhat.com>
24426 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24427 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24428 gimple_assign_load_p.
24430 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24432 PR middle-end/80692
24433 * real.c (do_compare): Give decimal_do_compare preference over
24434 comparing just the signs.
24436 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24438 * doc/md.texi (Canonicalization of Instructions): Describe the
24439 canonical form of instructions that inherently set a condition
24442 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24444 PR middle-end/80775
24445 * tree-cfg.c: Move deletion of unreachable case statements to after
24446 the merging of consecutive case labels.
24448 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24450 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24451 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24452 restoring of callee-saved registers.
24454 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24456 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24457 * config/visium/visium.c (single_set_and_flags): Likewise.
24458 * config/visium/visium.md (Substitutions): Likewise.
24460 2017-05-17 Martin Liska <mliska@suse.cz>
24462 * cfg.c: Introduce dump_flags_t type and
24463 use it instead of int type.
24465 * cfghooks.c: Likewise.
24466 * cfghooks.h (struct cfg_hooks): Likewise.
24467 * cfgrtl.c: Likewise.
24468 * cfgrtl.h: Likewise.
24469 * cgraph.c (cgraph_node::get_body): Likewise.
24470 * coretypes.h: Likewise.
24471 * domwalk.c: Likewise.
24472 * domwalk.h: Likewise.
24473 * dumpfile.c (struct dump_option_value_info): Likewise.
24474 (dump_enable_all): Likewise.
24475 (dump_switch_p_1): Likewise.
24476 (opt_info_switch_p): Likewise.
24477 * dumpfile.h (enum tree_dump_index): Likewise.
24478 (struct dump_file_info): Likewise.
24479 * genemit.c: Likewise.
24480 * generic-match-head.c: Likewise.
24481 * gengtype.c (open_base_files): Likewise.
24482 * gimple-pretty-print.c: Likewise.
24483 * gimple-pretty-print.h: Likewise.
24484 * graph.c (print_graph_cfg): Likewise.
24485 * graphite-scop-detection.c (dot_all_sese): Likewise.
24486 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24487 * loop-unroll.c (report_unroll): Likewise.
24488 * passes.c (pass_manager::register_one_dump_file): Likewise.
24489 * print-tree.c: Likewise.
24490 * statistics.c: Likewise.
24491 * tree-cfg.c: Likewise.
24492 * tree-cfg.h: Likewise.
24493 * tree-dfa.c: Likewise.
24494 * tree-dfa.h: Likewise.
24495 * tree-dump.c (dump_function): Likewise.
24496 * tree-dump.h (struct dump_info): Likewise.
24497 * tree-pretty-print.c: Likewise.
24498 * tree-pretty-print.h: Likewise.
24499 * tree-ssa-live.c: Likewise.
24500 * tree-ssa-live.h: Likewise.
24501 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24502 * tree-vect-loop.c: Likewise.
24503 * tree-vect-slp.c: Likewise.
24505 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24506 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24508 PR tree-optimization/80457
24509 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24510 of all arguments to a statement as scalar_to_vec operations.
24511 (vectorizable_call): Adjust call to vect_model_simple_cost for
24513 (vectorizable_conversion): Likewise.
24514 (vectorizable_assignment): Likewise.
24515 (vectorizable_shift): Likewise.
24516 (vectorizable_operation): Likewise.
24517 (vectorizable_comparison): Likewise.
24518 (vect_is_simple_cond): Record the def types for operands.
24519 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24520 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24521 for statement argument count.
24523 2017-05-16 Carl Love <cel@us.ibm.com>
24525 * config/rs6000/rs6000-c: Add support for built-in functions
24526 vector unsigned long long vec_bperm (vector unsigned long long,
24527 vector unsigned char)
24528 vector signed long long vec_mule (vector signed int,
24530 vector unsigned long long vec_mule (vector unsigned int,
24531 vector unsigned int)
24532 vector signed long long vec_mulo (vector signed int,
24534 vector unsigned long long vec_mulo (vector unsigned int,
24535 vector unsigned int)
24536 vector signed char vec_sldw (vector signed char,
24537 vector signed char,
24539 vector unsigned char vec_sldw (vector unsigned char,
24540 vector unsigned char,
24542 vector signed short vec_sldw (vector signed short,
24543 vector signed short,
24545 vector unsigned short vec_sldw (vector unsigned short,
24546 vector unsigned short,
24548 vector signed int vec_sldw (vector signed int,
24551 vector unsigned int vec_sldw (vector unsigned int,
24552 vector unsigned int,
24554 vector signed long long vec_sldw (vector signed long long,
24555 vector signed long long,
24557 vector unsigned long long vec_sldw (vector unsigned long long,
24558 vector unsigned long long,
24560 * config/rs6000/rs6000-c: Add support for built-in functions
24561 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24562 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24563 * doc/extend.texi: Update the built-in documentation for the
24564 new built-in functions.
24566 2017-05-16 Marek Polacek <polacek@redhat.com>
24570 * tree.c (save_expr): Don't fold the expression.
24572 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24574 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24575 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24576 and (?*y,m). Update insn attributes.
24578 2017-05-16 Martin Liska <mliska@suse.cz>
24580 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24581 flags argument of print_gimple_stmt, print_gimple_expr,
24582 print_generic_stmt and print_generic_expr.
24583 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24584 * coretypes.h: Likewise.
24585 * except.c (dump_eh_tree): Likewise.
24586 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24587 * gimple-pretty-print.h: Likewise.
24588 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24589 (backprop::push_to_worklist): Likewise.
24590 (backprop::pop_from_worklist): Likewise.
24591 (backprop::process_use): Likewise.
24592 (backprop::intersect_uses): Likewise.
24593 (note_replacement): Likewise.
24594 * gimple-ssa-store-merging.c
24595 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24596 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24597 (pass_store_merging::execute): Likewise.
24598 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24599 (ssa_base_cand_dump_callback): Likewise.
24600 (dump_incr_vec): Likewise.
24601 (replace_refs): Likewise.
24602 (replace_mult_candidate): Likewise.
24603 (create_add_on_incoming_edge): Likewise.
24604 (create_phi_basis): Likewise.
24605 (insert_initializers): Likewise.
24606 (all_phi_incrs_profitable): Likewise.
24607 (introduce_cast_before_cand): Likewise.
24608 (replace_one_candidate): Likewise.
24609 * gimplify.c (gimplify_expr): Likewise.
24610 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24611 (set_rename): Likewise.
24612 (rename_uses): Likewise.
24613 (copy_loop_phi_nodes): Likewise.
24614 (add_close_phis_to_merge_points): Likewise.
24615 (copy_loop_close_phi_args): Likewise.
24616 (copy_cond_phi_args): Likewise.
24617 (graphite_copy_stmts_from_block): Likewise.
24618 (translate_pending_phi_nodes): Likewise.
24619 * graphite-poly.c (print_pdr): Likewise.
24620 (dump_gbb_cases): Likewise.
24621 (dump_gbb_conditions): Likewise.
24622 (print_scop_params): Likewise.
24623 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24624 (build_cross_bb_scalars_use): Likewise.
24625 (gather_bbs::before_dom_children): Likewise.
24626 * hsa-dump.c (dump_hsa_immed): Likewise.
24627 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24628 (get_replacement_map): Likewise.
24629 * ipa-inline-analysis.c (dump_condition): Likewise.
24630 (estimate_function_body_sizes): Likewise.
24631 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24632 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24633 * ipa-prop.c (ipa_dump_param): Likewise.
24634 (ipa_print_node_jump_functions_for_edge): Likewise.
24635 (ipa_modify_call_arguments): Likewise.
24636 (ipa_modify_expr): Likewise.
24637 (ipa_dump_param_adjustments): Likewise.
24638 (ipa_dump_agg_replacement_values): Likewise.
24639 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24640 * ipa-pure-const.c (check_stmt): Likewise.
24641 (pass_nothrow::execute): Likewise.
24642 * ipa-split.c (execute_split_functions): Likewise.
24643 * omp-offload.c (dump_oacc_loop_part): Likewise.
24644 (dump_oacc_loop): Likewise.
24645 * trans-mem.c (tm_log_emit): Likewise.
24646 (tm_memopt_accumulate_memops): Likewise.
24647 (dump_tm_memopt_set): Likewise.
24648 (dump_tm_memopt_transform): Likewise.
24649 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24650 (print_loop): Likewise.
24651 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24652 (chkp_gather_checks_info): Likewise.
24653 (chkp_get_check_result): Likewise.
24654 (chkp_remove_check_if_pass): Likewise.
24655 (chkp_use_outer_bounds_if_possible): Likewise.
24656 (chkp_reduce_bounds_lifetime): Likewise.
24657 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24658 (chkp_mark_completed_bounds): Likewise.
24659 (chkp_register_incomplete_bounds): Likewise.
24660 (chkp_mark_invalid_bounds): Likewise.
24661 (chkp_maybe_copy_and_register_bounds): Likewise.
24662 (chkp_build_returned_bound): Likewise.
24663 (chkp_get_bound_for_parm): Likewise.
24664 (chkp_build_bndldx): Likewise.
24665 (chkp_get_bounds_by_definition): Likewise.
24666 (chkp_generate_extern_var_bounds): Likewise.
24667 (chkp_get_bounds_for_decl_addr): Likewise.
24668 * tree-chrec.c (chrec_apply): Likewise.
24669 * tree-data-ref.c (dump_data_reference): Likewise.
24670 (dump_subscript): Likewise.
24671 (dump_data_dependence_relation): Likewise.
24672 (analyze_overlapping_iterations): Likewise.
24673 * tree-inline.c (expand_call_inline): Likewise.
24674 (tree_function_versioning): Likewise.
24675 * tree-into-ssa.c (dump_defs_stack): Likewise.
24676 (dump_currdefs): Likewise.
24677 (dump_names_replaced_by): Likewise.
24678 (dump_update_ssa): Likewise.
24679 (update_ssa): Likewise.
24680 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24681 * tree-parloops.c (build_new_reduction): Likewise.
24682 (try_create_reduction_list): Likewise.
24683 (ref_conflicts_with_region): Likewise.
24684 (oacc_entry_exit_ok_1): Likewise.
24685 (oacc_entry_exit_single_gang): Likewise.
24686 * tree-pretty-print.h: Likewise.
24687 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24688 (get_scalar_evolution): Likewise.
24689 (add_to_evolution): Likewise.
24690 (get_loop_exit_condition): Likewise.
24691 (analyze_evolution_in_loop): Likewise.
24692 (analyze_initial_condition): Likewise.
24693 (analyze_scalar_evolution): Likewise.
24694 (instantiate_scev): Likewise.
24695 (number_of_latch_executions): Likewise.
24696 (gather_chrec_stats): Likewise.
24697 (final_value_replacement_loop): Likewise.
24698 (scev_const_prop): Likewise.
24699 * tree-sra.c (dump_access): Likewise.
24700 (disqualify_candidate): Likewise.
24701 (create_access): Likewise.
24702 (reject): Likewise.
24703 (maybe_add_sra_candidate): Likewise.
24704 (create_access_replacement): Likewise.
24705 (analyze_access_subtree): Likewise.
24706 (analyze_all_variable_accesses): Likewise.
24707 (sra_modify_assign): Likewise.
24708 (initialize_constant_pool_replacements): Likewise.
24709 (find_param_candidates): Likewise.
24710 (decide_one_param_reduction): Likewise.
24711 (replace_removed_params_ssa_names): Likewise.
24712 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24713 * tree-ssa-copy.c (dump_copy_of): Likewise.
24714 (copy_prop_visit_cond_stmt): Likewise.
24715 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24716 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24717 (record_equivalences_from_stmt): Likewise.
24718 * tree-ssa-dse.c (compute_trims): Likewise.
24719 (delete_dead_call): Likewise.
24720 (delete_dead_assignment): Likewise.
24721 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24722 (forward_propagate_into_cond): Likewise.
24723 (pass_forwprop::execute): Likewise.
24724 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24725 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24727 (move_computations_worker): Likewise.
24728 (execute_sm): Likewise.
24729 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24730 (remove_exits_and_undefined_stmts): Likewise.
24731 (remove_redundant_iv_tests): Likewise.
24732 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24733 (adjust_iv_update_pos): Likewise.
24734 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24735 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24736 (value_replacement): Likewise.
24737 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24738 * tree-ssa-pre.c (print_pre_expr): Likewise.
24739 (get_representative_for): Likewise.
24740 (create_expression_by_pieces): Likewise.
24741 (insert_into_preds_of_block): Likewise.
24742 (eliminate_insert): Likewise.
24743 (eliminate_dom_walker::before_dom_children): Likewise.
24744 (eliminate): Likewise.
24745 (remove_dead_inserted_code): Likewise.
24746 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24747 * tree-ssa-reassoc.c (get_rank): Likewise.
24748 (eliminate_duplicate_pair): Likewise.
24749 (eliminate_plus_minus_pair): Likewise.
24750 (eliminate_not_pairs): Likewise.
24751 (undistribute_ops_list): Likewise.
24752 (eliminate_redundant_comparison): Likewise.
24753 (update_range_test): Likewise.
24754 (optimize_range_tests_var_bound): Likewise.
24755 (optimize_vec_cond_expr): Likewise.
24756 (rewrite_expr_tree): Likewise.
24757 (rewrite_expr_tree_parallel): Likewise.
24758 (linearize_expr): Likewise.
24759 (break_up_subtract): Likewise.
24760 (linearize_expr_tree): Likewise.
24761 (attempt_builtin_powi): Likewise.
24762 (attempt_builtin_copysign): Likewise.
24763 (transform_stmt_to_copy): Likewise.
24764 (transform_stmt_to_multiply): Likewise.
24765 (dump_ops_vector): Likewise.
24766 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24767 (print_scc): Likewise.
24768 (set_ssa_val_to): Likewise.
24769 (visit_reference_op_store): Likewise.
24770 (visit_use): Likewise.
24771 (sccvn_dom_walker::before_dom_children): Likewise.
24772 (run_scc_vn): Likewise.
24773 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24775 (expr_hash_elt::print): Likewise.
24776 (const_and_copies::pop_to_marker): Likewise.
24777 (const_and_copies::record_const_or_copy_raw): Likewise.
24778 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24779 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24780 (dump_predicates): Likewise.
24781 (find_uninit_use): Likewise.
24782 (warn_uninitialized_phi): Likewise.
24783 (pass_late_warn_uninitialized::execute): Likewise.
24784 * tree-ssa.c (verify_vssa): Likewise.
24785 (verify_ssa): Likewise.
24786 (maybe_optimize_var): Likewise.
24787 * tree-vrp.c (dump_value_range): Likewise.
24788 (dump_all_value_ranges): Likewise.
24789 (dump_asserts_for): Likewise.
24790 (register_edge_assert_for_2): Likewise.
24791 (vrp_visit_cond_stmt): Likewise.
24792 (vrp_visit_switch_stmt): Likewise.
24793 (vrp_visit_stmt): Likewise.
24794 (vrp_visit_phi_node): Likewise.
24795 (simplify_cond_using_ranges_1): Likewise.
24796 (fold_predicate_in): Likewise.
24797 (evrp_dom_walker::before_dom_children): Likewise.
24798 (evrp_dom_walker::push_value_range): Likewise.
24799 (evrp_dom_walker::pop_value_range): Likewise.
24800 (execute_early_vrp): Likewise.
24802 2017-05-16 Richard Biener <rguenther@suse.de>
24804 * dwarf2out.c (loc_list_from_tree_1): Do not create
24805 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24807 2017-05-16 Richard Biener <rguenther@suse.de>
24809 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24811 (note_variable_value_in_expr): If we resolved the decl ref
24812 do not push to the stack.
24814 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24816 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24817 operations in fast-math mode.
24818 (vaddq_f16): Likewise.
24819 (vmul_f16): Likewise.
24820 (vmulq_f16): Likewise.
24821 (vsub_f16): Likewise.
24822 (vsubq_f16): Likewise.
24823 * config/arm/neon.md (add<mode>3): New.
24825 (fma:<VH:mode>3): New. Also remove outdated comment.
24828 2017-05-16 Martin Liska <mliska@suse.cz>
24832 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24833 (odr_types_equivalent_p): Likewise.
24835 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24837 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24839 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24842 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24843 non-interunit SSE move alternatives with '?'.
24844 (zero-extendsidi peephole2): New peephole to skip intermediate
24845 general register in SSE zero-extend sequence.
24847 2017-05-15 Jeff Law <law@redhat.com>
24849 * reorg.c (relax_delay_slots): Create a new variable to hold
24850 the temporary target rather than clobbering TARGET_LABEL.
24852 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24853 missing argument to extract_bit_field call.
24854 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24856 2017-05-15 Martin Liska <mliska@suse.cz>
24859 * gcc.c (process_command): Do not allow empty argument of -o option.
24861 2017-05-15 Renlin Li <renlin.li@arm.com>
24863 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24864 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24865 * config/aarch64/constraints.md (Usf): Add long call check.
24866 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24867 (call_value): Likewise.
24868 (sibcall): Likewise.
24869 (sibcall_value): Likewise.
24871 (call_value_insn): New.
24872 (sibcall_insn): Update rtx pattern.
24873 (sibcall_value_insn): Likewise.
24874 (call_internal): Remove.
24875 (call_value_internal): Likewise.
24876 (sibcall_internal): Likewise.
24877 (sibcall_value_internal): Likewise.
24878 (call_reg): Likewise.
24879 (call_symbol): Likewise.
24880 (call_value_reg): Likewise.
24881 (call_value_symbol): Likewise.
24883 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24886 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24888 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24890 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24891 compatible with CCGOCmode and with CCZmode.
24893 2017-05-14 Martin Sebor <msebor@redhat.com>
24895 PR middle-end/77671
24896 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24897 (gimple_fold_builtin_snprintf): Same.
24898 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24899 (gimple_fold_builtin_snprintf): Same.
24900 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24901 of character types.
24902 (is_call_safe): New function.
24903 (try_substitute_return_value): Call it.
24904 (try_simplify_call): New function.
24905 (pass_sprintf_length::handle_gimple_call): Call it.
24907 2017-05-14 Martin Sebor <msebor@redhat.com>
24909 PR middle-end/80669
24910 * builtins.c (expand_builtin_stpncpy): Simplify.
24912 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24914 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24915 * config/i386/i386.h
24916 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24917 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24918 (struct machine_function): Add new members call_ms2sysv,
24919 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24920 (struct machine_frame_state): New fields sp_realigned and
24921 sp_realigned_offset.
24922 * config/i386/i386.c
24923 (enum xlogue_stub): New enum.
24924 (enum xlogue_stub_sets): New enum.
24925 (class xlogue_layout): New class.
24926 (struct ix86_frame): New fields stack_realign_allocate_offset,
24927 stack_realign_offset and outlined_save_offset. Modify comments to
24928 detail stack layout when using out-of-line stubs.
24929 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24930 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24931 -mcall-ms2sysv-xlogues.
24932 (stub_managed_regs): New static variable.
24933 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24934 registers managed by out-of-line stub.
24935 (disable_call_ms2sysv_xlogues): New function.
24936 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24937 m->call_ms2sysv when appropriate and compute frame layout for
24939 (sp_valid_at, fp_valid_at): New inline functions.
24940 (choose_basereg): New function.
24941 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24943 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24944 Use align parameter of choose_baseaddr to generated aligned SSE movs
24946 (pro_epilogue_adjust_stack): Modify to track
24947 machine_frame_state::sp_realigned.
24948 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24949 (ix86_nsaved_sseregs): Likewise.
24950 (ix86_emit_save_regs): Likewise.
24951 (ix86_emit_save_regs_using_mov): Likewise.
24952 (ix86_emit_save_sse_regs_using_mov): Likewise.
24953 (get_scratch_register_on_entry): Likewise.
24954 (gen_frame_set): New function.
24955 (gen_frame_load): Likewise.
24956 (gen_frame_store): Likewise.
24957 (emit_outlined_ms2sysv_save): Likewise.
24958 (emit_outlined_ms2sysv_restore): Likewise.
24959 (ix86_expand_prologue): Modify stack re-alignment code and call
24960 emit_outlined_ms2sysv_save when appropriate.
24961 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24962 parameter rtx_insn *insn, which allows the function to be used to only
24963 generate the notes.
24964 (ix86_expand_epilogue): Modify validity checks of frame and stack
24965 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24966 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24967 * config/i386/predicates.md
24968 (save_multiple): New predicate.
24969 (restore_multiple): Likewise.
24970 * config/i386/sse.md
24971 (save_multiple<mode>): New pattern.
24972 (save_multiple_realign<mode>): Likewise.
24973 (restore_multiple<mode>): Likewise.
24974 (restore_multiple_and_return<mode>): Likewise.
24975 (restore_multiple_leave_return<mode>): Likewise.
24976 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24978 2017-05-14 Julia Koval <julia.koval@intel.com>
24980 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24981 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24982 (__builtin_ia32_xsetbv): New builtins.
24983 * config/i386/i386.c (ix86_expand_special_args_builtin):
24985 (ix86_expand_builtin): Special expand for new intrinsics.
24986 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24987 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24988 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24990 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24992 * cfganal.c (inverted_post_order_compute): Change argument type
24994 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24995 * df-core.c (rest_of_handle_df_initialize): Adjust.
24996 (rest_of_handle_df_finish): Likewise.
24997 (df_analyze_1): Likewise.
24998 (df_analyze): Likewise.
24999 (loop_inverted_post_order_compute): Change argument to be a vec *.
25000 (df_analyze_loop): Adjust.
25001 (df_get_n_blocks): Likewise.
25002 (df_get_postorder): Likewise.
25003 * df.h (struct df_d): Change field to be a vec.
25004 * lcm.c (compute_laterin): Adjust.
25005 (compute_available): Likewise.
25006 * lra-lives.c (lra_create_live_ranges_1): Likewise.
25007 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
25008 * tree-ssa-pre.c (compute_antic): Likewise.
25010 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25012 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
25013 (depth_first_search::depth_first_search): Change structure init
25014 function to this constructor.
25015 (depth_first_search::add_bb): Rename function to this member.
25016 (depth_first_search::execute): Likewise.
25017 (flow_dfs_compute_reverse_finish): Adjust.
25019 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25021 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
25022 (longest_simple_path): Likewise.
25023 * shrink-wrap.c (spread_components): Likewise.
25024 (disqualify_problematic_components): Likewise.
25025 (emit_common_heads_for_components): Likewise.
25026 (emit_common_tails_for_components): Likewise.
25027 (insert_prologue_epilogue_for_components): Likewise.
25029 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25031 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
25034 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25036 * df-core.c (df_set_blocks): Start using auto_bitmap.
25037 (df_compact_blocks): Likewise.
25038 * df-problems.c (df_rd_confluence_n): Likewise.
25039 * df-scan.c (df_insn_rescan_all): Likewise.
25040 (df_process_deferred_rescans): Likewise.
25041 (df_update_entry_block_defs): Likewise.
25042 (df_update_exit_block_uses): Likewise.
25043 (df_entry_block_bitmap_verify): Likewise.
25044 (df_exit_block_bitmap_verify): Likewise.
25045 (df_scan_verify): Likewise.
25046 * lra-constraints.c (lra_constraints): Likewise.
25047 (undo_optional_reloads): Likewise.
25048 (lra_undo_inheritance): Likewise.
25049 * lra-remat.c (calculate_gen_cands): Likewise.
25050 (do_remat): Likewise.
25051 * lra-spills.c (assign_spill_hard_regs): Likewise.
25052 (spill_pseudos): Likewise.
25053 * tree-ssa-pre.c (bitmap_set_and): Likewise.
25054 (bitmap_set_subtract_values): Likewise.
25056 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25058 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
25059 management with auto_bitmap.
25060 (fix_inter_tick): Likewise.
25061 (fix_recovery_deps): Likewise.
25062 * ira.c (add_store_equivs): Likewise.
25063 (find_moveable_pseudos): Likewise.
25064 (split_live_ranges_for_shrink_wrap): Likewise.
25065 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
25066 (rtx_reuse_manager::seen_def_p): Likewise.
25067 (rtx_reuse_manager::set_seen_def): Likewise.
25068 * print-rtl.h (class rtx_reuse_manager): Likewise.
25070 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25072 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
25074 (migrate_btr_def): Likewise.
25075 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
25076 * df-core.c (loop_post_order_compute): Likewise.
25077 (loop_inverted_post_order_compute): Likewise.
25078 * hsa-common.h: Likewise.
25079 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
25080 * init-regs.c (initialize_uninitialized_regs): Likewise.
25081 * ipa-inline.c (resolve_noninline_speculation): Likewise.
25082 (inline_small_functions): Likewise.
25083 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
25084 * ira.c (combine_and_move_insns): Likewise.
25085 (build_insn_chain): Likewise.
25086 * loop-invariant.c (find_invariants): Likewise.
25087 * lower-subreg.c (propagate_pseudo_copies): Likewise.
25088 * predict.c (tree_predict_by_opcode): Likewise.
25089 (predict_paths_leading_to): Likewise.
25090 (predict_paths_leading_to_edge): Likewise.
25091 (estimate_loops_at_level): Likewise.
25092 (estimate_loops): Likewise.
25093 * shrink-wrap.c (try_shrink_wrapping): Likewise.
25094 (spread_components): Likewise.
25095 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
25096 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
25097 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
25098 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
25099 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
25100 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
25101 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
25102 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
25103 (mark_threaded_blocks): Likewise.
25104 (thread_through_all_blocks): Likewise.
25105 * tree-ssa.c (verify_ssa): Likewise.
25106 (execute_update_addresses_taken): Likewise.
25107 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
25109 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25111 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
25113 (post_order_compute): Likewise.
25114 (inverted_post_order_compute): Likewise.
25115 (pre_and_rev_post_order_compute_fn): Likewise.
25117 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25119 * genrecog.c (int_set::int_set): Explicitly construct our
25120 auto_vec base class.
25121 * vec.h (auto_vec::auto_vec): New constructor.
25123 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25125 * bitmap.h (class auto_bitmap): New constructor taking
25126 bitmap_obstack * argument.
25128 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
25130 * bitmap.h (class auto_bitmap): Change type of m_bits to
25131 bitmap_head, and adjust ctor / dtor and member operators.
25133 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
25135 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
25136 when returned register mode doesn't match original mode.
25138 2017-05-12 Jeff Law <law@redhat.com>
25139 Jakub Jelinek <jakub@redhat.com>
25141 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
25142 we look for cc setter after the compare-elim changes.
25143 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
25144 within the vector to match what compare-elim now expects.
25145 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
25146 (xorsi3_flags, one_cmplsi2_flags): Likewise.
25148 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
25149 after the compare-elim changes.
25150 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
25151 the vector to match what compare-elim now expects.
25152 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
25153 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
25154 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
25155 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
25156 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
25158 * config/visium/visium.c (single_set_and_flags): Fix where
25159 we look for cc setter after the compare-elim changes.
25160 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
25161 with the vector to match what compare-elim now expects.
25162 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
25163 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
25164 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
25165 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
25166 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
25167 (neg<mode>2_insn_set_overflow): Likewise.
25169 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
25171 PR middle-end/79794
25172 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
25173 maybe_expand_insn call, set ops[0].target. If still set after call,
25174 set alt_rtl. Add extra arg to recursive calls.
25175 (extract_bit_field): Add alt_rtl argument. Pass to
25177 * expmed.h (extract_bit_field): Fix prototype.
25178 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
25179 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
25180 to extract_bit_field_calls.
25181 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
25182 Pass alt_rtl to extract_bit_field calls.
25183 * calls.c (store_unaligned_arguments_into_psuedos)
25184 load_register_parameters): Pass extra NULL to extract_bit_field calls.
25185 * optabs.c (maybe_legitimize_operand): Clear op->target when call
25187 * optabs.h (struct expand_operand): Add target bitfield.
25189 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25191 * compare-elim.c (try_eliminate_compare): Canonicalize
25192 operation with embedded compare to
25193 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
25194 (set (reg) (operation)].
25196 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
25198 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
25201 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
25202 cost of adding a carry flag for ADC instruction.
25203 [case MINUS]: Ignore the cost of subtracting a carry flag
25204 for SBB instruction.
25206 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
25208 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
25210 * config/rs6000/bmiintrin.h: New file.
25211 * config/rs6000/bmi2intrin.h: New file.
25212 * config/rs6000/x86intrin.h: New file.
25214 2017-05-12 Jeff Law <law@redhat.com>
25216 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
25219 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
25221 PR middle-end/80707
25222 * tree-cfg.c: Remove cfg edges of unreachable case statements.
25224 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25226 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25227 early expansion of vector divide builtins.
25228 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
25229 builtins identified as having unsigned arguments.
25231 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25233 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
25234 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
25235 expansion of vector logical operations (and, andc, or, xor,
25238 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25240 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
25241 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
25243 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
25245 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
25246 early GIMPLE expansion of vector multiplies.
25248 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25250 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
25251 TARGET_HAVE_MOVT conditional.
25252 (movt splitter): Likewise.
25254 2017-05-12 Richard Biener <rguenther@suse.de>
25256 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
25257 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25258 Fold all stmts not inplace.
25260 2017-05-12 Richard Biener <rguenther@suse.de>
25262 PR tree-optimization/80713
25263 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
25264 inserted_exprs bit for not removed stmts.
25266 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
25268 PR middle-end/69921
25269 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
25270 parallelized" attribute for parallelized OpenACC kernels.
25271 * omp-offload.c (execute_oacc_device_lower): Use it.
25273 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
25274 Set "oacc kernels" attribute.
25275 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
25276 parameter. Adjust all users.
25277 (oacc_fn_attrib_kernels_p): Remove function.
25278 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
25279 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
25280 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
25281 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
25282 assert "oacc kernels" attribute is set.
25284 2017-05-11 Carl Love <cel@us.ibm.com>
25286 * config/rs6000/rs6000-c: Add support for built-in functions
25287 vector unsigned char vec_popcnt (vector signed char)
25288 vector unsigned char vec_popcnt (vector unsigned char)
25289 vector unsigned short vec_popcnt (vector signed short)
25290 vector unsigned short vec_popcnt (vector unsigned short)
25291 vector unsigned int vec_popcnt (vector signed int)
25292 vector unsigned int vec_popcnt (vector unsigned int)
25293 vector unsigned long long vec_popcnt (vector signed long long)
25294 vector unsigned long long vec_popcnt (vector unsigned long long)
25295 vector signed long long vec_slo (vector signed long long,
25296 vector signed char)
25297 vector signed long long vec_slo (vector signed long long,
25298 vector unsigned char)
25299 vector unsigned long long vec_slo (vector unsigned long long,
25300 vector signed char)
25301 vector unsigned long long vec_slo (vector unsigned long long,
25302 vector unsigned char)
25303 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
25304 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
25305 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
25306 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
25307 * doc/extend.texi: Update the built-in documentation file for the
25308 new built-in functions.
25310 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
25312 * attribs.h (sorted_attr_string): Move machine independent
25313 functions for target clone support from the i386 port to common
25314 code. Rename ix86_function_versions to common_function_versions.
25315 Rename make_name to make_unique_name.
25316 (common_function_versions): Likewise.
25317 (make_unique_name): Likewise.
25318 (make_dispatcher_decl): Likewise.
25319 (is_function_default_version): Likewise.
25320 * attribs.c (attr_strcmp): Likewise.
25321 (sorted_attr_string): Likewise.
25322 (common_function_versions): Likewise.
25323 (make_unique_name): Likewise.
25324 (make_dispatcher_decl): Likewise.
25325 (is_function_default_version): Likewise.
25326 * config/i386/i386.c (attr_strcmp): Likewise.
25327 (sorted_attr_string): Likewise.
25328 (ix86_function_versions): Likewise.
25329 (make_name): Likewise.
25330 (make_dispatcher_decl): Likewise.
25331 (is_function_default_version): Likewise.
25332 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
25334 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25337 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
25338 Account for direct move costs for vec_construct of integer
25341 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
25344 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
25345 (UNSPEC_STX_ATOMIC): Ditto.
25346 (loaddi_via_sse): New insn.
25347 (storedi_via_sse): Ditto.
25348 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
25349 Update corresponding peephole2 patterns.
25350 (atomic_storedi_fpu): Ditto.
25352 2017-05-11 Julia Koval <julia.koval@intel.com>
25354 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
25355 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
25357 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
25358 (__builtin_ia32_rsqrt14ss_mask): New builtins.
25359 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
25361 2017-05-11 Nathan Sidwell <nathan@acm.org>
25363 * graphite-poly.c: Include dumpfile.h.
25365 * dumpfle.h (dump_function): Declare here ...
25366 * tree-dump.h (dump_function): ... not here.
25367 * dumpfile.c: #include tree-cfg.h.
25368 (dump_function): Move here from ...
25369 * tree-dump.c (dump_function): ... here.
25370 * gimplify.c: #include splay-tree.h, not tree-dump.h.
25371 * graphite-poly.c: Don't include tree-dump.h.
25372 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
25373 * print-tree.c: Likewise.
25374 * stor-layout.c: Likewise.
25375 * tree-nested.c: Likewise.
25377 * dumpfile.c (dump_start): Use TDF_FLAGS.
25378 (dump_enable_all): Fix TDF_KIND check thinko.
25380 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
25382 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25383 array entries to represent two legal parameterizations of the
25384 overloaded __builtin_cmpb function, as represented by the
25385 P6_OV_BUILTIN_CMPB constant.
25386 (altivec_resolve_overloaded_builtin): Add special case handling
25387 for the __builtin_cmpb function, as represented by the
25388 P6_OV_BUILTIN_CMPB constant.
25389 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
25390 (BU_P6_64BIT_2): New macro.
25391 (BU_P6_OVERLOAD_2): New macro
25392 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
25393 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
25394 (CMPB): Add overload support to represent both 32-bit and 64-bit
25395 compare-bytes function.
25396 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
25397 support for TARGET_CMPB.
25398 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
25399 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25400 documentation of the __builtin_cmpb overloaded built-in function.
25402 2017-05-11 Richard Biener <rguenther@suse.de>
25404 PR tree-optimization/80705
25405 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
25406 bases are not vectorizable.
25408 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25410 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
25411 when counting register pressure.
25413 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25415 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
25416 (iv_ca_more_deps): Renamed to ...
25417 (iv_ca_compare_deps): ... this.
25418 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25420 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25422 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25424 (determine_group_iv_costs): ... here.
25425 (find_inv_vars_cb): Record inv var if it's not recorded before.
25427 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25429 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25430 (get_shiftadd_cost): Ditto.
25432 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25434 * tree-ssa-address.c: Include header file.
25435 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25437 (add_to_parts): Refactor.
25438 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25439 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25442 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25444 PR tree-optimization/53090
25445 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25447 (extract_cond_operands): Detect condition with IV on both sides
25448 and return COMP_IV_EXPR_2.
25449 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25450 (rewrite_use_compare): Simplify by removing call to function
25451 extract_cond_operands.
25453 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25455 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25456 (extract_cond_operands): Detect condition comparing against non-
25457 invariant bound and return appropriate enum value.
25458 (find_interesting_uses_cond): Update use of extract_cond_operands.
25459 Handle its return value accordingly.
25460 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25462 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25464 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25465 nonlinear iv_use computation in loop invariant sensitive way.
25467 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25469 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25470 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25472 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25474 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25475 (dump_cand): Support iv_cand.inv_exprs.
25476 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25478 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25481 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25483 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25485 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25486 as local function. Include necessary header files.
25487 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25489 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25491 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25493 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25495 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25496 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25497 RSHIFT_EXPR and BIT_NOT_EXPR.
25499 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25501 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25502 (adjust_setup_cost): New parameter supporting round up adjustment.
25503 (struct address_cost_data): Delete.
25504 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25505 (split_address_cost, ptr_difference_cost): Delete.
25506 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25507 (struct ainc_cost_data): New struct.
25508 (get_address_cost_ainc): New function.
25509 (get_address_cost, get_computation_cost): Reimplement.
25510 (determine_group_iv_cost_address): Record inv_expr for all uses of
25512 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25513 (iv_ca_has_deps): Reimplemented to ...
25514 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25516 (iv_ca_extend): Call iv_ca_more_deps.
25518 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25520 * tree-ssa-address.c (struct mem_address): Move to header file.
25521 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25522 * tree-ssa-address.h (struct mem_address): Move from C file.
25523 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25525 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25527 * tree-affine.h (aff_combination_type): New interface.
25528 (aff_combination_zero_p): Remove static.
25529 (aff_combination_const_p): New interface.
25530 (aff_combination_singleton_var_p): New interfaces.
25532 2017-05-11 Richard Biener <rguenther@suse.de>
25534 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25535 Skip unreachable blocks and destinations.
25536 (eliminate): Move stmt removal and fixup ...
25537 (fini_eliminate): ... here. Skip inserted exprs.
25538 (pass_pre::execute): Move fini_pre after fini_eliminate.
25539 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25540 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25541 PRE to get rid of dead code that has invalid SSA form and
25542 split critical edges again.
25544 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25546 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25548 2017-05-11 Richard Biener <rguenther@suse.de>
25550 * passes.c (execute_function_todo): Verify loops if they are
25551 said to be up-to-date.
25552 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25553 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25555 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25558 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25559 handle calling assemble_external ourself.
25562 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25563 modes with zero size. Enhance comment.
25565 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25567 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25568 built-ins for vec_xl and vec_xst with short and char pointer
25571 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25573 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25574 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25575 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25576 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25577 (_mm_maskz_min_round_ss): New intrinsics.
25578 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25579 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25580 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25581 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25582 (__builtin_ia32_minss_mask_round): New builtins.
25583 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25584 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25585 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25587 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25588 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25590 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25593 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25595 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25596 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25597 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25598 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25599 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25600 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25601 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25602 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25603 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25604 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25605 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25606 (__builtin_ia32_mulss_mask_round): New builtins.
25607 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25608 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25609 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25611 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25612 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25614 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25617 2017-05-10 Julia Koval <julia.koval@intel.com>
25619 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25620 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25621 (_mm256_setr_m128i): New intrinsics.
25623 2017-05-10 Julia Koval <julia.koval@intel.com>
25625 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25626 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25627 (_mm_maskz_rcp14_ss): New intrinsics.
25628 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25629 (__builtin_ia32_rcp14ss_mask): New builtins.
25630 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25632 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25634 PR tree-optimization/51513
25635 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25636 (assert_unreachable_fallthru_edge_p): Use it.
25637 (group_case_labels_stmt): Likewise.
25638 * tree-cfg.h: Prototype it.
25639 * stmt.c: Include cfghooks.h and tree-cfg.h.
25640 (emit_case_dispatch_table) <gap_label>: New local variable.
25641 Use it to fill dispatch table gaps.
25642 Test for default_label before updating probabilities.
25643 (expand_case) <default_label>: Remove unneeded initialization.
25644 Test for unreachable default case statement and remove its edge.
25645 Set default_label accordingly.
25646 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25648 2017-05-10 Carl Love <cel@us.ibm.com>
25650 * config/rs6000/rs6000-c: Add support for built-in functions
25651 vector signed char vec_neg (vector signed char)
25652 vector signed short int vec_neg (vector short int)
25653 vector signed int vec_neg (vector signed int)
25654 vector signed long long vec_neg (vector signed long long)
25655 vector float vec_neg (vector float)
25656 vector double vec_neg (vector double)
25657 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25659 * config/rs6000/altivec.h: Add define for vec_neg
25660 * doc/extend.texi: Update the built-in documentation for the
25661 new built-in functions.
25663 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25665 PR tree-optimization/77644
25666 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25668 2017-05-10 Nathan Sidwell <nathan@acm.org>
25670 * dumpfile.h (TDI_lang_all): New.
25671 (TDF_KIND): New. Renumber others
25672 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25674 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25676 (get_dump_file_name): Adjust suffix generation.
25677 (dump_enable_all): Use TDF_KIND.
25678 * doc/invoke.texi (-fdump-lang-all): Document.
25680 * dumpfile.h: Tabify.
25682 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25685 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25686 Move member access before delete.
25688 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25690 * tree-inline.c (expand_call_inline): Split block at stmt
25693 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25696 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25697 are now unused after splitting mov{sf,sd}_hardfloat.
25698 (f32_lr2): Likewise.
25699 (f32_lm): Likewise.
25700 (f32_lm2): Likewise.
25701 (f32_li): Likewise.
25702 (f32_li2): Likewise.
25703 (f32_lv): Likewise.
25704 (f32_sr): Likewise.
25705 (f32_sr2): Likewise.
25706 (f32_sm): Likewise.
25707 (f32_sm2): Likewise.
25708 (f32_si): Likewise.
25709 (f32_si2): Likewise.
25710 (f32_sv): Likewise.
25711 (f32_dm): Likewise.
25712 (f32_vsx): Likewise.
25713 (f32_av): Likewise.
25714 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25715 For movsf, order stores so the VSX stores occur before the GPR
25716 store which encourages the register allocator to use a traditional
25717 FPR instead of a GPR. For movsd, order the stores so that the GPR
25718 store comes before the VSX stores to allow the power6 to work.
25719 This is due to the power6 not having a 32-bit integer store
25720 instruction from a FPR.
25721 (movsf_hardfloat): Likewise.
25722 (movsd_hardfloat): Likewise.
25724 2017-05-09 Martin Sebor <msebor@redhat.com>
25726 PR translation/80280
25727 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25730 PR translation/80280
25731 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25732 data member added in r247778.
25733 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25735 2017-05-09 Nathan Sidwell <nathan@acm.org>
25737 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25739 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25742 2017-05-09 Marek Polacek <polacek@redhat.com>
25744 * doc/invoke.texi: Fix typo.
25746 2017-05-09 Richard Biener <rguenther@suse.de>
25748 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25749 (vrp_val_is_min): Likewise.
25750 (set_value_range_to_value): Likewise.
25751 (set_value_range_to_nonnegative): Likewise.
25752 (gimple_assign_nonzero_p): Likewise.
25753 (gimple_stmt_nonzero_p): Likewise.
25754 (vrp_int_const_binop): Likewise. Remove unreachable case.
25755 (adjust_range_with_scev): Adjust comments.
25756 (compare_range_with_value): Likewise.
25757 (extract_range_from_phi_node): Likewise.
25758 (test_for_singularity): Likewise.
25760 2017-05-09 Richard Biener <rguenther@suse.de>
25762 * tree-vrp.c (get_single_symbol): Add assert that we don't
25763 get overflowed constants as invariant part.
25764 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25765 checks. Use wi::cmp instead of recursing for integer constants.
25766 (compare_values): Just ignore whether we assumed undefined
25767 overflow instead of failing the compare.
25768 (extract_range_for_var_from_comparison_expr): Add comment before the
25769 TREE_NO_WARNING sets.
25770 (test_for_singularity): Likewise.
25771 (extract_range_from_comparison): Do not disable optimization
25772 when we assumed undefined overflow.
25773 (extract_range_basic): Remove init of unused var.
25775 2017-05-09 Richard Biener <rguenther@suse.de>
25777 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25778 (extract_range_from_multiplicative_op_1): Adjust.
25779 (extract_range_from_binary_expr_1): Use int_const_binop.
25781 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25784 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25785 rs6000_store_data_bypass_p in seven define_bypass directives and
25786 in several comments.
25787 * config/rs6000/rs6000-protos.h: Add prototype for
25788 rs6000_store_data_bypass_p function.
25789 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25790 function implements slightly different (rs6000-specific) semantics
25791 than store_data_bypass_p, returning false rather than aborting
25792 with assertion error when arguments do not satisfy the
25793 requirements of store data bypass.
25794 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25795 rs6000_store_data_bypass_p.
25797 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25799 * config/xtensa/xtensa-protos.h
25800 (xtensa_initial_elimination_offset): New declaration.
25801 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25802 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25803 macro definition, add case for FRAME_POINTER_REGNUM when
25804 FRAME_GROWS_DOWNWARD.
25805 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25806 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25807 xtensa_initial_elimination_offset.
25809 2017-05-08 Nathan Sidwell <nathan@acm.org>
25811 * doc/invoke.texi: Alphabetize -fdump options.
25813 2017-05-08 Martin Sebor <msebor@redhat.com>
25815 PR translation/80280
25816 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25818 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25820 * target.def (compute_frame_layout): New optional target hook.
25821 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25822 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25823 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25825 * reload1.c (verify_initial_elim_offsets): Likewise.
25826 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25827 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25828 (arm_get_frame_offsets): Split up into this ...
25829 (arm_compute_frame_layout): ... and this function.
25831 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25833 * config/aarch64/constraints.md (Usa): New constraint.
25834 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25836 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25838 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25839 with_multilib_list after it has been checked.
25841 2017-05-08 Richard Biener <rguenther@suse.de>
25843 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25844 (bitmap_set_subtract_values): Likewise.
25846 2017-05-08 Richard Biener <rguenther@suse.de>
25848 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25849 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25851 (gimple_stmt_nonzero_warnv_p): Rename to ...
25852 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25854 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25855 (extract_range_basic): Adjust, do not disable propagation on
25856 strict overflow sensitive simplification.
25857 (vrp_visit_cond_stmt): Likewise.
25859 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25861 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25862 body size unconditionally.
25864 2017-05-07 Jeff Law <law@redhat.com>
25867 2017-05-06 Jeff Law <law@redhat.com>
25868 PR tree-optimization/78496
25869 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25872 PR tree-optimization/78496
25873 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25874 (simplify_stmt_using_ranges): Call it.
25875 (vrp_dom_walker::before_dom_children): Extract equivalences
25876 from an ASSERT_EXPR with an equality comparison against a
25879 2017-05-06 Jeff Law <law@redhat.com>
25881 PR tree-optimization/78496
25882 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25885 PR tree-optimization/78496
25886 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25887 (simplify_stmt_using_ranges): Call it.
25888 (vrp_dom_walker::before_dom_children): Extract equivalences
25889 from an ASSERT_EXPR with an equality comparison against a
25892 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25894 * lra-constraints.c (lra_copy_reg_equiv): New function.
25895 (split_reg): Use it to copy equivalence information from the
25896 original register to the spill register.
25898 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25900 PR rtl-optimization/75964
25901 * simplify-rtx.c (simplify_const_relational_operation): Remove
25902 invalid handling of comparisons of integer ABS.
25904 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25906 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25907 initialize to zero.
25908 (init_regs): Remove declaration.
25909 (function_arg_advance_32): Initialize error_p as boolean variable.
25911 2017-05-05 Nathan Sidwell <nathan@acm.org>
25913 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25914 lines. Use for (;;).
25916 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25918 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25919 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25920 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25921 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25922 VF=2 that require versioning.
25924 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25926 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25929 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25931 * diagnostic.h (diagnostic_override_option_index): Convert from
25932 macro to inline function.
25934 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25936 * diagnostic.c (last_module_changed_p): New function.
25937 (set_last_module): New function.
25938 (diagnostic_report_current_module): Convert macro usage to
25939 the above functions.
25940 * diagnostic.h (diagnostic_context::last_module): Strengthen
25941 from const line_map * to const line_map_ordinary *.
25942 (diagnostic_last_module_changed): Delete macro.
25943 (diagnostic_set_last_module): Delete macro.
25945 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25947 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25948 with diagnostic_report_diagnostic.
25949 (diagnostic_n_impl_richloc): Likewise.
25950 * diagnostic.h (report_diagnostic): Delete macro.
25951 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25952 with diagnostic_report_diagnostic.
25953 * substring-locations.c (format_warning_va): Likewise.
25955 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25957 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25958 save/restor of format_spec. Move option-printing code to...
25959 (print_option_information): ...this new function, and
25960 reimplement by simply printing to the pretty_printer,
25961 rather than appending to the format string.
25963 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25965 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25966 handling logic into...
25967 (update_effective_level_from_pragmas): ...this new function.
25969 2017-05-04 Andrew Waterman <andrew@sifive.com>
25971 * config/riscv/riscv.opt (mstrict-align): New option.
25972 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25973 (SLOW_UNALIGNED_ACCESS): Define.
25974 (riscv_slow_unaligned_access): Declare.
25975 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25977 (riscv_slow_unaligned_access): New variable.
25978 (rocket_tune_info): Set slow_unaligned_access to true.
25979 (optimize_size_tune_info): Set slow_unaligned_access to false.
25980 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25981 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25982 (riscv_option_override): Set riscv_slow_unaligned_access.
25983 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25985 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25987 * config/riscv/riscv.md: Unify indentation.
25989 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25994 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25996 (extendsi<mode>2): Add support for doing sign extension via
25997 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25998 don't have ISA 3.0 instructions.
25999 (extendsi<mode>2 splitter): Likewise.
26000 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
26001 generate the normal insns since SImode can now go in vector
26002 registers. Disallow the special UNSPECs needed for previous
26003 machines to hide SImode being used. Add new insns
26004 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
26005 (fix_trunc<mode>si2_stfiwx): Likewise.
26006 (fix_trunc<mode>si2_internal): Likewise.
26007 (fixuns_trunc<mode>si2): Likewise.
26008 (fixuns_trunc<mode>si2_stfiwx): Likewise.
26009 (fctiw<u>z_<mode>_smallint): Likewise.
26010 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
26011 of floating point to 32-bit integer from doing a direct move to
26012 the GPR registers to do a store.
26013 (fctiwz_<mode>): Break long line.
26015 2017-05-05 Bin Cheng <bin.cheng@arm.com>
26017 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
26018 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
26019 (addr_list, addr_offset_valid_p): New.
26020 (split_address_groups): Check offset validity with above function.
26021 (gt-tree-ssa-loop-ivopts.h): Include header file.
26023 2017-05-05 Nathan Sidwell <nathan@acm.org>
26025 * config.gcc (arm*-*-*): Add missing 'fi'.
26027 2017-05-05 Steve Ellcey <sellcey@cavium.com>
26029 * doc/invoke.texi (-fopt-info): Explicitly say order of options
26030 included in -fopt-info does not matter.
26031 * doc/optinfo.texi (-fopt-info): Fix description of default
26032 behavour. Explicitly say order of options included in -fopt-info
26035 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26037 * config.gcc: Allow combinations of aprofile and rmprofile values for
26038 --with-multilib-list.
26039 * config/arm/t-multilib: New file.
26040 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
26041 variables. Remove setting of ISA and floating-point ABI in
26042 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
26043 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
26044 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
26045 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
26047 * config/arm/t-rmprofile: Likewise except for the matches changes.
26048 * doc/install.texi (--with-multilib-list): Document the combination of
26049 aprofile and rmprofile values and warn about pitfalls in doing that.
26051 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26053 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
26054 (movdi_aarch64): Likewise.
26056 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26058 PR tree-optimization/80632
26059 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
26061 (build_arrays): Initialize it for virtual phis.
26062 (fix_phi_nodes): Use it for virtual phis.
26064 PR tree-optimization/80558
26065 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
26066 [x, y] op z into [x op, y op z] for op & or | if conditions
26069 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
26070 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26073 * config/arm/arm.md (use_literal_pool): Remove.
26074 (64-bit immediate split): No longer takes cost into consideration
26075 if arm_disable_literal_pool is enabled.
26076 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
26077 used when arm_disable_literal_pool is enabled.
26078 (arm_max_const_double_inline_cost): Remove use of
26079 arm_disable_literal_pool.
26080 (push_minipool_fix): Add assert.
26081 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
26082 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
26083 (no_literal_pool_sf_immediate): New.
26085 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26087 PR tree-optimization/80613
26088 * tree-ssa-dce.c (propagate_necessity): Remove cases for
26089 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
26091 2017-05-05 Richard Biener <rguenther@suse.de>
26093 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
26095 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
26097 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
26098 of this flag from insn conditions due to removal from r247495.
26100 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
26102 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
26104 (arm_early_store_addr_dep_ptr): Likewise.
26105 * config/arm/aarch-common-protos.h
26106 (arm_early_load_addr_dep_ptr): Add prototype.
26107 (arm_early_store_addr_dep_ptr): Likewise.
26108 * config/arm/cortex-a53.md: Add new bypasses.
26110 2017-05-05 Jakub Jelinek <jakub@redhat.com>
26112 * tree.c (next_type_uid): Change type to unsigned.
26113 (type_hash_canon): Decrement back next_type_uid if
26114 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
26115 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
26118 2017-05-04 Martin Sebor <msebor@redhat.com>
26120 * builtins.c: Fix a trivial typo in a comment.
26122 PR middle-end/79234
26123 * builtins.c (check_sizes): Adjust to handle reading past the end.
26124 Avoid printing excessive upper bound of ranges. Use %E to print
26125 tree nodes instead of converting them to %wu.
26126 (expand_builtin_memchr): New function.
26127 (compute_dest_size): Rename...
26128 (compute_objsize): ...to this.
26129 (expand_builtin_memcpy): Adjust.
26130 (expand_builtin_mempcpy): Adjust.
26131 (expand_builtin_strcat): Adjust.
26132 (expand_builtin_strcpy): Adjust.
26133 (check_strncat_sizes): Adjust.
26134 (expand_builtin_strncat): Adjust.
26135 (expand_builtin_strncpy): Adjust and simplify.
26136 (expand_builtin_memset): Adjust.
26137 (expand_builtin_bzero): Adjust.
26138 (expand_builtin_memcmp): Adjust.
26139 (expand_builtin): Handle memcmp.
26140 (maybe_emit_chk_warning): Check strncat just once.
26142 2017-05-04 Martin Sebor <msebor@redhat.com>
26144 PR preprocessor/79214
26145 PR middle-end/79222
26146 PR middle-end/79223
26147 * builtins.c (check_sizes): Add inlining context and issue
26148 warnings even when -Wno-system-headers is set.
26149 (check_strncat_sizes): Same.
26150 (expand_builtin_strncat): Same.
26151 (expand_builtin_memmove): New function.
26152 (expand_builtin_stpncpy): Same.
26153 (expand_builtin): Handle memmove and stpncpy.
26155 2017-05-04 Bin Cheng <bin.cheng@arm.com>
26157 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
26158 which is not used any more.
26160 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26162 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
26164 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26166 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
26167 (cortexa53_tunings): Likewise.
26168 (cortexa57_tunings): Likewise.
26169 (cortexa72_tunings): Likewise.
26170 (cortexa73_tunings): Likewise.
26172 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26174 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
26175 Set loop alignment to 8.
26177 2017-05-04 Martin Sebor <msebor@redhat.com>
26179 PR translation/80280
26180 * builtins.c (expand_builtin_object_size): Add missing quoting to
26181 %D and like directives.
26182 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
26183 (hsa_type_for_tree_type): Same.
26184 (verify_function_arguments): Same.
26185 * symtab.c (symbol_table::change_decl_assembler_name): Same.
26186 * varasm.c (get_section): Same.
26189 2017-05-04 Martin Sebor <msebor@redhat.com>
26191 PR translation/80280
26192 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
26194 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
26196 * config/aarch64/aarch64.c (generic_addrcost_table):
26197 Change HI/TI mode setting.
26199 2017-05-04 Martin Jambor <mjambor@suse.cz>
26201 PR tree-optimization/80622
26202 * tree-sra.c (comes_initialized_p): New function.
26203 (build_accesses_from_assign): Only set write lazily when
26204 comes_initialized_p is false.
26205 (analyze_access_subtree): Use comes_initialized_p.
26206 (propagate_subaccesses_across_link): Assert !comes_initialized_p
26207 instead of testing for PARM_DECL.
26209 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26211 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
26212 constraint on operand 0 to allow more general addressing modes.
26213 Adjust output template.
26214 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
26216 * config/aarch64/aarch64-protos.h
26217 (aarch64_address_valid_for_prefetch_p): Declare prototype.
26218 * config/aarch64/constraints.md (Dp): New address constraint.
26219 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
26222 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
26224 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
26225 update use of estimate_ipcp_clone_size_and_time.
26226 (estimate_local_effects): Update use of
26227 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
26228 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
26229 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
26230 Return nonspecialized time.
26232 2017-05-04 Richard Biener <rguenther@suse.de>
26234 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
26235 for the last VUSE which def dominates the PHI. Directly call
26237 (get_continuation_for_phi_1): Remove.
26239 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
26241 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
26242 to explain the use of truncating division. Cap the number of
26243 iterations to the maximum given by nb_iterations_upper_bound,
26246 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26248 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
26249 * configure: Regenerate.
26250 * config.in: Regenerate.
26251 * config/i386/driver-mingw32.c: new file.
26252 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
26253 * config.host: Link driver-mingw32.o on MinGW host.
26254 * doc/install.texi: Document new --enable-mingw-wildcard configure
26257 2017-05-04 Marek Polacek <polacek@redhat.com>
26259 PR tree-optimization/80612
26260 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
26262 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26263 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
26265 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
26266 (movt splitter): Likewise.
26267 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
26268 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
26269 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
26270 block for Thumb-1 with MOVT.
26271 (thumb2_legitimate_address_p): Move code block ...
26272 (can_avoid_literal_pool_for_label_p): ... into this new function.
26273 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
26275 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
26276 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
26277 "M-profile targets with the MOVT instruction".
26279 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
26281 * config/arm/arm-builtins.c (arm_init_builtins): Rename
26282 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
26283 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
26285 2017-05-04 Martin Liska <mliska@suse.cz>
26287 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
26288 variable cond_code.
26290 2017-05-04 Richard Biener <rguenther@suse.de>
26292 * tree.c (array_at_struct_end_p): Handle arrays at struct
26293 end with flexarrays more conservatively. Refactor and treat
26294 arrays of arrays or aggregates more strict. Fix
26295 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
26296 * tree.c (array_at_struct_end_p): Adjust prototype.
26297 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
26298 * gimple-fold.c (get_range_strlen): Likewise.
26299 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
26301 2017-05-04 Richard Biener <rguenther@suse.de>
26303 PR tree-optimization/31130
26304 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
26306 (supports_overflow_infinity): Likewise.
26307 (is_negative_overflow_infinity): Likewise.
26308 (is_positive_overflow_infinity): Likewise.
26309 (is_overflow_infinity): Likewise.
26310 (stmt_overflow_infinity): Likewise.
26311 (overflow_infinity_range_p): Likewise.
26312 (usable_range_p): Remove as always returning true.
26313 (make_overflow_infinity): Remove.
26314 (negative_overflow_infinity): Likewise.
26315 (positive_overflow_infinity): Likewise.
26316 (avoid_overflow_infinity): Likewise.
26317 (set_value_range): Adjust accordingly.
26318 (set_value_range_to_nonnegative): Likewise, remove now unused
26319 overflow_infinity arg.
26320 (vrp_operand_equal_p): Adjust.
26321 (update_value_range): Likewise.
26322 (range_int_cst_singleton_p): Likewise.
26323 (operand_less_p): Likewise.
26324 (compare_values_warnv): Likewise.
26325 (extract_range_for_var_from_comparison_expr): Likewise.
26326 (vrp_int_const_binop): Likewise.
26327 (zero_nonzero_bits_from_vr): Likewise.
26328 (extract_range_from_multiplicative_op_1): Likewise.
26329 (extract_range_from_binary_expr_1): Likewise.
26330 (extract_range_from_unary_expr): Likewise.
26331 (extract_range_from_comparison): Likewise.
26332 (extract_range_basic): Likewise.
26333 (adjust_range_with_scev): Likewise.
26334 (compare_ranges): Likewise.
26335 (compare_range_with_value): Likewise.
26336 (dump_value_range): Likewise.
26337 (test_for_singularity): Likewise, remove strict_overflow_p parameter
26339 (simplify_cond_using_ranges): Adjust.
26341 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26343 * brig-builtins.def: Added a builtin for class_f64.
26344 * builtin-types.def: Added a builtin type needed by class_f64.
26346 2017-05-03 Jason Merrill <jason@redhat.com>
26348 * timevar.def: Add TV_CONSTEXPR.
26350 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26352 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
26354 2017-05-03 Martin Jambor <mjambor@suse.cz>
26356 * ipa-prop.c (ipa_update_after_lto_read): Removed.
26357 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
26358 * ipa-cp.c (ipcp_propagate_stage): Do not call
26359 ipa_update_after_lto_read.
26360 * ipa-inline.c (ipa_inline): Likewise.
26362 2017-05-03 Martin Jambor <mjambor@suse.cz>
26364 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
26365 tag. Added a default constructor and a destructor.
26366 (ipa_edge_args_sum_t): New class;
26367 (ipa_edge_args_sum): Declare.
26368 (ipa_edge_args_vector): Remove declaration.
26369 (IPA_EDGE_REF): Use ipa_edge_args_sum.
26370 (ipa_free_edge_args_substructures): Remove declaration.
26371 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
26372 (ipa_edge_args_info_available_for_edge_p): Likewise.
26373 * ipa-prop.c (ipa_edge_args_vector): Removed.
26374 (edge_removal_hook_holder): Likewise.
26375 (edge_duplication_hook_holder): Likewise.
26376 (ipa_edge_args_sum): New variable.
26377 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
26378 ipa_edge_args_vector.
26379 (ipa_free_edge_args_substructures): Likewise.
26380 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
26381 ipa_edge_args_vector.
26382 (ipa_edge_removal_hook): Turned into method
26383 ipa_edge_args_sum_t::remove.
26384 (ipa_edge_duplication_hook): Turned into method
26385 ipa_edge_args_sum_t::duplicate.
26386 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
26387 registering edge hooks.
26388 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
26389 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
26390 ipa_edge_args_sum instead of ipa_edge_args_vector.
26391 * ipa-profile.c (ipa_profile): Likewise.
26393 2017-05-03 Martin Jambor <mjambor@suse.cz>
26395 * symbol-summary.h (function_summary): New method exists.
26396 (function_summary::symtab_removal): Deallocate through release.
26397 (call_summary): New class.
26398 (gt_ggc_mx): New overload.
26399 (gt_pch_nx): Likewise.
26400 (gt_pch_nx): Likewise.
26402 2017-05-03 Jeff Law <law@redhat.com>
26404 PR tree-optimization/78496
26405 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
26406 from simplify_cond_using_ranges. Split off code to walk
26407 backwards through casts into ...
26408 (simplify_cond_using_ranges_2): New function.
26409 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
26410 (execute_vrp): After identifying jump threads, call
26411 simplify_cond_using_ranges_2.
26413 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
26416 * ipa-inline.h (inline_summary): Add ctor.
26417 (create_ggc): Do not use ggc_cleared_alloc.
26419 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26420 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26422 * gcc.c (handle_braces): Support escaping in switch matching
26424 * doc/invoke.texi (Spec Files): Document it.
26425 Remove superfluous @code markup in items.
26427 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26429 * diagnostic-show-locus.c (struct column_range): New struct.
26430 (get_affected_columns): New function.
26431 (get_printed_columns): New function.
26432 (struct correction): New struct.
26433 (correction::ensure_capacity): New function.
26434 (correction::ensure_terminated): New function.
26435 (struct line_corrections): New struct.
26436 (line_corrections::~line_corrections): New dtor.
26437 (line_corrections::add_hint): New function.
26438 (layout::print_trailing_fixits): Reimplement in terms of the new
26440 (selftest::test_overlapped_fixit_printing): New function.
26441 (selftest::diagnostic_show_locus_c_tests): Call it.
26443 2017-05-03 Nathan Sidwell <nathan@acm.org>
26445 Canonicalize canonical type hashing
26446 * tree.h (type_hash_canon_hash): Declare.
26447 * tree.c (type_hash_list, attribute_hash_list): Move into
26448 type_hash_canon_hash.
26449 (build_type_attribute_qual_variant): Break out hash code calc into
26450 type_hash_canon_hash.
26451 (type_hash_canon_hash): New. Generic type hash computation.
26452 (build_range_type_1, build_array_type_1, build_function_type,
26453 build_method_type_directly, build_offset_type, build_complex_type,
26454 make_vector_type): Call it.
26456 2017-05-03 Richard Biener <rguenther@suse.de>
26458 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26459 When all DRs have unknown misaligned do not always peel
26460 when there is a store but apply the same costing model as if
26461 there were only loads.
26463 2017-05-03 Richard Biener <rguenther@suse.de>
26466 PR tree-optimization/80492
26467 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26468 compare_base_decls returning dont-know properly.
26470 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26472 * config/arm/iterators.md (CCSI): New mode iterator.
26473 (arch): New mode attribute.
26474 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26475 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26476 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26477 code iterator for success result mode.
26478 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26479 the corresponding new insn generators.
26481 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26484 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26485 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26487 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26489 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26490 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26491 (DDR_A): Wrap DDR argument in brackets.
26492 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26493 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26494 (DDR_REVERSED_P): Likewise.
26496 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26498 PR tree-optimization/79472
26499 * tree-switch-conversion.c (struct switch_conv_info): Add
26500 contiguous_range and default_case_nonstandard fields.
26501 (collect_switch_conv_info): Compute contiguous_range and
26502 default_case_nonstandard fields, don't clear final_bb if
26503 contiguous_range and only the default case doesn't have the required
26505 (check_all_empty_except_final): Set default_case_nonstandard instead
26506 of failing if contiguous_range and the default case doesn't have empty
26508 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26509 and only the default case doesn't have the required constants. Skip
26511 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26512 if default_case_nonstandard.
26513 (build_constructors): Build constant 1 just once. Assert that default
26514 values aren't inserted in between cases if contiguous_range. Skip
26516 (build_arrays): Skip virtual phis.
26517 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26518 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26519 Handle virtual phis.
26520 (gen_inbound_check): Handle default_case_nonstandard case.
26521 (process_switch): Adjust check_final_bb caller. Call
26522 gather_default_values with the first non-default case instead of
26523 default case if default_case_nonstandard.
26525 2017-05-02 Nathan Sidwell <nathan@acm.org>
26527 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26528 check. Fix formatting.
26530 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26532 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26533 errors when comparing specialized and unspecialized times.
26535 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26537 * diagnostic-show-locus.c
26538 (layout::should_print_annotation_line_p): Make private.
26539 (layout::print_annotation_line): Make private.
26540 (layout::annotation_line_showed_range_p): Make private.
26541 (layout::show_ruler): Make private.
26542 (layout::print_source_line): Make private. Pass in line and
26543 line_width, rather than calling location_get_source_line. Drop
26545 (layout::print_leading_fixits): New method.
26546 (layout::print_any_fixits): Rename to...
26547 (layout::print_trailing_fixits): ...this, and make private.
26548 Don't print newline fixits.
26549 (diagnostic_show_locus): Move logic for printing one row into...
26550 (layout::print_line): ...this new function. Move the
26551 location_get_source_line call and error-handling from
26552 print_source_line to here. Call print_leading_fixits, and rename
26553 print_any_fixits to print_trailing_fixits.
26554 (selftest::test_fixit_insert_containing_newline): Update now that
26555 newlines are partially supported.
26556 (selftest::test_fixit_insert_containing_newline_2): New test.
26557 (selftest::test_fixit_replace_containing_newline): Update comments.
26558 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26559 * edit-context.c (class added_line): New class.
26560 (class edited_line): Describe newline handling in comment.
26561 (edited_line::actually_edited_p): New method.
26562 (edited_line::print_content): Delete redundant decl.
26563 (edited_line::m_predecessors): New field.
26564 (edited_file::print_content): Call edited_line::print_content.
26565 (edited_file::print_diff): Update to support newlines.
26566 (edited_file::print_diff_hunk): Likewise.
26567 (edited_file::print_run_of_changed_lines): New function.
26568 (edited_file::print_diff_line): Convert to...
26569 (print_diff_line): ...this.
26570 (edited_file::get_effective_line_count): New function.
26571 (edited_line::edited_line): Initialize new field m_predecessors.
26572 (edited_line::~edited_line): Clean up m_predecessors.
26573 (edited_line::apply_fixit): Handle newlines.
26574 (edited_line::get_effective_line_count): New function.
26575 (edited_line::print_content): New function.
26576 (edited_line::print_diff_lines): New function.
26577 (selftest::test_applying_fixits_insert_containing_newline): New
26579 (selftest::test_applying_fixits_replace_containing_newline): New
26581 (selftest::insert_line): New function.
26582 (selftest::test_applying_fixits_multiple_lines): Add example of
26584 (selftest::edit_context_c_tests): Call the new tests.
26586 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26588 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26589 parameter cand. Update dump information.
26590 (get_computation_cost): Update uses.
26592 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26594 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26595 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26596 (get_computation_at, rewrite_use_address): Update use of
26597 get_computation_aff.
26599 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26601 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26602 (get_computation): Delete.
26603 (get_computation_cost): Implement like get_computation_cost_at.
26604 Use get_computation_at.
26605 (get_computation_cost_at): Delete.
26606 (rewrite_use_nonlinear_expr): Use get_computation_at.
26607 (rewrite_use_compare, remove_unused_ivs): Ditto.
26609 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26611 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26613 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26615 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26616 (ivopts_global_cost_for_size): Rename parameter and update uses.
26617 (iv_ca_recount_cost): Update uses.
26618 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26619 candidates seperately in n_invs and n_cands.
26620 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26622 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26624 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26625 (find_inv_vars_cb): New.
26626 (find_depends): Renamed to ...
26627 (find_inv_vars): ... this.
26628 (add_candidate_1, force_var_cost): Call find_inv_vars.
26629 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26631 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26633 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26634 inv_vars. Add inv_exprs.
26635 (struct iv_cand): Rename depends_on to inv_vars.
26636 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26637 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26638 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26639 (dump_cand): Dump inv_vars.
26640 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26641 (record_invariant, find_depends, add_candidate_1): Ditto.
26642 (set_group_iv_cost, force_var_cost): Ditto.
26643 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26644 (get_computation_cost_at, get_computation_cost): Ditto.
26645 (determine_group_iv_cost_generic): Ditto.
26646 (determine_group_iv_cost_address): Ditto.
26647 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26648 (determine_group_iv_costs): Ditto.
26649 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26650 (iv_ca_set_remove_invariants): Renamed to ...
26651 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26652 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26653 (iv_ca_set_add_invariants): Renamed to ...
26654 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26655 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26656 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26657 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26658 (create_new_ivs): Remove useless dump.
26660 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26662 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26664 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26665 (iv_ca_set_no_cp, create_new_iv): Ditto.
26667 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26669 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26671 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26673 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26674 function tree_check2.
26676 2017-05-02 Martin Liska <mliska@suse.cz>
26678 * doc/gcov.texi: Add missing preposition.
26679 * gcov.c (function_info::function_info): Properly fill up
26680 all member variables.
26682 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26684 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26686 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26688 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26690 2017-05-02 Martin Liska <mliska@suse.cz>
26693 * lto-streamer-in.c (lto_read_tree_1): Remove
26694 LTO_STREAMER_DEBUG.
26695 * lto-streamer.c (struct tree_hash_entry): Likewise.
26696 (struct tree_entry_hasher): Likewise.
26697 (tree_entry_hasher::hash): Likewise.
26698 (tree_entry_hasher::equal): Likewise.
26699 (lto_streamer_init): Likewise.
26700 (lto_orig_address_map): Likewise.
26701 (lto_orig_address_get): Likewise.
26702 (lto_orig_address_remove): Likewise.
26703 * lto-streamer.h: Likewise.
26704 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26705 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26707 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26709 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26710 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26711 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26712 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26713 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26714 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26715 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26716 (mm_maskz_sub_ss): New intrinsics.
26717 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26718 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26719 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26720 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26721 (__builtin_ia32_subss_mask_round): New builtins.
26722 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26723 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26724 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26726 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26727 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26729 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26732 2017-05-02 Martin Jambor <mjambor@suse.cz>
26734 PR tree-optimization/78687
26735 * tree-sra.c (access): New field parent.
26736 (process_subtree_disqualification): New function.
26737 (disqualify_candidate): Call it.
26738 (build_accesses_from_assign): Reset write flag if creating an
26740 (build_access_subtree): Fill in parent field and also prpagate
26741 down grp_write flag.
26742 (create_artificial_child_access): New parameter set_grp_write, set
26743 grp_write to its value.
26744 (propagate_subaccesses_across_link): Also propagate grp_write flag
26746 (propagate_all_subaccesses): Push the closest parent back to work
26747 queue if add_access_to_work_queue returned true.
26749 2017-05-02 Richard Biener <rguenther@suse.de>
26751 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26752 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26753 -fstrict-overflow documentation.
26754 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26755 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26756 flag_strict_overflow.
26757 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26758 * lto-opts.c (lto_write_options): Do not stream it.
26759 * lto-wrapper.c (merge_and_complain): Do not handle it.
26760 * opts.c (default_options_table): Do not set -fstrict-overflow.
26761 (finish_options): Likewise do not clear it when sanitizing.
26762 * simplify-rtx.c (simplify_const_relational_operation): Do not
26763 test flag_strict_overflow.
26765 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26767 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26768 using enabled attribute.
26769 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26770 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26771 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26772 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26773 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26774 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26775 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26776 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26777 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26778 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26780 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26782 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26784 2017-05-02 Richard Biener <rguenther@suse.de>
26786 PR tree-optimization/80591
26788 2017-04-10 Richard Biener <rguenther@suse.de>
26790 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26793 2017-05-02 Richard Biener <rguenther@suse.de>
26795 PR tree-optimization/80549
26796 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26797 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26798 headers if they do not have a preheader.
26800 2017-05-02 Martin Liska <mliska@suse.cz>
26803 * common.opt: Fix typo.
26804 * doc/invoke.texi: Likewise.
26806 2017-05-01 Jan Beulich <jbeulich@suse.com>
26808 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26809 swapping, add (x,x,m,x,n) alternative.
26811 2017-05-01 Nathan Sidwell <nathan@acm.org>
26813 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26814 unnecessary unadjusted_alignment check.
26816 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26819 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26821 * gimplify.c (gimplify_cilk_detach): New function.
26822 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26823 * tree-core.h: Document EXPR_CILK_SPAWN.
26824 * tree.h (EXPR_CILK_SPAWN): Define.
26826 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26828 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26829 to use new fixit_hint representation, using the "replace" logic.
26830 (get_line_span_for_fixit_hint): Likewise.
26831 (layout::print_any_fixits): Likewise.
26832 (selftest::test_one_liner_many_fixits): Rename to...
26833 (selftest::test_one_liner_many_fixits_1): ...this, and update
26834 comment and expected output to reflect that the multiple fix-it
26835 hints are now consolidated into one insertion.
26836 (selftest::test_one_liner_many_fixits_2): New test.
26837 (selftest::test_diagnostic_show_locus_one_liner): Update for
26839 (selftest::test_fixit_consolidation): Update for fix-it API
26841 * diagnostic.c (print_parseable_fixits): Likewise.
26842 * edit-context.c (edited_line::m_line_events): Convert from
26843 auto_vec <line_event *> to auto_vec <line_event>.
26844 (class line_event): Convert from abstract base class to a concrete
26845 class, taking over the role of replace_event.
26846 (class insert_event): Delete.
26847 (class replace_event): Rename to class line_event. Convert to
26849 (edit_context::add_fixits): Reimplement.
26850 (edit_context::apply_insert): Delete.
26851 (edit_context::apply_replace): Rename to...
26852 (edit_context::apply_fixit): ...this. Convert to half-open range.
26853 (edited_file::apply_insert): Delete.
26854 (edited_file::apply_replace): Rename to...
26855 (edited_file::apply_fixit): ...this.
26856 (edited_line::~edited_line): Drop deletion of events.
26857 (edited_line::apply_insert): Delete.
26858 (edited_line::apply_replace): Rename to...
26859 (edited_line::apply_fixit): ...this. Convert to half-open range.
26860 Update for change to type of m_line_events.
26861 * edit-context.h (edit_context::apply_insert): Delete.
26862 (edit_context::apply_replace): Rename to...
26863 (edit_context::apply_fixit): ...this.
26865 2017-05-01 Martin Sebor <msebor@redhat.com>
26867 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26870 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26873 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26874 __get_cpuid_max returns 0.
26875 (__get_cpuid_count): Ditto.
26877 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26879 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26880 replacement expression is another instance of one of its arguments.
26882 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26885 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26886 check for stack push/pop autoinc.
26887 * config/i386/i386.c (ix86_agi_dependent): Return false
26888 if the only reason why modified_in_p returned true is that
26889 addr is SP based and set_insn is a push or pop.
26891 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26893 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26896 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26899 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26900 (account_size_time): Use two predicates - exec_pred and
26902 (evaluate_conditions_for_known_args): Compute both clause and
26904 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26905 (inline_summary_t::duplicate): Update.
26906 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26908 (compute_inline_parameters): Likewise.
26909 (estimate_edge_size_and_time): Update caluclation of time.
26910 (estimate_node_size_and_time): Compute both time and nonspecialized
26912 (estimate_ipcp_clone_size_and_time): Update.
26913 (inline_merge_summary): Update.
26914 (do_estimate_edge_time): Update.
26915 (do_estimate_edge_size): Update.
26916 (do_estimate_edge_hints): Update.
26917 (inline_read_section, inline_write_summary): Stream both new predicates.
26918 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26920 (compute_inlined_call_time): Cleanup.
26921 (big_speedup_p): Update.
26922 (edge_badness): Update.
26923 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26924 (size_time_entry): Replace predicate by exec_predicate and
26925 nonconst_predicate.
26926 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26927 (estimate_edge_time): Return also nonspec_time.
26928 (reset_edge_growth_cache): Update.
26930 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26932 PR rtl-optimization/80491
26933 * ifcvt.c (noce_process_if_block): When looking for x setter
26934 with missing else_bb, don't check only the insn right before
26935 cond_earliest, but look for the last insn that x is modified in
26936 within the same bb.
26938 PR rtl-optimization/80491
26939 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26941 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26943 PR tree-optimization/80487
26944 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26946 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26948 PR tree-optimization/79697
26949 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26950 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26951 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26953 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26954 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26956 2017-04-28 Martin Sebor <msebor@redhat.com>
26958 PR tree-optimization/80523
26959 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26960 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26962 (maybe_warn, format_directive, parse_directive): Use new functions.
26963 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26965 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26967 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26969 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26971 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26972 target_header_dir): Set correctly.
26973 * configure: Regenerated.
26974 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26975 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26976 instead of SYSTEM_HEADER_DIR.
26978 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26980 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26981 (estimate_local_effects): Likewise.
26982 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26983 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26984 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26985 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26986 do_estimate_edge_time, estimate_edge_time): Likewise.
26987 * ipa-inline-analysis.c (estimate_node_size_and_time,
26988 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26989 (estimate_time_after_inlining): Remove.
26991 2017-04-28 Martin Liska <mliska@suse.cz>
26993 * doc/gcov.texi: Enhance documentation of gcov.
26995 2017-04-28 Martin Liska <mliska@suse.cz>
26997 * doc/gcov.texi: Sort options in alphabetic order.
26998 * doc/gcov-dump.texi: Likewise.
26999 * doc/gcov-tool.texi: Likewise.
27000 * gcov.c (print_usage): Likewise.
27001 * gcov-dump.c (print_usage): Likewise.
27002 * gcov-tool.c (print_merge_usage_message): Likewise.
27003 (print_rewrite_usage_message): Likewise.
27004 (print_overlap_usage_message): Likewise.
27006 2017-04-28 Martin Liska <mliska@suse.cz>
27008 PR gcov-profile/53915
27009 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
27011 2017-04-28 Martin Liska <mliska@suse.cz>
27013 PR gcov-profile/79891
27014 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
27015 is marked by compiler as living on a line.
27016 (get_cycles_count): Remove usage of the union.
27017 (output_intermediate_file): Likewise.
27018 (find_source): Fix GNU coding style.
27019 (accumulate_line_counts): Remove old non-all block mode.
27020 (output_lines): Remove usage of the union.
27021 * profile.c (output_location): Include all BBs, even if
27022 belonging to a same line (and file) as a previous BB.
27024 2017-04-28 Martin Liska <mliska@suse.cz>
27026 * gcov.c (process_args): Handle new argument 'w'.
27027 (read_graph_file): Assign ID to BBs.
27028 (output_branch_count): Display BB # if verbose flag is set.
27029 (output_lines): Likewise for arcs.
27030 (print_usage): Add '--verbose' option help.
27031 * doc/gcov.texi: Document --verbose (-w) option.
27033 2017-04-28 Martin Liska <mliska@suse.cz>
27035 * gcov.c (struct block_location_info): New struct.
27036 (process_file): Fill up the new structure.
27037 (read_graph_file): Replace usage of encoding by the newly added
27039 (add_line_counts): Likewise.
27040 (accumulate_line_counts): Remove usage of the union.
27041 (function_info::function_info): New function.
27042 (function_info::~function_info): Likewise.
27043 (process_file): Call delete instead of release_function.
27044 (release_function): Release the function.
27045 (release_structures): Call delete instead of release_function.
27046 (solve_flow_graph): Replace usage of num_blocks.
27047 (find_exception_blocks): Likewise.
27048 (output_lines): Fix GNU coding style.
27050 2017-04-28 Martin Liska <mliska@suse.cz>
27053 * coverage.c (coverage_remove_note_file): New function.
27054 * coverage.h: Declare the function.
27055 * toplev.c (finalize): Clean if an error has been seen.
27057 2017-04-28 Martin Liska <mliska@suse.cz>
27059 PR gcov-profile/80031
27060 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
27061 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
27062 * gcov.c (read_graph_file): Read just number of blocks.
27063 * profile.c (branch_prob): Do not stream 0 flags per a basic
27066 2017-04-28 Martin Liska <mliska@suse.cz>
27068 * gcov-dump.c (tag_*): Add new argument to declarations.
27069 (dump_gcov_file): Likewise.
27070 (tag_blocks): Add and use new argument depth.
27071 (tag_arcs): Likewise.
27072 (tag_lines): Likewise.
27073 (tag_counters): Likewise.
27074 (tag_summary): Likewise.
27075 (dump_working_sets): Use depth to do a proper indentation.
27077 2017-04-28 Jakub Jelinek <jakub@redhat.com>
27080 * cgraph.h (symtab_node::debug_symtab): No longer inline.
27081 * symtab.c (symtab_node::debug_symtab): Move definition here.
27083 2017-04-28 Richard Biener <rguenther@suse.de>
27085 * lto-streamer.h (LTO_major_version): Bump to 7.
27087 2017-04-28 Richard Biener <rguenther@suse.de>
27089 * tree-vrp.c (assert_info): New struct.
27090 (add_assert_info): New helper.
27091 (register_edge_assert_for_2): Refactor to add asserts to a vector
27093 (register_edge_assert_for_1): Likewise.
27094 (register_edge_assert_for): Likewise.
27095 (finish_register_edge_assert_for): New helper actually registering
27096 asserts where live on edge.
27097 (find_conditional_asserts): Adjust.
27098 (find_switch_asserts): Likewise.
27099 (evrp_dom_walker::try_find_new_range): Generalize.
27100 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
27102 2017-04-27 Marek Polacek <polacek@redhat.com>
27105 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
27106 arg10 and arg11 to itype.
27108 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
27110 * doc/extend.texi (Object Size Checking): Improve grammar.
27112 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
27115 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
27116 that the logic for permitting reciprocal estimates matches that
27119 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27122 * tree.c (type_cache_hasher::equal): Only compare
27123 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
27124 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
27125 non-aggregate element types.
27126 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
27127 about the flag on ARRAY_TYPEs in the comment, formatting fix.
27129 2017-04-27 Richard Biener <rguenther@suse.de>
27131 PR middle-end/80533
27132 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
27133 stripping ARRAY_REFs from MEM_EXPR make sure we're not
27134 keeping a reference to a trailing array.
27136 2017-04-27 Richard Biener <rguenther@suse.de>
27138 PR middle-end/80539
27139 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
27140 being in loop-closed SSA form conservatively.
27141 (chrec_fold_multiply_poly_poly): Likewise.
27143 2017-04-27 Tamar Christina <tamar.christina@arm.com>
27145 PR middle-end/79665
27146 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
27147 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
27149 2017-04-27 Jakub Jelinek <jakub@redhat.com>
27152 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
27153 (aarch64_function_arg_alignment): Return unsigned int again, but still
27154 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
27155 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27156 Don't emit -Wpsabi note.
27157 (aarch64_function_arg_boundary): Likewise.
27158 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27161 2017-04-26 Nathan Sidwell <nathan@acm.org>
27163 * tree.h (crc32_unsigned_n): Declare.
27164 (crc32_unsigned, crc32_unsigned): Make inline.
27165 * tree.c (crc32_unsigned_bits): Replace with ...
27166 (crc32_unsigned_n): ... this.
27167 (crc32_unsigned, crc32_byte): Remove.
27168 (crc32_string): Remove unnecessary braces.
27170 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27172 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
27173 * ipa-inline-analysis.c (MAX_TIME): Remove.
27174 (account_size_time): Use sreal for time.
27175 (dump_inline_summary): Update.
27176 (estimate_function_body_sizes): Update.
27177 (estimate_edge_size_and_time): Update.
27178 (estimate_calls_size_and_time): Update.
27179 (estimate_node_size_and_time): Update.
27180 (inline_merge_summary): Update.
27181 (inline_update_overall_summary): Update.
27182 (estimate_time_after_inlining): Update.
27183 (inline_read_section): Update.
27184 (inline_write_summary): Update.
27185 * ipa-inline.c (compute_uninlined_call_time): Update.
27186 (compute_inlined_call_time): Update.
27187 (recursive_inlining): Update.
27188 (inline_small_functions): Update.
27189 (dump_overall_stats): Update.
27190 * ipa-inline.h: Include sreal.h.
27191 (size_time_entry): Turn time to sreal.
27192 (inline_summary): Turn self_time nad time to sreal.
27194 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
27196 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
27198 (sreal::stream_out, sreal::stream_in): New.
27199 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
27201 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27203 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
27206 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
27209 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
27211 (dimode_scalar_chain::compute_convert_gain): Ditto.
27212 (dimode_scalar_chain::make_vector_copies): Ditto.
27213 (dimode_scalar_chain::convert_reg): Ditto.
27214 (dimode_scalar_chain::convert_insn): Ditto.
27215 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
27216 (VI248_AVX512BW_1): New mode iterator.
27217 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
27218 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
27221 2017-04-25 Martin Sebor <msebor@redhat.com>
27223 PR tree-optimization/80497
27224 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
27225 constants are representable in HOST_WIDE_INT.
27226 (parse_directive): Ditto.
27228 2017-04-25 Martin Sebor <msebor@redhat.com>
27231 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
27232 (new_zero_array): Adjust signature.
27233 (dom_info::dom_init): Used unsigned rather that size_t.
27234 (dom_info::dom_info): Same.
27236 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27237 Jakub Jelinek <jakub@redhat.com>
27240 * config/arm/arm.c: Include gimple.h.
27241 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27242 returns negative, increment ncrn only if it returned positive.
27243 (arm_needs_doubleword_align): Return int instead of bool,
27244 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
27245 members, but if there is any such non-FIELD_DECL
27246 > PARM_BOUNDARY aligned decl, return -1 instead of false.
27247 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
27248 returns negative, increment nregs only if it returned positive.
27249 (arm_setup_incoming_varargs): Likewise.
27250 (arm_function_arg_boundary): Emit -Wpsabi note if
27251 arm_needs_doubleword_align returns negative, return
27252 DOUBLEWORD_ALIGNMENT only if it returned positive.
27254 2017-04-25 Marek Polacek <polacek@redhat.com>
27257 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27258 first argument to type.
27260 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
27263 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
27264 type checks to test for compatibility instead of equality.
27266 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27267 Jakub Jelinek <jakub@redhat.com>
27270 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
27272 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
27273 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
27274 the alignment computation, but return their maximum in warn_alignment.
27275 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
27276 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
27278 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
27279 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
27282 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27284 * config/arc/simdext.md (dmpyh): Fix typo.
27286 2017-04-25 Richard Biener <rguenther@suse.de>
27288 PR tree-optimization/80492
27289 * alias.c (compare_base_decls): Handle registers with asm
27290 specification conservatively.
27291 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
27292 compare_base_decls returning dont-know properly.
27294 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27296 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
27297 (legitimate_offset_address_p): New function.
27298 (arc_legitimate_address_p): Use above function.
27300 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27302 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
27304 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27306 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
27307 ACCH registers whenever they are available.
27309 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27311 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
27312 double regs fix when not used.
27314 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27316 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
27318 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
27319 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
27321 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27323 * config/arc/arc.c (arc_output_addsi): Check for h-register class
27324 when emitting short ADD instructions.
27326 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
27328 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
27330 (cmpsi_cc_c_insn): Likewise.
27331 (cbranchsi4_scratch): Compute proper instruction length using
27332 compact_hreg_operand.
27333 * config/arc/predicates.md (compact_hreg_operand): New predicate.
27335 2017-04-25 Richard Biener <rguenther@suse.de>
27337 PR middle-end/80509
27338 * passes.c (pass_manager::pass_manager): Initialize
27339 m_name_to_pass_map.
27341 2017-04-25 Richard Biener <rguenther@suse.de>
27343 PR tree-optimization/79201
27344 * tree-ssa-sink.c (statement_sink_location): Handle calls.
27346 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27349 * config/s390/vector.md: Split MEM->GPR vector moves for
27350 non-s_operand addresses.
27352 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27355 * config/s390/predicates.md (reload_const_wide_int_operand): New
27357 * config/s390/s390.md ("movti"): Remove d/P alternative.
27358 ("movti_bigconst"): New pattern definition.
27360 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27363 * s390-protos.h (s390_expand_cs_hqi): Removed.
27364 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
27365 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
27366 modes as well as CCZ1mode and CCZmode.
27367 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
27368 signature of s390_emit_compare_and_swap.
27369 (s390_expand_cs_hqi): Likewise, make static.
27370 (s390_expand_cs_tdsi): Generate an explicit compare before trying
27371 compare-and-swap, in some cases.
27372 (s390_expand_cs): Wrapper function.
27373 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
27375 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
27376 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
27377 patterns for small and large integers. Forbid symref memory operands.
27378 Move expander to s390.c. Require cc register.
27379 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
27380 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
27381 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
27382 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
27383 symref memory operands. Remove CC mode and call s390_match_ccmode
27385 ("atomic_exchange<mode>"): Allow and implement all integer modes.
27387 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27389 * config/s390/s390.md (define_peephole2): New peephole to help
27390 combining the load-and-test pattern with volatile memory.
27392 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
27394 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
27395 with CCZmode for TARGET_Z196.
27397 2017-04-25 Jakub Jelinek <jakub@redhat.com>
27399 PR rtl-optimization/80501
27400 * combine.c (make_compound_operation_int): Set subreg_code to SET
27401 even for AND with mask of the sign bit of mode.
27403 PR rtl-optimization/80500
27404 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
27405 sum's initial value.
27407 2017-04-25 Julian Brown <julian@codesourcery.com>
27408 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27410 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
27412 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
27414 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
27416 2017-04-25 Julian Brown <julian@codesourcery.com>
27417 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27419 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27420 (thunderx2t99_sha): New Reservation.
27422 2017-04-25 Julian Brown <julian@codesourcery.com>
27423 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27425 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27426 type for 1-element load.
27428 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27430 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27432 2017-04-24 Martin Jambor <mjambor@suse.cz>
27434 PR tree-optimization/80293
27435 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27436 char arrays not totally scalarizable if it is false.
27437 (analyze_all_variable_accesses): Pass correct value in the new
27438 parameter. Add a statistics counter.
27440 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27442 PR middle-end/79931
27443 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27445 2017-04-24 Richard Biener <rguenther@suse.de>
27447 PR tree-optimization/80494
27448 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27449 out for complex types.
27451 2017-04-24 Richard Biener <rguenther@suse.de>
27453 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27454 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27455 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27456 (DFS): Adjust and never fail.
27457 (sccvn_dom_walker::fail): Remove.
27458 (sccvn_dom_walker::before_dom_children): Adjust.
27459 (run_scc_vn): Likewise and never fail.
27460 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27461 (pass_fre::execute): Likewise.
27463 2017-04-24 Richard Biener <rguenther@suse.de>
27465 PR tree-optimization/79725
27466 * tree-ssa-sink.c (statement_sink_location): Return whether
27467 failure reason was zero uses. Move that check later.
27468 (sink_code_in_bb): Deal with zero uses by removing the stmt
27471 2017-04-24 Richard Biener <rguenther@suse.de>
27474 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27475 pointer-based references.
27477 2017-04-24 Richard Biener <rguenther@suse.de>
27480 * pass_manager.h (pass_manager::operator new): Remove.
27481 (pass_manager::operator delete): Likewise.
27482 * passes.c (pass_manager::operator new): Remove.
27483 (pass_manager::operator delete): Likewise.
27484 (pass_manager::pass_manager): Zero individual pass members.
27486 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27489 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27490 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27491 Check "XEXP (src, 1)" operand here.
27492 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27493 Check "XEXP (src, 1)" operand here.
27494 (dimode_scalar_chain::make_vector_copies): Detect count register
27495 of a shift instruction. Zero extend count register from QImode
27496 to DImode to satisfy vector shift pattern count operand predicate.
27497 Substitute vector shift count operand with a DImode copy.
27498 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27501 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27503 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27504 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27505 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27506 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27507 (UNSPEC_NOREX_MEM): Remove definition.
27509 2017-04-21 Richard Biener <rguenther@suse.de>
27511 PR tree-optimization/79547
27512 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27513 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27514 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27515 without any constraints.
27517 2017-04-21 Richard Biener <rguenther@suse.de>
27519 PR tree-optimization/78847
27520 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27522 2017-04-21 Richard Biener <rguenther@suse.de>
27524 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27525 (build_distinct_type_copy): Likewise.
27526 (build_variant_type_copy): Likewise.
27527 * tree.c (build_qualified_type): Pass down mem-stat info.
27528 (build_distinct_type_copy): Likewise.
27529 (build_variant_type_copy): Likewise.
27531 2017-04-21 Richard Biener <rguenther@suse.de>
27533 PR tree-optimization/80237
27534 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27536 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27537 for a simplified result.
27539 2016-04-21 Richard Biener <rguenther@suse.de>
27541 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27542 sth as strict as a simple_iv but a chrec without symbols and an
27543 operand defined in the loop we are peeling (and not some subloop).
27544 (propagate_constants_for_unrolling): Propagate all constants.
27546 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27549 * config/i386/i386.c (print_reg): Remove assert for disalowed
27550 regno values, call output_operand_lossage instead.
27552 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27555 * config/i386/constraints.md (Yc): New register constraint.
27556 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27557 Use Yc constraint for alternative 2 of operand 0. Remove
27558 preferred_for_speed attribute.
27560 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27562 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27563 lastprivate clauses in SIMT case.
27565 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27567 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27569 2017-04-20 Richard Biener <rguenther@suse.de>
27571 PR tree-optimization/57796
27572 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27573 as N scalar stores.
27574 (vect_model_load_cost): Cost gathers as N scalar loads.
27576 2017-04-20 Richard Biener <rguenther@suse.de>
27578 * ggc-page.c (ggc_allocated_p): Rename to ...
27579 (safe_lookup_page_table_entry): ... this and return the lookup
27581 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27583 2017-04-20 Richard Biener <rguenther@suse.de>
27585 PR tree-optimization/80453
27586 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27587 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27588 from the conditions.
27589 (vn_phi_eq): Pass them down.
27590 (vn_phi_lookup): Record them.
27591 (vn_phi_insert): Likewise.
27593 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27595 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27596 uninitialized variable warning to avoid buffer overrun.
27598 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27601 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27602 is suppressed for '{ 0 }' in C.
27604 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27606 * BASE-VER: Set to 8.0.0.
27608 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27610 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27611 priority .init_array and .fini_array section with SECTION_NOTYPE
27614 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27616 PR middle-end/80423
27617 * tree.h (build_array_type): Add typeless_storage default argument.
27618 * tree.c (type_cache_hasher::equal): Also compare
27619 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27620 (build_array_type): Add typeless_storage argument, set
27621 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27623 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27624 (build_array_type): Likewise. Add typeless_storage argument.
27626 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27627 Jakub Jelinek <jakub@redhat.com>
27629 PR tree-optimization/80426
27630 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27631 operation on symbolic operands, also compute the overflow for the
27632 invariant part when the operation degenerates into a negation.
27634 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27637 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27638 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27641 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27643 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27646 * config/avr/avr.c (tree.h): Include it.
27647 (cgraph.h): Include it.
27648 (avr_encode_section_info): Don't warn for uninitialized progmem
27649 variable if it's just an alias.
27651 2017-04-19 Richard Biener <rguenther@suse.de>
27654 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27655 when needed by AutoPGO.
27657 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27660 * doc/lto.texi: Remove an extra 'that'.
27662 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27664 PR rtl-optimization/80429
27665 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27666 are only used in debug insns.
27668 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27669 Vladimir Makarov <vmakarov@redhat.com>
27671 * config/sparc/predicates.md (input_operand): Add comment. Return
27672 true for any memory operand when LRA is in progress.
27673 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27675 2017-04-18 Jeff Law <law@redhat.com>
27678 * mips.md ({return,simple_return}_internal): Do not overwrite
27681 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27683 PR tree-optimization/80443
27684 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27685 instead of adding 1, subtract -1 and similarly instead of subtracting
27688 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27690 PR rtl-optimization/80357
27691 * haifa-sched.c (tmp_bitmap): New variable.
27692 (model_recompute): Handle duplicate use records.
27693 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27694 (free_global_sched_pressure_data): Free it.
27696 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27699 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27700 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27701 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27702 instead of SYSTEM_HEADER_DIR.
27704 2017-04-18 Jeff Law <law@redhat.com>
27706 PR middle-end/80422
27707 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27708 predecessors after walking up the insn chain.
27710 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27713 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27714 sizetype type into debug info.
27716 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27719 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27720 unneeded test for TARGET_UPPER_REGS_SF.
27721 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27723 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27726 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27727 instead of gsi_after_labels.
27729 2017-04-18 Jeff Law <law@redhat.com>
27731 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27735 2017-04-13 Jeff Law <law@redhat.com>
27736 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27737 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27739 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27742 * config/avr/avr.c (intl.h): Include it.
27743 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27745 2017-04-18 Martin Liska <mliska@suse.cz>
27747 PR gcov-profile/78783
27748 * gcov-tool.c (gcov_output_files): Validate that destination
27749 file is either removed by the tool or by a user.
27751 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27752 Guy Benyei <guybe@mellanox.com>
27754 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27755 block, and do not negate it, the stored id is already negative.
27757 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27759 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27761 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27764 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27765 masks of options that should be turned off if the VSX vector
27766 options are turned off.
27767 (OTHER_P8_VECTOR_MASKS): Likewise.
27768 (OTHER_VSX_VECTOR_MASKS): Likewise.
27769 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27770 rs6000_disable_incompatible_switches to validate no type switches
27772 (rs6000_incompatible_switch): New function to disallow turning on
27773 other vector options if -mno-vsx, -mno-power8-vector, or
27774 -mno-power9-vector are specified.
27776 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27778 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27780 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27782 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27783 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27784 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27785 (ARG_POINTER_CFA_OFFSET): Likewise.
27787 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27789 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27790 conditions to take advantage of various optimizations.
27792 2017-04-13 Jeff Law <law@redhat.com>
27794 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27795 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27796 (zero_extendsidi2_dext): Likewise.
27798 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27801 * fold-const.c (fold_ternary_loc): Revert
27802 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27805 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27807 PR rtl-optimization/80343
27808 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27809 new scratch pseudo.
27811 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27814 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27815 to ubsan_encode_value.
27817 2017-04-13 Jeff Law <law@redhat.com>
27819 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27820 appearing in DEBUG_INSNs.
27822 2017-04-13 Martin Liska <mliska@suse.cz>
27824 PR gcov-profile/80413
27825 * gcov-io.c (gcov_write_string): Copy to buffer just when
27826 allocated size is greater than zero.
27828 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27831 * dwarf2out.c (decls_for_scope): Ignore declarations of
27832 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27834 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27837 * ipa-visibility.c (non_local_p): Fix typos.
27838 (localize_node): When localizing symbol in same comdat group,
27839 dissolve the group only when we know external symbols are going
27841 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27843 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27845 PR tree-optimization/79390
27846 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27847 order does not result in usable sequence, retry with reversed operand
27853 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27854 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27855 op0 instead of fold_convert_loc (loc, type, arg0).
27857 2017-04-12 Jeff Law <law@redhat.com>
27859 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27860 has a delay slot in the generated code.
27862 * config/cris/cris.md (cris_preferred_reload_class): Return
27863 GENNONACR_REGS rather than GENERAL_REGS.
27865 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27868 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27869 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27870 signedness of the result type.
27872 2017-04-12 Richard Biener <rguenther@suse.de>
27873 Jeff Law <law@redhat.com>
27875 PR tree-optimization/80359
27876 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27877 trim stores to TARGET_MEM_REFs.
27879 2017-04-12 Richard Biener <rguenther@suse.de>
27881 PR tree-optimization/79390
27882 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27883 threading case even more.
27885 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27888 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27889 for quad_address_p for TImode, instead of just not indexed_address.
27891 2017-04-12 Richard Biener <rguenther@suse.de>
27892 Bernd Edlinger <bernd.edlinger@hotmail.de>
27894 PR middle-end/79671
27895 * alias.c (component_uses_parent_alias_set_from): Handle
27896 TYPE_TYPELESS_STORAGE.
27897 (get_alias_set): Likewise.
27898 * tree-core.h (tree_type_common): Add typeless_storage flag.
27899 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27900 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27901 for types containing members with TYPE_TYPELESS_STORAGE.
27902 (place_field): Likewise.
27903 (layout_type): Likewise for ARRAY_TYPE.
27904 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27905 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27906 TYPE_TYPELESS_STORAGE.
27907 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27909 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27912 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27913 first argument to type.
27915 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27919 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27920 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27921 (rs6000_expand_binop_builtin): Likewise.
27922 (rs6000_expand_ternop_builtin): Likewise; also add missing
27923 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27924 vshasigma built-ins.
27925 * doc/extend.texi: Document that vec_xxpermdi's third argument
27926 must be a constant.
27928 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27930 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27931 Use shift_const cost parameter when calculating gain of STV shifts.
27933 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27935 PR rtl-optimization/70478
27936 * lra-constraints.c (process_alt_operands): Check memory for
27937 disfavoring memory insn operand.
27939 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27941 PR middle-end/80100
27942 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27943 left shift in unsigned HOST_WIDE_INT type.
27945 PR rtl-optimization/80385
27946 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27947 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27950 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27951 if they have any depend clauses.
27953 2017-04-11 Martin Liska <mliska@suse.cz>
27956 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27957 * ipa-split.c (split_function): Create a local comdat symbol
27958 if caller is in a comdat group.
27960 2017-04-11 Martin Liska <mliska@suse.cz>
27963 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27966 2017-04-11 Martin Sebor <msebor@redhat.com>
27968 PR middle-end/80364
27969 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27970 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27972 (directive::set_width, directive::set_precision, format_character):
27974 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27977 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27980 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27981 conflict, set target->arch_name instead of target->cpu_name.
27983 2017-04-11 Richard Biener <rguenther@suse.de>
27985 PR tree-optimization/80374
27986 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27987 build_zero_cst, remove fold_convertible_p check again.
27989 2017-04-11 Martin Liska <mliska@suse.cz>
27992 * ubsan.c (instrument_object_size): Do not instrument register
27995 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27998 * config/i386/i386-builtin-types.def
27999 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
28000 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
28001 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
28002 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
28003 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
28004 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
28005 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
28006 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
28007 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
28008 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
28009 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
28010 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
28011 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
28012 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
28013 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
28014 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
28015 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
28016 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
28017 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
28018 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
28019 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
28020 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
28021 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
28022 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
28023 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
28024 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
28025 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
28026 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
28027 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
28028 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
28029 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
28030 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
28031 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
28032 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
28033 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
28034 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
28035 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
28036 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
28037 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
28038 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
28039 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
28040 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
28041 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
28042 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
28043 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
28044 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
28046 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
28047 flag to second_arg_count, handle 4 argument function type _COUNT
28048 aliases, handle second_arg_count on second argument rather than last.
28050 2017-04-10 Jeff Law <law@redhat.com>
28052 PR tree-optimization/80374
28053 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
28054 record anything if we can not convert integer_zero_node to the
28057 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28060 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28061 Enhance special handling given to the TARGET_P9_MINMAX option in
28062 relation to certain other options.
28064 2017-04-10 Bin Cheng <bin.cheng@arm.com>
28066 PR tree-optimization/80153
28067 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
28068 remove POINTER_PLUS_EXPR's base part directly, rather than through
28071 2017-04-10 Richard Biener <rguenther@suse.de>
28072 Bin Cheng <bin.cheng@arm.com>
28074 PR tree-optimization/80153
28075 * tree-affine.c (aff_combination_to_tree): Get base pointer from
28076 the first element of pointer type aff_tree. Build result expr in
28078 (add_elt_to_tree): Convert to type unconditionally. Remove other
28079 fold_convert calls.
28080 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
28081 (rewrite_use_nonlinear_expr): Check invariant using iv information.
28083 2017-04-10 Richard Biener <rguenther@suse.de>
28085 * tree-ssa-structalias.c (find_func_aliases): Properly handle
28088 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
28090 PR rtl-optimization/70478
28091 * lra-constraints.c (curr_small_class_check): New.
28092 (update_and_check_small_class_inputs): New.
28093 (process_alt_operands): Update curr_small_class_check. Disfavor
28094 alternative insn memory operands. Check available regs for small
28097 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28100 * config/mips/mips.opt (-mvirt): Update description.
28101 * doc/invoke.texi (-mvirt): Likewise.
28103 2017-04-10 Richard Biener <rguenther@suse.de>
28105 PR middle-end/80362
28106 * fold-const.c (fold_binary_loc): Look at unstripped ops when
28107 looking for NEGATE_EXPR in -A / -B to A / B folding.
28109 2017-04-10 Martin Liska <mliska@suse.cz>
28111 PR gcov-profile/80224
28112 * gcov.c (print_usage): Fix usage string.
28113 (get_gcov_intermediate_filename): Remove.
28114 (output_gcov_file): Use both for normal and intermediate format.
28115 (generate_results): Do not initialize special file for
28116 intermediate format.
28118 2017-04-10 Richard Biener <rguenther@suse.de>
28120 PR tree-optimization/80304
28121 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
28124 2017-04-10 Nathan Sidwell <nathan@acm.org>
28127 * config/rs6000/rs6000.c (rs6000_vector_type): New.
28128 (rs6000_init_builtins): Use it.
28130 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28132 * config/arm/arm.md (<mrc>): Add mode to SET source.
28133 (<mrrc>): Likewise.
28135 2017-04-10 Richard Biener <rguenther@suse.de>
28137 PR middle-end/80344
28138 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
28140 2017-04-10 Jakub Jelinek <jakub@redhat.com>
28143 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
28144 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
28145 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
28146 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
28147 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
28148 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
28149 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
28150 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
28151 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
28152 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
28153 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
28154 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
28155 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
28156 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
28157 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
28158 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
28159 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
28160 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
28161 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
28162 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
28163 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
28164 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
28165 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
28167 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
28169 PR rtl-optimization/70478
28170 * lra-constraints.c: Reverse the last patch.
28172 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
28174 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
28175 Add comment for WCHAR_T.
28177 2017-04-08 Martin Liska <mliska@suse.cz>
28180 2017-04-07 Martin Liska <mliska@suse.cz>
28183 * ipa-split.c (split_function): Add function part to a same comdat
28186 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28189 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
28191 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
28193 * rs6000/rs6000.c (vec_load_pendulum): Rename...
28194 (vec_pairing): ...to this.
28195 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
28196 (rs6000_sched_init): Adjust for name change.
28197 (struct rs6000_sched_context): Likewise.
28198 (rs6000_init_sched_context): Likewise.
28199 (rs6000_set_sched_context): Likewise.
28201 2017-04-07 Jakub Jelinek <jakub@redhat.com>
28207 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
28209 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
28210 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
28211 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
28213 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
28215 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
28217 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28219 PR rtl-optimization/70703
28220 * ira-color.c (update_conflict_hard_regno_costs): Use
28221 int64_t instead of HOST_WIDE_INT.
28223 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
28225 PR rtl-optimization/70478
28226 * lra-constraints.c (process_alt_operands): Disfavor alternative
28227 insn memory operands.
28229 2017-04-07 Jeff Law <law@redhat.com>
28231 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
28232 CALL and NOTE_INSN_CALL_ARG_LOCATION.
28234 2017-04-07 Martin Liska <mliska@suse.cz>
28237 * config/aarch64/aarch64.c (aarch64_process_target_attr):
28238 Show error message instead of an ICE.
28240 2017-04-07 Martin Liska <mliska@suse.cz>
28243 * ipa-split.c (split_function): Add function part to a same comdat
28246 2017-04-07 Richard Biener <rguenther@suse.de>
28248 PR middle-end/80341
28249 * tree.c (get_unwidened): Also handle ! for_type case for
28251 * convert.c (do_narrow): Split out from ...
28252 (convert_to_integer_1): ... here. Do not pass final truncation
28253 type to get_unwidened for TRUNC_DIV_EXPR.
28255 2017-04-07 Richard Biener <rguenther@suse.de>
28257 * tree-affine.c (wide_int_ext_for_comb): Take type rather
28259 (aff_combination_const): Adjust.
28260 (aff_combination_scale): Likewise.
28261 (aff_combination_add_elt): Likewise.
28262 (aff_combination_add_cst): Likewise.
28263 (aff_combination_convert): Likewise.
28264 (add_elt_to_tree): Likewise. Remove unused argument.
28265 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
28267 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
28269 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
28271 * config/arm/arm.c (arm_default_short_enums): Use
28272 ARM_DEFAULT_SHORT_ENUMS.
28273 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
28275 2017-04-06 Jakub Jelinek <jakub@redhat.com>
28278 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
28279 members with redundant out-of-class redeclaration.
28281 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28284 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
28285 * config/i386/i386.md (*zero_extendsidi2):
28286 Add (?*x,*x) and (?*v,*v) alternatives.
28288 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
28291 * config/i386/i386.c (ix86_expand_builtin)
28292 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
28293 mode from insn data. Convert operands to insn operand mode.
28294 Copy operands that don't satisfy insn predicate to a register.
28296 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
28298 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
28301 2017-04-06 Richard Biener <rguenther@suse.de>
28303 PR tree-optimization/80334
28304 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
28305 preserve alignment of accesses.
28307 2017-04-06 Richard Biener <rguenther@suse.de>
28309 PR tree-optimization/80262
28310 * tree-sra.c (build_ref_for_offset): Preserve address-space
28312 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
28313 Drop useless address-space information on MEM_REF offsets.
28315 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
28317 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
28319 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28321 PR rtl-optimization/70703
28322 * ira-color.c (update_conflict_hard_regno_costs): Use
28323 HOST_WIDE_INT instead of long.
28325 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
28328 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
28329 not defined for x86_64 target. Add -mmmx target option when __SSE2__
28331 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
28332 for x86_64 target. Handle -m3dnowa option.
28334 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
28336 PR rtl-optimization/70703
28337 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
28338 (update_conflict_hard_regno_costs): Use long instead of unsigned
28339 arithmetic for cost calculation.
28341 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28342 Bernd Edlinger <bernd.edlinger@hotmail.de>
28345 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
28348 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
28351 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
28352 ptr_mode with Pmode throughout.
28353 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
28354 into probe_stack_range and use DImode.
28356 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
28359 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
28360 call_eh_return is true.
28362 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28364 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
28365 Initialize last_match_fntype_index.
28367 2017-04-05 Jakub Jelinek <jakub@redhat.com>
28370 * tree-nvr.c: Include internal-fn.h.
28371 (pass_return_slot::execute): Ignore internal calls without
28374 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28375 Richard Biener <rguenther@suse.de>
28378 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
28379 captures used multiple times, except for the last use.
28380 * generic-match-head.c: Include gimplify.h.
28382 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28384 PR tree-optimization/79390
28385 * target.h (struct noce_if_info): Declare.
28386 * targhooks.h (default_noce_conversion_profitable_p): Declare.
28387 * target.def (noce_conversion_profitable_p): New target hook.
28388 * ifcvt.h (struct noce_if_info): New type, moved from ...
28389 * ifcvt.c (struct noce_if_info): ... here.
28390 (noce_conversion_profitable_p): Renamed to ...
28391 (default_noce_conversion_profitable_p): ... this. No longer
28393 (noce_try_store_flag_constants, noce_try_addcc,
28394 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
28395 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
28396 instead of noce_conversion_profitable_p.
28397 * config/i386/i386.c: Include ifcvt.h.
28398 (ix86_option_override_internal): Don't override
28399 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
28400 (ix86_noce_conversion_profitable_p): New function.
28401 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
28402 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
28403 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
28404 * doc/tm.texi: Regenerated.
28406 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28408 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
28411 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
28414 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
28415 instructions for small multiply cores.
28417 2017-04-04 Jeff Law <law@redhat.com>
28419 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28421 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28422 that are not set by the loop over the elements.
28424 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28427 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28428 int mode, convert_modes it to mode as unsigned, otherwise use
28429 lowpart_subreg to mode rather than SImode.
28430 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28431 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28432 Use DImode instead of SImode for the shift count operand.
28433 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28436 2017-04-04 Richard Biener <rguenther@suse.de>
28438 PR middle-end/80281
28439 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28440 arithmetic done for the negate or the plus. Simplify.
28441 (A - (-B) -> A + B): Likewise.
28442 * fold-const.c (split_tree): Make sure to not negate pointers.
28444 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28446 PR rtl-optimization/60818
28447 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28448 a compare of comparisons with the thing compared if this results
28449 in a different machine mode.
28451 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28453 * alias.c (base_alias_check): Fix typo in comment.
28454 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28455 * cgraphunit.c (symbol_table::compile): Likewise.
28456 * collect2.c (maybe_run_lto_and_relink): Likewise.
28457 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28458 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28459 * config/avr/avr.c (avr_map_op_t): Likewise.
28460 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28461 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28462 * config/epiphany/epiphany.md (movcc): Likewise.
28463 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28464 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28466 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28467 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28468 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28469 * config/sh/sh.c (sh_rtx_costs): Likewise.
28470 * fold-const.c (fold_truth_andor): Likewise.
28471 * genautomata.c (collapse_flag): Likewise.
28472 * gengtype.h (struct type::u::s): Likewise.
28473 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28474 * input.c (FORMAT_AMOUNT): Likewise.
28475 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28476 (known_aggs_to_agg_replacement_list): Likewise.
28477 * ipa-inline-analysis.c: Likewise.
28478 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28479 * ipa-polymorphic-call.c
28480 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28481 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28482 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28484 * modulo-sched.c (apply_reg_moves): Likewise.
28485 * omp-expand.c (build_omp_regions_1): Likewise.
28486 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28487 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28488 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28489 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28490 * value-prof.c: Likewise.
28491 * var-tracking.c (val_reset): Likewise.
28493 2017-04-03 Richard Biener <rguenther@suse.de>
28495 PR tree-optimization/80275
28496 * fold-const.c (split_address_to_core_and_offset): Handle
28499 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28501 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28502 descriptors is at least equal to that of functions.
28504 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28506 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28508 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28511 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28512 (mov<IMOD4:mode>): New expander.
28513 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28515 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28517 PR rtl-optimization/79405
28518 * fwprop.c (propagations_left): New variable.
28519 (forward_propagate_into): Decrement it.
28520 (fwprop_init): Initialize it.
28521 (fw_prop): If the variable has reached zero, stop propagating.
28522 (fwprop_addr): Ditto.
28524 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28527 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28528 a FUNCTION_DECL, pass it as decl instead of origin to
28531 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28533 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28536 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28539 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28540 TARGET_VSX_SMALL_INTEGER.
28542 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28544 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28545 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28547 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28549 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28550 extraction from odd-numbered MSA register.
28552 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28554 PR middle-end/80173
28555 * expmed.c (store_bit_field_1): Don't attempt to create
28556 a word subreg out of hard registers wider than word if they
28557 have HARD_REGNO_NREGS of 1 for their mode.
28559 PR middle-end/80163
28560 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28561 conversions to integer types wider than word and pointer.
28564 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28565 (rtx_equal_for_cselib_p): Pass 0 to it.
28566 * cselib.c (cselib_hasher::equal): Likewise.
28567 (rtx_equal_for_cselib_1): Add depth argument. If depth
28568 is 128, don't look up VALUE locs and punt. Increment
28569 depth in recursive calls when walking VALUE locs.
28571 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28573 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28574 (make_gcov_file_name): Use the canonical path name for generating
28576 (read_line): Fix handling of files with ascii null bytes.
28578 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28580 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28581 to initialise a vector register instead
28582 of using a const_int.
28584 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28586 PR translation/80189
28587 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28588 diagnostic messages.
28590 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28593 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28594 (dfp_diex_<mode>): Update mode of operand 1.
28595 * doc/extend.texi (dxex, dxexq): Document change to return type.
28596 (diex, diexq): Document change to argument type.
28598 2017-03-30 Martin Jambor <mjambor@suse.cz>
28601 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28602 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28603 it reflects the signature changes performed at the callee side.
28604 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28605 to cgraph_build_function_type_skip_args.
28606 (build_function_decl_skip_args): Adjust call to the above function.
28608 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28611 * config/i386/sse.md
28612 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28613 register as dest whenever it is a MEM not rtx_equal_p to the
28614 corresponding dup operand, and when forcing into reg move the
28615 reg into the memory afterwards.
28616 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28617 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28618 for the force_reg mode.
28619 (avx512vl_vextractf128<mode>): Use register as dest either
28620 always when a MEM, or when it is a MEM not rtx_equal_p to the
28621 corresponding dup operand, or even not when it is a CONST_VECTOR
28622 depending on the mode and lo vs. hi.
28623 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28625 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28626 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28627 Likewise. Require that operands[2] is even.
28628 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28629 Remove extraneous parens. Require that operands[2] is a multiple
28631 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28632 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28633 disallow memory then.
28635 2017-03-30 Richard Biener <rguenther@suse.de>
28637 PR tree-optimization/77498
28638 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28639 to non-constants over backedges.
28641 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28643 PR rtl-optimization/80233
28644 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28645 as last_combined_insn. Do not test for BARRIER_P separately.
28647 2017-03-29 Andreas Schwab <schwab@suse.de>
28650 * calls.c (prepare_call_address): Convert funexp to Pmode before
28651 copying to temp reg.
28653 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28655 PR tree-optimization/80158
28656 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28657 Handle possible future case of more than one alternate
28659 (replace_rhs_if_not_dup): Likewise.
28660 (replace_one_candidate): Likewise.
28662 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28664 PR rtl-optimization/80193
28665 * ira.c (ira): Do not check allocation for LRA.
28667 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28669 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28670 (nvptx_output_simt_exit): Declare.
28671 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28672 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28673 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28674 (nvptx_declare_function_name): ...here. Emit declaration of local
28675 memory space buffer for omp_simt_enter insn.
28676 (nvptx_output_unisimt_switch): New.
28677 (nvptx_output_softstack_switch): New.
28678 (nvptx_output_simt_enter): New.
28679 (nvptx_output_simt_exit): New.
28680 * config/nvptx/nvptx.h (struct machine_function): New fields
28681 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28682 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28683 (UNSPECV_SIMT_EXIT): Ditto.
28684 (omp_simt_enter_insn): New insn.
28685 (omp_simt_enter): New expansion.
28686 (omp_simt_exit): New insn.
28687 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28689 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28690 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28691 (expand_GOMP_SIMT_EXIT): New.
28692 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28693 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28694 (GOMP_SIMT_EXIT): Ditto.
28695 * target-insns.def (omp_simt_enter): New insn.
28696 (omp_simt_exit): Ditto.
28697 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28699 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28700 (lower_rec_input_clauses): Likewise.
28701 (lower_lastprivate_clauses): Handle SIMT privatization.
28703 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28704 (ompdevlow_adjust_simt_enter): New.
28705 (find_simtpriv_var_op): New.
28706 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28707 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28709 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28710 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28711 (copy_decl_for_dup_finish): Ditto.
28713 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28715 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28718 * config/i386/i386.c (ix86_option_override_internal): Always
28719 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28721 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28723 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28725 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28727 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28728 mark new edge's irreducible flag accordign to it.
28729 (vect_do_peeling): Check loop preheader edge's irreducible flag
28730 and pass it to function slpeel_add_loop_guard.
28732 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28734 PR tree-optimization/80218
28735 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28736 Update block frequencies and counts.
28738 2017-03-28 Richard Biener <rguenther@suse.de>
28740 PR tree-optimization/78644
28741 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28742 of a simplification result we may not use it at all.
28744 2017-03-28 Richard Biener <rguenther@suse.de>
28747 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28748 without arguments, generate default definition of a SSA name.
28750 2017-03-28 Richard Biener <rguenther@suse.de>
28752 PR middle-end/80222
28753 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28754 TYPE_REF_CAN_ALIAS_ALL references.
28755 * fold-const.c (fold_indirect_ref_1): Likewise.
28757 2017-03-28 Martin Liska <mliska@suse.cz>
28760 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28761 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28763 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28764 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28766 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28767 (EXTRA_SPECS): Define.
28768 (SUBTARGET_EXTRA_SPECS): Likewise.
28769 (SUBTARGET_CPP_SPEC): Likewise.
28770 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28771 SUBTARGET_EXTRA_SPECS.
28772 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28774 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28776 * config/arc/simdext.md (vst64_insn): Update pattern.
28777 (vld32wh_insn): Likewise.
28778 (vld32wl_insn): Likewise.
28779 (vld64_insn): Likewise.
28780 (vld32_insn): Likewise.
28782 2017-03-28 Marek Polacek <polacek@redhat.com>
28785 * fold-const.c (fold_comparison): Use protected_set_expr_location
28786 instead of SET_EXPR_LOCATION.
28788 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28790 * tree.c (add_expr): Avoid name lookup warning.
28792 2017-03-27 Jeff Law <law@redhat.com>
28794 PR tree-optimization/80216
28795 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28796 function name. Limit recursion depth.
28797 (record_temporary_equivalences): Corresponding changes.
28799 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28801 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28804 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28807 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28809 * cfgcleanup.c (reg_note_cfa_p): New array.
28810 (insns_have_identical_cfa_notes): New function.
28811 (old_insns_match_p): Don't cross-jump in between /f
28812 and non-/f instructions. If both i1 and i2 are frame related,
28813 verify all CFA notes, their order and content.
28815 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28818 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28819 HImode and SImode with zero extend to DImode to one insn.
28820 (bswap<mode>2_extenddi): Likewise.
28821 (bswapsi2_extenddi): Likewise.
28822 (bswaphi2_extendsi): Likewise.
28823 (bswaphi2): Combine bswap HImode and SImode into one insn.
28824 Separate memory insns from swapping register.
28825 (bswapsi2): Likewise.
28826 (bswap<mode>2): Likewise.
28827 (bswaphi2_internal): Delete, no longer used.
28828 (bswapsi2_internal): Likewise.
28829 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28830 store, and gpr<-gpr swap insns.
28831 (bswap<mode>2_store): Likewise.
28832 (bswaphi2_reg): Register only splitter, combine with the splitter.
28833 (bswaphi2 splitter): Likewise.
28834 (bswapsi2_reg): Likewise.
28835 (bswapsi2 splitter): Likewise.
28836 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28837 the insns into load, store, and register/register insns.
28838 (bswapdi2_ldbrx): Likewise.
28839 (bswapdi2_load): Likewise.
28840 (bswapdi2_store): Likewise.
28841 (bswapdi2_reg): Likewise.
28843 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28845 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28846 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28848 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28851 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28853 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28854 special handling for target option conflicts between dform
28855 options (-mpower9-dform, -mpower9-dform-vector,
28856 -mpower9-dform-scalar) and -mno-direct-move.
28858 2017-03-27 Richard Biener <rguenther@suse.de>
28860 PR tree-optimization/80181
28861 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28863 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28865 * config/arc/predicates.md (move_double_src_operand): Replace the
28866 call to move_double_src_operand with a call to address_operand.
28868 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28870 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28871 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28872 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28874 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28876 * config/arc/predicates.md (long_immediate_loadstore_operand):
28877 Consider scaled addresses cases.
28879 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28881 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28882 restored when in interrupt.
28883 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28884 doesn't have delay slot.
28886 2017-03-27 Richard Biener <rguenther@suse.de>
28889 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28890 inlined thunk clones.
28892 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28895 * asan.c (instrument_derefs): Copy over last operand from
28896 original COMPONENT_REF to the new COMPONENT_REF with
28897 DECL_BIT_FIELD_REPRESENTATIVE.
28898 * ubsan.c (instrument_object_size): Likewise.
28900 2017-03-27 Richard Biener <rguenther@suse.de>
28902 PR tree-optimization/80170
28903 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28904 sure DR/SCEV didnt fold in constants we do not see when looking
28905 at the reference base alignment.
28907 2017-03-27 Richard Biener <rguenther@suse.de>
28909 PR middle-end/80171
28910 * gimple-fold.c (fold_ctor_reference): Properly guard against
28911 NULL return value from canonicalize_constructor_val.
28913 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28916 * config/i386/i386.c (ix86_expand_builtin)
28917 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28918 flags reg setting and flags reg using instructions.
28919 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28920 clobbering instructions to zero extend op2.
28922 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28924 * doc/install.texi (Configuration) <--with-aix-soname>:
28925 Update link to AIX ld.
28927 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28929 PR rtl-optimization/80160
28930 PR rtl-optimization/80159
28931 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28932 reg_alternate_class into account.
28934 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28937 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28938 to consider in curr_insn_transform.
28940 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28942 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28943 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28944 and emit_mode_inner.
28946 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28948 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28949 argument to the overloaded builtin variants. Use the new flag to
28950 deprecate certain builtin variants.
28951 * config/s390/s390-builtin-types.def: Add new builtin types.
28952 * config/s390/s390-builtins.h: Support new flags field for
28953 overloaded builtins.
28954 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28955 (s390_macro_to_expand): Enable vector float data type.
28956 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28957 builtins by incrementing the __VEC__ version number.
28958 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28960 (s390_resolve_overloaded_builtin): Emit error messages depending
28961 on the builtin flags.
28962 * config/s390/s390.c (s390_expand_builtin): Support additional
28963 flags argument. Change error message to match the messages
28964 emitted in s390-c.c.
28965 * config/s390/s390.md: New UNSPEC_* constants.
28966 (op_type): Add new instruction types.
28967 * config/s390/vecintrin.h: Add new builtins and test data class
28969 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28970 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28971 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28972 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28973 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28974 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28976 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28977 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28978 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28979 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28981 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28982 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28983 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28984 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28986 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28987 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28988 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28989 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28990 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28991 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28992 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28994 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28995 ("vec_scatter_element<V_HW_4:mode>_DI")
28996 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28997 ("vec_fpint<mode>", "vflls")
28998 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28999 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
29000 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
29001 ("*vec_cmphe<mode>_cc"): ... these.
29003 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
29004 mode constant instead of magic value.
29006 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29008 * config/s390/s390.c (s390_expand_vec_compare): Support other
29009 vector floating point modes than just V2DF.
29010 (s390_expand_vcond): Likewise.
29011 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
29012 (s390_cannot_change_mode_class): Prevent mode changes between TF
29013 and V1TF in vector registers.
29014 * config/s390/s390.md (DF, SF): New mode attributes.
29015 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
29016 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
29017 SFmode support for VRs.
29018 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
29020 (VFT, VF_HW): New mode iterators.
29021 (vw, sdx): New mode attributes.
29022 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
29023 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
29024 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
29025 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
29026 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
29027 also the new vector floating point modes. Renaming to ...
29029 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
29030 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
29031 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
29032 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
29033 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
29034 ("vec_unordered<mode>"): ... these.
29036 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
29037 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
29038 ("*vec_extendv2df"): New insn definitions.
29040 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29042 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
29043 ("mulditi3_2", "*muldi3_sign"): New patterns.
29044 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
29045 rename the pattern definition.
29047 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29049 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
29051 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
29053 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29055 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
29056 instruction if possible.
29057 * config/s390/vector.md (vec_halfnumelts): New mode
29059 ("*vec_vllezlf<mode>"): New pattern.
29061 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29063 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
29064 ("popcountv4si2", "popcountv2di2"): Rename to ...
29065 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
29066 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
29068 ("popcount<mode>2_vxe"): New pattern.
29070 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29072 * common/config/s390/s390-common.c (processor_flags_table): Add
29074 * config.gcc: Add arch12.
29075 * config/s390/driver-native.c (s390_host_detect_local_cpu):
29076 Default to arch12 for unknown CPU model numbers.
29077 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
29078 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
29079 PROCESSOR_max sanity check.
29080 * config/s390/s390-opts.h (enum processor_type): Add
29082 * config/s390/s390.c (processor_table): Add arch12.
29083 (s390_expand_builtin): Add check for B_VXE flag.
29084 (s390_issue_rate): Add PROCESSOR_ARCH12.
29085 (s390_get_sched_attrmask): Likewise.
29086 (s390_get_unit_mask): Likewise.
29087 (s390_sched_score): Enable z13 scheduling for arch12.
29088 (s390_sched_reorder): Likewise.
29089 (s390_sched_variable_issue): Likewise.
29090 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
29092 (s390_tune_attr): Use z13 scheduling also for arch12.
29093 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
29094 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
29095 (TARGET_VXE_P): New macros.
29096 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
29097 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
29098 * config/s390/s390.opt: Add arch12 as processor_type.
29100 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29102 * config/s390/s390.md
29103 ("fixuns_truncdddi2", "fixuns_trunctddi2")
29104 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
29105 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
29107 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
29108 Rename expanders to ...
29110 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
29111 ("fixuns_truncdddi2_emu"): ... these.
29113 ("fixuns_trunc<mode>si2_emu"): New expander.
29115 ("*fixuns_truncdfdi2_z13"): Rename to ...
29116 ("*fixuns_truncdfdi2_vx"): ... this.
29118 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29120 * config/s390/2964.md: Remove the single element vector compare
29121 instructions which are no longer used.
29122 * config/s390/s390.c (s390_select_ccmode): Remove handling of
29124 (s390_canonicalize_comparison): Remove handling of DFmode
29126 (s390_expand_vec_compare_scalar): Remove function.
29127 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
29128 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
29130 ("*cmp<mode>_ccs"): Add wfcdb instruction.
29132 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29134 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
29136 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
29137 will anyway by matched by mov<mode>_64dfp.
29139 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29141 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
29142 vlef/vstef. Add missing operand to vleif.
29144 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29146 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
29147 pair for all vector types with 64 bit elements.
29148 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
29149 * config/s390/vector.md (V_HW_64): ... here.
29150 (V_128_NOSINGLE): New mode iterator.
29151 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
29152 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
29153 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
29154 ("*vec_load_pairv2di"): Change to ...
29155 ("*vec_load_pair<mode>"): ... this one.
29157 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29159 * config/s390/constraints.md: Add comments.
29160 (jKK): Reject element sizes > 8 bytes.
29161 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
29163 * config/s390/s390.md: Add the s_operand checks formerly in
29164 s390_split_ok_p to various splitters where they are still
29166 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
29167 for 128 bit vectors. Plus two splitters.
29169 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29171 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
29174 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29177 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
29178 error if the boundary argument is not constant.
29180 2017-03-24 Jakub Jelinek <jakub@redhat.com>
29182 PR rtl-optimization/80112
29183 * loop-doloop.c (doloop_condition_get): Don't check condition
29184 if cmp isn't SET with IF_THEN_ELSE src.
29186 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29188 PR tree-optimization/80158
29189 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
29190 replacing a candidate statement, also replace it for the
29191 candidate's alternate interpretation.
29192 (replace_rhs_if_not_dup): Likewise.
29193 (replace_one_candidate): Likewise.
29195 2017-03-24 Richard Biener <rguenther@suse.de>
29197 PR tree-optimization/80167
29198 * graphite-isl-ast-to-gimple.c
29199 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
29201 (translate_isl_ast_to_gimple::get_rename): Likewise.
29203 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
29205 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
29206 handling of certain combinations of target options, including the
29207 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
29208 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
29210 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29213 * config/arm/arm.md (*load_multiple): Add reload_completed to
29214 matching condition.
29216 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29217 Richard Biener <rguenth@suse.de>
29219 PR tree-optimization/79908
29220 PR tree-optimization/80136
29221 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29222 been cast away, gimplify_and_add suffices.
29224 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
29226 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
29228 2017-03-23 Richard Biener <rguenther@suse.de>
29230 PR tree-optimization/80032
29231 * gimplify.c (gimple_push_cleanup): Forced unconditional
29232 cleanups still have to go to the conditional_cleanups
29235 2017-03-22 Jakub Jelinek <jakub@redhat.com>
29237 PR tree-optimization/80072
29238 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
29240 (next_operand_entry_id): Change type to unsigned int.
29241 (sort_by_operand_rank): Make sure to return the right return value
29242 even if unsigned fields are bigger than INT_MAX.
29243 (struct oecount): Change cnt and id type to unsigned int.
29244 (oecount_hasher::equal): Formatting fix.
29245 (oecount_cmp): Make sure to return the right return value
29246 even if unsigned fields are bigger than INT_MAX.
29247 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
29250 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
29251 TREE_READONLY on result if writing it more than once.
29254 * doc/invoke.texi (-fsanitize=thread): Document that with
29255 -fnon-call-exceptions atomics are not able to throw
29259 * tsan.c: Include tree-eh.h.
29260 (instrument_builtin_call): Call maybe_clean_eh_stmt or
29261 maybe_clean_or_replace_eh_stmt where needed.
29262 (instrument_memory_accesses): Add cfg_changed argument.
29263 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
29264 if it returned true.
29265 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
29267 PR rtl-optimization/63191
29268 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
29269 wrapper function, moved the whole old content into ...
29270 (ix86_delegitimize_address_1): ... this. New inline function.
29271 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
29272 true as last argument instead of ix86_delegitimize_address.
29274 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29276 * config/aarch64/aarch64.c (generic_branch_cost): Copy
29277 cortexa57_branch_cost.
29279 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
29281 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
29283 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29286 * doc/md.texi (Constraints): Document wA constraint.
29287 * config/rs6000/constraints.md (wA): New.
29288 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
29289 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
29290 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
29291 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
29293 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
29296 * gimplify.c (is_oacc_declared): New function.
29297 (oacc_default_clause): Use it to set default flags for acc declared
29298 variables inside parallel regions.
29299 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
29300 declared variables.
29301 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
29302 declare attribute to any decl as necessary.
29304 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
29307 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
29308 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
29309 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
29310 (arm_arch_lpae): This.
29311 * config/arm/arm.c (arm_arch7ve): Rename into ...
29312 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
29313 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
29316 2017-03-22 Martin Liska <mliska@suse.cz>
29319 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
29320 error message instead of an ICE.
29322 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29324 * doc/extend.texi (6.11 Additional Floating Types): Revise.
29326 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
29328 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29330 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29333 2017-03-21 Martin Sebor <msebor@redhat.com>
29335 * doc/extend.texi: Use "cannot" instead of "can't."
29336 * doc/hostconfig.texi: Same.
29337 * doc/install.texi: Same.
29338 * doc/invoke.texi: Same.
29339 * doc/loop.texi: Same.
29340 * doc/md.texi: Same.
29341 * doc/objc.texi: Same.
29342 * doc/rtl.texi: Same.
29343 * doc/tm.texi: Same.
29344 * doc/tm.texi.in: Same.
29345 * doc/trouble.texi: Same.
29347 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
29350 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
29351 (collect_checksum_attributes): Set it.
29352 (die_checksum_ordered): Use it.
29354 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29356 PR tree-optimization/79908
29357 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
29358 change: For a VA_ARG whose LHS has been cast away, use
29359 force_gimple_operand to construct the side effects.
29361 2017-03-21 David Malcolm <dmalcolm@redhat.com>
29363 PR translation/80001
29364 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
29365 more amenable to translation.
29366 (oacc_loop_auto_partitions): Likewise.
29368 2017-03-21 Marek Polacek <polacek@redhat.com>
29369 Martin Sebor <msebor@redhat.com>
29371 PR tree-optimization/80109
29372 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
29373 on INTEGRAL_TYPE_P.
29375 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29376 Segher Boessenkool <segher@kernel.crashing.org>
29379 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
29380 check reg_used_between_p between insn and one of succ or succ2
29381 depending on if succ is artificial insn not inserted into insn
29384 2017-03-21 Martin Liska <mliska@suse.cz>
29386 PR gcov-profile/80081
29387 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
29388 * doc/gcc.texi: Include gcov-dump stuff.
29389 * doc/gcov-dump.texi: New file.
29391 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
29393 PR rtl-optimization/79150
29394 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
29395 conditional jump, if the jump is the last insn of the loop.
29397 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29398 Richard Biener <rguenth@suse.de>
29400 PR tree-optimization/79908
29401 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
29402 been cast away, use force_gimple_operand to construct the side
29405 2017-03-21 Martin Liska <mliska@suse.cz>
29407 PR libfortran/79956
29408 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
29411 2017-03-21 Brad Spengler <spender@grsecurity.net>
29414 * plugin.c (htab_hash_plugin): New function.
29415 (add_new_plugin): Use it and adjust.
29416 (parse_plugin_arg_opt): Adjust.
29417 (init_one_plugin): Likewise.
29419 2017-03-21 Richard Biener <rguenther@suse.de>
29421 PR tree-optimization/80032
29422 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29423 if set force the cleanup to happen unconditionally.
29424 (gimplify_target_expr): Push inserted clobbers with force_uncond
29425 to avoid them being removed by control-dependent DCE.
29427 2017-03-21 Richard Biener <rguenther@suse.de>
29429 PR tree-optimization/80122
29430 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29431 va_arg_pack_len when the inlined call stmt requires pack
29433 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29435 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29438 * tsan.c (instrument_builtin_call): If the memory model argument
29439 is not a constant, assume it is valid.
29442 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29445 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29447 PR rtl-optimization/79910
29448 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29449 if its dest is used by an insn before I2 (other than the combined
29450 insns themselves, which are properly handled already).
29452 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29455 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29457 * combine.c (record_used_regs): New static function.
29458 (try_combine): Handle situations where there is an additional
29459 instruction between I2 and I3 which needs to have a LOG_LINK
29463 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29465 * combine.c (try_combine): Delete redundant i1 test. Call
29466 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29468 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29471 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29472 alternatives 13/14.
29474 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29476 PR tree-optimization/80054
29477 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29478 the optimization if a PHI or any of its arguments is not dominated
29479 by the candidate's basis. Use gphi* rather than gimple* as
29481 (replace_profitable_candidates): Clean up a gimple* variable that
29482 should be a gphi* variable.
29484 2017-03-20 Martin Sebor <msebor@redhat.com>
29487 * doc/extend.texi (attribute constructor): Document present limitation.
29489 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29492 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29493 __POWER9_VECTOR__ #ifdef control, change template definition to
29494 use Power9-specific built-in function.
29495 (vec_any_eq): Likewise.
29496 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29497 to control outcomes from this test.
29498 (vector_ae_<mode>p): For VEC_F modes, likewise.
29500 2017-03-20 Ian Lance Taylor <iant@google.com>
29502 * config/i386/i386.c (ix86_function_regparm): Save an extra
29503 register for -fsplit-stack with DECL_STATIC_CHAIN.
29505 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29508 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29509 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29511 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29513 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29515 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29518 2017-03-20 Marek Polacek <polacek@redhat.com>
29521 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29523 2017-03-20 Richard Biener <rguenther@suse.de>
29525 PR tree-optimization/80113
29526 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29527 allocate extra SSA name for PHI def.
29528 (add_close_phis_to_outer_loops): Likewise.
29529 (add_close_phis_to_merge_points): Likewise.
29530 (copy_loop_close_phi_args): Likewise.
29531 (copy_cond_phi_nodes): Likewise.
29533 2017-03-20 Martin Liska <mliska@suse.cz>
29535 PR middle-end/79753
29536 * tree-chkp.c (chkp_build_returned_bound): Do not build
29537 returned bounds for a LHS that's not a BOUNDED_P type.
29539 2017-03-20 Martin Liska <mliska@suse.cz>
29543 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29544 COMPLEX_CST and VECTOR_CST.
29546 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29549 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29550 target operand. A new splitter adds the clobber statement in case
29551 the target operand is dead anyway.
29553 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29555 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29556 to age-old versions of binutils and glibc.
29558 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29560 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29562 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29564 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29566 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29568 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29569 requirement for binutils 2.13.
29571 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29573 * combine.c (try_combine): Delete redundant i1 test. Call
29574 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29576 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29578 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29579 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29581 <riscv64-*-elf>: Re-arrange section
29582 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29583 <riscv32-*-linux>: Likewise.
29584 <riscv64-*-elf>: Likewise
29585 <riscv64-*-linux>: Likewise.
29587 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29590 * aarch64.opt(verbose-cost-dump): Fix typo.
29592 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29595 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29596 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29598 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29600 * reload.c (find_reloads): When reloading a nonoffsettable address,
29601 use RELOAD_OTHER for it and its address reloads.
29603 PR rtl-optimization/79910
29604 * combine.c (record_used_regs): New static function.
29605 (try_combine): Handle situations where there is an additional
29606 instruction between I2 and I3 which needs to have a LOG_LINK
29609 2017-03-17 Jeff Law <law@redhat.com>
29611 PR tree-optimization/71437
29612 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29613 conditional in the hash table first.
29614 (vrp_dom_walker::before_dom_children): Extract condition from
29615 ASSERT_EXPR. Record condition, its inverion and any implied
29616 conditions as well.
29618 2017-03-17 Marek Polacek <polacek@redhat.com>
29619 Markus Trippelsdorf <markus@trippelsdorf.de>
29621 PR tree-optimization/80079
29622 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29625 2017-03-17 Richard Biener <rguenther@suse.de>
29627 PR middle-end/80075
29628 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29629 Properly verify the LHS before the RHS possibly claims to be
29631 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29634 2017-03-17 Martin Jambor <mjambor@suse.cz>
29636 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29637 (List of -O2 options): Likewise.
29638 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29641 2017-03-17 Tom de Vries <tom@codesourcery.com>
29643 * gcov-dump.c (print_usage): Print bug_report_url.
29645 2017-03-17 Richard Biener <rguenther@suse.de>
29647 PR middle-end/80050
29648 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29649 (parser::peek): Likewise.
29651 2017-03-17 Richard Biener <rguenther@suse.de>
29653 PR tree-optimization/80048
29654 * sese.c (free_sese_info): Properly release rename_map and
29655 copied_bb_map elements.
29657 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29659 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29660 Add linked-list forward and backlinks. Insert on
29661 construction, remove on destruction.
29662 (class pass_store_merging): Add m_stores_head field.
29663 (pass_store_merging::terminate_and_process_all_chains):
29664 Iterate over m_stores_head list.
29665 (pass_store_merging::terminate_all_aliasing_chains):
29667 (pass_store_merging::execute): Check for debug stmts first.
29668 Push new chains onto the m_stores_head stack.
29670 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29673 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29674 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29675 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29677 2017-03-16 Jeff Law <law@redhat.com>
29679 PR tree-optimization/71437
29680 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29681 member function. Implementation moved into after_dom_children
29682 member function and into the threader's thread_outgoing_edges
29684 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29685 some code into new thread_outgoing_edges.
29686 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29687 definition. Simplify marker handling (do it here). Assume we always
29688 have the available expression and the const/copies tables.
29689 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29691 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29692 * tree-vrp.c (equiv_stack): No longer file scoped.
29693 (vrp_dom_walker): New class.
29694 (vrp_dom_walker::before_dom_children): New member function.
29695 (vrp_dom_walker::after_dom_children): Likewise.
29696 (identify_jump_threads): Setup domwalker. Use it rather than
29697 walking edges in a random order by hand. Simplify setup/finalization.
29698 (finalize_jump_threads): Remove.
29699 (vrp_finalize): Do not call identify_jump_threads here.
29700 (execute_vrp): Do it here instead and call thread_through_all_blocks
29703 PR tree-optimization/71437
29704 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29706 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29708 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29709 (dom_opt_dom_walker::thread_across_edge): Remove
29710 handle_dominating_asserts argument. All callers changed.
29711 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29712 changes. Remove calls to lhs_of_dominating_assert. Other
29713 uses of handle_dominating_asserts turn into unconditional code
29714 (simplify_control_stmt_condition_1): Likewise.
29715 (simplify_control_stmt_condition): Likewise.
29716 (thread_through_normal_block, thread_across_edge): Likewise.
29717 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29718 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29719 object if it is not an SSA_NAME.
29720 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29721 before calling into the VRP specific simplifiers.
29722 (identify_jump_threads): Remove handle_dominating_asserts
29725 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29728 * tree-diagnostic.c (default_tree_printer): No longer static.
29729 * tree-diagnostic.h (default_tree_printer): New prototype.
29731 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29733 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29734 Change ins into fmov.
29736 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29738 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29739 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29740 Use h_con constraint for operand 1.
29741 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29742 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29744 2017-03-15 Jeff Law <law@redhat.com>
29746 PR tree-optimization/71437
29747 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29748 (record_temporary_equivalences): Use it.
29750 PR tree-optimization/71437
29751 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29752 tree-ssa-scopedtables.
29753 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29754 (record_conditions, record_cond, vuse_eq): Likewise.
29755 (record_edge_info): Adjust to API tweak of record_conditions.
29756 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29757 (record_temporary_equivalences, optimize_stmt): Likewise.
29758 (eliminate_redundant_computations): Likewise.
29759 (record_equivalences_from_stmt): Likewise.
29760 * tree-ssa-scopedtables.c: Include options.h and params.h.
29761 (vuse_eq): New function, moved from tree-ssa-dom.c
29762 (build_and_record_new_cond): Likewise.
29763 (record_conditions): Likewise. Accept vector of conditions rather
29764 than edge_equivalence structure for first argument.
29765 for the first argument.
29766 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29767 from tree-ssa-dom.c.
29768 (avail_exprs_stack::record_cond): Likewise.
29769 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29770 from tree-ssa-dom.c.
29771 (avail_exprs_stack): Add new member functions lookup_avail_expr
29773 (record_conditions): Declare.
29775 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29778 * lra-constraints.c (process_alt_operands): Increase reject for
29779 reloading an input/output operand.
29781 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29784 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29785 insns to convert from signed/unsigned char/short to IEEE 128-bit
29787 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29789 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29792 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29793 subreg of inner mode for values already in registers.
29795 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29797 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29798 iteration reg is used after the loop.
29800 2017-03-14 Martin Sebor <msebor@redhat.com>
29802 PR tree-optimization/79800
29803 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29804 precision in negative-positive range.
29805 (format_floating): Call non-const overload with adjusted precision.
29807 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29810 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29813 2017-03-14 Martin Sebor <msebor@redhat.com>
29815 PR middle-end/80020
29816 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29817 * builtins.def (aligned_alloc): Use it.
29820 * Makefile.in (GTFILES): Add calls.c.
29821 * calls.c: Include "gt-calls.h".
29823 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29825 PR rtl-optimization/79728
29826 * regs.h (struct target_regs): New field
29827 x_contains_allocatable_regs_of_mode.
29828 (contains_allocatable_regs_of_mode): New macro.
29829 * reginfo.c (init_reg_sets_1): Initialize it, and change
29830 contains_reg_of_mode so it includes global regs as well.
29831 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29832 rather than contains_regs_of_mode.
29834 2017-03-14 Martin Liska <mliska@suse.cz>
29836 * doc/invoke.texi: Document options that can't be combined with
29837 -fcheck-pointer-bounds.
29839 2017-03-14 Martin Liska <mliska@suse.cz>
29841 PR middle-end/79831
29842 * doc/invoke.texi (-Wchkp): Document the option.
29844 2017-03-14 Martin Liska <mliska@suse.cz>
29846 * Makefile.in: Install gcov-dump.
29848 2017-03-14 Martin Liska <mliska@suse.cz>
29850 * multiple_target.c (expand_target_clones): Bail out for
29851 an invalid attribute.
29853 2017-03-14 Richard Biener <rguenther@suse.de>
29855 * alias.c (struct alias_set_entry): Pack properly.
29856 * cfgloop.h (struct loop): Likewise.
29857 * cse.c (struct set): Likewise.
29858 * ipa-utils.c (struct searchc_env): Likewise.
29859 * loop-invariant.c (struct invariant): Likewise.
29860 * lra-remat.c (struct cand): Likewise.
29861 * recog.c (struct change_t): Likewise.
29862 * rtl.h (struct address_info): Likewise.
29863 * symbol-summary.h (function_summary): Likewise.
29864 * tree-loop-distribution.c (struct partition): Likewise.
29865 * tree-object-size.c (struct object_size_info): Likewise.
29866 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29867 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29868 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29869 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29870 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29871 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29872 (struct _stmt_vec_info): Likewise.
29874 2017-03-14 Martin Liska <mliska@suse.cz>
29877 * multiple_target.c (create_dispatcher_calls): Check that
29878 a target can create a function dispatcher.
29880 2017-03-14 Martin Liska <mliska@suse.cz>
29883 * multiple_target.c (expand_target_clones): Drop local.local
29884 flag for default implementation.
29886 2017-03-14 Richard Biener <rguenther@suse.de>
29888 PR tree-optimization/80030
29889 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29891 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29893 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29894 gcc_fallthrough() instead of __attribute__((fallthrough));
29896 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29898 * doc/gcc.texi: Remove "up" link to (DIR).
29899 * doc/gccint.texi: Ditto.
29901 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29903 * doc/install.texi (Specific) <avr>: Remove reference to
29906 2017-03-13 Jeff Law <law@redhat.com>
29908 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29909 attribute rather than comments.
29911 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29912 match_scratch operand is highest.
29914 2017-03-13 Martin Liska <mliska@suse.cz>
29916 PR middle-end/78339
29917 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29918 is a CHKP clone, use original declaration.
29920 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29922 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29923 (arc_conditional_register_usage): Use a different allocation order
29924 when optimizing for size.
29925 * common/config/arc/arc-common.c (arc_option_optimization_table):
29926 Section anchors default on when optimizing for size.
29928 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29930 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29932 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29934 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29935 * config/arc/arc.md (cpu_facility): Add cd variant.
29936 (*movqi_insn): Add code density variant.
29937 (*movhi_insn): Likewise.
29938 (*movqi_insn): Likewise.
29939 (*addsi3_mixed): Likewise.
29940 (subsi3_insn): Likewise.
29942 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29944 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29946 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29948 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29949 expressions with MINUS and UNARY ops.
29951 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29954 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29956 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29957 between vec_select and vector argument.
29958 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29959 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29960 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29961 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29962 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29963 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29965 2017-03-13 Richard Biener <rguenther@suse.de>
29968 * params.def (vect-max-peeling-for-alignment): Fix typo.
29970 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29972 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29973 issue that only occurred with binutils below 2.18.
29975 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29977 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29978 refer to binutils 2.11/2.12 minimum.
29980 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29982 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29983 ftp.kernel.org and simplify binutils requirement.
29985 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29987 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29989 (Optimize Options): Ditto. Also remove redundancy.
29991 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29993 PR translation/79848
29994 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29996 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29997 to G_ to avoid double translation.
29999 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30001 PR translation/79923
30002 * auto-profile.c (get_combined_location): Convert leading
30003 character of diagnostics to lower case and remove trailing period.
30004 (read_profile): Likewise for various diagnostics.
30005 * config/arm/arm.c (arm_option_override): Remove trailing period
30006 from various diagnostics.
30007 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
30008 (msp430_expand_delay_cycles): Likewise.
30010 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30013 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
30014 full command-line argument, rather than just "str".
30015 (aarch64_validate_march): Likewise.
30016 (aarch64_validate_mtune): Likewise.
30018 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
30020 PR rtl-optimization/78911
30021 * lra-assigns.c (must_not_spill_p): New function.
30022 (spill_for): Use it.
30024 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30026 PR tree-optimization/79981
30027 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
30028 ATOMIC_COMPARE_EXCHANGE ifn result.
30029 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
30030 IFN_ATOMIC_COMPARE_EXCHANGE.
30032 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30035 * opts.c (parse_sanitizer_options): Add missing question mark to
30036 "did you mean" message.
30038 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30040 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
30042 (VMULEUH_UNS): Likewise.
30043 (VMULOUB_UNS): Likewise.
30044 (VMULOUH_UNS): Likewise.
30045 * config/rs6000/rs6000.c (builtin_function_type): Remove
30046 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
30048 2017-03-10 David Malcolm <dmalcolm@redhat.com>
30051 * read-rtl-function.c (function_reader::read_rtx_operand): Update
30052 x with result of extra_parsing_for_operand_code_0.
30053 (function_reader::extra_parsing_for_operand_code_0): Convert
30054 return type from void to rtx, returning x. When reading
30055 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
30056 larger size containing struct block_symbol.
30058 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
30060 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
30061 -mfloat128-hardware without -m64.
30063 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
30066 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
30067 entries to the case statement that marks unsigned arguments to
30068 overloaded functions.
30070 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30072 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
30073 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
30075 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
30078 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
30079 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
30081 2017-03-10 Martin Liska <mliska@suse.cz>
30085 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
30086 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
30089 2017-03-10 Olivier Hainque <hainque@adacore.com>
30091 * tree-switch-conversion (array_value_type): Start by resetting
30092 candidate type to it's main variant.
30094 2017-03-10 Jakub Jelinek <jakub@redhat.com>
30096 PR rtl-optimization/79909
30097 * combine.c (try_combine): Use simplify_replace_rtx on individual
30098 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
30099 of the whole CALL_INSN_FUNCTION_USAGE.
30101 PR tree-optimization/79972
30102 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
30103 get_range_info on SSA_NAMEs. Formatting fixes.
30105 2017-03-10 Richard Biener <rguenther@suse.de>
30106 Jakub Jelinek <jakub@redhat.com>
30108 PR tree-optimization/77975
30109 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
30110 edge to be constant.
30111 (get_val_for): For constant x return it. Formatting fix.
30112 (loop_niter_by_eval): Avoid pointless looping if the next iteration
30113 would use the same bases as the current one.
30115 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30117 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
30118 instead of vec_select for V1TImode.
30119 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
30121 (VSX_LE_128): Add V1TI to this mode iterator.
30122 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
30123 (*vsx_le_perm_store_<mode>): Likewise.
30124 (pre-reload splitter for VSX stores): Likewise.
30125 (post-reload splitter for VSX stores): Likewise.
30126 (*vsx_xxpermdi2_le_<mode>): Likewise.
30127 (*vsx_lxvd2x2_le_<mode>): Likewise.
30128 (*vsx_stxvd2x2_le_<mode>): Likewise.
30130 2017-03-09 Michael Eager <eager@eagercon.com>
30132 Correct failures with --enable-checking=yes,rtl.
30134 * config/microblaze/microblaze.c (microblaze_expand_shift):
30135 Replace GET_CODE test with CONST_INT_P and INTVAL test with
30136 test for const0_rtx.
30137 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
30138 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
30140 2017-03-09 Richard Biener <rguenther@suse.de>
30142 PR tree-optimization/79977
30143 * graphite-scop-detection.c (scop_detection::merge_sese):
30144 Handle the case of extra exits to blocks dominating the entry.
30146 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
30148 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30151 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
30153 PR rtl-optimization/79949
30154 * lra-constraints.c (process_alt_operands): Check memory when
30155 trying to predict a cycle. Print about the overall increase.
30157 2017-03-09 Richard Biener <rguenther@suse.de>
30159 PR middle-end/79971
30160 * gimple-expr.c (useless_type_conversion_p): Preserve
30161 TYPE_SATURATING for fixed-point types.
30163 2017-03-09 Richard Biener <rguenther@suse.de>
30166 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
30167 alignment of BLKmode params.
30169 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30172 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
30173 (VALL_NO_V2Q): Likewise.
30175 * config/aarch64/aarch64-simd.md
30176 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
30178 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
30179 VALL_NO_V2Q mode iterator.
30180 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
30182 2017-03-09 Martin Liska <mliska@suse.cz>
30184 PR tree-optimization/79631
30185 * tree-chkp-opt.c (chkp_is_constant_addr): Call
30186 tree_int_cst_sign_bit just for INTEGER constants.
30188 2017-03-09 Martin Liska <mliska@suse.cz>
30192 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
30195 2017-03-09 Marek Polacek <polacek@redhat.com>
30198 * tree.c (inchash::add_expr): Handle TREE_VEC.
30200 2017-03-09 Martin Liska <mliska@suse.cz>
30203 (chkp_narrow_size_and_offset): New function.
30204 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
30205 (void chkp_parse_bit_field_ref): New function.
30206 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
30207 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
30209 2017-03-09 Martin Liska <mliska@suse.cz>
30212 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
30213 (chkp_find_bounds_1): Remove gcc_unreachable.
30215 2017-03-09 Jakub Jelinek <jakub@redhat.com>
30218 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
30219 BUILT_IN_SYNC*, determine the access type from the size suffix and
30220 always build a MEM_REF with that type. Handle forgotten
30221 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
30224 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
30225 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
30226 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
30227 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
30228 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
30229 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
30230 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
30231 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
30232 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
30233 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
30234 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
30235 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
30236 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
30237 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
30238 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
30239 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
30240 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
30241 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
30242 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
30243 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
30244 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
30245 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
30246 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
30247 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
30248 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
30249 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
30250 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
30251 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
30252 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
30253 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
30254 definitions outside of __OPTIMIZE__ guarded section.
30257 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
30258 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
30259 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
30260 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
30263 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30265 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
30266 ("vfenez<mode>"): Add missing constraints.
30268 2017-03-08 Martin Sebor <msebor@redhat.com>
30271 * config/nds32/nds32.c (nds32_option_override):
30272 Fix misspelled diagnostic.
30274 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30277 * gimplify.c (gimplify_omp_for): Replace index var in outer
30278 taskloop statement with an artificial variable and add
30279 OMP_CLAUSE_PRIVATE clause for it.
30281 2017-03-08 Richard Biener <rguenther@suse.de>
30283 PR tree-optimization/79955
30284 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
30285 for accesses that are completely outside of the variable.
30287 2017-03-08 Andrew Haley <aph@redhat.com>
30289 PR tree-optimization/79943
30290 * tree-ssa-loop-split.c (compute_new_first_bound): When
30291 calculating the new upper bound, (END-BEG) should be added, not
30294 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30296 * config/avr/avr.md (setmemhi): Make sure match_dup
30297 operand number comes before match_scratch.
30299 2017-03-08 Richard Biener <rguenther@suse.de>
30301 PR tree-optimization/79920
30302 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
30303 with ncopies == 1 to ...
30304 (vect_transform_slp_perm_load): ... here. Properly compute
30305 all element loads by iterating VF times over the group. Do
30306 not handle ncopies (computed in a broken way) in
30307 vect_create_mask_and_perm.
30309 2017-03-08 Jakub Jelinek <jakub@redhat.com>
30312 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
30313 is a uniform vector, use uniform_vector_p return value instead of
30314 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
30316 2017-03-07 Marek Polacek <polacek@redhat.com>
30318 PR middle-end/79809
30319 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
30320 (alloca_call_type): Likewise.
30322 2017-03-07 Martin Liska <mliska@suse.cz>
30324 * gcov.c (process_args): Put comment to correct location.
30326 2017-03-07 Martin Liska <mliska@suse.cz>
30328 PR middle-end/68270
30329 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
30330 Use array_at_struct_end_p instead of DECL_CHAIN (field).
30331 (chkp_narrow_bounds_for_field): Likewise.
30332 (chkp_parse_array_and_component_ref): Pass one more argument to
30335 2017-03-07 Richard Biener <rguenther@suse.de>
30337 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
30340 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
30342 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
30343 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
30345 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30348 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
30349 to end of description.
30350 (PARAM_MAX_STORES_TO_MERGE): Likewise.
30352 2017-03-07 Jakub Jelinek <jakub@redhat.com>
30354 PR rtl-optimization/79901
30355 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
30357 (*avx512f_<code><mode>3<mask_name>): ... this.
30358 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
30359 iterator instead of VI8_AVX2_AVX512BW.
30361 PR rtl-optimization/79901
30362 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
30363 min/max expander, expand it using expand_vec_cond_expr.
30366 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
30369 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30372 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
30373 to void * for PCH reasons.
30374 * dwarf2out.c (output_loc_operands, output_die): Cast
30375 v.val_vec.array to unsigned char *.
30377 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
30380 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
30383 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
30385 PR rtl-optimization/79571
30386 * lra-constraints.c (process_alt_operands): Calculate static
30387 reject and subtract it from overall when only addresses will be
30390 2017-03-06 Julia Koval <julia.koval@intel.com>
30393 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
30394 incoming stack boundary to 128 for 64-bit targets.
30396 2017-03-06 Richard Biener <rguenther@suse.de>
30398 PR tree-optimization/79894
30399 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
30400 to NULL after folding it.
30402 2017-03-06 Richard Biener <rguenther@suse.de>
30404 PR tree-optimization/79824
30405 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
30406 check disabling peeling for gaps.
30408 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
30410 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
30411 attributes): Document gettimeofday.
30413 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30415 * config/s390/s390.c (s390_option_override_internal): Set
30416 PARAM_MIN_VECT_LOOP_BOUND
30418 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30420 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30421 * config/s390/s390.md: Likewise.
30423 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30426 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30427 (<avx2_avx512>_perm<mode>): Rename to ...
30428 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30430 (<avx512>_perm<mode>_mask): Rename to ...
30431 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30433 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30434 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30435 instead of VI8F_256_512.
30436 (avx512f_perm<mode>): New define_expand.
30437 (avx512f_perm<mode>_mask): Likewise.
30438 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30439 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30441 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30443 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30444 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30446 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30448 2017-03-06 Martin Liska <mliska@suse.cz>
30451 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30452 when having a SSA NAME w/o VAR_DECL assigned to it.
30454 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30456 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30457 msa_dpsub_<su>_d): Fix MODE for vec_select.
30459 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30461 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30463 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30465 2017-03-06 Richard Biener <rguenther@suse.de>
30467 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30468 * plugin.c (register_plugin_info): Likewise.
30469 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30471 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30473 * config/i386/sse.md (sse_storehps, sse_storelps,
30474 avx_<castmode><avxsizesuffix>_<castmode>,
30475 avx512f_<castmode><avxsizesuffix>_<castmode>,
30476 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30477 in condition that at least one operand is not a MEM.
30479 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30481 PR middle-end/79805
30482 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30483 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30485 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30486 gimple_call_nothrow_p flag based on whether original builtin can throw.
30487 If it can, emit following stmts on the fallthrough edge.
30488 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30489 don't create new bb if inserting just debug stmts on the edge, try to
30490 insert them on the fallthru bb or just reset debug stmts.
30492 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30495 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30496 restore recog_data (including the operand rtxes inside it) around
30497 the call to get_insn_template.
30499 2017-03-03 Martin Sebor <msebor@redhat.com>
30501 PR tree-optimization/79699
30502 * context.c (context::~context): Free MPFR caches to avoid
30503 a memory leak on program exit.
30505 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30507 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30508 Use wide_int::ulow () instead of .elt (0).
30510 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30512 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30513 (*pushxf): Limit oF constraint to 32bit targets and add oC
30514 constraint for 64bit targets.
30515 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30516 (*pushdf): Change rmF constraint to rmC.
30518 2017-03-03 Martin Liska <mliska@suse.cz>
30520 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30521 Remove unused variable.
30523 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30526 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30527 is a memory operand, increase num_memory.
30528 (ix86_expand_args_builtin): Likewise.
30530 2017-03-03 Jan Hubicka <jh@suse.cz>
30533 * ipa-devirt.c (maybe_record_node): Properly handle
30534 __cxa_pure_virtual visibility.
30536 2017-03-03 Martin Liska <mliska@suse.cz>
30538 PR tree-optimization/79803
30539 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30541 (pass_loop_prefetch::execute): Disabled optimization if an
30542 assumption about L1 cache size is not met.
30544 2017-03-03 Martin Liska <mliska@suse.cz>
30546 PR rtl-optimization/79574
30547 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30548 (hash_scan_set): Likewise.
30549 (dump_hash_table): Likewise.
30550 (hoist_code): Likewise.
30552 2017-03-03 Richard Biener <rguenther@suse.de>
30554 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30555 in place of uhigh ().
30556 (fixed_convert_from_real): Likewise.
30558 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30561 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30563 2017-03-03 Richard Biener <rguenther@suse.de>
30565 PR middle-end/79818
30566 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30567 TYPE_OVERFLOW_UNDEFINED check.
30569 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30571 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30573 (vector_ae_<mode>_p): Likewise.
30574 (vector_nez_<mode>_p): Likewise.
30575 (vector_ne_v2di_p): Likewise.
30576 (vector_ae_v2di_p): Likewise.
30577 (vector_ne_<mode>_p): Likewise.
30578 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30580 (vsx_tsqrt<mode>2_fe): Likewise.
30582 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30585 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30587 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30589 PR rtl-optimization/79780
30590 * cprop.c (one_cprop_pass): When second and further conditional trap
30591 in a single basic block is turned into an unconditional trap, turn it
30592 into a deleted note to avoid RTL verification failures.
30594 2017-03-02 Richard Biener <rguenther@suse.de>
30596 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30598 2017-03-02 Richard Biener <rguenther@suse.de>
30600 PR tree-optimization/79345
30602 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30603 param and abort the walk, returning -1 if it is hit.
30604 (walk_aliased_vdefs): Take a limit param and pass it on.
30605 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30606 defaulting to 0 and return a signed int.
30607 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30608 (check_defs): New helper.
30609 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30610 about uninitialized memory.
30611 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30612 bogus uninitialized warning.
30613 (fixed_convert_from_real): Likewise.
30615 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30617 PR tree-optimization/66768
30618 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30619 iv_use if base object can't be determined.
30621 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30623 PR tree-optimization/79345
30624 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30625 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30626 (get_pattern_stats): Initialize it.
30627 * genemit.c (gen_expand): Verify match_scratch numbers come after
30628 match_operand/match_dup numbers.
30629 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30630 match_scratch numbers.
30631 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30633 * config/s390/s390.md (trunctdsd2): Likewise.
30635 2017-03-02 Richard Biener <rguenther@suse.de>
30637 * wide-int.h (wide_int_storage::operator=): Implement in terms
30640 2017-03-02 Richard Biener <rguenther@suse.de>
30642 PR tree-optimization/79777
30643 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30644 the to insert expression to sth existing.
30646 2017-03-01 Martin Sebor <msebor@redhat.com>
30648 PR middle-end/79692
30649 * gimple-ssa-sprintf.c
30650 (directive::known_width_and_precision): New function.
30651 (format_integer): Use it.
30652 (get_mpfr_format_length): Consider the full range of precision
30653 when computing %g output with the # flag. Set the likely byte
30654 count to 3 rather than 1 when precision is indeterminate.
30655 (format_floating): Correct the lower bound of precision.
30657 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30659 * doc/invoke.texi: Document default code model for 64-bit Linux.
30661 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30664 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30665 udiv rather than div since input pattern is unsigned.
30667 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30669 * config/i386/i386.c (print_reg): Warn for values of
30670 unsupported size in integer register.
30672 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30675 * config/rs6000/predicates.md (current_file_function_operand): Do
30676 not allow self calls to be local if the function is replaceable.
30678 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30681 * config/rs6000/altivec.h (vec_ctz and others): Change the
30682 preprocessor macro that controls conditional compilation from
30683 _ARCH_PWR9 to __POWER9_VECTOR__.
30684 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30685 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30686 control (instead of _ARCH_PWR9 control) so that template
30687 definition uses power9-specific function.
30688 (vec_any_eq): Likewise.
30689 (vec_all_ne): Change macro definition to use a power9-specific
30690 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30691 _ARCH_PWR9 control).
30692 (vec_any_eq) Likewise.
30693 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30694 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30695 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30696 support for xvcmpnedp instruction.
30697 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30698 macro expansion so that Power9 implementation of vec_all_ne does
30699 not use the AltiVec predicate framework.
30700 (VCMPNEH_P): Likewise.
30701 (VCMPNEW_P): Likewise.
30702 (VCMPNED_P): Likewise.
30703 (VCMPNEFP_P): Likewise.
30704 (VCMPNEDP_P): Likewise.
30705 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30706 implementation of vec_any_eq to not use AltiVec predicate
30708 (VCMPAEH_P): Likewise.
30709 (VCMPAEW_P): Likewise.
30710 (VCMPAED_P): Likewise.
30711 (VCMPAEFP_P): Likewise.
30712 (VCMPAEDP_P): Likewise.
30713 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30714 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30715 not use the AltiVec predicate framework.
30716 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30717 of vec_any_eq to not use AltiVec predicate framework.
30718 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30719 support for predefined __POWER9_VECTOR__ macro to indicate that
30720 Power9 instruction selection is enabled.
30721 (altivec_overloaded_builtins): Remove extraneous
30722 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30723 function argument types RS6000_BTI_bool_V16QI and
30724 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30725 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30726 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30727 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30728 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30729 Power9 for implementations of vec_cmpne. Change the signature for
30730 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30731 (representing vec_all_ne) to remove the previously described first
30732 argument of type RS6000_BTI_INTSI, as this was an artifact of
30733 reliance on the AltiVec predicate framework, which is no longer
30734 used in the implementation of these functions. Add
30735 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30736 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30737 since, unlike the AltiVec predicate framework implementation, we
30738 do not share function descriptors between vec_alle and vec_anyeq.
30739 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30740 set of modes that receive special treatment even when
30741 TARGET_P9_VECTOR is true. The special treatment emits code that
30742 does not depend on Power9 instructions.
30743 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30744 define_expand to not rely on AltiVec predicate framework.
30745 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30747 (vector_ne_v2di_p): Change this define_expand to not rely on
30748 AltiVec predicate framework.
30749 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30751 (vector_ne_<mode>_p): Change this define_expand to not rely on
30752 AltiVec predicate framework.
30753 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30755 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30756 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30757 define_insn pattern.
30758 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30759 define_insn pattern because the xvcmpne<VSs>. instruction is not
30761 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30762 instruction is not supported.
30764 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30766 * config/nvptx/nvptx.c: Include intl.h.
30768 2017-03-01 Martin Jambor <mjambor@suse.cz>
30771 * ipa-prop.h (ipa_bits): Removed field known.
30772 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30773 to pointers. Adjusted their comments to warn about their sharing.
30774 (ipcp_transformation_summary): Change bits to a vector of pointers.
30775 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30776 (ipa_get_ipa_bits_for_value): Declare.
30777 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30778 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30779 (ipa_bits_hash_table): Likewise.
30780 (ipa_vr_ggc_hash_traits): Likewise.
30781 (ipa_vr_hash_table): Likewise.
30782 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30783 being pointers and vr_known being removed.
30784 (ipa_set_jf_unknown): Likewise.
30785 (ipa_get_ipa_bits_for_value): New function.
30786 (ipa_set_jfunc_bits): Likewise.
30787 (ipa_get_value_range): New overloaded functions.
30788 (ipa_set_jfunc_vr): Likewise.
30789 (ipa_compute_jump_functions_for_edge): Use the above functions to
30790 construct bits and vr parts of jump functions.
30791 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30792 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30794 (ipcp_grow_transformations_if_necessary): Also allocate
30795 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30797 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30798 them. Fix too long lines.
30799 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30800 vr_known being removed.
30801 (ipa_read_jump_function): Use new setter functions to construct bits
30802 and vr parts of jump functions or set them to NULL.
30803 (write_ipcp_transformation_info): Adjust for bits being pointers.
30804 (read_ipcp_transformation_info): Likewise.
30805 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30807 Include gt-ipa-prop.h.
30808 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30810 (ipcp_store_bits_results): Likewise.
30811 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30812 Do not write to existing jump functions but use a temporary instead.
30814 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30817 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30818 attempt to use its first operand as BIT_FIELD_REF base.
30820 2017-03-01 Richard Biener <rguenther@suse.de>
30822 PR middle-end/79721
30823 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30824 interpolating formula in wrapping arithmetic.
30825 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30827 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30829 PR tree-optimization/79734
30830 * tree-vect-generic.c (expand_vector_condition): Optimize
30831 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30832 Handle VEC_COND_EXPR where comparison has different inner width from
30833 type's inner width.
30835 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30837 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30838 markup, and similar issues. Remove @opindex entries for things
30839 that aren't options. Add missing -mmpy-option entries.
30841 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30843 PR tree-optimization/79737
30844 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30845 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30846 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30847 instead of byte_size. Formatting fix.
30848 (shift_bytes_in_array_right): Formatting fix.
30850 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30853 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30854 condition on optimize for the leaf function test.
30856 2017-02-28 Martin Liska <mliska@suse.cz>
30859 * read-rtl-function.c (function_reader::handle_unknown_directive):
30860 Bail out when one uses -flto.
30862 2017-02-28 Martin Liska <mliska@suse.cz>
30864 * common.opt: Replace space with tabular for options of <number>
30866 * config/i386/i386.opt: Show <number> value for
30867 -mlarge-data-threshold.
30868 * opts.c (print_filtered_help): Do not display number in hexadecimal
30871 2017-02-28 Martin Liska <mliska@suse.cz>
30873 * common.opt: Fix --help=option -Q for options which are of
30876 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30878 * config/i386/i386.c (print_reg): Error out for values
30879 of 8-bit size in invalid integer register.
30881 2017-02-28 Martin Sebor <msebor@redhat.com>
30883 PR tree-optimization/79691
30884 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30886 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30889 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30890 gcc_unreachable with output_operand_lossage.
30892 2017-02-28 Richard Biener <rguenther@suse.de>
30894 PR tree-optimization/79740
30895 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30897 (visit_nary_op): Insert the nary into the hashtable if we
30898 pattern-matched sth.
30899 * tree-ssa-pre.c (eliminate_insert): Robustify.
30901 2017-02-28 Richard Biener <rguenther@suse.de>
30903 PR middle-end/79731
30904 * fold-const.c (decode_field_reference): Reject out-of-bound
30907 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30909 * config/i386/i386.c: Include intl.h.
30910 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30911 instead of just cond ? "..." : "...".
30912 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30913 * coverage.c (read_counts_file): Likewise.
30914 * omp-offload.c: Include intl.h.
30915 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30916 of just cond ? "..." : "...".
30917 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30918 of just cond ? "..." : "...".
30920 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30923 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30925 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30926 'tune for' CPU name.
30927 * config/arm/arm-cpu-data.h: Regenerated.
30929 2017-02-28 Richard Biener <rguenther@suse.de>
30931 PR tree-optimization/79732
30932 * tree-inline.c (expand_call_inline): Do not shadow var.
30934 2017-02-28 Richard Biener <rguenther@suse.de>
30936 PR tree-optimization/79723
30937 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30938 address-space properly.
30940 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30942 * doc/optinfo.texi (Optimization groups): Fix option used for
30944 * doc/invoke.texi (-fopt-info): Document "omp".
30945 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30946 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30947 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30948 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30949 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30951 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30953 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30954 OPTGROUP_OMP with "omp".
30956 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30959 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30960 for arithmetic shift of unsigned V2DI.
30962 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30964 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30965 arc/linux.h headers.
30966 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30967 (LINK_SPEC): Likewise.
30968 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30969 (EXTRA_SPECS): Likewise.
30970 (STARTFILE_SPEC): Likewise.
30971 (ENDFILE_SPEC): Likewise.
30972 (LIB_SPEC): Likewise.
30973 (TARGET_SDATA_DEFAULT): Likewise.
30974 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30975 (MULTILIB_DEFAULTS): Likewise.
30976 (DWARF2_UNWIND_INFO): Likewise.
30977 * config/arc/big.h: New file.
30978 * config/arc/elf.h: Likewise.
30979 * config/arc/linux.h: Likewise.
30980 * config/arc/t-uClibc: Remove.
30982 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30984 PR tree-optimization/77536
30985 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30986 (tree_transform_and_unroll_loop): Use above function to compute the
30987 estimated niter of unrolled loop and use it when scaling profile.
30988 Also use count info rather than frequency if it's non-zero.
30989 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30990 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30991 (vect_transform_loop): Call above function.
30993 2017-02-27 Richard Biener <rguenther@suse.de>
30995 PR tree-optimization/45397
30996 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30997 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30998 (visit_nary_op): Add pattern matching for CSEing sign-changed
30999 or truncated operations with wider ones.
31001 2017-02-27 Richard Biener <rguenther@suse.de>
31003 PR tree-optimization/79690
31004 * tree-vect-stmts.c (vectorizable_store): Use vector type
31005 built from the DR with address-space.
31007 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
31009 * doc/invoke.texi (Optimize Options): Refine the description
31010 of asan-use-after-return.
31012 2017-02-25 Alan Modra <amodra@gmail.com>
31014 PR rtl-optimization/79584
31015 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
31016 base, not ad->base_term, the reg within base. Remove assertion
31017 that ad->base == ad->base_term. Replace gen_int_mode using
31018 bogus mode with const0_rtx.
31020 2017-02-25 Jakub Jelinek <jakub@redhat.com>
31022 PR middle-end/79396
31023 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
31024 FMA_EXPR like tcc_binary or tcc_unary.
31026 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
31029 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
31031 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
31032 (output_loc_operands): Handle DW_OP_call_ref and
31033 DW_OP_GNU_variable_value.
31034 (struct variable_value_struct): New type.
31035 (struct variable_value_hasher): Likewise.
31036 (variable_value_hash): New variable.
31037 (string_types): Remove.
31038 (copy_loc_descr): New function.
31039 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
31040 (prepend_loc_descr_to_each): New function.
31041 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
31042 instead of add_loc_descr_to_each if the first argument is single
31043 location list and the second has multiple.
31044 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
31045 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
31046 when looking for variable value which doesn't have other location info.
31047 (loc_list_from_tree): Formatting fix.
31048 (gen_array_type_die): Simplify DW_AT_string_length handling.
31049 (adjust_string_types): Remove.
31050 (gen_subprogram_die): Don't call adjust_string_types nor test/set
31051 string_types. Call resolve_variable_values.
31052 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
31053 (resolve_addr_in_expr): Likewise. Add A argument.
31054 (copy_deref_exprloc): Remove deref argument. Adjust for the
31055 original expression being DW_OP_GNU_variable_value with optionally
31056 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
31057 optionally after it.
31058 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
31059 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
31060 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
31061 (variable_value_hasher::hash, variable_value_hasher::equal): New
31063 (resolve_variable_value_in_expr, resolve_variable_value,
31064 resolve_variable_values, note_variable_value_in_expr,
31065 note_variable_value): New functions.
31066 (dwarf2out_early_finish): Call note_variable_value on all toplevel
31069 2017-02-24 Jakub Jelinek <jakub@redhat.com>
31072 * opts.h (handle_generated_option): Add GENERATED_P argument.
31073 * opts-common.c (handle_option): Adjust function comment.
31074 (handle_generated_option): Add GENERATED_P argument, pass it to
31076 (control_warning_option): Pass false to handle_generated_option
31078 * opts.c (maybe_default_option): Pass true to handle_generated_option
31080 * optc-gen.awk: Likewise.
31082 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31084 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
31085 a REG, look at the REG it is a SUBREG of.
31086 (splitter for cmpeqsi_t): Ditto.
31088 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31090 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
31091 the special USEs with the pattern of the insn, not the insn itself.
31093 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
31096 * doc/invoke.texi: Document -mload-store-pairs.
31098 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31099 Sandra Loosemore <sandra@codesourcery.com>
31101 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
31102 argument isn't a CONST_INT.
31103 (nios2_alternate_compare_const): Assert op is a CONST_INT.
31104 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
31105 (nios2_validate_compare): Bypass alternate compare logic if *op2
31106 is not a CONST_INT.
31107 (ldstwm_operation_p): Return false if first_base is not a REG or
31108 if first_offset is not a CONST_INT.
31110 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31112 * config/cris/cris.md: Use correct operand in a define_peephole2.
31114 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31116 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
31118 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
31120 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
31121 this_insn if it is an INSN or JUMP_INSN.
31122 (force_offsettable): Look at base, not at addr.
31123 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
31124 on things that aren't necessarily CONST_INTs.
31126 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
31128 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
31129 -mfpmath=sse is the default also for x86-32 targets with SSE2
31130 instruction set when @option{-ffast-math} is enabled
31132 2017-02-24 Jeff Law <law@redhat.com>
31134 PR rtl-optimizatoin/79286
31135 * ira.c (update_equiv_regs): Drop may_trap_p exception to
31138 2017-02-24 Richard Biener <rguenther@suse.de>
31140 PR tree-optimization/79389
31141 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
31144 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
31146 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
31147 function comment to reflect reality.
31148 (loop_exits_before_overflow): Fix typo in function description.
31150 2017-02-24 Richard Biener <rguenther@suse.de>
31152 PR tree-optimization/79389
31153 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
31154 properly that a threading opportunity exists. Detect conditional
31155 copy/constant propagation opportunities.
31157 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
31159 * config/visium/visium.md (type): Add trap.
31160 (b): New mode attribute.
31161 (*btst): Rename into...
31162 (*btst<mode>): ...this and adjust.
31163 (*cbranchsi4_btst_insn): Rename into...
31164 (*cbranch<mode>4_btst_insn): ...this and adjust.
31165 (trap): New define_insn.
31167 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31169 PR tree-optimization/79389
31170 * ifcvt.c (struct noce_if_info): Add rev_cond field.
31171 (noce_reversed_cond_code): New function.
31172 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
31173 reversed_comparison_code. Formatting fix.
31174 (noce_try_store_flag): Test rev_cond != NULL in addition to
31175 reversed_comparison_code.
31176 (noce_try_store_flag_constants): Likewise.
31177 (noce_try_store_flag_mask): Likewise.
31178 (noce_try_addcc): Use rev_cond if non-NULL instead of
31179 reversed_comparison_code.
31180 (noce_try_cmove_arith): Likewise. Formatting fixes.
31181 (noce_try_minmax, noce_try_abs): Clear rev_cond.
31182 (noce_find_if_block): Initialize rev_cond.
31183 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
31184 instead of false as last argument never attempt to reverse it
31187 2017-02-23 Bin Cheng <bin.cheng@arm.com>
31189 PR tree-optimization/79663
31190 * tree-predcom.c (combine_chains): Process refs in reverse order
31191 only for ZERO length chains, and add explaining comment.
31193 2017-02-23 Jeff Law <law@redhat.com>
31195 PR tree-optimization/79578
31196 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
31197 in call to operand_equal_p.
31199 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
31202 * config/i386/cpuid.h: Fix another undefined behavior.
31204 2017-02-23 Richard Biener <rguenther@suse.de>
31206 PR tree-optimization/79683
31207 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
31208 vector types for data-refs.
31210 2017-02-23 Martin Liska <mliska@suse.cz>
31212 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
31214 2017-02-23 Jakub Jelinek <jakub@redhat.com>
31216 PR middle-end/79665
31217 * internal-fn.c (get_range_pos_neg): Moved to ...
31218 * tree.c (get_range_pos_neg): ... here. No longer static.
31219 * tree.h (get_range_pos_neg): New prototype.
31220 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
31221 are known to be in between 0 and signed maximum inclusive, try to
31222 expand both unsigned and signed divmod and use the cheaper one from
31225 2017-02-22 Jeff Law <law@redhat.com>
31227 PR tree-optimization/79578
31228 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
31229 to compare base operands.
31231 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
31234 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
31235 gpc_reg_operand instead of fpr_reg_operand.
31237 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
31239 * config/mips/mips.c (mips_return_in_memory): Force FP
31240 vector types to be returned in memory for o32 ABI.
31242 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31244 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
31245 instead of DW_TAG_member for static data member declarations and don't
31246 set no_linkage_name for static inline data members.
31247 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
31250 2017-02-22 Martin Liska <mliska@suse.cz>
31252 * doc/invoke.texi: Replace inequality signs with square brackets
31255 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
31258 * lra-constraints.c (simplify_operand_subreg): Handle
31259 WORD_REGISTER_OPERATIONS targets.
31261 2017-02-22 Jakub Jelinek <jakub@redhat.com>
31264 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
31265 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
31266 elimination by swapping fld*.
31268 2017-02-22 Richard Biener <rguenther@suse.de>
31270 PR tree-optimization/79673
31271 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
31272 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
31273 irrelevant address-space qualifiers and avoiding a
31274 ADDR_SPACE_CONVERT_EXPR from fold_convert.
31276 2017-02-22 Richard Biener <rguenther@suse.de>
31278 PR tree-optimization/79666
31279 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
31280 to not symbolically negate if that may introduce undefined
31283 2017-02-22 Martin Liska <mliska@suse.cz>
31286 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
31287 * data-streamer-out.c (streamer_write_gcov_count_stream):
31289 * value-prof.c (stream_out_histogram_value): Make assert more
31290 precise based on type of counter.
31292 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
31295 * config/i386/i386.md (standard_x87sse_constant_load splitter):
31296 Use nonimmediate_operand instead of memory_operand for operand 1.
31297 (float-extend standard_x87sse_constant_load splitter): Ditto.
31299 2017-02-21 Jeff Law <law@redhat.com>
31301 PR tree-optimization/79621
31302 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
31303 blocks with edges to themselves.
31305 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31308 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
31309 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
31310 Use gimple_call_builtin_p.
31313 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
31314 on temporarily removed DEBUG_INSNs.
31316 PR tree-optimization/79649
31317 * tree-loop-distribution.c (classify_partition): Give up on
31318 non-generic address space loads/stores.
31320 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
31322 * doc/loop.texi (Loop manipulation): Remove nonexistent
31323 tree_ssa_loop_version from the documentation.
31324 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
31326 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31329 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
31330 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
31331 * config/rs6000/rs6000.c: Include except.h.
31332 (rs6000_expand_split_stack_prologue): Call
31333 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
31335 2017-02-21 Martin Jambor <mjambor@suse.cz>
31338 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
31339 have been analyzed.
31341 2017-02-21 Martin Jambor <mjambor@suse.cz>
31343 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
31344 for backward compatibility only.
31345 * doc/invoke.texi (Option Summary): Remove all references to
31346 -fipa-cp-alignment.
31348 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
31352 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31354 * lra-constraints.c (curr_insn_transform): Handle
31355 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31357 2017-02-21 Martin Liska <mliska@suse.cz>
31359 * config/i386/i386.opt: Replace -masm-dialect with -masm.
31361 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
31363 PR translation/79638
31364 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
31366 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
31369 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
31370 (arm_function_ok_for_sibcall): Return false for an indirect call by
31371 descriptor if all the argument registers are used.
31372 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
31373 alignment of the function.
31375 2017-02-21 Jakub Jelinek <jakub@redhat.com>
31377 PR tree-optimization/61441
31378 * simplify-rtx.c (simplify_const_unary_operation): For
31379 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
31380 the sNaN unmodified.
31382 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31384 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31385 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31386 instead of SYSTEM_HEADER_DIR.
31388 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
31389 Martin Liška <mliska@suse.cz>
31391 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
31392 Fix typos and grammar, use active voice, and clarify.
31394 2017-02-20 Marek Polacek <polacek@redhat.com>
31396 PR middle-end/79537
31397 * gimplify.c (gimplify_expr): Handle unused *&&L;.
31400 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
31402 2017-02-20 Jakub Jelinek <jakub@redhat.com>
31405 * config/i386/i386.c (ix86_expand_builtin): Handle
31406 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
31407 ix86_builtins_isa[fcode].isa as a requirement of those
31408 flags and any other flag in the bitmask.
31409 (ix86_init_mmx_sse_builtins): Use 0 instead of
31410 ~OPTION_MASK_ISA_64BIT as mask.
31411 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
31412 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
31413 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
31414 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
31416 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31419 * lra-constraints.c (split_reg): Check requested split mode
31420 is supported by the register.
31422 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31424 * lra-constraints.c (simplify_operand_subreg): Remove early
31427 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31430 * lra-constraints.c (curr_insn_transform): Tighten condition
31431 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31433 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31436 * lra-constraints.c (curr_insn_transform): Handle
31437 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31439 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31442 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31444 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31446 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31449 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31452 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31454 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31457 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31459 * final.c (last_columnnum, override_columnnum): New variables.
31460 (final_start_function): Set last_columnnum, pass it to begin_prologue
31461 hook and pass 0 to dwarf2out_begin_prologue.
31462 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31463 to source_line debug hook.
31464 (notice_source_line): Compute last_columnnum and for debug_column_info
31465 return true on column changes.
31466 * debug.h (struct gcc_debug_hooks): Add column argument to
31467 source_line and begin_prologue hooks.
31468 (debug_nothing_int_charstar_int_bool): Remove prototype.
31469 (debug_nothing_int_int_charstar,
31470 debug_nothing_int_int_charstar_int_bool): New prototypes.
31471 (dwarf2out_begin_prologue): Add column argument.
31472 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31473 begin_prologue hooks.
31474 (debug_nothing_int_charstar_int_bool): Remove.
31475 (debug_nothing_int_int_charstar,
31476 debug_nothing_int_int_charstar_int_bool): New functions.
31477 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31478 through to dwarf2out_source_line.
31479 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31480 (dwarf2out_source_line): Add column argument, emit it if requested.
31481 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31483 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31484 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31485 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31486 through to dwarf2out_begin_prologue.
31487 (vmsdbgout_source_line): Add column argument, pass it through to
31488 dwarf2out_source_line.
31489 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31490 dbxout_source_line caller.
31491 (dbxout_source_line): Add column argument.
31493 * common.opt (gno-column-info, gcolumn-info): New options.
31494 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31495 (check_die): Also test for multiple DW_AT_decl_column attributes.
31496 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31497 DW_AT_decl_column if requested.
31498 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31500 (gen_variable_die): Likewise.
31501 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31502 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31505 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31506 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31507 (ix86_handle_option): Handle OPT_m3dnowa.
31508 * doc/invoke.texi (-m3dnowa): Document.
31509 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31510 -m3dnowa instead of -m3dnow -march=athlon.
31513 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31514 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31516 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31519 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31520 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31521 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31522 generator for vsx_xxpermdi_<mode>_be.
31523 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31524 force big-endian semantics.
31525 (vsx_xxpermdi_<mode>_be): New define_expand with same
31526 implementation as previous version of vsx_xxpermdi_<mode>.
31528 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31530 PR tree-optimization/79327
31531 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31532 variable, its initialization and use.
31534 2017-02-17 Julia Koval <julia.koval@intel.com>
31536 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31537 (OPTION_MASK_ISA_PKU_UNSET): New.
31538 (ix86_handle_option): Handle -mrdpid.
31539 * config/i386/cpuid.h (bit_RDPID): New.
31540 * config/i386/driver-i386.c (host_detect_local_cpu):
31541 Detect RDPID feature.
31542 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31543 * config/i386/i386-c.c (ix86_target_macros_internal):
31545 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31546 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31547 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31548 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31549 * config/i386/i386.md (define_insn "rdpid"): New.
31550 * config/i386/i386.opt Add -mrdpid.
31551 * config/i386/immintrin.h (_rdpid_u32): New.
31553 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31555 PR rtl-optimization/79541
31556 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31557 instead of transforming it into USE.
31559 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31561 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31562 If HONOR_SNANS (SFmode) force the input to a register.
31563 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31564 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31565 an frsp or similar insn.
31567 2017-02-17 Martin Liska <mliska@suse.cz>
31569 PR rtl-optimization/79577
31570 * params.def (selsched-max-sched-times): Increase minimum to 1.
31572 2017-02-17 Martin Liska <mliska@suse.cz>
31574 PR rtl-optimization/79574
31575 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31577 2017-02-17 Martin Liska <mliska@suse.cz>
31579 PR tree-optimization/79529
31580 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31581 ssa_defined_default_def_p to handle cases which are implicitly
31583 * tree-ssa.c (ssa_defined_default_def_p): New function.
31584 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31585 which are implicitly defined.
31586 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31588 2017-02-17 Richard Biener <rguenther@suse.de>
31590 PR middle-end/79576
31591 * params.def (max-ssa-name-query-depth): Limit to 10.
31593 2017-02-17 Richard Biener <rguenther@suse.de>
31595 PR tree-optimization/79552
31596 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31599 2017-02-17 Richard Biener <rguenther@suse.de>
31602 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31604 2017-02-17 Marek Polacek <polacek@redhat.com>
31606 PR middle-end/79536
31607 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31608 (fold_negate_expr): New wrapper.
31610 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31612 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31613 Correct terminology and de-emphasize pre-standard behavior.
31615 2017-02-16 Alan Modra <amodra@gmail.com>
31617 PR rtl-optimization/79286
31618 * ira.c (def_dominates_uses): New function.
31619 (update_equiv_regs): Don't create an equivalence for insns that
31620 may trap where the register def does not dominate the use.
31622 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31624 PR rtl-optimization/78127
31625 * lra.c (lra): Call lra_eliminate before finish the loop after
31628 2017-02-16 Richard Biener <rguenther@suse.de>
31630 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31632 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31633 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31634 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31635 (isl_val_int_from_wi): New function.
31636 (extract_affine_gmp): Rename to ...
31637 (extract_affine_wi): ... this, take a widest_int.
31638 (extract_affine_int): Just wrap extract_affine_wi.
31639 (add_param_constraints): Use isl_val_int_from_wi.
31640 (add_loop_constraints): Likewise, and extract_affine_wi.
31642 2017-02-15 Jeff Law <law@redhat.com>
31644 PR middle-end/79521
31645 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31646 ira_init_register_move_cost_if_necessary.
31648 2017-02-15 Martin Sebor <msebor@redhat.com>
31650 PR middle-end/32003
31651 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31652 removed in a prior commit.
31654 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31656 PR tree-optimization/79347
31657 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31658 counters during peeling.
31660 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31662 * Makefile.in (site.exp): Remove "set ISLVER".
31664 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31667 * real.c (real_from_integer): Call real_convert even for decimal.
31669 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31672 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31674 2017-02-14 Andrew Pinski <apinski@cavium.com>
31676 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31677 cores and change the partno/implementer to be correct.
31678 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31679 the 'B" as the implementer.
31680 * config/aarch64/aarch64-tune.md: Regenerate.
31682 2017-02-14 Carl Love <cel@us.ibm.com>
31684 * config/rs6000/rs6000.c: Add case statement entry to make the
31685 xvcvuxdsp built-in argument unsigned.
31686 * config/rs6000/vsx.md: Fix the source and return operand types so they
31687 match the instruction definitions from the ISA document. Fix typo
31688 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31691 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31694 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31695 member early_clobber_alts.
31696 * lra-lives.c (reg_early_clobber_p): New.
31697 (process_bb_lives): Use it.
31698 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31699 (debug_operand_data): Initialize early_clobber_alts.
31700 (setup_operand_alternative): Set up early_clobber_alts.
31701 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31702 alternatives to new_insn_reg.
31703 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31705 (lra_update_insn_regno_info): Pass the new arg.
31707 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31709 PR middle-end/79505
31710 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31711 (new_oacc_loop_raw): Don't clear already cleared fields.
31714 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31715 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31716 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31718 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31721 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31723 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31726 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31727 the extra instruction to the right place to store 128-bit constant
31730 2017-02-14 Martin Sebor <msebor@redhat.com>
31732 PR middle-end/79448
31733 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31734 warning for strings of unknown length.
31736 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31738 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31740 2017-02-14 Jeff Law <law@redhat.com>
31743 * ira-costs.c (scan_one_insn): Initialize register move costs
31744 for pseudos seen in USE/CLOBBER insns.
31746 PR tree-optimization/79095
31747 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31748 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31749 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31750 if the operands are known to be not equal, then the resulting range
31752 (intersect_ranges): If the new range is ~[0,0] and the old range is
31753 wide, then prefer ~[0,0].
31754 * tree-vrp.c (overflow_comparison_p_1): New function.
31755 (overflow_comparison_p): New function.
31756 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31757 if NAME is used in an overflow test.
31758 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31759 overflow check that can be expressed as an equality test, then adjust
31760 ops to be that equality test.
31762 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31764 * config/s390/s390-builtin-types.def: Remove flags argument.
31765 * config/s390/s390.c (s390_init_builtins): Likewise.
31767 2017-02-14 Martin Liska <mliska@suse.cz>
31769 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31770 vector. Fix trailing white spaces.
31772 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31774 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31777 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31779 PR rtl-optimization/68664
31780 * config/arm/arm.c (arm_sched_can_speculate_insn):
31781 New function. Declare prototype.
31782 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31784 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31786 PR rtl-optimization/68664
31787 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31789 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31791 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31793 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31794 max skip bytes for function, loop and jump.
31796 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31798 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31799 ABS_EXPR for gimple dump.
31801 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31804 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31806 PR tree-optimization/79408
31807 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31808 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31809 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31810 also if rhs1 is INTEGER_CST.
31812 2017-02-14 Richard Biener <rguenther@suse.de>
31814 PR middle-end/79432
31815 * tree-into-ssa.c (insert_phi_nodes): When the function can
31816 have abnormal edges rewrite SSA names with broken use-def
31817 dominance out of SSA and register them for PHI insertion.
31819 2017-02-13 Martin Sebor <msebor@redhat.com>
31821 PR middle-end/79496
31822 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31823 clearing info.nowrite flag when snprintf size argument is a range.
31825 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31827 * cprop.c (cprop_jump): Add missing space in string literal.
31828 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31829 (get_constraint_for_component_ref): Likewise.
31830 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31831 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31832 * lra-constraints.c (process_alt_operands): Likewise.
31833 * ipa-inline.c (inline_small_functions): Likewise.
31834 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31835 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31836 * trans-mem.c (diagnose_tm_1_op): Likewise.
31837 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31838 (grid_parallel_clauses_gridifiable): Likewise.
31840 * config/nvptx/mkoffload.c (process): Add space in between
31843 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31844 "MOD4_SSE_REGS" and "ALL_REGS".
31846 * spellcheck.c (test_data): Add , in between "foo" and "food".
31848 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31851 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31852 boundary crossing check and subsequent code generation agree.
31854 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31856 * config/aarch64/aarch64.c (has_memory_op): Delete.
31857 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31860 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31862 PR rtl-optimization/79388
31863 PR rtl-optimization/79450
31864 * combine.c (distribute_notes): When removing TEM_INSN for which
31865 corresponding dest has last value recorded, invalidate that last
31868 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31870 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31871 of explicit '@'. Add missing assembly comment marker on branch costs
31874 2017-02-13 Nathan Sidwell <nathan@acm.org>
31876 * gengtype-lex.l (<in_struct>): Add '/'.
31878 2017-02-13 Martin Liska <mliska@suse.cz>
31881 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31883 2017-02-13 Richard Biener <rguenther@suse.de>
31885 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31887 * configure: Re-generate.
31888 * config.in: Likewise.
31889 * graphite-dependences.c: Simplify as if
31890 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31891 * graphite-isl-ast-to-gimple.c: Likewise.
31892 * graphite-optimize-isl.c: Likewise.
31893 * graphite-poly.c: Likewise.
31894 * graphite-sese-to-poly.c: Likewise.
31895 * graphite.h: Likewise.
31896 * toplev.c: Include isl/version.h and use isl_version () for
31897 printing the ISL version.
31898 * doc/install.texi: Update ISL requirement.
31900 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31902 * doc/standards.texi (Standards): Update reference to
31905 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31907 * doc/extend.texi (Named Address Spaces): sourceware.org now
31909 * doc/install.texi (Binaries): Ditto.
31912 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31914 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31915 standard terminology "stringize"/"stringizing" throughout.
31916 * doc/cppinternals.texi: Likewise.
31918 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31920 * doc/extend.texi: Fix some spelling mistakes and typos.
31921 * doc/invoke.texi: Likewise.
31923 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31926 * params.def (inline-min-speedup) Change from 10 to 8.
31928 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31930 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31933 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31936 * ipa-inline-analysis.c (get_minimal_bb): New function.
31937 (record_modified): Use it.
31938 (remap_edge_change_prob): Handle also ancestor functions.
31940 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31942 * doc/contrib.texi (Contributors): Remove broken link into
31943 the Mauve CVS repository.
31945 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31947 PR middle-end/79454
31948 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31949 result computation whenever lhs doesn't have vector mode, not
31950 just when it has BLKmode.
31952 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31954 * doc/makefile.texi (profiledbootstrap): Refer to the
31955 installation instructions only in textual form.
31957 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31960 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31962 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31964 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31965 (Specific): Update mingw-w64 reference.
31967 (Specific): Remove broken link to Renesas RX processor.
31969 2017-02-10 Richard Biener <rguenther@suse.de>
31971 * toplev.c (process_options): Do not mention obsolete graphite
31972 options when printing sorry message about missing graphite support.
31973 Mention -floop-nest-optimize.
31975 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31977 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31978 (vtst_p16): Likewise.
31979 (vtstq_p8): Likewise.
31980 (vtstq_p16): Likewise.
31982 (vtstq_p64): Likewise.
31983 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31984 (vset_lane_p64): New.
31985 (vsetq_lane_p64): New.
31987 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31989 PR tree-optimization/79411
31990 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31991 stmt operands are SSA_NAMEs used in abnormal phis.
31992 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31995 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31998 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
32001 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
32003 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
32005 2017-02-09 Jakub Jelinek <jakub@redhat.com>
32007 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
32011 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
32012 not arbitrary TREE_CONSTANT.
32015 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
32016 "omp declare target link" attribute unless is_global_var.
32017 * omp-offload.c (find_link_var_op): Likewise.
32019 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
32020 Chung-Lin Tang <cltang@codesourcery.com>
32022 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
32024 (gimplify_adjust_omp_clauses): Don't delete TILE.
32025 (gimplify_omp_for): Deal with TILE.
32026 * internal-fn.c (expand_GOACC_TILE): New function.
32027 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
32029 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
32030 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
32032 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
32033 avoid DIV for outermost collapse var.
32034 (expand_oacc_for): Insert tile element loop as needed. Adjust.
32035 Remove out of date comments, fix whitespace.
32036 * omp-general.c (omp_extract_for_data): Deal with tiling.
32037 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
32038 adjust OLF_DIM_BASE value.
32039 (struct omp_for_data): Add tiling field.
32040 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
32041 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
32042 for auto loops. Remove default auto determining, moved to
32043 oacc_loop_fixed_partitions.
32044 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
32045 stmts, add e_mask field.
32046 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
32047 (oacc_thread_numbers): Use oacc_dim_call.
32048 (oacc_xform_tile): New.
32049 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
32050 (finish_oacc_loop): Adjust for ifns vector.
32051 (oacc_loop_discover_walk): Append loop abstraction sites to list,
32052 add case for GOACC_TILE fns.
32053 (oacc_loop_xform_loop): Delete.
32054 (oacc_loop_process): Iterate over call list directly, and add
32055 handling for GOACC_TILE fns.
32056 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
32058 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
32059 vector partitioning to outer loops. Assign 2 partitions to loops
32060 when available. Add TILE handling.
32061 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
32062 (execite_oacc_device_lower): Process GOACC_TILE fns,
32063 ignore unknown specs.
32064 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
32065 * tree.c (omp_clause_num_ops): Adjust TILE ops.
32066 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
32068 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
32070 * configure.ac (ACX_BUGURL): Update.
32071 * configure: Regenerate.
32073 2017-02-09 Richard Biener <rguenther@suse.de>
32075 PR tree-optimization/69823
32076 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
32077 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
32079 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32081 * config/arc/arc-c.def: Add __NPS400__ definition.
32082 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
32083 (TARGET_NPS400): Define.
32085 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
32087 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
32089 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
32090 pointer, arch_info.
32091 (arc_cpu_types): Fill the arch_info field with a pointer into the
32092 arc_arch_types table.
32093 (arc_selected_cpu): Declare.
32094 * config/arc/arc.c (arc_selected_cpu): Make global.
32095 (arc_selected_arch): Delete.
32096 (arc_base_cpu): Delete.
32097 (arc_override_options): Remove references to deleted variables,
32098 update access to arch information.
32099 (ARC_OPT): Update access to arch information.
32100 (ARC_OPTX): Likewise.
32101 * config/arc/arc.h (arc_base_cpu): Remove declaration.
32102 (TARGET_ARC600): Update access to arch information.
32103 (TARGET_ARC601): Likewise.
32104 (TARGET_ARC700): Likewise.
32105 (TARGET_EM): Likewise.
32106 (TARGET_HS): Likewise.
32107 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
32110 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
32113 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
32114 condition/operands for integer GE/LE/GEU/LEU operations.
32116 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
32118 PR translation/79397
32119 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
32122 2017-02-08 Martin Jambor <mjambor@suse.cz>
32125 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
32126 whether allocation happened.
32127 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
32128 nothing was allocated.
32130 2017-02-08 Jakub Jelinek <jakub@redhat.com>
32132 PR tree-optimization/79408
32133 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
32134 constant, but SSA_NAME with a known integer range, use the minimum
32135 of that range instead of op1 to determine if modulo can be replaced
32136 with its first operand.
32138 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32140 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
32142 2017-02-08 Richard Biener <rguenther@suse.de>
32144 PR tree-optimization/71824
32145 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32146 Check all loops contained in the merged region.
32148 2017-02-07 Andrew Pinski <apinski@cavium.com>
32150 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
32152 2017-02-07 Andrew Pinski <apinski@cavium.com>
32154 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
32155 (thunderxt88): Likewise.
32156 (thunderxt81): Disable LSE and change v8.1 to v8.
32157 (thunderxt83): Likewise.
32159 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32160 Richard Biener <rguenther@suse.de>
32162 PR middle-end/79399
32163 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
32164 type from int to size_t.
32165 * ira-costs.c (struct_costs_size): Change type from int to size_t.
32167 2017-02-07 Jakub Jelinek <jakub@redhat.com>
32169 PR rtl-optimization/79386
32170 * cprop.c (bypass_conditional_jumps): Initialize
32171 bypass_last_basic_block already before splitting bbs after
32172 unconditional traps...
32173 (bypass_conditional_jumps): ... rather than here.
32176 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
32177 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
32178 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
32179 fix -masm=intel patterns.
32181 2017-02-07 Richard Biener <rguenther@suse.de>
32183 PR tree-optimization/79256
32184 PR middle-end/79278
32185 * builtins.c (get_object_alignment_2): Use min_align_of_type
32186 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
32187 and ADJUST_FIELD_ALIGN.
32189 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
32191 * doc/tm.texi: Regenerate.
32192 * stor-layout.c (layout_decl): Adjust.
32193 (update_alignment_for_field): Likewise.
32194 (place_field): Likewise.
32195 (min_align_of_type): Likewise.
32196 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
32197 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
32198 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
32199 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
32200 * config/frv/frv.c (frv_adjust_field_align): Likewise.
32201 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
32202 * config/i386/i386.c (x86_field_alignment): Likewise.
32203 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
32204 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
32205 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
32206 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
32207 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
32208 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
32212 2017-01-30 Richard Biener <rguenther@suse.de>
32214 PR tree-optimization/79256
32215 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32216 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32219 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
32221 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
32222 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
32223 builtins to SImode and emit a zero-extend, if necessary.
32225 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32227 * docs/invoke.texi (RISC-V Options): Alphabetize.
32229 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32231 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
32234 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
32236 * config/riscv/riscv.c: New file.
32237 * common/config/riscv/riscv-common.c: Likewise.
32238 * config.gcc: Likewise.
32239 * config/riscv/constraints.md: Likewise.
32240 * config/riscv/elf.h: Likewise.
32241 * config/riscv/generic.md: Likewise.
32242 * config/riscv/linux.h: Likewise.
32243 * config/riscv/multilib-generator: Likewise.
32244 * config/riscv/peephole.md: Likewise.
32245 * config/riscv/pic.md: Likewise.
32246 * config/riscv/predicates.md: Likewise.
32247 * config/riscv/riscv-builtins.c: Likewise.
32248 * config/riscv/riscv-c.c: Likewise.
32249 * config/riscv/riscv-ftypes.def: Likewise.
32250 * config/riscv/riscv-modes.def: Likewise.
32251 * config/riscv/riscv-opts.h: Likewise.
32252 * config/riscv/riscv-protos.h: Likewise.
32253 * config/riscv/riscv.h: Likewise.
32254 * config/riscv/riscv.md: Likewise.
32255 * config/riscv/riscv.opt: Likewise.
32256 * config/riscv/sync.md: Likewise.
32257 * config/riscv/t-elf-multilib: Likewise.
32258 * config/riscv/t-linux: Likewise.
32259 * config/riscv/t-linux-multilib: Likewise.
32260 * config/riscv/t-riscv: Likewise.
32261 * configure.ac: Likewise.
32262 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
32263 Waterman as RISC-V maintainers.
32264 * doc/install.texi: Add RISC-V entries.
32265 * doc/invoke.texi: Add RISC-V options section.
32266 * doc/md.texi: Add RISC-V constraints section.
32267 * configure: Regenerated.
32269 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
32272 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
32273 false values to be constant vectors with all 0 or all 1 bits set.
32274 (vcondu<mode><mode>): Likewise.
32275 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
32277 (fpmask_comparison_operator): Update comment.
32278 (vecint_comparison_operator): New predicate.
32279 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
32280 vector conditionals when the true and false values are constant
32281 vectors with all 0 bits or all 1 bits set.
32283 2017-02-06 Martin Sebor <msebor@redhat.com>
32285 PR tree-optimization/79376
32286 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
32288 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
32290 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
32291 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
32292 to simplify split condition.
32294 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32296 * omp-expand.c (oxpand_omp_atomic_fetch_op,
32297 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
32300 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
32302 PR rtl-optimization/68664
32303 * target.def (can_speculate_insn): New hook.
32304 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
32305 * doc/tm.texi: Regenerate.
32306 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
32307 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
32308 (rs6000_sched_can_speculate_insn): New function.
32310 2017-02-06 Jakub Jelinek <jakub@redhat.com>
32312 PR tree-optimization/79284
32313 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
32314 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
32315 vectorizable_mask_load_store, vectorizable_operation,
32316 vect_is_simple_cond, get_same_sized_vectype): Use it instead
32317 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
32318 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
32319 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
32320 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
32321 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32322 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
32323 is_gimple_assign (stmt). Replace another such test with
32324 is_gimple_assign (stmt).
32326 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32329 * config/avr/avr.c (rtl-iter.h): Include it.
32330 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
32331 (avr_legitimate_combined_insn): ...and implementation.
32333 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
32335 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
32336 * config/s390/s390.c (s390_const_operand_ok)
32337 (s390_canonicalize_comparison, s390_extract_part)
32338 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
32339 (s390_contiguous_bitmask_p, s390_rtx_costs)
32340 (legitimize_pic_address): Likewise.
32341 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
32342 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
32343 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
32344 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
32345 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
32347 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
32349 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
32350 REGNO($0) == REGNO($1).
32352 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32354 * config/s390/linux.h(SIZE_TYPE): Add comment.
32356 2017-02-06 Julian Brown <julian@codesourcery.com>
32357 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32358 Virendra Pathak <virendra.pathak@broadcom.com>
32360 * config/aarch64/aarch64-cores.def: Change the scheduler
32362 * config/aarch64/aarch64.md: Include thunderx2t99.md.
32363 * config/aarch64/thunderx2t99.md: New file.
32365 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32367 * doc/standards.texi (Go Language): Update link to language
32370 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32372 * tree-eh.c (lower_resx): Sanitize profile.
32373 (cleanup_empty_eh_move_lp): Likewise.
32375 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
32378 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
32380 * cfgloopmanip.h (loop_version): Update prototype.
32381 * modulo-sched.c (sms_schedule): Update call of loop_version.
32382 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
32383 * tree-parloops.c (gen_parallel_loop): Likewise.
32384 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
32385 * tree-ssa-loop-split.c (split_loop): Likewise.
32386 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
32387 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
32389 2017-02-05 Martin Liska <mliska@suse.cz>
32392 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
32394 (print_operand_address): Initialize a struct to zero.
32396 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32398 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
32399 garbage collector only in textual form.
32401 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
32403 * doc/extend.texi (x86 specific memory model extensions for
32404 transactional memory): Simplify a phrase.
32406 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
32409 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
32410 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
32411 (atomic_storedi_1): Likewise.
32413 2017-02-04 Jakub Jelinek <jakub@redhat.com>
32415 PR tree-optimization/79338
32416 * tree-parloops.c (gather_scalar_reductions): Don't call
32417 vect_analyze_loop_form for loop->inner before destroying loop's
32420 2017-02-03 Martin Sebor <msebor@redhat.com>
32422 PR tree-optimization/79327
32423 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32424 when precision has resulted in leading zeros.
32425 (format_integer): Adjust the likely counter to assume an unknown
32426 argument that may be zero is non-zero.
32428 2017-02-03 Jason Merrill <jason@redhat.com>
32431 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32432 avoid copying non-taken branch.
32434 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32436 PR tree-optimization/79340
32437 * tree-vect-loop.c (vectorizable_reduction): Release
32438 vec_defs elements after safe_splicing them into other vectors.
32441 PR tree-optimization/79327
32442 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32443 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32445 (format_integer): Use wide_int_to_tree instead of build_int_cst
32446 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32447 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32448 of shortest and longest sequence.
32450 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32452 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32453 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32455 2017-02-03 Walter Lee <walt@tilera.com>
32458 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32459 after initial stackframe link reg save.
32460 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32462 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32465 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32466 wu for stxssp alternative.
32468 2017-02-03 Martin Sebor <msebor@redhat.com>
32470 PR tree-optimization/79352
32471 * gimple-fold.c (get_range_strlen): Add argument.
32472 (get_range_strlen): Change return type to bool.
32473 (get_maxval_strlen): Pass in a dummy argument.
32474 * gimple-fold.h (get_range_strlen): Change return type to bool.
32475 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32476 * tree.h (array_at_struct_end_p): Add argument.
32477 * tree.c (array_at_struct_end_p): Handle it.
32479 2017-02-03 Martin Liska <mliska@suse.cz>
32482 * multiple_target.c (create_dispatcher_calls): Redirect edge
32483 from a caller of a dispatcher.
32484 (expand_target_clones): Make the clones local.
32485 (ipa_target_clone): Do both target clones and resolvers.
32486 (ipa_dispatcher_calls): Remove the pass.
32487 (pass_dispatcher_calls::gate): Likewise.
32488 (make_pass_dispatcher_calls): Likewise.
32489 * passes.def (pass_target_clone): Put as very first IPA early
32492 2017-02-03 Martin Liska <mliska@suse.cz>
32494 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32495 in case of a function with ifunc attribute.
32497 2017-02-03 Martin Liska <mliska@suse.cz>
32499 * cgraph.c (cgraph_node::dump): Dump function version info.
32500 * symtab.c (symtab_node::dump_base): Add missing new line.
32502 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32504 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32505 (ifcombine_ifandif): Use it.
32507 2017-02-03 Martin Liska <mliska@suse.cz>
32509 * doc/invoke.texi: Document default value for
32510 use-after-scope-direct-emission-threshold.
32512 2017-02-03 Martin Liska <mliska@suse.cz>
32514 PR tree-optimization/79339
32515 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32516 (format_floating): Likewise.
32518 2017-02-03 Martin Liska <mliska@suse.cz>
32521 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32523 (ipa_node_params_t::remove): Likewise.
32524 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32525 initialization from removed ipa_node_params_t::insert.
32526 (ipa_node_params::~ipa_node_params): Move from removed
32527 ipa_node_params_t::release.
32528 * symbol-summary.h (symbol_summary::m_released): New member.
32529 Do not release a summary twice. Do not allow to call finalizer
32530 for types of a summary that live in GGC memory.
32532 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32534 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32537 2017-02-02 Martin Sebor <msebor@redhat.com>
32539 PR middle-end/79275
32540 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32541 (format_string): Tighten up the range of output for non-constant
32542 strings and correct the expected range for wide non-constant strings.
32544 2017-02-02 Martin Sebor <msebor@redhat.com>
32546 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32548 PR middle-end/32003
32549 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32551 (-fdump-tree-@var): Add to index and document how to come up
32552 with pass-specific option and dump file names.
32553 (-fdump-passes): Clarify where to look for output.
32555 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32557 PR middle-end/77445
32558 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32559 statistics of the analyzed path; allow threading for speed when
32560 any of BBs along the path are optimized for speed.
32562 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32564 PR middle-end/78468
32565 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32566 settings of the virtual registers.
32569 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32571 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32572 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32575 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32577 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32578 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32580 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32582 * config/s390/s390.md: Add missing comments with the expanded
32584 * config/s390/vector.md: Likewise.
32585 * config/s390/vx-builtins.md: Likewise.
32587 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32590 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32591 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32592 conditions on a single line.
32594 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32596 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32597 __S390_VX__ to __VX__.
32599 2017-02-01 Andrew Pinski <apinski@cavium.com>
32601 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32602 stmt_info to record_stmt_cost.
32603 (vect_get_known_peeling_cost): Pass stmt_info if known to
32605 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32606 cpu_vector_cost field into
32607 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32608 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32609 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32610 splitting of scalar_stmt_cost and vec_stmt_cost.
32611 (thunderx_vector_cost): Likewise.
32612 (cortexa57_vector_cost): LIkewise.
32613 (exynosm1_vector_cost): Likewise.
32614 (xgene1_vector_cost): Likewise.
32615 (thunderx2t99_vector_cost): Improve after the splitting of the two
32617 (aarch64_builtin_vectorization_cost): Update for the splitting of
32618 scalar_stmt_cost and vec_stmt_cost.
32620 2017-02-01 Torvald Riegel <triegel@redhat.com>
32621 Richard Henderson <rth@redhat.com>
32623 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32624 conditional on existance of a fast atomic load.
32625 * optabs-query.c (can_atomic_load_p): New function.
32626 * optabs-query.h (can_atomic_load_p): Declare it.
32627 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32628 no fast atomic load is available for the particular size of access.
32629 (expand_atomic_compare_and_swap): Likewise.
32630 (expand_atomic_load): Likewise.
32631 (expand_atomic_store): Likewise.
32632 (expand_atomic_fetch_op): Likewise.
32633 * testsuite/lib/target-supports.exp
32634 (check_effective_target_sync_int_128): Remove x86 because it provides
32635 no fast atomic load.
32636 (check_effective_target_sync_int_128_runtime): Likewise.
32638 2017-02-01 Richard Biener <rguenther@suse.de>
32640 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32641 (graphite_transform_loops): Provide opt-info for optimized nests.
32642 * tree-parloop.c (parallelize_loops): Provide opt-info for
32643 parallelized loops.
32645 2017-02-01 Richard Biener <rguenther@suse.de>
32647 PR middle-end/79315
32648 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32649 was not set before.
32651 2017-02-01 Richard Biener <rguenther@suse.de>
32653 PR tree-optimization/71824
32654 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32655 Verify the loops are valid in the merged SESE region.
32656 (scop_detection::can_represent_loop_1): Check analyzing the
32657 evolution of the number of iterations in the region succeeds.
32659 2017-01-31 Ian Lance Taylor <iant@golang.org>
32661 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32662 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32664 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32666 PR preprocessor/79210
32667 * input.c (get_substring_ranges_for_loc): Replace line_width
32668 assertion with error-handling.
32670 2017-01-31 Richard Biener <rguenther@suse.de>
32672 PR tree-optimization/77318
32673 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32674 (create_pw_aff_from_tree): Take loop parameter.
32675 (add_condition_to_pbb): Pass loop of the condition to
32676 create_pw_aff_from_tree.
32678 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32680 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32681 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32683 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32687 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32689 (convert_int_to_float128): Likewise.
32690 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32691 (convert_int_to_float128): Likewise.
32692 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32693 (UNSPEC_IEEE128_CONVERT): Likewise.
32694 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32695 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32696 Use local variables for IBM extended format.
32697 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32698 (fix_trunc<mode>si2_fprs): Likewise.
32699 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32700 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32701 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32702 to know that we can now have integers of all sizes in vector
32704 (fix<uns>_<mode>di2_hw): Likewise.
32705 (float<uns>_<mode>si2_hw): Likewise.
32706 (fix_<mode>si2_hw): Likewise.
32707 (fixuns_<mode>si2_hw): Likewise.
32708 (float<uns>_<mode>di2_hw): Likewise.
32709 (float_<mode>di2_hw): Likewise.
32710 (float_<mode>si2_hw): Likewise.
32711 (floatuns_<mode>di2_hw): Likewise.
32712 (floatuns_<mode>si2_hw): Likewise.
32713 (xscvqp<su>wz_<mode>): Delete, no longer used.
32714 (xscvqp<su>dz_<mode>): Likewise.
32715 (xscv<su>dqp_<mode>): Likewise.
32716 (ieee128_mfvsrd_64bit): Likewise.
32717 (ieee128_mfvsrd_32bit): Likewise.
32718 (ieee128_mfvsrwz): Likewise.
32719 (ieee128_mtvsrw): Likewise.
32720 (ieee128_mtvsrd_64bit): Likewise.
32721 (ieee128_mtvsrd_32bit): Likewise.
32723 2017-01-31 Martin Liska <mliska@suse.cz>
32726 * ipa-prop.c (ipa_free_all_node_params): Call release method
32727 instead of ~sumbol_summary to not to trigger double times
32730 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32732 PR tree-optimization/71691
32733 * bitmap.h (class auto_bitmap): New.
32734 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32735 is_maybe_undefined instead of ssa_undefined_value_p.
32737 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32739 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32740 __S390_ARCH_LEVEL__ to __ARCH__.
32742 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32744 PR tree-optimization/79267
32745 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32746 if should_remove_lhs_p is true.
32748 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32751 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32752 (add_alignment_attribute): New.
32753 (base_type_die): Add alignment attribute.
32754 (subrange_type_die): Likewise.
32755 (modified_type_die): Likewise.
32756 (gen_array_type_die): Likewise.
32757 (gen_descr_array_type_die: Likewise.
32758 (gen_enumeration_type_die): Likewise.
32759 (gen_subprogram_die): Likewise.
32760 (gen_variable_die): Likewise.
32761 (gen_field_die): Likewise.
32762 (gen_ptr_to_mbr_type_die): Likewise.
32763 (gen_struct_or_union_type_die): Likewise.
32764 (gen_subroutine_type_die): Likewise.
32765 (gen_typedef_die): Likewise.
32766 (base_type_cmp): Compare alignment attribute.
32768 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32771 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32772 (setb_unsigned) New pattern for setb with CCUNS.
32773 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32774 subfc./subfe sequence to avoid overflow problems. Generate a
32775 shorter sequence with cmpld/setb for power9.
32776 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32777 for generating subfc. instruction.
32778 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32779 now uses this instruction.
32781 2017-01-30 Ian Lance Taylor <iant@google.com>
32784 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32785 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32787 2017-01-30 Martin Sebor <msebor@redhat.com>
32789 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32790 Move constant to the right of a relational operator.
32791 (get_mpfr_format_length, format_character, format_string): Ditto.
32792 (should_warn_p, maybe_warn): Same.
32794 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32796 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32799 * asan.c (get_translation_unit_decl): Remove function.
32800 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32802 2017-01-30 Martin Liska <mliska@suse.cz>
32804 PR gcov-profile/79259
32805 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32806 -fprofile-generate.
32808 2017-01-30 Martin Liska <mliska@suse.cz>
32811 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32812 Initialize variables with NULL value.
32814 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32817 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32819 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32821 2017-01-30 Richard Biener <rguenther@suse.de>
32823 PR tree-optimization/79276
32824 * tree-vrp.c (process_assert_insertions): Properly adjust common
32825 when removing a duplicate.
32827 2017-01-30 Richard Biener <rguenther@suse.de>
32829 PR tree-optimization/79256
32830 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32831 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32833 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32835 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32838 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32839 ("*r<noxa>sbg_<mode>_sll_bitmask")
32840 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32841 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32842 Use contiguous_bitmask_nowrap_operand.
32844 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32847 * config/rs6000/altivec.h (vec_xl): Revise #define.
32848 (vec_xst): Likewise.
32850 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32852 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32854 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32856 PR rtl-optimization/79194
32857 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32858 traps before call to bypass_conditional_jumps.
32860 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32862 PR tree-optimization/71374
32863 * lra-constraints.c (check_conflict_input_operands): New.
32864 (match_reload): Use it.
32866 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32869 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32870 account to calculate conflict_set.
32872 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32874 PR rtl-optimization/78559
32875 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32876 other_insn in combine.
32878 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32880 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32881 uint16_type_node for BT_UINT16.
32883 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32885 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32886 "RTL Tests" to menu.
32887 (GIMPLE Tests): New node.
32888 (RTL Tests): New node.
32890 2017-01-27 Richard Biener <rguenther@suse.de>
32892 PR tree-optimization/79245
32893 * tree-loop-distribution.c (distribute_loop): Apply cost
32894 modeling also to detected patterns.
32896 2017-01-27 Richard Biener <rguenther@suse.de>
32898 PR tree-optimization/71433
32899 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32900 (compare_assert_loc): New function.
32901 (process_assert_insertions): Sort and optimize assert locations
32902 to remove duplicates and push down identical assertions on
32903 edges to their destination block.
32905 2017-01-27 Richard Biener <rguenther@suse.de>
32907 PR tree-optimization/79244
32908 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32909 out SSA names even if abnormal.
32911 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32913 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32914 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32915 instead of MPFR_RNDN.
32917 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32920 * arm.c (arm_option_override): Don't call build_target_option_node
32921 until after doing all option overrides.
32922 (arm_valid_target_attribute_tree): Likewise.
32924 2017-01-27 Martin Liska <mliska@suse.cz>
32926 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32927 for {cd}tors and C++ {cd}tors.
32929 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32931 * config/s390/s390.md ("*setmem_long_and")
32932 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32934 2017-01-26 Martin Sebor <msebor@redhat.com>
32936 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32939 2017-01-26 Martin Sebor <msebor@redhat.com>
32941 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32942 HAVE_DFmode before using XFmode or DFmode.
32943 (parse_directive): Avoid using the z length modifier to avoid
32944 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32946 PR middle-end/78703
32947 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32948 to accept adjustment as an array.
32949 (get_int_range): New function.
32950 (struct directive): Make width and prec arrays.
32951 (directive::set_width, directive::set_precision): Call get_int_range.
32952 (format_integer, format_floating): Handle width and precision ranges.
32953 (format_string, parse_directive): Same.
32955 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32958 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32959 comdat_type_p set, just clone them, but keep the children in the
32963 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32964 which have direct callers with -fvar-tracking-assignments enabled
32966 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32967 inside of type units.
32969 2017-01-26 Martin Sebor <msebor@redhat.com>
32971 PR middle-end/78703
32972 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32974 (struct format_result): Replace number_chars, number_chars_min,
32975 and number_chars_max with a single member of struct result_range.
32977 (format_result::operator+=): Adjust.
32978 (struct fmtresult): Remove bounded. Handle likely and unlikely
32980 (fmtresult::adjust_for_width_or_precision): New function.
32981 (fmtresult:type_max_digits): New function.
32982 (bytes_remaining): Handle likely and unlikely counters.
32983 (min_bytes_remaining): Remove.
32984 (format_percent): Simplify.
32985 (format_integer, format_floating): Set likely and unlikely counters.
32986 (get_string_length, format_character, format_string): Same.
32987 (format_plain, should_warn_p): New function.
32988 (maybe_warn): Call should_warn_p. Update diagnostic messages
32989 and handle those for all directives, including plain strings.
32990 (format_directive): Handle likely and unlikely counters.
32991 Remove unnecessary quoting from diagnostics. Add an informational
32993 (add_bytes): Remove.
32994 (pass_sprintf_length::compute_format_length): Simplify.
32995 (try_substitute_return_value): Handle likely and unlikely counters.
32997 2017-01-26 Carl Love <cel@us.ibm.com>
32999 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
33000 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
33002 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
33005 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
33006 endianess for subregs into account.
33007 * lra-constraints.c (lra_constraints): Do risky transformations
33008 always on the first iteration.
33009 * lra-lives.c (check_pseudos_live_through_calls): Add arg
33010 last_call_used_reg_set.
33011 (process_bb_lives): Define and use last_call_used_reg_set.
33012 * lra.c (lra): Always continue after lra_constraints on the first
33015 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
33017 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
33019 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
33021 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33023 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
33024 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
33025 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
33026 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
33027 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
33028 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
33029 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
33030 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
33031 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
33033 2017-01-26 Marek Polacek <polacek@redhat.com>
33036 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
33037 for the third operand.
33039 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33041 PR middle-end/79236
33042 * omp-low.c (struct omp_context): Add simt_stmt field.
33043 (scan_omp_for): Return omp_context *.
33044 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
33045 context to the _simt_ SIMD stmt.
33046 (lower_omp_for): For combined SIMD with sibling _simt_
33047 SIMD, make sure to use the same decls in _looptemp_
33048 clauses as in the sibling.
33050 2017-01-26 David Sherwood <david.sherwood@arm.com>
33052 PR middle-end/79212
33053 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
33056 2017-01-26 Jakub Jelinek <jakub@redhat.com>
33059 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
33060 emit fld b; fld a; if possible.
33062 * brig-builtins.def: Update copyright years.
33063 * config/arm/arm_acle_builtins.def: Update copyright years.
33065 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
33068 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
33069 constraint instead of o for the stxsd instruction.
33071 2017-01-25 Carl Love <cel@us.ibm.com>
33073 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
33074 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
33076 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
33078 * doc/invoke.texi (C++ Dialect Options): Fix typo.
33080 2017-01-25 Richard Biener <rguenther@suse.de>
33082 PR tree-optimization/69264
33083 * target.def (vector_alignment_reachable): Improve documentation.
33084 * doc/tm.texi: Regenerate.
33085 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
33087 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
33088 earlier changes with respect to TYPE_USER_ALIGN.
33089 (vector_alignment_reachable_p): Likewise. Improve dumping.
33091 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33094 * config/arm/arm.md (xordi3): Force constant operand into a register
33097 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33099 * doc/invoke.texi (-fstore-merging): Correct default optimization
33100 levels at which it is enabled.
33101 (-O): Move -fstore-merging from list to...
33104 2017-01-25 Richard Biener <rguenther@suse.de>
33107 * omp-expand.c: Include debug.h.
33108 (expand_omp_taskreg): Make sure to generate early debug before
33109 outlining anything from a function.
33110 (expand_omp_target): Likewise.
33111 (grid_expand_target_grid_body): Likewise.
33113 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
33116 * asan.c (get_translation_unit_decl): New function.
33117 (asan_add_global): Extract modules file name from globals
33118 TRANSLATION_UNIT_DECL name.
33120 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
33123 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
33124 for long calls with APCS frame and VFP.
33126 2017-01-24 David Malcolm <dmalcolm@redhat.com>
33128 * cfg.c (original_copy_tables_initialized_p): New function.
33129 * cfg.h (original_copy_tables_initialized_p): New decl.
33130 * cfgrtl.c (relink_block_chain): Guard the call to
33131 free_original_copy_tables with a call to
33132 original_copy_tables_initialized_p.
33133 * cgraph.h (symtab_node::native_rtl_p): New decl.
33134 * cgraphunit.c (symtab_node::native_rtl_p): New function.
33135 (symtab_node::needed_p): Don't assert for early assembly output
33136 for __RTL functions.
33137 (cgraph_node::finalize_function): Set "force_output" for __RTL
33139 (cgraph_node::analyze): Bail out early for __RTL functions.
33140 (analyze_functions): Update assertion to support __RTL functions.
33141 (cgraph_node::expand): Bail out early for __RTL functions.
33142 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
33144 * function.h (struct function): Update comment for field
33146 * gimple-expr.c: Include "tree-pass.h".
33147 (gimple_has_body_p): Return false for __RTL functions.
33148 * Makefile.in (OBJS): Add run-rtl-passes.o.
33149 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
33151 (gcc::pass_manager::get_clean_slate): New accessor.
33152 * passes.c: Include "insn-addr.h".
33153 (should_skip_pass_p): Add logging. Update logic for running
33154 "expand" to be compatible with both __GIMPLE and __RTL. Guard
33155 property-provider override so it is only done for gimple passes.
33157 (skip_pass): New function.
33158 (execute_one_pass): Call skip_pass when skipping passes.
33159 * read-md.c (md_reader::read_char): Support filtering
33160 the input to a subset of line numbers.
33161 (md_reader::md_reader): Initialize fields
33162 m_first_line and m_last_line.
33163 (md_reader::read_file_fragment): New function.
33164 * read-md.h (md_reader::read_file_fragment): New decl.
33165 (md_reader::m_first_line): New field.
33166 (md_reader::m_last_line): New field.
33167 * read-rtl-function.c (function_reader::create_function): Only
33168 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
33169 curr_properties. Set DECL_INITIAL to a dummy block.
33170 (read_rtl_function_body_from_file_range): New function.
33171 * read-rtl-function.h (read_rtl_function_body_from_file_range):
33173 * run-rtl-passes.c: New file.
33174 * run-rtl-passes.h: New file.
33176 2017-01-24 Jeff Law <law@redhat.com>
33178 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
33181 2017-01-24 Bin Cheng <bin.cheng@arm.com>
33183 PR tree-optimization/79159
33184 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
33185 (record_nonwrapping_iv): Improve boundary using above function if no
33186 value range information.
33188 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
33189 Martin Jambor <mjambor@suse.cz>
33191 * brig-builtins.def: New file.
33192 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
33193 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
33194 (DEF_HSAIL_SAT_BUILTIN): Likewise.
33195 (DEF_HSAIL_INTR_BUILTIN): Likewise.
33196 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
33197 * builtin-types.def (BT_INT8): New.
33198 (BT_INT16): Likewise.
33199 (BT_UINT8): Likewise.
33200 (BT_UINT16): Likewise.
33201 (BT_FN_ULONG): Likewise.
33202 (BT_FN_UINT_INT): Likewise.
33203 (BT_FN_UINT_ULONG): Likewise.
33204 (BT_FN_UINT_LONG): Likewise.
33205 (BT_FN_UINT_PTR): Likewise.
33206 (BT_FN_ULONG_PTR): Likewise.
33207 (BT_FN_INT8_FLOAT): Likewise.
33208 (BT_FN_INT16_FLOAT): Likewise.
33209 (BT_FN_UINT32_FLOAT): Likewise.
33210 (BT_FN_UINT16_FLOAT): Likewise.
33211 (BT_FN_UINT8_FLOAT): Likewise.
33212 (BT_FN_UINT64_FLOAT): Likewise.
33213 (BT_FN_UINT16_UINT32): Likewise.
33214 (BT_FN_UINT32_UINT16): Likewise.
33215 (BT_FN_UINT16_UINT16_UINT16): Likewise.
33216 (BT_FN_INT_PTR_INT): Likewise.
33217 (BT_FN_UINT_PTR_UINT): Likewise.
33218 (BT_FN_LONG_PTR_LONG): Likewise.
33219 (BT_FN_ULONG_PTR_ULONG): Likewise.
33220 (BT_FN_VOID_UINT64_UINT64): Likewise.
33221 (BT_FN_UINT8_UINT8_UINT8): Likewise.
33222 (BT_FN_INT8_INT8_INT8): Likewise.
33223 (BT_FN_INT16_INT16_INT16): Likewise.
33224 (BT_FN_INT_INT_INT): Likewise.
33225 (BT_FN_UINT_FLOAT_UINT): Likewise.
33226 (BT_FN_FLOAT_UINT_UINT): Likewise.
33227 (BT_FN_ULONG_UINT_UINT): Likewise.
33228 (BT_FN_ULONG_UINT_PTR): Likewise.
33229 (BT_FN_ULONG_ULONG_ULONG): Likewise.
33230 (BT_FN_UINT_UINT_UINT): Likewise.
33231 (BT_FN_VOID_UINT_PTR): Likewise.
33232 (BT_FN_UINT_UINT_PTR: Likewise.
33233 (BT_FN_UINT32_UINT64_PTR): Likewise.
33234 (BT_FN_INT_INT_UINT_UINT): Likewise.
33235 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
33236 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
33237 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
33238 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
33239 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
33240 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
33241 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
33242 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
33243 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
33244 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
33245 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
33246 * doc/frontends.texi: List BRIG FE.
33247 * doc/install.texi (Testing): Add BRIG tesring requirements.
33248 * doc/invoke.texi (Overall Options): Mention BRIG.
33249 * doc/standards.texi (Standards): Doucment BRIG HSA version.
33251 2017-01-24 Richard Biener <rguenther@suse.de>
33253 PR translation/79208
33254 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
33256 2017-01-24 Martin Jambor <mjambor@suse.cz>
33259 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
33260 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
33261 and known_contexts.
33263 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
33265 PR middle-end/79123
33266 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
33267 casts from signed to unsigned really don't have a range.
33269 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
33271 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
33272 GMP_RNDx for compatiblity.
33274 2017-01-24 Martin Liska <mliska@suse.cz>
33277 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
33278 that would prevent us to call alloca with -1 as argument.
33280 2017-01-24 Jakub Jelinek <jakub@redhat.com>
33282 * dwarf2out.c (output_compilation_unit_header, output_file_names):
33283 Avoid -Wformat-security warning.
33285 2017-01-23 Andrew Pinski <apinski@cavium.com>
33287 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
33290 2017-01-23 Martin Sebor <msebor@redhat.com>
33292 PR middle-end/78703
33293 * gimple-ssa-sprintf.c (warn_level): New global.
33294 (format_integer): Use it here and throughout the rest of the file.
33295 Use the same switch to compute sign as base.
33296 (maybe_warn): New function.
33297 (format_directive): Factor out warnings into maybe_warn.
33298 Add debugging output. Use warn_level.
33299 (add_bytes): Use warn_level.
33300 (pass_sprintf_length::compute_format_length): Add debugging output.
33301 (try_substitute_return_value): Same.
33302 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
33304 PR middle-end/78703
33305 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
33306 (struct fmtresult, format_integer, format_floating): Adjust.
33307 (fmtresult::fmtresult): Set max correctly in two argument ctor.
33308 (get_string_length, format_string,format_directive): Same.
33309 (pass_sprintf_length::compute_format_length): Same.
33310 (try_substitute_return_value): Simplify slightly.
33312 PR middle-end/78703
33313 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
33314 (fmtresult::operator+=): Outlined.
33315 (struct fmtresult): Add ctors.
33316 (struct conversion_spec): Rename...
33317 (struct directive): ...to this. Add and remove data members.
33318 (directive::set_width, directive::set_precision): New functions.
33319 (format_percent): Use fmtresult ctor.
33320 (get_width_and_precision): Remove.
33321 (format_integer): Make naming changes. Avoid computing width and
33323 (format_floating): Same. Adjust indentation.
33324 (format_character, format_none): New functions.
33325 (format_string): Moved character handling to format_character.
33326 (format_directive): Remove arguments, change return type.
33327 (parse_directive): New function.
33328 (pass_sprintf_length::compute_format_length): Move directive
33329 parsing to parse_directive.
33331 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33333 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
33334 (assign_assembler_name_if_needed): ... this.
33335 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
33336 (assign_assembler_name_if_needed): ... this.
33337 (free_lang_data_in_cgraph): Adjust callers.
33338 * cgraphunit.c (cgraph_node::analyze): Likewise.
33339 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
33342 2017-01-23 Richard Biener <rguenther@suse.de>
33344 PR tree-optimization/79088
33345 PR tree-optimization/79188
33346 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
33347 resetting loop bounds after last path deletion. Reset loop
33348 bounds of the target loop, make code match the comments.
33349 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
33350 Make sure loops need no fixups.
33352 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33354 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
33355 exponent support with double type for first argument.
33356 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
33357 type returned by __builtin_vec_extract_sig,
33358 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
33359 functions from "vector int" to "vector unsigned int" or from
33360 "vector long long int" to "vector unsigned long long int".
33361 Changed type returned by __builtin_vec_extract_exp,
33362 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
33363 functions from "vector int" to "vector unsigned int" or from
33364 "vector long long int" to "vector unsigned long long int".
33365 Changed return type of __builtin_vec_test_data_class,
33366 __builtin_vec_test_data_class_sp, and
33367 __builtin_vec_test_data_class_dp from "vector int" to
33368 "vector bool int" or from "vector long long int" to "vector bool
33369 long long int" and changed second argument type from "unsigned
33370 int" to "int". Added new overloaded function forms "vector float
33371 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
33372 "vector float __builtin_vec_insert_exp_sp (vector float, vector
33373 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
33374 double, vector unsigned long long int)" and "vector double
33375 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
33376 long int)". Changed return type of
33377 __builtin_scalar_test_data_class and
33378 __builtin_scalar_test_data_class_sp and
33379 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
33380 int" and changed second argument from "unsigned int" to "int".
33381 Changed type returned by __builtin_scalar_test_neg,
33382 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
33383 from "int" to "bool int". Added new overloaded function form
33384 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
33385 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
33386 exponent double-precision with floating point first argument.
33387 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
33388 documentation of scalar_test_data_class, scalar_test_neg,
33389 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
33390 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
33391 vec_test_data_class built-in functions to reflect refinements in
33392 their type signatures.
33394 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
33396 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
33398 (aarch64_elf_asm_destructor): Likewise.
33400 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
33402 PR rtl-optimization/78634
33403 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
33404 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
33405 * ifcvt.c (noce_try_cmove): Add missing cost check.
33407 PR rtl-optimization/71724
33408 * combine.c (if_then_else_cond): Look for situations where it is
33409 beneficial to undo the work of one of the recursive calls.
33411 2017-01-23 Bin Cheng <bin.cheng@arm.com>
33413 PR tree-optimization/70754
33414 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
33415 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
33416 combined stmt before it if not NULL.
33417 (combine_chains): Process refs reversely and compute dominance point
33420 2017-01-23 Martin Liska <mliska@suse.cz>
33422 PR tree-optimization/79196
33423 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33424 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33426 (strlen_optimize_stmt): Call the renamed function.
33428 2017-01-23 Michael Matz <matz@suse.de>
33430 PR tree-optimization/78384
33431 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33433 2017-01-23 Richard Biener <rguenther@suse.de>
33435 PR tree-optimization/79186
33436 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33437 both incoming edges before moving an assert.
33439 2017-01-23 Martin Jambor <mjambor@suse.cz>
33441 * ipa-prop.c (load_from_param_1): Removed.
33442 (load_from_unmodified_param): Bits from load_from_param_1 put back
33444 (load_from_param): Removed.
33445 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33446 with stmt. Reverted back to use of load_from_unmodified_param.
33448 2017-01-23 Martin Jambor <mjambor@suse.cz>
33451 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33452 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33453 field a pointer to garbage collected vector, mark lattices and
33454 ipcp_orig_node with GTY((skip)).
33455 (ipa_get_param_count): Adjust to descriptors being a pointer.
33456 (ipa_get_param): Likewise.
33457 (ipa_get_type): Likewise.
33458 (ipa_get_param_move_cost): Likewise.
33459 (ipa_set_param_used): Likewise.
33460 (ipa_get_controlled_uses): Likewise.
33461 (ipa_set_controlled_uses): Likewise.
33462 (ipa_is_param_used): Likewise.
33463 (ipa_node_params_t): Move into garbage collector. New methods insert
33465 (ipa_node_params_sum): Annotate wth GTY(()).
33466 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33468 (ipa_load_from_parm_agg): Adjust declaration.
33469 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33470 * ipa-profile.c (ipa_profile): Likewise.
33471 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33472 (ipa_populate_param_decls): Make descriptors parameter garbage
33474 (ipa_dump_param): Adjust to descriptors being a pointer.
33475 (ipa_alloc_node_params): Likewise.
33476 (ipa_initialize_node_params): Likewise.
33477 (load_from_param_1): Make descriptors parameter garbage collected.
33478 (load_from_unmodified_param): Likewise.
33479 (load_from_param): Likewise.
33480 (ipa_load_from_parm_agg): Likewise.
33481 (ipa_node_params::~ipa_node_params): Removed.
33482 (ipa_free_all_node_params): Remove call to delete operator.
33483 (ipa_node_params_t::insert): New.
33484 (ipa_node_params_t::remove): Likewise.
33485 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33486 copy known_csts and known_contexts vectors.
33487 (ipa_read_node_info): Adjust to descriptors being a pointer.
33488 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33490 (ipcp_transform_function): Make descriptors variable garbage
33493 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33495 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33496 * config/i386/avx512dqintrin.h: Ditto.
33497 * config/i386/avx512fintrin.h: Ditto.
33498 * config/i386/i386.c: Handle new builtins.
33499 * config/i386/i386-builtin.def: Add new builtins.
33500 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33501 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33503 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33504 Martin Liska <mliska@suse.cz>
33506 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33507 * asan.c (asan_expand_poison_ifn): Support stores and use
33508 appropriate ASAN report function.
33509 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33510 * internal-fn.def (ASAN_POISON_USE): Declare.
33511 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33512 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33513 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33514 ASAN_POISON calls w/o LHS.
33515 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33516 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33517 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33518 * gimplify.c (asan_poison_variables): Add attribute
33519 use_after_scope_memory to variables that really needs to live
33521 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33522 having the attribute.
33524 2017-01-23 Martin Liska <mliska@suse.cz>
33526 * asan.c (create_asan_shadow_var): New function.
33527 (asan_expand_poison_ifn): Likewise.
33528 * asan.h (asan_expand_poison_ifn): New declaration.
33529 * internal-fn.c (expand_ASAN_POISON): Likewise.
33530 * internal-fn.def (ASAN_POISON): New builtin.
33531 * sanopt.c (pass_sanopt::execute): Expand
33532 asan_expand_poison_ifn.
33533 * tree-inline.c (copy_decl_for_dup_finish): Make function
33535 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33536 * tree-ssa.c (is_asan_mark_p): New function.
33537 (execute_update_addresses_taken): Rewrite local variables
33538 (identified just by use-after-scope as addressable) into SSA.
33540 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33542 * doc/install.texi (Specific): opensource.apple.com uses https
33543 now. Remove trailing slash.
33545 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33547 * README.Portability: Remove note on an Irix compatibility issue.
33549 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33551 * gcov.c (INCLUDE_ALGORITHM): Define.
33552 (INCLUDE_VECTOR): Define.
33553 No longer include <vector> and <algorithm> directly.
33555 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33557 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33559 * doc/invoke.texi (Code Gen Options): Ditto.
33561 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33564 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33566 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33568 rtl-optimization/79125
33569 * cprop.c (local_cprop_pass): Handle cases where we make an
33570 unconditional trap.
33572 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33576 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33577 read from, for big endian.
33579 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33581 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33582 register pauth builtins for LP64 only.
33584 2017-01-20 Marek Polacek <polacek@redhat.com>
33587 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33590 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33592 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33594 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33595 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33596 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33598 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33601 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33602 in big-endian mode when they are not a single duplicated value.
33604 2017-01-20 Richard Biener <rguenther@suse.de>
33606 * BASE-VER: Bump to 7.0.1.
33608 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33610 * omp-low.c (omplow_simd_context): New struct. Use it...
33611 (lower_rec_simd_input_clauses): ...here and...
33612 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33613 references to idx, lane, max_vf, is_simt.
33615 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33617 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33620 2017-01-20 Martin Jambor <mjambor@suse.cz>
33622 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33623 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33625 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33626 (GTFILES): Rename hsa.c to hsa-common.c.
33627 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33628 * hsa-dump.c: Likewise.
33629 * hsa-gen.c: Likewise.
33630 * hsa-regalloc.c: Likewise.
33631 * ipa-hsa.c: Likewise.
33632 * omp-expand.c: Likewise.
33633 * omp-low.c: Likewise.
33634 * toplev.c: Likewise.
33636 2017-01-20 Marek Polacek <polacek@redhat.com>
33639 * doc/invoke.texi: Document -Wduplicated-branches.
33640 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33641 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33642 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33643 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33644 return 0 only when not OEP_LEXICOGRAPHIC.
33645 (fold_build_cleanup_point_expr): Use the expression
33646 location when building CLEANUP_POINT_EXPR.
33647 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33648 * tree.c (add_expr): Handle error_mark_node.
33650 2017-01-20 Martin Liska <mliska@suse.cz>
33653 * tree-profile.c (init_ic_make_global_vars): Do not call
33655 (gimple_init_gcov_profiler): Likewise.
33657 2017-01-20 Martin Liska <mliska@suse.cz>
33660 * cgraph.h (maybe_create_reference): Remove argument and
33662 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33664 * ipa-cp.c (create_specialized_node): Likewise.
33665 * symtab.c (symtab_node::maybe_create_reference): Handle
33666 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33668 2017-01-20 Martin Liska <mliska@suse.cz>
33670 * read-rtl-function.c (function_reader::create_function): Use
33671 build_decl instread of build_decl_stat.
33673 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33675 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33676 * config/i386/avx512dqintrin.h: Ditto.
33677 * config/i386/avx512fintrin.h: Ditto.
33678 * config/i386/i386-builtin-types.def: Add new types.
33679 * config/i386/i386.c: Handle new types.
33680 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33681 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33682 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33683 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33684 (__builtin_ia32_kshiftridi): New.
33685 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33687 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33691 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33692 define to rs6000_init_stack_protect_guard.
33693 (rs6000_init_stack_protect_guard): New function.
33695 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33696 Yunqiang Su <yunqiang.su@imgtec.com>
33698 * config.gcc (supported_defaults): Add madd4.
33699 (with_madd4): Add validation.
33700 (all_defaults): Add madd4.
33701 * config/mips/mips.opt (mmadd4): New option.
33702 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33704 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33706 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33707 (ISA_HAS_FUSED_MADD4): Likewise.
33708 * doc/invoke.texi (-mmadd4): Document the new option.
33709 * doc/install.texi (--with-madd4): Document the new option.
33711 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33713 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33714 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33715 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33716 (aarch64_init_pauth_hint_builtins): New.
33717 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33718 (aarch64_expand_builtin): Expand new builtins.
33720 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33722 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33723 * combine-stack-adj.c (no_unhandled_cfa): Handle
33724 REG_CFA_TOGGLE_RA_MANGLE.
33725 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33726 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33727 info for return address signing.
33728 (aarch64_expand_epilogue): Likewise.
33730 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33732 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33733 * config/aarch64/aarch64-protos.h
33734 (aarch64_return_address_signing_enabled): New declaration.
33735 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33737 (aarch64_expand_prologue): Sign return address before it's pushed onto
33739 (aarch64_expand_epilogue): Authenticate return address fetched from
33741 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33742 (aarch64_attributes): New function attributes for "sign-return-address".
33743 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33744 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33745 ("*do_return"): Generate combined instructions according to key index.
33746 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33747 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33749 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33750 * config/aarch64/aarch64.opt (msign-return-address=): New.
33751 * doc/extend.texi (AArch64 Function Attributes): Documents
33752 "sign-return-address=".
33753 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33755 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33757 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33758 overall option summary.
33760 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33762 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33763 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33764 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33765 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33767 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33769 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33770 -mpower9-minmax by default for -mcpu=power9.
33771 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33772 128-bit floating point.
33774 2017-01-20 Alan Modra <amodra@gmail.com>
33776 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33777 optimizing for size.
33779 2017-01-20 Alan Modra <amodra@gmail.com>
33782 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33783 for strcmp and strncmp from corresponding builtin decl.
33785 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33787 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33788 instead of i386/rtems-64.h.
33789 * config/i386/rtems-64.h: Remove.
33791 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33795 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33797 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33799 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33801 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33802 Change int to HOST_WIDE_INT.
33803 * config/aarch64/aarch64-protos.h
33804 (aarch64_simd_gen_const_vector_dup): Likewise.
33805 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33807 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33809 * langhooks-def.h (lhd_type_for_size): New decl.
33810 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33811 * langhooks.c (lhd_type_for_size): New function, taken from
33814 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33816 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33817 define_bypass for CR latency.
33818 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33819 (power9-alu2): Add define_bypass for CR latency.
33821 (power9-mul): Update insn latency.
33822 (power9-mul-compare): Update insn latency, bypass latency and remove
33825 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33827 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33829 * config/aarch64/aarch64.md
33830 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33831 aarch64_nopcrelative_literal_loads.
33832 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33834 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33836 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33837 TARGET_LOONGSON_3A.
33838 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33840 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33843 * config.gcc (supported_defaults): Add lxc1-sxc1.
33844 (with_lxc1_sxc1): Add validation.
33845 (all_defaults): Add lxc1-sxc1.
33846 * config/mips/mips.opt (mlxc1-sxc1): New option.
33847 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33849 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33850 __mips_no_lxc1_sxc1.
33851 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33852 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33853 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33855 2017-01-19 Richard Biener <rguenther@suse.de>
33857 PR tree-optimization/72488
33858 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33859 sure to restore SSA info.
33860 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33862 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33864 PR rtl-optimization/79121
33865 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33866 of the inner type when shifting an extended value.
33868 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33871 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33872 interposable aliases.
33874 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33877 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33878 Use the evmergelohi instruction.
33879 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33880 (mov_sitf_e500_subreg8_2_be): Likewise.
33881 (mov_sitf_e500_subreg12_2_le): Likewise.
33882 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33883 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33884 (mov_sitf_e500_subreg8_2_le): Likewise.
33885 (mov_sitf_e500_subreg12_2_be): Likewise.
33887 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33889 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33890 attribute from vecsimple to vecperm.
33891 (altivec_vbpermq2): Likewise.
33893 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33896 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33898 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33899 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33900 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33901 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33902 case where N arg is SIZE_MAX.
33903 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33904 (cmpstrsi): Add pattern.
33906 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33908 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33909 __builtin_vec_revb builtins.
33910 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33911 built-in functions to support generation of the ISA 3.0 XXBR<x>
33912 vector byte reverse instructions.
33913 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33914 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33915 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33916 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33917 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33918 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33919 (P9V_BUILTIN_VEC_REVB): Likewise.
33920 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33921 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33922 (p9_xxbrq_v16qi): Likewise.
33923 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33924 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33925 (p9_xxbrh_v8hi): Likewise.
33926 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33927 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33928 vec_revb built-in functions.
33930 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33932 PR rtl-optimization/78952
33933 * config/i386/i386.md (any_extract): New code iterator.
33934 (*insvqi_2): Use any_extract for source operand.
33935 (*insvqi_3): Use any_shiftrt for source operand.
33937 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33939 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33941 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33943 2017-01-18 Matthias Klose <doko@ubuntu.com>
33945 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33947 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33949 * config/rs6000/altivec.h (vec_bperm): Change #define.
33950 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33951 (altivec_vbpermq2): New define_insn.
33952 (altivec_vbpermd): Likewise.
33953 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33954 function interface.
33955 (VBPERMD): Likewise.
33956 (VBPERM): New polymorphic function interface.
33957 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33958 Add entries for P9V_BUILTIN_VEC_VBPERM.
33959 * doc/extend.texi: Add interfaces for vec_bperm.
33961 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33963 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33964 first letter of error messages.
33965 (s390_resolve_overloaded_builtin): Likewise.
33966 * config/s390/s390.c (s390_expand_builtin): Likewise.
33967 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33968 (s390_valid_target_attribute_inner_p): Likewise.
33969 * config/s390/s390.md ("tabort"): Likewise.
33971 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33973 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33974 (ISA_AVOID_DIV_HILO): New macro.
33975 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33976 (ISA_HAS_DDIV): Likewise.
33978 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33980 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33982 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33984 * doc/invoke.texi (fabi-version): Spelling fix.
33986 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33989 * doc/invoke.texi (fabi-version): Mention mangling fix for
33992 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33995 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33997 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
34000 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
34001 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
34003 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
34005 (stack_protect_test): Ditto.
34006 * config/rs6000/rs6000.opt (mstack-protector-guard=,
34007 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
34009 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
34010 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
34011 -mstack-protector-guard-offset=.
34012 (RS/6000 and PowerPC Options): Ditto.
34014 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34016 * config/i386/i386.h (MASK_CLASS_P): New define.
34017 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
34018 there are no registers from different register sets also when
34019 mask registers are used. Update function comment.
34020 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
34021 to (*k/*r) and (*k/*km) alternatives.
34023 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
34025 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
34026 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
34027 (EH_RETURN_HANDLER_RTX): New define.
34028 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
34029 Force frame pointer in EH return functions.
34030 (aarch64_expand_epilogue): Add barrier for eh_return.
34031 (aarch64_final_eh_return_addr): Remove.
34032 (aarch64_eh_return_handler_rtx): New function.
34033 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
34035 (aarch64_eh_return_handler_rtx): New prototype.
34037 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34039 * config/rs6000/altivec.h (vec_rlmi): New #define.
34040 (vec_vrlnm): Likewise.
34041 (vec_rlnm): Likewise.
34042 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
34043 (UNSPEC_VRLNM): Likewise.
34044 (VIlong): New mode iterator.
34045 (altivec_vrl<VI_char>mi): New define_insn.
34046 (altivec_vrl<VI_char>nm): Likewise.
34047 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
34049 (VRLDNM): Likewise.
34050 (RLNM): New polymorphic function entry.
34051 (VRLWMI): New monomorphic function entry.
34052 (VRLDMI): Likewise.
34053 (RLMI): New polymorphic function entry.
34054 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
34055 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
34056 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
34059 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34062 * dwarf2out.c (field_byte_offset): Restore the
34063 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
34064 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
34065 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
34068 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34071 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
34073 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34076 * dwarf2out.c (add_data_member_location_attribute): For constant
34077 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
34078 instead of DW_AT_data_member_location, DW_AT_bit_offset and
34079 DW_AT_byte_size attributes.
34081 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
34083 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
34084 after forcing to constant memory when the code model is medium.
34086 2017-01-17 Julia Koval <julia.koval@intel.com>
34089 * config/i386/avx512fintrin.h
34090 (_mm512_i32gather_ps): Change __addr type to void const*.
34091 (_mm512_mask_i32gather_ps): Ditto.
34092 (_mm512_i32gather_pd): Ditto.
34093 (_mm512_mask_i32gather_pd): Ditto.
34094 (_mm512_i64gather_ps): Ditto.
34095 (_mm512_mask_i64gather_ps): Ditto.
34096 (_mm512_i64gather_pd): Ditto.
34097 (_mm512_mask_i64gather_pd): Ditto.
34098 (_mm512_i32gather_epi32): Ditto.
34099 (_mm512_mask_i32gather_epi32): Ditto.
34100 (_mm512_i32gather_epi64): Ditto.
34101 (_mm512_mask_i32gather_epi64): Ditto.
34102 (_mm512_i64gather_epi32): Ditto.
34103 (_mm512_mask_i64gather_epi32): Ditto.
34104 (_mm512_i64gather_epi64): Ditto.
34105 (_mm512_mask_i64gather_epi64): Ditto.
34106 (_mm512_i32scatter_ps): Change __addr type to void*.
34107 (_mm512_mask_i32scatter_ps): Ditto.
34108 (_mm512_i32scatter_pd): Ditto.
34109 (_mm512_mask_i32scatter_pd): Ditto.
34110 (_mm512_i64scatter_ps): Ditto.
34111 (_mm512_mask_i64scatter_ps): Ditto.
34112 (_mm512_i64scatter_pd): Ditto.
34113 (_mm512_mask_i64scatter_pd): Ditto.
34114 (_mm512_i32scatter_epi32): Ditto.
34115 (_mm512_mask_i32scatter_epi32): Ditto.
34116 (_mm512_i32scatter_epi64): Ditto.
34117 (_mm512_mask_i32scatter_epi64): Ditto.
34118 (_mm512_i64scatter_epi32): Ditto.
34119 (_mm512_mask_i64scatter_epi32): Ditto.
34120 (_mm512_i64scatter_epi64): Ditto.
34121 (_mm512_mask_i64scatter_epi64): Ditto.
34122 * config/i386/avx512pfintrin.h
34123 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
34124 (_mm512_mask_prefetch_i32gather_ps): Ditto.
34125 (_mm512_mask_prefetch_i64gather_pd): Ditto.
34126 (_mm512_mask_prefetch_i64gather_ps): Ditto.
34127 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
34128 (_mm512_prefetch_i32scatter_ps): Ditto.
34129 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
34130 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
34131 (_mm512_prefetch_i64scatter_pd): Ditto.
34132 (_mm512_prefetch_i64scatter_ps): Ditto.
34133 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
34134 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
34135 * config/i386/avx512vlintrin.h
34136 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
34137 (_mm_mmask_i32gather_ps): Ditto.
34138 (_mm256_mmask_i32gather_pd): Ditto.
34139 (_mm_mmask_i32gather_pd): Ditto.
34140 (_mm256_mmask_i64gather_ps): Ditto.
34141 (_mm_mmask_i64gather_ps): Ditto.
34142 (_mm256_mmask_i64gather_pd): Ditto.
34143 (_mm_mmask_i64gather_pd): Ditto.
34144 (_mm256_mmask_i32gather_epi32): Ditto.
34145 (_mm_mmask_i32gather_epi32): Ditto.
34146 (_mm256_mmask_i32gather_epi64): Ditto.
34147 (_mm_mmask_i32gather_epi64): Ditto.
34148 (_mm256_mmask_i64gather_epi32): Ditto.
34149 (_mm_mmask_i64gather_epi32): Ditto.
34150 (_mm256_mmask_i64gather_epi64): Ditto.
34151 (_mm_mmask_i64gather_epi64): Ditto.
34152 (_mm256_i32scatter_ps): Change __addr type to void*.
34153 (_mm256_mask_i32scatter_ps): Ditto.
34154 (_mm_i32scatter_ps): Ditto.
34155 (_mm_mask_i32scatter_ps): Ditto.
34156 (_mm256_i32scatter_pd): Ditto.
34157 (_mm256_mask_i32scatter_pd): Ditto.
34158 (_mm_i32scatter_pd): Ditto.
34159 (_mm_mask_i32scatter_pd): Ditto.
34160 (_mm256_i64scatter_ps): Ditto.
34161 (_mm256_mask_i64scatter_ps): Ditto.
34162 (_mm_i64scatter_ps): Ditto.
34163 (_mm_mask_i64scatter_ps): Ditto.
34164 (_mm256_i64scatter_pd): Ditto.
34165 (_mm256_mask_i64scatter_pd): Ditto.
34166 (_mm_i64scatter_pd): Ditto.
34167 (_mm_mask_i64scatter_pd): Ditto.
34168 (_mm256_i32scatter_epi32): Ditto.
34169 (_mm256_mask_i32scatter_epi32): Ditto.
34170 (_mm_i32scatter_epi32): Ditto.
34171 (_mm_mask_i32scatter_epi32): Ditto.
34172 (_mm256_i32scatter_epi64): Ditto.
34173 (_mm256_mask_i32scatter_epi64): Ditto.
34174 (_mm_i32scatter_epi64): Ditto.
34175 (_mm_mask_i32scatter_epi64): Ditto.
34176 (_mm256_i64scatter_epi32): Ditto.
34177 (_mm256_mask_i64scatter_epi32): Ditto.
34178 (_mm_i64scatter_epi32): Ditto.
34179 (_mm_mask_i64scatter_epi32): Ditto.
34180 (_mm256_i64scatter_epi64): Ditto.
34181 (_mm256_mask_i64scatter_epi64): Ditto.
34182 (_mm_i64scatter_epi64): Ditto.
34183 (_mm_mask_i64scatter_epi64): Ditto.
34184 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
34185 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
34186 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
34187 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
34188 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
34189 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
34190 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
34191 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
34192 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
34193 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
34194 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
34195 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
34196 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
34197 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
34198 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
34199 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
34200 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
34201 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
34202 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
34203 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
34204 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
34205 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
34206 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
34207 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
34208 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
34209 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
34210 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
34211 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
34212 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
34213 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
34214 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
34215 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
34216 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
34217 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
34218 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
34219 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
34220 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
34221 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
34222 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
34223 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
34224 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
34225 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
34226 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
34227 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
34228 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
34229 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
34230 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
34231 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
34232 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
34233 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
34234 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
34235 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
34236 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
34237 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
34238 definitions accordingly.
34240 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
34241 Kuan-Lin Chen <kuanlinchentw@gmail.com>
34244 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
34247 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
34250 * ira-conflicts.c (ira_build_conflicts): Update total conflict
34251 hard regs for inner regno.
34253 2017-01-17 Martin Liska <mliska@suse.cz>
34256 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
34257 assumption and add comment.
34259 2017-01-17 Nathan Sidwell <nathan@acm.org>
34261 * ipa-visibility.c (localize_node): New function, broken out of ...
34262 (function_and_variable_visibility): ... here. Call it.
34264 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
34266 PR middle-end/77445
34267 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
34268 correctly set frequency of oudgoing edge.
34269 (duplicate_thread_path): Fix profile updating.
34271 2017-01-17 Jakub Jelinek <jakub@redhat.com>
34274 * configure.ac: Add GCC_BASE_VER.
34275 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
34276 version from BASE-VER file.
34277 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
34278 (gcc.o): Depend on $(BASEVER).
34279 * common.opt (dumpfullversion): New option.
34280 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
34281 * doc/invoke.texi: Document -dumpfullversion.
34282 * doc/install.texi: Document --with-gcc-major-version-only.
34283 * configure: Regenerated.
34285 2017-01-17 Richard Biener <rguenther@suse.de>
34287 PR tree-optimization/71433
34288 * tree-vrp.c (register_new_assert_for): Merge same asserts
34289 on all incoming edges.
34290 (process_assert_insertions_for): Handle insertions at the
34293 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
34295 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
34296 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
34298 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
34301 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
34304 2017-01-17 Alan Modra <amodra@gmail.com>
34307 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
34308 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
34309 symbolic stack limit when pic.
34311 2017-01-16 Martin Sebor <msebor@redhat.com>
34313 PR tree-optimization/78608
34314 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
34316 2017-01-16 Jeff Law <law@redhat.com>
34319 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
34320 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
34321 for several include directories that may be relative to sysroot.
34322 * config/i386/x-mingw32 (gplus_includedir): Define.
34323 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
34324 (native_system_includedir): Likewise.
34325 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
34326 override if TARGET_SYSTEM_ROOT is defined.
34327 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
34329 PR tree-optimization/79090
34330 PR tree-optimization/33562
34331 PR tree-optimization/61912
34332 PR tree-optimization/77485
34333 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
34334 and computed trims into the dump file.
34336 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
34338 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
34340 2017-01-16 Jakub Jelinek <jakub@redhat.com>
34343 * gimplify.c (gimplify_init_constructor): If want_value and
34344 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
34348 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
34349 sequence. Formatting fixes.
34350 (doloop_optimize): Formatting fixes.
34353 * gcc.c (debug_level_greater_than_spec_func): New function.
34354 (static_spec_functions): Add debug-level-gt spec function.
34355 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
34357 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34358 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
34359 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
34360 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
34361 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
34362 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
34364 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
34366 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
34367 QImode fixups to general and mask registers only.
34369 2017-01-16 Carl Love <cel@us.ibm.com>
34371 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
34372 for built-in functions
34373 vector signed char vec_nabs (vector signed char)
34374 vector signed short vec_nabs (vector signed short)
34375 vector signed int vec_nabs (vector signed int)
34376 vector signed long long vec_nabs (vector signed long long)
34377 vector float vec_nabs (vector float)
34378 vector double vec_nabs (vector double)
34379 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
34381 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
34382 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
34383 * doc/extend.texi: Update the documentation file for the new built-in
34386 2017-01-16 Martin Sebor <msebor@redhat.com>
34388 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
34391 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34393 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
34394 UNSPEC_VSX__XXSPLTD to require special splat handling.
34396 2017-01-16 David Malcolm <dmalcolm@redhat.com>
34399 * system.h: Poison strndup.
34401 2017-01-16 Alan Modra <amodra@gmail.com>
34404 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
34407 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
34409 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
34411 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
34413 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
34414 call recog here. Assert that INSN_CODE (insn) is non-negative.
34416 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34419 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34421 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34422 in the currently scheduled RTL fragment.
34424 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34426 PR rtl-optimization/78751
34427 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34430 2017-01-14 Jeff Law <law@redhat.com>
34432 PR tree-optimization/79090
34433 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34434 variable length stores.
34435 (compute_trims): Delete dead assignment to *trim_tail.
34436 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34439 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34441 PR rtl-optimization/78626
34442 PR rtl-optimization/78727
34443 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34444 of a block, and split such blocks after everything else is finished.
34446 2017-01-14 Alan Modra <amodra@gmail.com>
34449 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34450 target legitimate_combined_insn.
34451 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34452 (rs6000_legitimate_combined_insn): New function.
34453 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34455 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34456 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34457 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34459 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34461 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34463 2017-01-13 Jeff Law <law@redhat.com>
34465 PR tree-optimization/33562
34466 PR tree-optimization/61912
34467 PR tree-optimization/77485
34468 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34470 (delete_dead_assignment): Likewise.
34471 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34472 statement to delete_dead_call and delete_dead_assignment.
34474 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34477 * substring-locations.c (format_warning_va): Strengthen case 1 so
34478 that both endpoints of the substring must be within the format
34479 range for just the substring to be printed.
34481 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34483 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34484 * config/i386/i386.c (ix86_target_string): Add missing options
34485 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34486 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34487 flags_other and ix86_target_other to flags2_other. Display unknown
34489 (ix86_valid_target_attribute_inner_p): Add missing options and
34490 reorder options by implied ISAs, as in ix86_target_string.
34492 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34494 * hash-table.h (hash_table::too_empty_p): New function.
34495 (hash_table::expand): Use it.
34496 (hash_table::traverse): Likewise.
34497 (hash_table::empty_slot): Use sizeof (value_type) instead of
34498 sizeof (PTR) to convert bytes to elements. Shrink the table
34499 if the current size is excessive for the current number of
34502 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34504 * ira-costs.c (record_reg_classes): Break from the inner loop
34505 early once alt_fail is known to be true. Update outer loop
34506 handling accordingly.
34508 2017-01-13 Jeff Law <law@redhat.com>
34510 * tree-ssa-dse.c (decrement_count): New function.
34511 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34512 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34513 when we know the partially dead statement is a mem* function.
34515 PR tree-optimization/61912
34516 PR tree-optimization/77485
34517 * tree-ssa-dse.c: Include expr.h.
34518 (maybe_trim_constructor_store): New function.
34519 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34521 PR tree-optimization/33562
34522 PR tree-optimization/61912
34523 PR tree-optimization/77485
34524 * doc/invoke.texi: Document new dse-max-object-size param.
34525 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34526 * tree-ssa-dse.c: Include params.h.
34527 (dse_store_status): New enum.
34528 (initialize_ao_ref_for_dse): New, partially extracted from
34530 (valid_ao_ref_for_dse, normalize_ref): New.
34531 (setup_live_bytes_from_ref, compute_trims): Likewise.
34532 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34533 (maybe_trim_partially_dead_store): Likewise.
34534 (maybe_trim_complex_store): Likewise.
34535 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34536 Track what bytes live from the original store. Return tri-state
34537 for dead, partially dead or live.
34538 (dse_dom_walker): Add constructor, destructor and new private members.
34539 (delete_dead_call, delete_dead_assignment): New extracted from
34541 (dse_optimize_stmt): Make a member of dse_dom_walker.
34542 Use initialize_ao_ref_for_dse.
34544 PR tree-optimization/33562
34545 PR tree-optimization/61912
34546 PR tree-optimization/77485
34547 * sbitmap.h (bitmap_count_bits): Prototype.
34548 (bitmap_clear_range, bitmap_set_range): Likewise.
34549 * sbitmap.c (bitmap_clear_range): New function.
34550 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34552 2017-01-13 Martin Liska <mliska@suse.cz>
34555 * function.c (set_cfun): Add new argument force.
34556 * function.h (set_cfun): Likewise.
34557 * ipa-inline-transform.c (inline_call): Use the function when
34558 strict alising from is dropped for function we inline to.
34560 2017-01-13 Richard Biener <rguenther@suse.de>
34562 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34563 for dumping GIMPLE INTEGER_CSTs.
34565 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34567 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34568 to 201112L since C++17.
34570 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34573 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34574 if -fsanitize=kernel-address is present.
34576 2017-01-13 Richard Biener <rguenther@suse.de>
34578 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34579 as _Literal ( type ) number in case usual suffixes do not
34580 preserve all information.
34582 2017-01-13 Richard Biener <rguenther@suse.de>
34584 PR tree-optimization/77283
34585 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34586 and ssa-iterators.h.
34587 (is_feasible_trace): Implement a cost model based on joiner
34590 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34593 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34594 char or short to __float128/_Float128 directly.
34596 2017-01-12 Martin Sebor <msebor@redhat.com>
34598 to -Wformat-overflow.
34599 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34600 (min_bytes_remaining): Same.
34601 (get_string_length): Same.
34602 (format_string): Same.
34603 (format_directive): Same.
34605 (pass_sprintf_length::handle_gimple_call): Same.
34607 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34609 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34610 info.nowrite calls with no lhs that can't throw. Return bool
34611 whether gsi_remove has been called or not.
34612 (pass_sprintf_length::handle_gimple_call): Return bool whether
34613 try_substitute_return_value called gsi_remove. Formatting fix.
34614 (pass_sprintf_length::execute): Don't use gsi_remove if
34615 handle_gimple_call returned true.
34618 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34619 be removed due to side-effects, don't remove following barrier nor
34620 turn the successor edge into fallthru edge.
34622 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34625 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34626 element-reversing loads and stores as not swappable.
34628 2017-01-12 Nathan Sidwell <nathan@acm.org>
34629 Nicolai Stange <nicstange@gmail.com>
34631 * combine.c (try_combine): Don't ignore result of overlap checking
34632 loop. Combine overlap & asm check into single loop.
34634 2017-01-12 Richard Biener <rguenther@suse.de>
34636 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34637 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34639 2017-01-12 Richard Biener <rguenther@suse.de>
34641 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34642 and TS_TARGET_OPTION directly derive from TS_BASE.
34643 * tree-core.h (tree_optimization_option): Derive from tree_base.
34644 (tree_target_option): Likewise.
34646 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34648 * config/i386/i386.c (memory_address_length): Increase len
34649 only when rip_relative_addr_p returns false.
34651 2017-01-11 Julia Koval <julia.koval@intel.com>
34653 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34654 (OPTION_MASK_ISA_SGX_SET): New.
34655 (ix86_handle_option): Handle OPT_msgx.
34656 * config.gcc: Added sgxintrin.h.
34657 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34658 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34659 * config/i386/i386.c (ix86_target_string): Add -msgx.
34661 (ix86_option_override_internal): Handle new options.
34662 (ix86_valid_target_attribute_inner_p): Add sgx.
34663 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34664 * config/i386/i386.opt: Add msgx.
34665 * config/i386/sgxintrin.h: New file.
34666 * config/i386/x86intrin.h: Add sgxintrin.h.
34668 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34671 * fold-const.c (maybe_nonzero_address): Return 1 for function
34673 (tree_single_nonzero_warnv_p): Don't handle function local objects
34677 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34680 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34683 * opts.c: Include "spellcheck.h"
34684 (struct string_fragment): New struct.
34685 (struct edit_distance_traits<const string_fragment &>): New
34687 (get_closest_sanitizer_option): New function.
34688 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34690 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34692 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34694 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34695 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34696 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34697 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34698 for initial die_offset if dwarf_split_debug_info.
34699 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34700 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34702 (output_skeleton_debug_sections): Formatting fix. Use
34703 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34704 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34706 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34708 * config/arm/cortex-a53.md: Add bypasses for
34709 cortex_a53_r2f_cvt.
34710 (cortex_a53_r2f): Only use for transfers.
34711 (cortex_a53_f2r): Likewise.
34712 (cortex_a53_r2f_cvt): Add reservation for conversions.
34713 (cortex_a53_f2r_cvt): Likewise.
34715 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34717 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34718 to all inlined functions, change static to extern.
34720 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34723 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34725 (arm_assemble_integer): Likewise.
34727 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34729 * config.gcc: Use new awk script to check CPU, FPU and architecture
34730 parameters for --with-... options.
34731 * config/arm/parsecpu.awk: New file
34732 * config/arm/arm-cpus.in: New file.
34733 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34735 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34737 * config/arm/t-arm: Update dependency rules.
34738 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34739 of processing .def files.
34740 * config/arm/genopt.sh: Deleted.
34741 * config/arm/gentune.sh: Deleted.
34742 * config/arm/arm-cores.def: Deleted.
34743 * config/arm/arm-arches.def: Deleted.
34744 * config/arm/arm-fpus.def: Deleted.
34745 * config/arm/arm-tune.md: Regenerated.
34746 * config/arm/arm-tables.opt: Regenerated.
34747 * config/arm/arm-cpu.h: New generated file.
34748 * config/arm/arm-cpu-data.h: New generated file.
34749 * config/arm/arm-cpu-cdata.h: New generated file.
34751 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34754 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34756 (input_varpool_node): Unpack dynamically_initialized bit.
34758 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34760 PR rtl-optimization/79032
34761 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34762 the alignment of the adjusted memory reference against that of MODE,
34763 instead of the alignment of the original memory reference.
34765 2017-01-11 Martin Jambor <mjambor@suse.cz>
34767 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34769 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34770 decorated functions.
34772 2017-01-11 Richard Biener <rguenther@suse.de>
34774 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34775 set range/nonnull info for PHI results. Do not set it on
34776 stmts marked for removal.
34778 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34780 * expr.c (store_field): In the bitfield case, fetch the return value
34781 from the registers before applying a single big-endian adjustment.
34782 Always do a final load for a BLKmode value not larger than a word.
34784 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34787 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34788 that we correctly handle column numbers greater than
34789 LINE_MAP_MAX_COLUMN_NUMBER.
34791 2017-01-10 Martin Sebor <msebor@redhat.com>
34793 PR middle-end/78245
34794 * gimple-ssa-sprintf.c (get_destination_size): Call
34795 {init,fini}object_sizes.
34796 * tree-object-size.c (addr_object_size): Adjust.
34797 (pass_through_call): Adjust.
34798 (pass_object_sizes::execute): Adjust.
34799 * tree-object-size.h (fini_object_sizes): Declare.
34801 2017-01-10 Martin Sebor <msebor@redhat.com>
34803 PR tree-optimization/78775
34804 * builtins.c (get_size_range): Move...
34805 * calls.c: ...to here.
34806 (alloc_max_size): Accept zero argument.
34807 (operand_signed_p): Remove.
34808 (maybe_warn_alloc_args_overflow): Call get_size_range.
34809 * calls.h (get_size_range): Declare.
34811 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34813 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34814 from TI's devices.csv file as of September 2016.
34815 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34817 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34819 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34820 * doc/invoke.texi: Likewise.
34821 * doc/md.texi: Likewise.
34822 * doc/objc.texi: Likewise.
34824 2017-01-10 Joshua Conner <joshconner@google.com>
34826 * config/arm/fuchsia-elf.h: New file.
34827 * config/fuchsia.h: New file.
34828 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34829 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34831 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34833 2016-01-10 Richard Biener <rguenther@suse.de>
34835 PR tree-optimization/79034
34836 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34837 Propagate out degenerate PHIs in the joiner.
34839 2017-01-10 Martin Liska <mliska@suse.cz>
34841 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34842 (sort_congruence_classes_by_decl_uid): Likewise.
34843 (sort_congruence_class_groups_by_decl_uid): Likewise.
34844 (sem_item_optimizer::merge_classes): Sort class, groups in these
34845 classes and members in the groups by DECL_UID of declarations.
34846 This would make merge operations stable.
34848 2017-01-10 Martin Liska <mliska@suse.cz>
34850 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34851 usage of m_classes_vec.
34852 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34853 (sem_item_optimizer::get_group_by_hash): Likewise.
34854 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34855 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34856 (sem_item_optimizer::verify_classes): Likewise.
34857 (sem_item_optimizer::process_cong_reduction): Likewise.
34858 (sem_item_optimizer::dump_cong_classes): Likewise.
34859 (sem_item_optimizer::merge_classes): Likewise.
34860 * ipa-icf.h (congruence_class_hash): Rename from
34861 congruence_class_group_hash. Remove declaration of m_classes_vec.
34863 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34865 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34866 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34867 * config.gcc: Add avx512vpopcntdqintrin.h.
34868 * config/i386/avx512vpopcntdqintrin.h: New.
34869 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34870 * config/i386/i386-builtin-types.def: Add new types.
34871 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34872 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34873 __builtin_ia32_vpopcountq_v8di_mask): New.
34874 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34875 __AVX512VPOPCNTDQ__.
34876 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34877 (PTA_AVX512VPOPCNTDQ): Define.
34878 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34879 TARGET_AVX512VPOPCNTDQ_P): Define.
34880 * config/i386/i386.opt: Add mavx512vpopcntdq.
34881 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34882 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34884 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34886 PR middle-end/77484
34887 * predict.def (PRED_CALL): Set to 67.
34889 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34891 * expr.c (store_field): In the bitfield case, if the value comes from
34892 a function call and is of an aggregate type returned in registers, do
34893 not modify the field mode; extract the value in all cases if the mode
34894 is BLKmode and the size is not larger than a word.
34896 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34899 * config/i386/cpuid.h: Fix undefined behavior.
34901 2017-01-04 Jeff Law <law@redhat.com>
34903 PR tree-optimization/79007
34904 PR tree-optimization/67955
34905 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34906 conservative for pt.null when flag_non_call_exceptions is on.
34908 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34910 PR translation/79019
34911 PR translation/79020
34912 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34913 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34914 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34916 * config/avr/avr.opt (maccumulate-args): Likewise.
34917 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34918 * common.opt (freport-bug): Likewise.
34919 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34920 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34921 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34922 translatable string.
34923 * config/i386/i386.c (function_value_32): Likewise.
34924 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34925 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34927 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34928 * common/config/msp430/msp430-common.c (msp430_handle_option):
34930 * symtab.c (symtab_node::verify_base): Likewise.
34931 * opts.c (set_debug_level): Likewise.
34932 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34933 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34934 missing whitespace to translatable strings.
34935 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34936 * config/sh/superh.h: Likewise.
34937 * config/i386/xopintrin.h: Likewise.
34938 * config/i386/znver1.md: Likewise.
34939 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34940 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34941 * double-int.h (struct double_int): Likewise.
34942 * double-int.c (div_and_round_double): Likewise.
34943 * wide-int.cc: Likewise.
34944 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34945 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34946 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34947 (crossjumps_occurred): ... this.
34948 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34951 PR tree-optimization/78899
34952 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34953 returning bool return struct loop *, NULL for failure and the new
34955 (versionable_outer_loop_p): Don't version outer loop if it has
34956 dont_vectorized bit set.
34957 (tree_if_conversion): When versioning outer loop, ensure
34958 tree_if_conversion is performed also on the inner loop of the
34959 non-vectorizable outer loop copy.
34960 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34961 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34962 prevent vectorization of it.
34963 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34964 the outer loop vectorization of the non-scalar version is attempted
34965 before vectorization of the inner loop in scalar version. If
34966 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34967 vectorization of its inner loop.
34968 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34969 has 2 inner loops, rename also on edges from bb whose single pred
34970 is outer_loop->header. Fix typo in function comment.
34972 2017-01-09 Martin Sebor <msebor@redhat.com>
34975 * asan.c (asan_emit_stack_protection): Increase local buffer size
34976 to avoid snprintf truncation warning.
34978 2017-01-09 Andrew Pinski <apinski@cavium.com>
34980 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34981 to reference thunderx2t99 for the tuning structure
34982 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34984 (thunderx2t99_extra_costs): This.
34985 * config/aarch64/aarch64-tune.md: Regenerate.
34986 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34987 (vulcan_addrcost_table): This.
34988 (vulcan_regmove_cost): Rename to ...
34989 (thunderx2t99_regmove_cost): This.
34990 (vulcan_vector_cost): Rename to ...
34991 (thunderx2t99_vector_cost): this.
34992 (vulcan_branch_cost): Rename to ...
34993 (thunderx2t99_branch_cost): This.
34994 (vulcan_tunings): Rename to ...
34995 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34996 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34998 2017-01-09 Martin Jambor <mjambor@suse.cz>
35002 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
35003 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
35004 (propagate_vr_accross_jump_function): Use the above function for all
35005 value range computations for pass-through jump functions and type
35006 converasion from explicit value range values.
35007 (ipcp_propagate_stage): Do not attempt to deduce types of formal
35008 parameters from TYPE_ARG_TYPES.
35009 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
35010 (ipa_write_node_info): Stream type of the actual argument.
35011 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
35013 2017-01-09 Martin Liska <mliska@suse.cz>
35016 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
35017 (lookup_compiler): Do not show error message with have_E.
35019 2017-01-09 Jakub Jelinek <jakub@redhat.com>
35021 PR tree-optimization/78938
35022 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
35023 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
35024 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
35025 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
35028 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35030 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
35033 2017-01-09 Richard Biener <rguenther@suse.de>
35035 PR tree-optimization/78997
35036 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
35037 name condition properly.
35039 2017-01-09 Richard Biener <rguenther@suse.de>
35042 * dwarf2out.c (is_cxx): New overload with context.
35043 (is_naming_typedef_decl): Use it.
35045 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35047 * invoke.texi (Option Summary): Correct spacing in option lists
35048 and add line breaks to fix over-long lines.
35050 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35052 PR middle-end/17660
35054 * extend.texi (Common Variable Attributes): Add xref to GCC
35055 Internals manual to explain mode attribute keywords.
35057 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
35060 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
35061 and Preprocessor Options.
35062 (Options for Linking): Document -pthread here....
35063 (RS/6000 and PowerPC Options): ...not here.
35064 (Solaris 2 Options): ...or here.
35065 * doc/cppopts.texi: Document -pthread.
35067 2017-01-08 Martin Sebor <msebor@redhat.com>
35069 PR middle-end/77708
35070 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
35071 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
35072 New member functions.
35073 (format_directive): Used them.
35075 (pass_sprintf_length::handle_gimple_call): Same.
35076 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
35077 to avoid truncation for any argument.
35078 (extract_affine_mul): Same.
35079 * tree.c (get_file_function_name): Same.
35081 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35083 PR middle-end/77484
35084 * predict.def (PRED_INDIR_CALL): Set to 86.
35086 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35088 PR preprocessor/54124
35089 * doc/cppopts.texi: Reformat -d subtable to list the full name
35090 of the options. Add cross-reference to the docs for the general
35091 compiler -d options.
35092 * doc/invoke.texi (Developer Options): Add cross-reference to the
35093 preprocessor-specific -d option documentation.
35095 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35097 PR preprocessor/13498
35098 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
35099 redudant material, and reflect new command-line options.
35100 (System Headers): Likewise.
35102 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
35104 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
35105 -isystem, and -idirafter. Copy-edit.
35106 * doc/cppopts.texi: Copy-edit. Remove contradiction about
35107 default for -ftrack-macro-expansion. Delete obsolete and
35108 badly-formatted implementation details about -fdebug-cpp output.
35109 * doc/cppwarnopts.texi: Copy-edit.
35111 2017-01-07 David Malcolm <dmalcolm@redhat.com>
35114 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
35115 that the transition from a max line width >= 1<<10 to narrower
35116 lines works correctly.
35118 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
35120 * doc/options.texi (PerFunction): New.
35121 * opt-functions.awk (switch_flags): Map both Optimization and
35122 PerFunction to CL_OPTIMIZATION.
35123 * opth-gen.awk: Test for PerFunction flag along with
35125 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
35126 it only when the latter is present. Skip those that don't in
35127 the hash function generator.
35128 * common.opt (fvar-tracking): Mark as PerFunction instead of
35130 (fvar-tracking-assignments): Likewise.
35131 (fvar-tracking-assignments-toggle): Likewise.
35132 (fvar-tracking-uninit): Likewise.
35134 2017-01-07 Jakub Jelinek <jakub@redhat.com>
35136 PR translation/79018
35137 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
35140 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
35143 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
35144 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
35145 TARGET_LONG_JUMP_TABLE_OFFSETS.
35146 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
35147 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
35148 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
35149 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
35150 * config/m68k/m68k.md (tablejump expander): Likewise.
35151 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
35152 TARGET_LONG_JUMP_TABLE_OFFSETS.
35153 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
35154 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
35156 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
35157 David Holsgrove <david.holsgrove@xilinx.com>
35159 * common/config/microblaze/microblaze-common.c
35160 (TARGET_EXCEPT_UNWIND_INFO): Remove.
35161 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
35163 * config/microblaze/microblaze.c (microblaze_must_save_register)
35164 (microblaze_expand_epilogue, microblaze_return_addr): Handle
35166 (microblaze_eh_return): New function.
35167 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
35168 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
35169 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
35170 * config/microblaze/microblaze.md (eh_return): New pattern.
35172 2017-01-06 Jakub Jelinek <jakub@redhat.com>
35174 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
35175 GCC_DIAGNOSTIC_STRINGIFY): Define.
35177 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
35179 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35181 * config/arm/arm.md (<mcrr>): New.
35183 * config/arm/arm.c (arm_arch5te): New.
35184 (arm_option_override): Set arm_arch5te.
35185 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
35187 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
35188 (arm_mcrr_qualifiers): ... this. New.
35189 (MRRC_QUALIFIERS): Define to...
35190 (arm_mrrc_qualifiers): ... this. New.
35191 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
35193 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
35194 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
35195 (MRRCI, mrrc, MRRC): New.
35196 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
35197 VUNSPEC_MRRC2): New.
35199 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35201 * config/arm/arm.md (<mcr>): New.
35203 * config/arm/arm.c (arm_coproc_builtin_available): Add
35204 support for mcr, mrc, mcr2 and mrc2.
35205 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
35206 (arm_mcr_qualifiers): ... this. New.
35207 (MRC_QUALIFIERS): Define to ...
35208 (arm_mrc_qualifiers): ... this. New.
35209 (MCR_QUALIFIERS): Define to ...
35210 (arm_mcr_qualifiers): ... this. New.
35211 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
35213 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
35214 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
35215 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
35216 VUNSPEC_MRC2): New.
35218 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35220 * config/arm/arm.md (*ldc): New.
35224 * config/arm/arm.c (arm_coproc_builtin_available): Add
35225 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
35226 (arm_coproc_ldc_stc_legitimate_address): New.
35227 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
35228 'qualifier_const_pointer'.
35229 (LDC_QUALIFIERS): Define to...
35230 (arm_ldc_qualifiers): ... this. New.
35231 (STC_QUALIFIERS): Define to...
35232 (arm_stc_qualifiers): ... this. New.
35233 * config/arm/arm-protos.h
35234 (arm_coproc_ldc_stc_legitimate_address): New.
35235 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
35236 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
35237 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
35238 stc2, stcl, stc2l): New.
35239 * config/arm/constraints.md (Uz): New.
35240 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
35241 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
35242 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
35243 VUNSPEC_STC2L): New.
35245 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35247 * config/arm/arm.md (<cdp>): New.
35248 * config/arm/arm.c (neon_const_bounds): Rename this ...
35249 (arm_const_bounds): ... this.
35250 (arm_coproc_builtin_available): New.
35251 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
35252 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
35253 (CDP_QUALIFIERS): Define to...
35254 (arm_cdp_qualifiers): ... this. New.
35256 (arm_expand_builtin_args): Add case for 6 arguments.
35257 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
35258 (arm_const_bounds): ... this.
35259 (arm_coproc_builtin_available): New.
35260 * config/arm/arm_acle.h (__arm_cdp): New.
35262 * config/arm/arm_acle_builtins.def (cdp): New.
35264 * config/arm/iterators.md (CDPI,CDP,cdp): New.
35265 * config/arm/neon.md: Rename all 'neon_const_bounds' to
35266 'arm_const_bounds'.
35267 * config/arm/types.md (coproc): New.
35268 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
35269 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
35270 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
35271 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
35273 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35275 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
35276 (UBINOP_QUALIFIERS): New.
35278 (acle_builtin_data): New. Change comment.
35279 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
35280 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
35281 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
35282 arm_acle_builtins.def.
35283 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
35284 (arm_init_acle_builtins): New.
35285 (CRC32_BUILTIN): Remove.
35286 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
35287 crc32cb, crc32ch and crc32cw.
35288 (arm_init_crc32_builtins): Remove.
35289 (arm_init_builtins): Use arm_init_acle_builtins rather
35290 than arm_init_crc32_builtins.
35291 (arm_expand_acle_builtin): New.
35292 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
35293 * config/arm/arm_acle_builtins.def: New.
35295 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
35297 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
35298 (arm_builtin_datum): ... this.
35299 (arm_init_neon_builtin): Rename to ...
35300 (arm_init_builtin): ... this. Add a new parameters PREFIX
35301 and USE_SIG_IN_NAME.
35302 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
35303 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
35304 'arm_builtin_datum'.
35305 (arm_init_vfp_builtins): Likewise.
35306 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
35307 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
35308 (arm_expand_neon_args): Rename to ...
35309 (arm_expand_builtin_args): ... this. Rename builtin_arg
35310 enum values and differentiate between ARG_BUILTIN_MEMORY
35311 and ARG_BUILTIN_NEON_MEMORY.
35312 (arm_expand_neon_builtin_1): Rename to ...
35313 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
35314 values, arm_expand_builtin_args and add bool parameter NEON.
35315 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
35316 (arm_expand_vfp_builtin): Likewise.
35317 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
35319 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35321 PR middle-end/77484
35322 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
35323 * predict.c (tree_estimate_probability_bb): Reverse direction of
35324 polymorphic call predictor.
35326 2017-01-06 David Malcolm <dmalcolm@redhat.com>
35328 * passes.c (execute_one_pass): Split out pass-skipping logic into...
35329 (determine_pass_name_match): ...this new function and...
35330 (should_skip_pass_p): ...this new function.
35332 2017-01-06 Nathan Sidwell <nathan@acm.org>
35334 * ipa-visibility.c (function_and_variable_visibility): Reformat
35335 comments and long lines. Remove extrneous if.
35336 * symtab.c (symtab_node::make_decl_local): Fix code format.
35337 (symtab_node::set_section_for_node): Fix comment typo.
35339 2017-01-06 Martin Liska <mliska@suse.cz>
35342 * lra-constraints.c: Rename invariant to lra_invariant.
35343 * predict.c (set_even_probabilities): Initialize e to NULL.
35345 2017-01-05 Martin Sebor <msebor@redhat.com>
35347 PR tree-optimization/78910
35348 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
35349 (format_integer): Correct off-by-one error in the handling
35350 of precision with negative numbers in signed conversions..
35352 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
35354 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
35356 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35358 PR tree-optimization/71016
35359 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
35360 factor_out_conditional_conversion. Formatting fix.
35361 (factor_out_conditional_conversion): Add cond_stmt argument.
35362 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
35363 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
35366 2017-01-05 David Malcolm <dmalcolm@redhat.com>
35368 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
35369 read-rtl-function.o, and selftest-rtl.o.
35370 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
35371 (selftest::aarch64_test_loading_full_dump): New function.
35372 (selftest::aarch64_run_selftests): New function.
35373 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
35374 selftest::aarch64_run_selftests.
35375 * config/i386/i386.c
35376 (selftest::ix86_test_loading_dump_fragment_1): New function.
35377 (selftest::ix86_test_loading_call_insn): New function.
35378 (selftest::ix86_test_loading_full_dump): New function.
35379 (selftest::ix86_test_loading_unspec): New function.
35380 (selftest::ix86_run_selftests): Call the new functions.
35381 * emit-rtl.c (maybe_set_max_label_num): New function.
35382 * emit-rtl.h (maybe_set_max_label_num): New decl.
35383 * function.c (instantiate_decls): Guard call to
35384 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
35385 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
35387 * gensupport.c (gen_reader::gen_reader): Pass "false"
35388 for new "compact" param of rtx_reader.
35389 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
35390 rather than an empty string for NULL strings.
35391 * read-md.c: Potentially include config.h rather than bconfig.h.
35392 Wrap include of errors.h with #ifdef GENERATOR_FILE.
35393 (have_error): New global, copied from errors.c.
35394 (md_reader::read_name): Rename to...
35395 (md_reader::read_name_1): ...this, adding "out_loc" param,
35396 and converting "missing name or number" to returning false, rather
35398 (md_reader::read_name): Reimplement in terms of read_name_1.
35399 (md_reader::read_name_or_nil): New function.
35400 (md_reader::read_string): Handle "(nil)" by returning NULL.
35401 (md_reader::md_reader): Add new param "compact".
35402 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
35403 (md_reader::read_file): New method.
35404 * read-md.h (md_reader::md_reader): Add new param "compact".
35405 (md_reader::read_file): New method.
35406 (md_reader::is_compact): New accessor.
35407 (md_reader::read_name): Convert return type from void to file_location.
35408 (md_reader::read_name_or_nil): New decl.
35409 (md_reader::read_name_1): New decl.
35410 (md_reader::m_compact): New field.
35411 (noop_reader::noop_reader): Pass "false" for new "compact" param
35413 (rtx_reader::rtx_reader): Add new "compact" param.
35414 (rtx_reader::read_rtx_operand): Make virtual and convert return
35415 type from void to rtx.
35416 (rtx_reader::read_until): New decl.
35417 (rtx_reader::handle_any_trailing_information): New virtual function.
35418 (rtx_reader::postprocess): New virtual function.
35419 (rtx_reader::finalize_string): New virtual function.
35420 (rtx_reader::m_in_call_function_usage): New field.
35421 (rtx_reader::m_reuse_rtx_by_id): New field.
35422 * read-rtl-function.c: New file.
35423 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35424 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35425 (selftest::verify_three_block_rtl_cfg): New decl.
35426 * read-rtl-function.h: New file.
35427 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35428 For host, include function.h, memmodel.h, and emit-rtl.h.
35429 (one_time_initialization): New function.
35430 (struct compact_insn_name): New struct.
35431 (compact_insn_names): New array.
35432 (find_code): Handle insn codes in compact dumps.
35433 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35434 (bind_subst_iter_and_attr): Likewise.
35435 (add_condition_to_string): Likewise.
35436 (add_condition_to_rtx): Likewise.
35437 (apply_attribute_uses): Likewise.
35438 (add_current_iterators): Likewise.
35439 (apply_iterators): Likewise.
35440 (initialize_iterators): Guard usage of apply_subst_iterator with
35441 #ifdef GENERATOR_FILE.
35442 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35443 (md_reader::read_mapping): Likewise.
35444 (add_define_attr_for_define_subst): Likewise.
35445 (add_define_subst_attr): Likewise.
35446 (read_subst_mapping): Likewise.
35447 (check_code_iterator): Likewise.
35448 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35450 (one_time_initialization): New function.
35451 (rtx_reader::read_until): New method.
35452 (read_flags): New function.
35453 (parse_reg_note_name): New function.
35454 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35455 Handle reuse_rtx ids.
35456 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35457 Add parsing support for RTL dumps, mirroring the special-cases in
35458 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35459 values, and calling handle_any_trailing_information.
35460 (rtx_reader::read_rtx_operand): Convert return type from void
35461 to rtx, returning return_rtx. Handle case 'e'. Call
35462 finalize_string on XSTR and XTMPL fields.
35463 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35464 "(nil)" values were omitted. Call the postprocess vfunc on the
35466 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35467 class ctor. Initialize m_in_call_function_usage. Call
35468 one_time_initialization.
35469 * rtl-tests.c (selftest::test_uncond_jump): Call
35470 set_new_first_and_last_insn.
35471 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35472 * selftest-rtl.c: New file.
35473 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35474 (selftest::get_insn_by_uid): New decl.
35475 * selftest-run-tests.c (selftest::run_tests): Call
35476 read_rtl_function_c_tests.
35477 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35478 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35481 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35483 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35484 operands in a special way. Assert that pos+len <= mode precision.
35486 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35488 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35489 3 argument Alias with unlimited for the negative form.
35490 (fno-vect-cost-model): Removed.
35492 2017-01-05 Martin Liska <mliska@suse.cz>
35494 * hsa-gen.c (gen_hsa_divmod): New function.
35495 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35497 2017-01-05 Martin Liska <mliska@suse.cz>
35500 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35503 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35505 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35506 small constant length operands.
35508 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35510 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35511 between loop iterations.
35513 2017-01-05 Martin Liska <mliska@suse.cz>
35516 * gimplify.c (gimplify_decl_expr): Compare to
35517 asan_poisoned_variables instread of checking flags.
35518 (gimplify_target_expr): Likewise.
35519 (gimplify_expr): Likewise.
35520 (gimplify_function_tree): Conditionally initialize
35521 asan_poisoned_variables.
35523 2017-01-04 Jeff Law <law@redhat.com>
35525 PR tree-optimizatin/78812
35526 * rtl.h (contains_mem_rtx_p): Prototype.
35527 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35528 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35529 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35530 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35531 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35533 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35535 * input.c (assert_char_at_range): Default-initialize actual_range.
35537 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35539 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35540 to match the caller.
35542 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35544 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35545 insns after final jump in test to emit dummy move.
35547 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35549 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35550 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35552 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35554 * multiple_target.c (create_dispatcher_calls): Init e_next.
35555 * tree-ssa-loop-split.c (split_loop): Init border.
35556 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35559 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35564 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35565 (altivec_register_operand): Do not return true if the operand
35566 contains a SUBREG mixing SImode and SFmode.
35567 (vsx_register_operand): Likewise.
35568 (vsx_reg_sfsubreg_ok): New predicate.
35569 (vfloat_operand): Do not return true if the operand contains a
35570 SUBREG mixing SImode and SFmode.
35571 (vint_operand): Likewise.
35572 (vlogical_operand): Likewise.
35573 (gpc_reg_operand): Likewise.
35574 (int_reg_operand): Likewise.
35575 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35576 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35577 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35579 (rs6000_emit_move_si_sf_subreg): New helper function.
35580 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35581 fixup SUBREGs involving SImode and SFmode.
35582 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35583 numbers for the new peephole2 optimization.
35584 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35585 the GLIBC math library that do AND/IOR/XOR operations on single
35586 precision floating point.
35587 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35588 target macros to say whether we need to avoid SUBREGs mixing
35590 (TARGET_ALLOW_SF_SUBREG): Likewise.
35591 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35592 (UNSPEC_SI_FROM_SF): Likewise.
35593 (iorxor): Change spacing.
35594 (and_ior_xor): New iterator for AND, IOR, and XOR.
35595 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35596 (movdi_from_sf_zero_ext): Likewise.
35597 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35598 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35599 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35600 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35601 (fms<mode>4): Likewise.
35602 (fnma<mode>4): Likewise.
35603 (fnms<mode>4): Likewise.
35604 (nfma<mode>4): Likewise.
35605 (nfms<mode>4): Likewise.
35607 2017-01-04 Marek Polacek <polacek@redhat.com>
35610 * doc/invoke.texi: Document -Wpointer-compare.
35612 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35614 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35617 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35618 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35621 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35624 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35625 documentation of the powerpc_popcntb_ok attribute.
35626 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35627 code to issue warning messages if a requested CPU configuration is
35628 not supported by the binary (assembler and loader) toolchain.
35629 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35630 made to define a built-in function that has been disabled.
35631 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35632 made to define a built-in function that has been disabled.
35633 (altivec_init_builtins): Add comment explaining why definition
35634 of the DST built-in functions is not preceded by an assertion
35635 check. Add assertions to prevent ICE if attempts are made to
35636 define an altivec predicate or an abs* built-in function that has
35638 (htm_init_builtins): Add comment explaining why definition of the
35639 htm built-in functions is not preceded by an assertion check.
35641 2017-01-04 Jeff Law <law@redhat.com>
35643 PR tree-optimizatin/67955
35644 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35645 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35646 the points-to solution does not include pt_null. Use DECL_PT_UID
35649 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35651 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35652 Use gen_int_mode instead of gen_lopwart for const_int operands.
35654 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35656 PR tree-optimization/71563
35657 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35658 out of range value - has low bits known to be zero.
35660 2017-01-04 Alan Modra <amodra@gmail.com>
35662 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35663 * configure: Regenerate.
35664 * config.in: Regenerate.
35666 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35669 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35670 a substring of the message, but strcmp with the whole message. Ifdef
35671 ENABLE_NLS, translate the message first using dgettext.
35673 2017-01-03 Jeff Law <law@redhat.com>
35675 PR tree-optimizatin/78856
35676 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35677 (mark_threaded_blocks): Remove code to truncate thread paths that
35678 cross multiple loop headers. Instead invalidate the cached loop
35679 iteration information and handle case of a thread path walking
35680 into an irreducible region.
35682 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35685 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35686 assertions. Add support for doing the signbit if the IEEE 128-bit
35687 floating point value is in a GPR.
35688 * config/rs6000/rs6000.md (Fsignbit): Delete.
35689 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35690 Update the length attribute if the value is in a GPR.
35691 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35692 the sign or zero extension instruction, since the value is always 0/1.
35693 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35696 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35697 extracting SImode to a GPR register so that we can generate a
35698 store, limit the vector to be in a traditional Altivec register
35699 for the vextuwrx instruction.
35701 2017-01-03 Ian Lance Taylor <iant@google.com>
35703 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35705 2017-01-03 Martin Sebor <msebor@redhat.com>
35707 PR tree-optimization/78696
35708 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35709 precision. Use MPFR for %f for greater fidelity. Correct handling
35711 (pass_sprintf_length::compute_format_length): Set width and precision
35712 specified by asrerisk to void_node for vararg functions.
35713 (try_substitute_return_value): Adjust dump output.
35715 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35717 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35719 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35721 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35722 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35723 -mlra/-mno-lra was passed to the compiler.
35725 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35727 PR rtl-optimization/65618
35728 * emit-rtl.c (try_split): Move initialization of "before" and
35729 "after" to just before the call to emit_insn_after_setloc.
35731 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35733 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35735 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35737 * dwarf2out.c (gen_enumeration_type_die): When
35738 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35740 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35742 PR tree-optimization/78965
35743 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35744 Change first argument from const call_info & to call_info &. For %n
35745 set info.nowrite to false.
35747 PR middle-end/78901
35748 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35749 possibly throwing calls.
35751 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35752 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35753 and fns handling, rather than in a separate case SSA_NAME.
35755 2017-01-02 Jeff Law <law@redhat.com>
35757 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35758 fixes for first_period and second_period variables.
35760 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35763 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35764 (*insvqi_1): New insn pattern.
35765 (*insvqi_1_mem_rex64): Ditto.
35766 (*insvqi_2): Ditto.
35767 (*insvqi_3): Rename from *insvqi.
35769 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35771 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35773 * doc/cfg.texi (Edges): Remove reference to Java.
35774 (Maintaining the CFG): Ditto.
35776 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35778 PR middle-end/77674
35779 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35780 transparent aliases.
35782 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35784 PR middle-end/77484
35785 * predict.def (PRED_CALL): Update hitrate.
35786 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35787 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35788 into direct/indirect/polymorphic variants.
35790 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35792 Update copyright years.
35794 * gcc.c (process_command): Update copyright notice dates.
35795 * gcov-dump.c (print_version): Ditto.
35796 * gcov.c (print_version): Ditto.
35797 * gcov-tool.c (print_version): Ditto.
35798 * gengtype.c (create_file): Ditto.
35799 * doc/cpp.texi: Bump @copying's copyright year.
35800 * doc/cppinternals.texi: Ditto.
35801 * doc/gcc.texi: Ditto.
35802 * doc/gccint.texi: Ditto.
35803 * doc/gcov.texi: Ditto.
35804 * doc/install.texi: Ditto.
35805 * doc/invoke.texi: Ditto.
35807 Copyright (C) 2017 Free Software Foundation, Inc.
35809 Copying and distribution of this file, with or without modification,
35810 are permitted in any medium without royalty provided the copyright
35811 notice and this notice are preserved.